THE MC 240: ACHIEVING MAXIMUM PERFORMANCE FROM AN ICONIC DESIGN

[dcgillespie]

INTRODUCTION

The McIntosh Unity Coupled output stage — as typified by the McIntosh MC 240 — is recognized world wide as the pentacle of achievement in pentode amplifier design, producing at once the most power, least distortion, and highest efficiency of all pentode arrangements. These combined qualities simply cannot be matched by any other practical pentode configuration. I have a very nice example of the last version of this amplifier, and recently had it on the bench for a checkup.

After basic component and voltage checks, these amplifiers are often given their final health exam in the classic Mac Clinic style: THD and power response are checked over the entire audio band at rated power output with both channels driven. If these results are typical and within Mac’s guaranteed specifications, then invariably all is deemed to be good.

A couple of small signal tubes were replaced, and then my own unit met Mac’s distortion specs right up to maximum power output at a very healthy 56 watts per channel with both channels driven — the extra margin of power output developed being one of the most recognized but unspecified characteristics of McIntosh amplifiers.

With this checkup however I did further testing, which uncovered some other performance opportunities worth pursuing. A discussion of them follows, along with an explanation of the measures available for resolve. For those wishing to follow along, a copy of the final schematic is provided at the end, with all component references made relative to it.


INFRASONIC TRANSIENT STABILITY

Transient triggering of the amplifier produced less than desirable stabilization after a transient signal was injected. On a scope, this could be seen in the form of decaying oscillations of the base line for about 2 cycles before settling again. The amplitude of the trailing oscillations are in direct proportion to the amplitude of the transient signal applied. At high power levels, such instability can be hard on the tubes, steals power, and is obviously not an original component of the transient applied.

This conduct was traced to the combination of using a significant amount of NFB (21.5 db) within in the global feedback loop, coupled with the presence of four LF poles occurring within the loop as well. Without careful design attention, such a configuration can become unstable rather quickly at low frequencies.

Mac did a good job of stabilizing its design by adding partial direct coupling between the phase inverter and driver stages (via R19,20,42,43), to effectively eliminate one of the LF poles. With a slight re-configuration as to how the NFB is reinserted into the input stage, an additional LF pole can effectively be eliminated as well, which then further stabilizes the loop to produce the rapid settling desired.

This can be accomplished as follows:

1. Disconnect the leads from the NFB networks (R11,C2 & R34,C12) at their existing terminals on the main tag board, and reconnect them to the tag board terminals directly above them connected to pins 3&8 of V1 respectively.

2. Remove R12 and R35.

3. Change C3 and C13 to 2200 uF @ 6 vdc caps.

The effect of this change is to prevent the feedback signal from passing through C3 and C13, while still maintaining the original gain of the AF Amplifier stage V1, and providing the same amount of NFB as originally applied. The input sensitivity of the amplifier is unchanged as a result of this modification, but it does produce a very slight (and insignificant) reduction in both plate and cathode voltages in both sections of V1 (~ 5 volts and 0.1 volts respectively).

COMMENT: As originally designed, the 240 could hardly be considered “unstable” by any means. But with a greater understanding as to how stability impacts sound quality, and the availability of component values in smaller packages today, it makes for a practical modification that can achieve even better LF stability than before.


HIGH POWER TRANSIENT DISTORTION

Applying a full power audio sine wave transient signal to both channels from a quiescent state displayed serious initial clipping, before quickly rising to an undistorted full power waveform. This “blocking” distortion could only be eliminated by backing the transient power burst down to 39 watts per channel. Mac never provided a specification for undistorted transient power, but if the amplifier could not reach a transient power of at least rated power output without the blocking distortion occurring, then it would seem that something was not quite right.

It was also noted that:

1. When the blocking distortion occurred, the two channels recovered at different rates.

2. If the transient was applied in rapid fire succession, the clipping only occurred on the first transient — not succeeding ones. But wait a few moments, and this conduct would repeat itself again.

3. If only one channel (either one) was fed the transient, the blocking distortion did not occur.

The conclusion drawn then is that when instantaneous power is required in both channels that approaches rated power output (think Telarc 1812 cannon shots for example), all that extra undistorted steady state power that the 240 is capable of producing becomes questionable as to whether it really translates into any useful headroom under such demanding dynamic conditions. This is not meant to criticize the 240’s design, but does suggest that the long understood extra headroom provided may need to be looked at in a new light. That begins with understanding why the amplifier’s capable power output is so much greater than rated power output to begin with.

Mac has always championed conservative ratings, but in this case, the answer goes back to a decision Mac made dating back to the release of the original MC 30 amplifier. Before that time, Mac’s original Unity Coupled 6L6 amplifier — the 50W-1 — was their mainstay product (not counting the short lived A-116). It was rated for 50 watts output, and carried the standard Mac performance guarantee. So what happened to cause the same basic output stage to all of a sudden become rated for just 30 watts?


THE UNDERLYING PROBLEM

By conventional practice, the 240 and its brethren are a complex, high performance design. But such designs demand that every portion of the circuit achieve the same high level of performance if the ultimate capability of the end product is to be realized. If not, compromises result, and performance suffers.

But compromises are a part of any practical commercial design. For the 240, those compromises exist in the power supply, where its capabilities ultimately cannot keep up with the the high performance demands of the audio circuits under maximum high power transient conditions. More specifically:

When a transient signal causes both amplifiers to rise from quiescent conditions to maximum power output in both channels, it rather instantly produces about a 55 volt drop in available B+ voltage — typically from 460 vdc to 405 vdc on today’s average AC line voltage — when all is operating properly. There are two areas within the amplifier circuits proper that are affected by this — one more minor in effect, but the other much more seriously.

The discussion that follows actually pertains to most vintage Mac amplifiers, but it specifically relates to the 240 unless otherwise noted.


ISSUE #1: The Direct Coupled Connection From V1 To V2 & V5

The large drop in B+ voltage causes a proportional drop in the 140 volt supply for V1. The plate voltage in both sections of this tube then drops, as does the grid voltage on pin 2 of both V2 & V5, thanks to the direct coupled connection to them. V2 & V5 are balanced differential amplifiers, but only remain balanced if the grid voltage for each of their sections remains the same. But C4/14 at grid pin 7 of V2 & V5 respectively produce a long time constant relative to ground at those terminals. Any transient shift in supply voltage for V1 then is delayed in appearing at pin 7 of these tubes. Sudden changes in the B+ supply then upsets the balance of V2 & V5 until these capacitors have time to settle in at the new operating voltage presented to them. Because of the partial direct coupling employed from the outputs of these stages, this temporary imbalance then upsets the balance of the succeeding push-pull stages further downstream as well.

This condition can be corrected by disconnecting C4 and C14 from ground, and connecting them to the 140 volt supply source. The two differential amplifiers still operate exactly as before, but now their balance will be unaffected by large changes in the B+ supply voltage, since the caps holding the pin 7 grids at AC ground potential now reference AC ground by way of the 140 volt source. Therefore, changes in the 140 supply source now affect both grids equally, and balance is maintained regardless of any supply voltage deviations occurring.

