The End of an Era: The Fisher 55-A

The Bias Regulator

So with the secrets of the driver section unlocked, the bias regulator circuit is up to bat now.

A few commercial designs, as well as more than a few seasoned experimenters have all toyed with or executed some form of bias regulation in equipment they've worked with. After all, it makes sense, right? Stabilizing the bias voltage must stabilize the operating point of the output stage, which must make for better performance. In reality however, unless such schemes are planned out very carefully, they can actually do more harm than good.

If you stabilize just the negative bias voltage, but not the (positive) voltages going to the other elements of a tube, it can make for radical shifts in the operating point depending on the prevailing AC line conditions: When the B+ voltage rises but the bias voltage doesn't, output tube quiescent current can rise significantly, leading to potentially overheating the output tubes. If the B+ falls but the bias doesn't, quiescent current can fall significantly, leading to increased distortion, and reduced power output. But that appears to be exactly what Fisher is doing here, right?

Well, no, they're not. The concerns regarding regulation of the bias voltage as outlined, relate to any amplifier. And while it may appear that against the backdrop just presented that Fisher made a big boo-boo, that's not what this particular bias regulator is doing.

The concept of bias regulation is virtually always made from the standpoint of regulating the SOURCE of bias voltage. But in this case, the regulator is actually quite invisible in that regard: As the B+ rises and falls, so will the bias voltage presented to the output stage. So what is the regulator actually doing then? In these amplifiers, the regulator is actually SINKING current -- current generated by the driver stage!

As long as the amplifier is operating in either in Class A or Class AB1, the bias voltage is a product of whatever the regulator sets it at, where at that point, it will then fluctuate with line voltage movement just as any of the other operating voltages within the amplifier do -- the point being that the regulator does not react to changing voltage input applied to it. However, when the amplifier shifts into Class AB2 mode, then the regulator goes to work.

Remember that in Class AB2 mode, the driver stage is developing POWER, because of the current drawn by the output tube grids during that mode. Power is a product of voltage AND current. In Class A or Class AB1 mode, there is no current drawn by the grids, and therefore no power produced by the driver stage; there is only a simple voltage swing being applied to the output tube grids from the driver stage. So when the driver stage is actually dumping power into the output stage, what is the circuit path that the current flow takes?

There are three elements in the grid current circuit path: There is the grid cathode path within the output tube, with the cathode effectively being grounded. Then, there is the driver circuit itself which is supplying the source of the current. And finally, there is the bias regulator, of which its output (the plate) is grounded. So there is now a complete circuit path created, with all three of these elements connected in series. But why is the bias regulator needed?

As with any series circuit, if any element in the circuit path represents a high impedance, then it impedes the flow of current in the complete circuit. We saw that happen in the driver circuit, where the losses in the driver transformer were great enough, that it took C10 and C11 to lower the driving impedance to the point where the driver could then develop enough current flow in the circuit path to push the output tube grids positive, and commence Class AB2 operation. That's why I spelled out the difference in driver stage internal impedance both with C10/C11 in, and out of the circuit.

And so it is with the bias supply as well. If -- as a source -- it represents a high impedance, then it will impede the flow of current in grid current circuit, and prevent the current from flowing in the grid of the output tube, preventing Class AB2 operation. In such a scenario, as Class AB2 operation tried to commence, the negative bias voltage supplied by the bias supply would actually increase (become more negative) due to the high impedance the supply represents, with the bias supply effectively absorbing the driver output voltage that was intended to be applied as a positive voltage across the grid/cathode element within the output tube. In such a case, the grid/cathode elements within the output tubes simply become rectifiers, rectifying the excessive driver voltage so that the resulting negative voltage created augments the negative voltage generated by the bias supply itself, causing a greater negative voltage to appear across the supply.

THAT then is exactly what V1B and the bias regulator circuit is designed to prevent from happening. It provides a low impedance path at the output of the bias supply, that can sink (or pass if you will) the current supplied by the driver stage during Class AB2 operation to ground. With the impedance of the driver stage itself reduced to a small value, and the impedance of the bias supply reduced to a small value as well, then it means a low impedance path now exists from the output of the driver stage, to the grid/cathode elements within the output tubes, allowing current to flow across them, and Class AB2 operation to take place.

So with the operation of all the unique design elements of these amplifiers laid bare now, is there anything that can be addressed in them today to improve the performance that sprang from their drawing board over 60 years ago? It turns out there is, with one of them being quite significant. It doesn't change the character of the amplifier at all. It just gives you more of the character it already has.

But that's next time.

Dave
 
So essentially the 12AU7 bias regulator