Continued

 

[dcgillespie]

ISSUE #2: Driver Section Design

This issue is the heart of the matter, and for that, the design elements involved first need to be identified:

With few exceptions, the McIntosh Unity Coupled output stage operates in Class B, which invariably means that the stage will draw grid current under maximum output conditions. In the MC 240, this current is supplied by a 12AX7 cathode follower driver stage inserted between (what becomes) the 12BH7 pre-driver stage, and the output stage. R/C coupling is used between the pre-driver and driver stages, and direct coupling from the driver stage to the output stage to facilitate the necessary flow of grid current.

Additionally, the UC output stage inherently requires a very high drive voltage to achieve maximum power output. To achieve the necessary voltage level, both the driver and pre-driver stages are powered from a connection to the output tube plates, creating a “bootstrap” circuit for the driver section. This bootstrap connection allows the AC signals at the output tube plates to assist these stages in developing the very high drive levels required by the UC output stage to develop full power output.

Finally, because the output stage operates in Class B, the OPTs are unusually large to provide the huge amounts of inductance necessary for excellent LF performance from this class of operation. To take full advantage of their LF capabilities then, the coupling networks between the pre-driver and driver stages (C7,8,17,18 & R25,26,48,49) are made quite large as well. This prevents any significant phase shift from occurring at low frequencies, which maximizes the effectiveness of the plate coupled bootstrap connection, and ultimately ensures plenty of low distortion LF drive is available for the UC output stage under full power conditions.

The achilles heel of the design however is that because of the bootstrap connections, it also subjects the driver section tubes to the same fluctuations in B+ voltage that affects the output stage as well. With the plate bootstrap connections, large R/C networks coupling the pre-driver and driver stages, and direct coupling into the output stage, all the ingredients are now in place for the driver section to produce the blocking distortion created when full power transients cause a large drop in power supply voltage to occur. Consider:

A large drop in output stage B+ produces a large drop in pre-driver plate voltages as well. Because of the large time constant coupling the pre-driver and driver stages, this produces a large negative pulse appearing at the grids of the driver stage, which their cathodes follow. The negative pulse produced there is then applied to the output tube control grids via the direct coupled connection to them.

The negative peak delivered to the output stage control grids — in addition to the already reduced output stage B+ (specifically for the screen grids) — drives these tubes very negative into partial cutoff during a time when they should be conducting, causing the blocking distortion to occur. If the full power condition continues, the operating potential across C7,8,17,18 then re-stabilizes at the new operating voltage, and as that happens, the initially clipped sine wave rises to full undistorted amplitude again.


DISCUSSION

High power transient distortion in McIntosh amplifiers using this driver design is not widely recognized — but it’s not being presented here as a new found concern, either. Frank and Gordon were aware of this issue from the very beginning, even to the point of addressing the transient problems that driver stage R/C coupling can cause in their original article introducing the Unity Coupled output stage design in the December 1949 issue of Audio Engineering magazine.

It is because of this problem, that the original McIntosh 6L6 amplifier (the 50W-1, beginning in 1949) used a driver transformer for coupling into the output stage, because it effectively prevents the effects of imperfect power supply regulation from affecting the bias voltage presented to the output stage. As a result, the 50W-1 — using virtually identical output stage operating conditions as the MC-30 and MC-40/240 amplifiers — could be rated as a 50 watt amplifier with the standard McIntosh performance guarantee. It couldn’t produce any more total power, but could produce more usable power.

But the driver transformer presented its own problems, so by the mid 50s, Mac abandoned its use ultimately in favor of the long standing 12BH7/12AX7 pre-driver/driver design it developed — which of course does includes R/C coupling, along with the required plate bootstrap connections as well. With it however, the very transient problem Frank and Gordon originally warned of then became a reality.

As a result, Mac then had to de-rate the power output capability of the original amplifier’s design accordingly, if performance was to still be guaranteed under both steady state and transient conditions. Accordingly, the 50 watt power output capability of the original 50W-1, became just a 30 watt rating for the MC-30, despite both amplifiers employing virtually identical output stage operating conditions. And just like that, the seeds of Mac’s abundant headroom reputation were sown.

Over the ensuing years since the MC 30 was introduced, Mac took a number of steps to help minimize the effects of the transient distortion that the new driver design introduced. These include:

1. Providing power supply regulation in some mono amplifiers for the AF Amplifier input stage, via the (originally) unused section of the twin-triode tube used for that function. Since it is direct coupled to the phase inverter stage, minimizing any plate voltage variations at the first stage would help minimize any transient issues further down the line. This wasn’t real effective against the larger problem however, and was ultimately abandoned in favor of a simple shunt regulating resistor in succeeding designs.

2. By 1960, a bootstrap connection was added to the cathode side of the driver stage as well. This helped to ensure maximum driver stage performance, and became a mainstay in future amplifier designs.

3. By 1961, SS power supplies helped to improve power supply regulation, so that the MC 30 could now become the MC 40 (and the MC 60 the MC 75), while still maintaining Mac’s performance guarantee. This move also made practical the design of the MC 240 that is the subject of this thread.

4. In the final version of the MC 40 and MC 240, the phase inverter stage was redesigned so that it could operate from the same B+ source as all the other (succeeding) push-pull stages do. This meant that the effects of high power transients on the power supply affected all the push-pull stages equally, further helping to minimize any transient issues between these stages.

All of these changes were primarily detail improvements, although #3 was significant enough to allow rerating amplifier power through the introduction of new models. Still however, the final 6L6 Class amplifiers could not be rated at the power of the original 50W-1 — even in mono form — if the Mac performance guarantee was to hold. So in the end, 40 watts was still the maximum rated power output — up from that of the MC 30, but still less than that of the 50W-1, since the effects of the R/C coupled driver design still ruled the day.

Continued

 

[dcgillespie]

A COMPLICATING FACTOR: Output stage quiescent current

Output stage quiescent current also plays a significant role in the high power transient issue as well. McIntosh never specified what that current should be — only typical output tube terminal voltages as shown on their schematics. Therefore, when checking that the output stage is properly biased, it is typically verified by ensuring the correct bias voltage is applied to the tubes when a Variac is used to adjust the AC line voltage to produce a B+ of 435 vdc. If not, a small resistance is either added or adjusted in the bias rectifier circuit to make it so if that rectifier has been replaced with silicon. With the proper supply and tube element voltages present, and correct component values and well matched tubes installed, then little thought is given to the actual quiescent current level produced.

That level would only be correct however if the tube set installed also has the same bias characteristics that Mac specified their tubes to have. And therein lies the problem: Modern production tubes have become very good in many respects today, but their bias characteristics are often still all over the map compared to an original bogey characteristic. Therefore, it’s a bit of a crap shoot as to whether the actual quiescent current produced from a set of modern production tubes will be anywhere close to that which Mac intended — even though component and voltage values are correct. As a result, in today’s audio environment, there (now) needs be some capability to at least measure, and hopefully adjust output stage quiescent current — if only to account for modern tube manufacturing tolerances.

But in Mac’s case, adjust it to what? In the absence of any definitive information, a pretty good educated guess can be made using a couple of standard approaches:

1. Use information from the 240’s voltage and resistance chart and convert it directly into current flow. This suggests that the total quiescent current for each tube is 26.9 mA. Or —

2. Apply information from the same chart (including the cathode bias voltage) to GE’s family of plate curves published for their 6L6GC, where it will be seen that the total quiescent current per tube will be 36 mA.

These approaches result in two different indicated quiescent current levels, with both results derived from Mac’s own information. In reality, any value within this range could be used, but it will be shown why the first result is closest to that intended, and also why it’s always more important to achieve a proper quiescent current level than an indicated bias voltage. This has a very unique importance with these amplifiers.

The McIntosh UC transformer design was specifically developed to operate an output stage in Class B, as an all out effort to achieve high quality, high power output from that mode of operation, without generating the attending notch distortion that more conventional transformer design produces in that mode. As a Class B design then, the quiescent operating current is inherently very low by definition, and thanks to Mac’s UC output configuration, a McIntosh Class B output stage is also inherently rather insensitive to current flow versus distortion produced — the actual current flow largely having only a very limited impact on THD and IMD. The low quiescent current flow then reduces heat, increases tube life, and eliminates the need for a bias control.

This wide latitude of acceptable quiescent current level is one of the primary reasons why Mac can offer the performance guarantee it does with these amplifiers. But if this is true, then why the concern for determining the proper level and achieving it? In this case, it’s not because of steady state performance concerns but rather, because of the impact it has on power supply performance. Consider:

All SS full wave power supplies with a capacitor input filter exhibit a DC output voltage that reaches 1.414 X the RMS value of the AC voltage applied to it under no load conditions (2.828 X RMS for voltage doubler designs like the 240 uses). As current is drawn from the supply, output voltage falls rather quickly to a value of about 1.33 (or 2.66 for VD designs) X the RMS value when 25% of the supply’s total current capacity is drawn. This is the first tier of regulation, and typically represents about 94%, where (with the 240) the output voltage drops by about 0.20 volt/mA drawn. The 240 employs a 600 mA total capacity HV supply, so that means the output voltage has dropped 30 volts when 150 mA is drawn from the supply.

This is actually rather poor performance for a 600 mA supply capability, but after that point, regulation shifts gears to a second tier, where it improves considerably. Technically, regulation drops further to 88% on this tier, but since this is over the remainder of the supply’s current capability (450 mA), the output voltage only drops by about .122 volt/mA.

When this power supply characteristic is then considered along with the transient distortion that Mac’s driver section causes from less than ideal power supply regulation, a salient point emerges: For maximum performance from the MC 240, the total quiescent current draw of the complete amplifier must be at least 150 mA. Ensuring that this minimum current draw is met under quiescent conditions then means that the power supply is operating in its most linearly regulated region (second tier) when the amplifier is developing power output — and that then minimizes the impact it has on the high power transient distortion issue.

This fact alone then becomes the main consideration in establishing what the optimum output tube quiescent current should be for the MC 240: If the current draw of everything but the four output tubes (50 mA) is subtracted from the 150 mA minimum total current draw requirement, it leaves 100 mA for the two output stages to operate from. Spread over the four tubes, this equals 25 mA per tube, or very close to the level of quiescent current derived directly from Mac’s voltage and resistance chart.

This is no doubt how Mac intended their UC output stage to operate in the 240, producing a Pd of just over 10 watts per tube. The optimum output stage quiescent current then has little to do with maximizing steady state performance capability, but everything to do with minimizing transient performance issues. All that remains then is to determine how best to implement measurement and adjustment capability into the amplifier, so the proper current flow can be achieved.

It was decided that the best way measure output stage quiescent current is to install a 1Ω 1% 0.25 watt resistor in series with each red OPT B+ lead. This lead powers not only the output tubes, but a complete “output section”, including the driver and pre-driver tubes as well. Therefore, if the current draw of the latter two tubes (17.4 mA) is subtracted from the total current draw flowing through these resistors, then the current flow of the output stage they serve would be accurately known. If matched output tubes are installed (and they should be), then half of the remaining current flow will be the quiescent current passing through each output tube. A total current of 68 mA (.068 vdc) through each of these resistors produces the desired output stage quiescent current draw.


RESULT

In the initial testing of my 240 however, I noted that the otherwise excellent JJ 6L6GC tubes installed were only drawing about 15 mA each under quiescent conditions. Operating directly from my 121.5 vac line voltage, this produced a 6.58 vac heater voltage, but an elevated B+ of 465 vdc, due largely to the fact that the complete amplifier was only drawing about 110 mA total from the supply. This was with the amplifier biased as described earlier, and shows the fallacy of relying on bias voltage to achieve optimum performance. It was also noted that in this condition, the amplifier could only deliver 39 watts cleanly and instantly in each channel with both channels driven, which again shows why steady state testing is incomplete testing for these amplifiers.

This situation improved when a good, well match (by pairs) quad of classic RCA 6L6GC tubes from the 240’s original time period were installed, because overall they drew more quiescent current in the amplifier as designed than the JJ tubes did. The basic blocking distortion issue was still present with these tubes of course, but it was not nearly as pronounced because they moved the transient clipping point to a level closer to that of maximum power output — now above the 40 watt rating of the amplifier (to ~45 watts) — allowing the amplifier to then truly meet its rated specifications.

The JJ tubes were then reinstalled, and quiescent current was adjusted to 68 mA total per output section by shunting an appropriate resistor across R61 (~430K in my case). The quiescent B+ then fell to 460 vdc, and Pd rose slightly to 10.9 watts. The improvement this made was immediate and obvious. Yes, there was still blocking distortion present like that with the RCA tubes, but nothing like these tubes produced earlier. It was not nearly so pronounced, did not appear until about 45 watts of power was instantly demanded in both channels (also like the RCA tubes) and, it recovered much more quickly when it occurred. Steady state performance however simply yawned at all of this, improving very little if at all.

Setting the 240’s actual output stage quiescent current level (versus bias voltage) then is more important than widely recognized — too little current and high power transient distortion increases rapidly. But too much and you’re simply wasting tube life.

As a side note, because of their size, the 1Ω resistors added to facilitate accurate current measurement also act as an excellent fuse in case anything should go south in either output section — certainly providing more protection for the tubes and transformers than the added 3.2A Slo-Blo B+ fuse does. They also handle sustained maximum power output in each channel with ease as well. Because of the double benefit these resistors provide, their inclusion is highly recommended.

Continued

 

[dcgillespie]

TWO OTHER OBSERVATIONS WORTH NOTING

1. It was mentioned earlier that when high power transient clipping occurred, the two channels recovered from the event at different rates, with the right channel always lagging behind the left in recovery. This observation was unaffected by using different tubes or swapping them between channels.

The time lag was ultimately traced to the usually excellent black Sprague caps Mac used in the 240, which normally hold their value very well, and rarely leak.

All of these caps in the right channel were removed and tested for value and leakage at rated voltage. All passed both of these tests with flying colors with one exception: Compared to new modern caps of the same voltage rating and value, the removed caps caused the eye tube on my Heath IT-28 cap checker to open more slowly when the leakage test was performed — not in a lazy way as outright leakage might indicate, just a little more slowly. All of them acted this way. They were all replaced with new caps of modern manufacture, which then caused both channels to recover from a clipping event at an absolutely identical rate.

What is notable regarding this is that while the two channels now recovered at an identical rate, the left channel still had all its original black Sprague caps installed……

It is likely that the removed caps were in the very earliest stages of becoming compromised (pre-leaking), causing the slower recovery rate. The caps at C17 & C18 are particularly sensitive to this, as reinstalling the original caps in this location caused the recovery disparity to return. Their value was dead on, and they weren’t leaking by traditional standards. But they were causing a performance issue none the less that traditional evaluation methods couldn’t detect. It was only by direct comparison of the two channels during recovery from a transient event, or by comparison of the original caps with equivalent modern caps on the cap checker that their compromised condition could be observed.

Since the unequal recovery rate is directly related to the condition of the coupling caps installed, testing for equal recovery time after a full power transient clipping event is an effective tool and indicator in evaluating the overall condition of the film caps installed.

2. And then there’s this — not directly related, but still related to the maximum power output capability of the amplifier: The original current limiter device installed is large enough such that under full power conditions, it heats even more to actually increase the AC voltage applied to the power transformer, over that it receives under otherwise typical quiescent conditions. This causes maximum power output capability to rise over that from an initial indication, adding its own element of delay and recovery to this investigation. It’s not a big deal — just an observation, but was the reason I paused between each transient testing event, so that its effects were consistent with each test conducted.


TRANSIENT DISTORTION SUMMARY

There is no simple modification that can be easily installed to completely eliminate the high power transient distortion from occurring in the original design. From the perspective of the amplifier circuit proper, there’s technically nothing wrong with its design to cause it — just the very real consideration it requires for the performance of the power supply powering it. As a result, this problem then lies squarely at the feet of the conventional power supply design used. Within that limitation then, there are two important take away points:

1. In addition to the regular steady state testing normally done to prove out the performance of a vintage McIntosh amplifier, it is equally important to conduct high power transient testing as well to ensure that none of its effects appear down within the rated power output region of the amplifier.

2. Ensure the output stage operates with the correct quiescent current as determined, versus bias voltage given. With nothing more than a DVM and installing the suggested resistors, it is the best way to ensure that all the total performance Mac designed into these amplifiers is achieved when using modern tubes under both steady state and transient conditions.

The only way to completely eliminate the transient distortion issue is to regulate the power supply, which of course is a much more involved project. But even that starts with the discussions and modifications suggested here. For those interested, that effort is discussed along with other related topics in the “more” section after this presentation.

It is unfortunate that I don’t have the ability to upload any video pics I took of the conduct displayed by my 240 pertaining to this thread, since videos cannot be uploaded directly to this site. However, anyone or shop with a scope, generator, and appropriate load resistors can easily verify the presence of the transient distortion issue described. On the other hand, listening tests were very enlightening.


LISTENING TESTS

Is any of this going to make any difference for the typical MC240 user?
The answer to that question depends on the demands placed on it.

For the person driving Klipsch (type) speakers (like me) to normal listening levels using typical modern source material, probably not. I could not distinguish any difference in sound from before to after the quiescent current was adjusted in my amplifier in that particular scenario. This of course would be completely consistent with the Mac design, where amplifier performance within the rated power level has little dependence on output tube quiescent current level.

This effort however was not so much about performance within that range as it was that beyond it — and more importantly, how the performance in that region — when entered — impacts the overall sonic presentation. What was needed then was a few digital cannon shot comparisons, and that’s where differences were clearly heard.

The test started with the amplifier in it’s original un-adjusted form, with the power output level adjusted to produce an indicated instantaneous power output of 40 watts in each channel at the initial crack of the last cannon shot on Telarc’s recording. This was audibly compared to the sound produced under the exact same conditions using my ARC D115/V70 100 watt per channel amplifier. As it should be and would be expected, the ARC’s presentation was clean, authoritative, and very real in it’s presentation. The Mac’s presentation was notably anemic and asthmatic by comparison, lacking the authority presented by the ARC, with a rather harsh breakup component embedded in the sound. This is exactly what I was seeing on the scope earlier, where the amplifier couldn’t even quite make 40 watts in each channel before the transient clipping began, and when it did begin, was very apparent and hard.

The quiescent bias was then adjusted as discussed and the same test run again. This time, there was notable improvement, with the breakup element gone that was previously heard on the initial cannon crack. The presentation was clean and undistorted, and you wouldn’t have any indication that the amplifier was nearly reaching its limits if you didn’t know the details of the test. Pushed just slightly further however and breakup was again apparent. But this time, it was not nearly so impolite in its presence, with more of an indication of the power limit reaching a padded wall rather than a brick wall.

So for little more than installing two measurement resistors, and as a result, a bias adjustment resistor, the amplifier now not only meets McIntosh steady state performance specs like before, but more importantly, now also sounds and measures like it meets them under demanding dynamic conditions, too. But again, is this a worthwhile effort to undertake for the average MC-240 user?

Since most Mac owners appreciate these amplifiers well enough for the ingenuity, engineering, and production effort that went into producing them, they usually also want to know that all the performance their own example is producing is actually being met as well. From that standpoint then, and the limited effort required, I would say absolutely so.

For me however, there’s a more personal piece to this story, which I would invite you to check out in the more section that follows this presentation.


QUICKIES: HUM, FREQUENCY RESPONSE & STARTUP PROTECTION

These are some modifications I performed long ago on my unit to address opportunities noted at the time. You may find them helpful:

1. Like all of Mac’s vacuum tube basic stereo amplifiers, the 240 uses an unbalanced heater supply with respect to ground. This can cause a low level buzz in the speakers particularly if input level controls are set to some mid-level setting, or the stereo inputs are selected. The buzz can be eliminated by converting the heater system to a balanced system with respect to ground:

A. Disconnect the existing heater system ground by removing the short bare wire connecting terminals 4&5 of socket V1 to a nearby ground lug. The center post of the socket however must remain grounded.

B. Install two 220Ω 0.25 watt resistors — one from pin 9 to the center post connection, and one from pins 4&5 to the center post connection. Keep the resistor leads short and the resistors trimmed close to the socket terminals.

2. The Stereo Input attenuating ladder causes a slight reduction in HF response versus that displayed at the Twin or Mono input jacks when their level controls are up full. This can be corrected by connecting a 5 pF cap in parallel with each 200K resistor (R4,5) in the input attenuating ladder.

3. At turn on, as the 12AU7 tubes (V2, V5) warm up, there can be a nearly 200 volt positive potential between the grids and cathodes within these tubes during this time. This causes their grids to conduct current (which they are not designed to do) until they have fully warmed and their normal bias voltage is present.

The high grid potential can be eliminated by connecting a 1N4148 diode between pins 2 & 3 of V2 & V5, with the banded end connected to pin 3 of each tube.

Continued

 

[dcgillespie]

EPILOGUE

This thread was never intended to be a “how to” piece, since modification of any vintage McIntosh equipment should be carefully considered before implementation. It is therefore more of an information piece for each to evaluate as to their own opinion of worth and/or need. I was able to install all the elements of this discussion and more into my own MC 240, without any physical alterations required. Therefore, all can be completely reversed easily enough if desired.

The other modification I’ve developed is the addition of power supply regulators, that eliminate the transient distortion issue once and for all. Details on that are covered in the more section along with additional listening tests conducted with that effort. I have provided some pics showing how all of this work was installed in my own unit for those who may be interested.

Happy Listening!

Dave

MC 240 Final design, input section:

MC 240 Final design, output section:

Resistor side of tag board showing removal of R12 and R35:


Capacitor side of the tag board showing C3 and C13 changed to 2200 uF 6V capacitors, and twisted NFB leads removed from their original tag board connection terminals directly below where they are connected now, to their new location shown:


Installation of new 1Ω 1% 0.25 watt precision resistors for measuring current draw in each output "section".


5pF caps installed across R4 and R5:


Showing installation of heater circuit balancing resistors and one of the protection diodes for V2 and V5. V2 is shown:

 

[dcgillespie]

BUT WAIT — THERE’S MORE!


LISTENING TESTS PART I: FROM A LONG TIME AGO……..

I’ve lived in GA for the last 55 years, but if you (can) rewind back to 1967, I was starting my first year of HS (10th Grade) in Indianapolis. It was still a fairly new facility (opened in 63), and at the time, one of the best of the best facility wise. To help facilitate the greater community’s needs, it had a first rate auditorium that was not only used for the school’s needs, but also for shows and performances that came to the area. As a result, it had the best of the best equipment installed for that day as well — two Voice of the Theater speaker cabs on each wing of the stage, sound and lighting booths up high at the rear with large open air windows opening out into the auditorium — and McIntosh equipment to power the sound.

Between the auditorium and the language lab (that supported three language rooms) there were at least eight pieces of Mac equipment, and the morning announcement system? — yep, it was there also, with not just one or two, but three MC275s running flat out balls to the walls every morning for the announcements or general paging. These three amps powered quite literally 500 speakers installed throughout the facility on a 70V constant voltage distribution system. There were speakers in every class room, restroom, lab room, lecture hall, hall way, walk way, office complex, and up and downstairs common areas. I could go on, but you get the picture. All of this equipment — in the language labs, auditorium, and announcement system — was all very professionally installed with numerous large prewired equipment racks, connecting jack boards, and endless lengths of cable bundles — and I was one of the lucky kids that got to work with all of it.

I say work with, because I never actually had to work on any of the Mac gear, because the silly stuff never broke in the entire three years I was there. None of it — other than to perform regular tube replacement schedules. I actually did work on other equipment there, but my duties with the Mac gear were more akin to configuring all of it for the various needs as they arose in the three different systems. But — it all adds up to a lot of very real exposure to the quality, dependability, and capability of Mac equipment as a teenager.

I mention all of this to make this point: the sound in the language labs and auditorium was absolutely first rate, but the announcement system? Well it was simply godawful: Certainly intelligible and loud enough, but anemic in presentation, sounding nothing like the quality of the other two installations. Overloaded? Obviously — but not in a normal overloaded kind of way. I know what normal output stage overload sounds like. This sound had a harsher edge along with a pulsing element to it as the announcing tone bars would first be struck, and then the spoken announcements would begin. With the less than rhythmic way the tones were (manually) played, and the words were spoken — dollars to doughnuts, this harsh, pulsing sound was the audible result of the Mac transient distortion discussed here — and I heard it every school day for three straight years.

Now granted, these were 275s, not 240s. But these two amplifier designs are otherwise so close to each other that if they were kissing cousins, their relationship would be illegal. The overload characteristics are the same between these two amplifiers, because the conditions causing the overload are the same between them.

Now obviously, the announcement system could have used more horsepower. I figure each speaker was only receiving (on average) a little under a watt at best on a good day. Two more MC 275s were really needed to truly get the job done. But it was what it was and for me, what it was ended up serving to impress early on what happens when vintage Macs are pushed into dynamically overloaded conditions. I’ll pause now for the next section, and pick this back up again in Part II.


THE ULTIMATE SOLUTION: POWER SUPPLY REGULATION

I have total respect for the McIntosh brand and its products. But they could only do so much with what was available back in the day these amplifiers were produced. Today, possibilities for improvement exist that the Mac team of yore could only dream of — and where possibilities exist, I’m the kind of person who hates to leave performance on the table. I never modify for the sake of modification, but when such a real and demonstrable opportunity exists like eliminating the transient clipping distortion discussed here, I’m one who will take that challenge on. Regulating the 240’s power supply then is such a challenge.

The question was how to do it without physically modifying the 240 in any way, or changing the underside appearance from its neat, well thought out look, to one of looking like a rat’s nest when done. Most experienced hobbyists who’ve been around a while have seen modification projects that ended up looking like that, and just cringe. Such a result with a piece of McIntosh equipment would simply be criminal.

The obvious solution then was to employ an external unit that would plug into the 240 via a repurposed constant voltage output connection socket. The early 240s that also include the preamp power socket can use it as an even better option. It’s highly doubtful that the constant voltage output socket is ever used in this day and age, and even less chance that the preamp power socket is used either, so really either one could be used. If the existing connections are carefully removed from the needed terminals (on either socket), then they can always be reconnected if needed in the future with nobody the wiser for it.

It was also desired to devise the regulator circuit connections so that either the regulator could be plugged in for improved performance or, a dummy plug could be inserted in its place instead that would then return the amplifier to its box stock design — albeit with its power supply connections passing through the chosen socket and dummy plug.

Proper regulation requires that both the B+ supply and bias supply voltages be regulated, so a rather simple circuit was devised for each regulator, with the bias regulator regulating only the negative source serving the driver stage grids. Both regulators provide for variable control of their set point, and both use a 3 terminal constant current regulator to produce a constant reference voltage. The B+ regulator also uses a 135 watt 500 volt 9A power Mosfet as a pass element. Between the measurement resistors and the Mosfet, there is less than a one volt drop inserted between the power supply and the amplifiers when they are drawing maximum current, which allows the amplifier to produce fully as much power whether the regulator is in the circuit, or not. The Mosfet dissipates just over 10 watts worst case, so appropriate heat sinking is required.

Since this project was primarily a trial effort, the whole thing was built in that spirit with junk box parts, and whatever I had available on hand that would work. The resulting regulator box then isn’t the prettiest thing, and has way more heat sink than needed, but it gets the job done, and its performance has been quite gratifying:

From a properly adjusted and stabilized quiescent current condition, to an instantaneous 110 watt output level combined between both channels, the regulated B+ — which has been set at 405 vdc — drops just one volt to 404 volts for a very brief second before rising back to 405 volts again (thanks to the current limiter’s heating), while the regulated bias voltage doesn’t change at all. On the scope, a maximum power output sine waveform is instantly (and fully) there — with no sign or even a hint of clipping, no matter how, when, or how many times the signal is applied. When the signal is removed, the base line returns to zero instantly, without any bounce or fanfare. So the regulator circuit works, and works exceedingly well, with the heatsink only becoming slightly warm at best after numerous hours of operation.

With the regulator, the output tubes can actually operate with a lower Pd, because while (for now) the same quiescent current set point is being used as with non-regulated operation, the plate voltage is about 55 volts less, so the Pd drops to under 10 watts per tube at 9.6 watts each.

Ultimately however, because of the Unity Coupled output stage configuration and the Class B operation it provides for, with the regulator, it is possible to operate the output stages with very little quiescent current at all. Understand that in the 240, the quiescent current drawn by the output stage really only serves two purposes:

1. It causes the amplifier to draw enough quiescent current so as to allow it to effectively operate only from the power supply’s second tier regulation level whenever it’s developing any power output, and —

2. It ensures there is enough quiescent current still flowing to maintain Class B operation when both channels are at a full boil and the power supply B+ voltage has sagged to its lowest point.

When operating from the regulator however, neither of these concerns arise since the operating voltages remain constant regardless of the power output being developed. As a result, quiescent current can be set to the lowest practical level that will maintain Class B operation at all power levels. It hasn’t been tried yet, but it is quite likely that the amplifier will fully meet its distortion specifications with the output tube plates dissipating as little as 5 watts each.

Also, with the regulator connected, all B+ voltages throughout the amplifier circuits are about 88% of what they would normally be without the regulator connected — or effectively, at the same level they would be in the stock amplifier when developing maximum power output in both channels. Input sensitivity and NFB levels for the regulated amplifier are essentially unchanged from that of non-regulated operation.

Finally, operation with the regulator produces less heat generation under the chassis from the lower operating voltages, and also quieter operation as well since the regulator filters the B+ voltage >60X over that of the stock design under quiescent conditions.

With the regulator finished, it was back to the listening room again to see what audible difference its inclusion might make.


LISTENING TESTS PART II: TODAY, WITH REGULATED OPERATION

PRELUDE

Understand that with the regulator, the amplifier technically operates quite differently. As a Class B design, current drawn by the output stage is directly proportional to the amount of signal applied. Therefore, the power supply voltage is always moving somewhere between a high of 460 volts and a low of 405 volts, depending on the amount of power being produced. In the stock design, the amplifier circuits are powered directly from the power supply, and therefore are modulated by its fluctuations while amplifying the signal. The push-pull design and significant amount of NFB employed help the audio circuits reject that modulation, and do their jobs well in spite of the voltage variations presented.

When the regulator is connected however, all the audio circuits then operate from a constant voltage source regardless of the amount of power produced. Because the operating voltages remain stable, these circuits then don’t have to constantly adjust to a varying power supply voltage all the time, so there can be no modulation of the audio signal by it. Without any modulation then, no transient clipping distortion can be produced, and the feedback circuits then function only to lower distortion produced by the amplifier circuits themselves.

Since this action affects amplifier performance at all signal levels then, a second set of listening tests was set up to compare regulated and unregulated performance of the amplifier, with the amplifier operating with the same optimum quiescent current level in both cases.

Continued

 

[dcgillespie]

THE TESTS

In the first test conducted, routine listening levels were used on a variety of modern source materials. With the regulator in the circuit, there was a perception of better definition and separation of instruments in the mid bass to upper regions, but overall, performance was very similar to that of the stock amplifier. At the low end however, the differences were impressive:

It’s not that there was any more bass, but what was there simply had more energy, more impact, and of course better definition as well. Kick drums, bass organ pipes, bass guitars, electronic bass instruments, and even upright basses were just more real sounding and alive in presentation. In presenting this improvement, the bass region also kept its own space, without impacting the regions above it because of its new found muscle.

This is not inconsistent with countless reports and manufacturer trends over the years — even with McIntosh themselves — where increased power supply capacity improves LF performance and definition in a very notable way. This is just one more benefit then that regulation of the 240’s power supply provides.

Now I’ll be the first one to question if any real differences are heard unless they can be verified through well conducted double blind testing. But this difference was so noticeable from the get-go that it was hard to deny the results of multiple comparison tests.

In the second test, I mimicked my old high school observations: I compared the regulated and unregulated amplifier when a microphone became the source, and the amplifier was run at an absolute full boil in mono mode while a passage of speech was read. The level was not one of gross overload, but enough that overload was clearly and routinely present with the speech, just as it was so long ago. The output to the speaker was padded down of course, and the material was read in a similar manner to the way the announcements were read so long ago (with different voices, pauses, added emphasis, etc.).

With the unregulated amplifier, it was real enough compared to my high school days to make me think I was back in home room again: There was the anemic and pulsing sound I had heard so long ago all over again….

When the regulator was connected however, overload was still present of course, but now it took on a stronger, more classic overload characteristic: the old, overly irritating elements of the original presentation were greatly diminished, replaced by a more intelligible sound that was somewhat louder, far less irritating, and simply hitting its peak output capability. When breakup occurred, it was notably smoother, perfectly coinciding with the overload (no pulsing), and had far less audible impact on the spoken word. The difference was striking and undeniable.


END

So there you have it. What started out as a routine check of my 240’s performance ended up as a full blown investigation of it. Why these simple exercises always seem to end up that way for me, I’ll never know.

One last thought about these modifications however is this: when compared by weight and footprint size, a stock 240 sort of gets lost between an MC 225, and an MC 275. That is, while both of those amplifiers receive unending praise for their sonic performances, the 225 does so certainly for its capabilities, but also for its capabilities compared to its diminutive physical size. The 275 does for its capabilities as well, but also for its capability of sheer grunt power. But for the extra weight and size of the stock 240, it doesn’t have the diminutive size, nor grunt power advantage of its siblings. Being Mac’s first stereo amplifier then, it doesn’t have the advantages that either of those later efforts offer.

With power supply regulation however, the increased usable power output helps the 240 take its proper place in the pecking order with its younger two siblings. So modified, it now approaches nearly double the useable power of the 225, while producing enough power to make a 275 think twice about slacking off on the job.

At the time these units were first manufactured, the audio world could hardly conceive of digitally recorded cannon shots, so the chances of a 240 being instantly driven beyond its rated capabilities in both channels in a typical home installation with the sources and speakers of its day were slim to none (and Slim left town). But today, we do have such recordings, 747s taking off in our living rooms, and a whole lot more in between what was, and what now is.

In this new age of audio then, making a few improvements to this iconic classic will allow it to make the utmost of its performance capabilities, and the most of Frank and Gordon’s historic work in developing it.


RELATED TOPICS OF INTEREST

1. MC 75/275 AND THE ADDITIONAL OPT WINDING

The design of the MC 60 was such that — for all intents and purposes — it was little more than a bigger transformer set and bigger output tubes with an otherwise MC 30 input and driver section. Moving to SS rectification allowed these amplifiers to morph into the MC 40 and MC 75, and ultimately the MC 240 and MC 275 as well.

Development of the stereo units brought greater standardization between and detail improvements to both series of amplifiers, but while that was all the 6L6 amplifiers received, the new 75 watt amplifiers received an additional tweak in the form of an additional secondary winding added to the OPT, that was now trifilar wound along with the two primary windings.

The new winding powers the driver stage, and provides the exact same amount of bootstrap feedback to it as if the plates of this stage were connected directly to the output tube plates as in the 240. The winding also powers a new driver tube type — changed from the 12AX7 of the MC 60s to a higher current 12AZ7/12AT7 for the 75 watt amplifiers — and thanks to this dedicated winding, the stage could also be powered from a much lower B+ voltage source as well. This additional attention given to the 75 watt amplifiers is likely the cause of their delayed release over that of the MC 240. But why was the extra winding and higher current driver tube included?

Every reference I’ve ever seen regarding this feature suggests the winding was included to be able to operate the driver tube from a lower B+ voltage, in an effort to be a little “kinder” to that tube — referring to how much voltage appears across it in the McIntosh circuit. While this is mostly true, if that was truly the end game, it begs the question: why wasn’t it added to the 240’s transformers as well? They’re hardly chopped liver! Answer? Because “kindness” wasn’t the end game.

The third winding was necessary for the 75 watt amplifiers because with their bigger output tubes and now increased power output capability, they draw more grid current at a full boil than the the 60 watt amplifiers do, which exceeds that which the 12AX7 tube can provide. For the 40 and 60 watt amplifiers, the 12AX7 could provide the necessary grid current. But with the greater output of the 75 watt amplifiers, the big tubes draw more grid current than a 12AX7 can supply, so a switch was made to the higher current 12AZ7/12AT7 tubes.

But while these tubes could supply the increased grid current required, their Pd rating is not much greater than that of the 12AX7. Therefore, if operated directly from the output tube plates, the increased current flow coupled with the output tube level plate voltage would over-dissipate the tubes in grid current operation, and lead to shortened life.

To resolve that problem then, Mac included the third winding, allowing the driver stage to be powered from a lower voltage source. This then eliminated the excessive dissipation issue, while still providing the full bootstrap feedback needed to properly drive the output stage.

In modern incarnations of this amplifier, this extra winding was dropped in favor of using Zener diodes to drop the output tube plate level voltage to the driver stage. With this move, the driver plates could source their B+ and bootstrap feedback from the output tube plates again, so the extra winding was no longer needed.

Also of note, while the modern versions of this amplifier do not employ active power supply regulation, they do include a seriously increased B+ reservoir, which no doubt helps to address the transient distortion issue.


2. SIMILARITIES (OF SORTS) BETWEEN THE A-116, AND MC 225 (Revised 09/01/25)

The A-116 (no suffix) uses an entirely different front end design employing a split load phase inverter and optional on-board preamp option, while the 225 uses the more traditional front end configuration of it’s bigger brothers. Both of these designs use the boot strapped 12BH7 driver stage and Unity Coupled output stage, but both have the 12BH7 directly driving the output stage without the aid of an intervening cathode follower stage. For the A-116, this may have been an early trial effort that ultimately didn’t make the final cut for Mac’s longstanding driver topology going forward, while for the 225, it was by all appearances an effort to make for a simplified design. The net effect of these simplified driver designs however is that without the aid of the cathode follower stage, the driver circuit cannot produce any output stage grid current, which slightly limits the maximum power output these designs can produce.

But even though these amplifiers share the same output and driver stage topology, operating voltages indicate that the output stage class of operation between them is apparently quite different. Based on schematic voltages for the A-116, it suggests that the output stage in this amplifier operates well into Class AB1 territory, while the 225 operates in Class B like it’s more modern bigger brothers. As with the 6L6s of the MC 240 however, there is confusion between how tube data suggests 7591 tubes should operate in the 225, and how they actually do.

Typical operating conditions published for push-pull 7591s by tube manufacturers indicates that with fixed bias operation, a screen voltage of 400 volts, a plate voltage at least that high, and a g1 voltage of -21 volts, Class AB1 operation is produced, with a quiescent plate + screen current of 38 mA per tube. These conditions are nearly identical to those given for the 225, where 390 volts plate and screen and -21 volts for g1 are specified. With identical g1 voltage and only a 10 volt B+ difference between these two scenarios then, a quick glance at this data in the tube manual would suggest that the 225 is also a Class AB1 design. Except that this particular (and widely published) operating condition is at odds with the grid curves also published by manufacturers of the tube.

Extrapolating from Sylvania’s published curves, operation with 400 volts plate and screen and -21 volts on g1 produces an indicated total plate + screen current flow of only 27 mA — or about 70% of that which the typical operation data indicates. So again, as with the 6L6, there is conflicting tube data provided, although this far down on the curves, some error is likely to be encountered. Except this time, the published grid curves in fact coincide perfectly with the 7591’s operation in the 225, whereas the relevant curves for the 6L6 in the 240 do not. In this case, it’s the published typical operating conditions that would seem to be out of sync with reality.

Using data generated from the published grid curves then and Mac supplied voltage information — and accounting for the cathode bias generated as well, the correct total quiescent output tube current in the 225 then is 20 mA per tube. With direct calculation, using Mac’s voltage and resistance data, it indicates that total quiescent current flow is also 20 mA per tube, so the grid curves and Mac’s information agree. With a driver stage current of 8.3 mA, this sets 49 mA as the optimum total quiescent current draw per output “section”, as described in the 240’s discussion.

Unlike the 240 however, there is no high power transient issue with the 225, as without the direct coupled cathode follower driver, and using a greatly reduced R/C coupling factor into the output stage, the transient problem does not occur. That makes the actual quiescent current setting less of a concern in the 225 than with the 240, although ensuring the correct setting will ensure that maximum performance and tube life are both achieved at the same time.

Against this however, the 7591 has basically double the Gm of the 6L6, so manufacturing tolerances have a greater impact on this tube’s grid characteristic. All else being equal then, this will produce a wider variation in actual quiescent current produced from one tube example to the next. Therefore, in addition to using a matched quad of output tubes, it would also be a good idea to install the same 1Ω precision current measuring resistors in the 225, as is recommended for the 240 as well. Actual quiescent current can then be adjusted by appropriate adjustment of R51.

---------------------------------------------------------------------------------------------------------------------------------------------------------------------

The regulator connected to the MC240. I know, it's not pretty. But it was a junk box trial effort and served the purpose. The dummy plug is shown next to the regulator. When it is used, the power supply connections within the amplifier are exactly as Mac designed them. If there is interest in the regulator design, I will be happy to draw up a schematic and post it at a later date:

Underside. This pic was taken at a time the search was on for the dissimilar recovery time between the two channels from a transient overload event. All of the wiring for the regulator is installed, showing that the look remains largely as Mac built it:


One wire was removed and taped off, and others added to convert the Constant Voltage output connection socket into a socket for connecting the regulator or dummy plugs:

 

[damacman]

Fascinating read Dave. I'll have to revisit for a 2nd pass to get it all. Wow.

 

[62caddy]

I read the entire thing. Much of it is beyond me but the findings are interesting and helps to explain Mc's conservative output ratings.

As I understand it, output within THD spec is the most restricted only for the first transient while maximum output for successive transients is closer to the real-world continuous output (ie ~ 20% above rated spec). Of course, none of this is critical if transients never exceed 39 wpc so it's likely to be a non-factor in most situations as probably few push elderly tube equipment to its limits and the likelihood such equipment would be paired with relatively sensitive loudspeakers.

I liked the part about the three MC275s being used in the school's PA system. I wonder where those units wound up!

Thank you for the detailed analysis. The only question I have is whether the early SS models have the same characteristics.

 

[Paul K]

OK...this is a lot to intake.
I am certain this answers why with my probes some of the JJ 6L6GC tubes were displaying 13 or so ma when I tested them.
So this seems like a weird question but in those cases is it best to shoot for a quad that reads approx the nominal you indicated (26 or so ma) and not use the JJs in this application? (Or any that exhibit low current draw?)

 

[rustnrot]

Brilliant Dave!

This work certainly wins the "What did I do on my summer vacation" contest.

Your work and documentation is always first rate and this certainly is one of the best R&D projects you've done.

I'm looking forward to more details of this, including the regulated power supply schematic.

 

[KeithTr]

This post reminds me a little of a friend who used to design satellite propulsion systems. He once showed me a paper he was presenting at a conference and I got through about one of ten pages before I asked him: How many people are going to be at this conference? - he said about 200. I then asked, How many people are going to understand your paper? - and he replied: At Least a Hundred!

A lot of this is well beyond my limited knowledge, but it's one of the reasons I really appreciate this site. Nobody has to do this or has any obligation to help - and yet so many here do. I could probably read this ten times and not get it all, but the fact that it's here at all is deserving of a big thank you.

Kind regards, Keith

 

[SaturationPt]

A great summary of some excellent work Dave, thank you for spending the time to put it together here for those of us less motivated and less knowledgeable.

Unfortunately for me, I will probably start to dig into my MC75 restorations and begin the “analysis paralysis” phase looking for similar opportunities.

 

[ramjet]

Very thorough and informative work Dave. For me, a better understanding of why the Mc225 is different, possibly also why it is held in such high regard. Also, testing and analysis of how important a power supply is to the quality of audio power output. I would be curious if by testing either the Mc40 or Mc75 would show if they are affected by transient distortion to the same level as their stereo siblings; that being one power supply and one driven channel vs two.

 

[John James]

Way, way over my head, but I certainly appreciate the time and effort it took to put all that down on "paper".
Thank you, Sir!

 

[TSmith8605]

Well written and very informative. Actual engineering.

When's the book coming out? I'll buy a copy.

Thanks for posting.

 

[dcgillespie]

Thanks for the responses! I didn't know how much would generated in responses, but the goal was to inform, more so than to generate a huge response.

Dance -- Thanks for taking a stab at it. Nothing worse than reading some long something you don't really understand. For my part, I did try to write it in a way that the most of the community could benefit from it -- not too technical, but not so general as to be of no help, either.

Caddy -- Thanks for reading all the way through -- I know it was a long read, but --

Paul -- I knew it was a lot of intake, but I also realize that this is Mac territory, and if you're going to come with anything but accolades, then you better come prepared to back up your message. I hardly mean that against anyone here, or even Mac itself. Mac did sooo much right with these designs, that anything less than an accolade needs to be thoroughly explained, and that's really what much of the presentation is -- an explanation of what was happening, why, and what could be done -- more than it was an endless list of design deficiencies. In it's time, it would be hard to quibble with any aspect of the design at all. The opportunities I raise are the result of time (from better understanding as to how amplifier anomalies affect reproduction), and technology, that simply wasn't available in that day.

As for what output tubes to use, the main point to understand is that brand, construction, vintage, yada, yada, yada of the tubes used falls out of the equation as long as the tubes are set to idle at the correct quiescent current. To summarize, as long as each output section (which includes output tubes, driver and pre-driver stage are drawing 68 mA, (which equates to about 25 mA per output tube), then it doesn't matter what tubes are used as long as they are good. In my amplifier, I would never bet my life or the farm on being able to tell the difference between the RCA black plates and the JJ tubes -- as long as they're idling at the same, optimum quiescent current, their performance is virtually identical.

Rust -- I'll get to work on the regulator scheme I devised soon.

Keith -- Thanks for the kind words. I knew going into it that it may too long for folks to get through. But I'm hoping they love their vintage Macs enough to see it through. I tried to write it as clear and keep it as interesting as possible to hopefully minimize reader fatigue......

Sat -- Get to work! Those 75's are worth every minute put into them!

Ram -- I have the same questions about the mono brethern, but I would bet that they are affected the same way. Actually, the stereo amplifiers have a bit of an advantage in that if a high power transient hits just one channel, there is no transient clipping at all in the driven channel, because the un-driven channel acts as a sort of regulator for the power supply when that happens: the undriven channel reduces its current draw when power supply voltages droop, and so acts to minimize the droop. I didn't explain why, but commented on this when I indicated that with only one channel driven, there was no transient clipping present.

The mono units have no such advantage -- but I've never had one here to check how they perform under transient conditions. Maybe SaturationPt will be able to give us some input on that.

Seems pretty impossible to truly mimic a mono 40 with a 240 power supply powering just one channel -- a condition where we already know the clipping won't occur.

Dave

 

[SaturationPt]

Thanks Dave.

I don’t recommend anyone “holding his breath” waiting for me to complete my 75s, they haven’t moved in a while. I need to retire and get to it!

Thanks again for your detailed summary, I have already bookmarked it for review when I restart my project.

I think I’d like this added (as a link) to the Hints and Kinks section if we can.

<edit> Thanks Mods!

 

[sberger]

I love my 240. It was acquired from Audio Classics a few years back and has been trouble free. That said, I shall DL this incredible piece by Prof. Dave and save it for that time when I may need somebody with the skills to use it will do so in making this wonderful amp ready for future generations to come.

Thank you Dave!

 

[psychoaudio]

Dang, this Is epic. Thank you for all your contributions Dave.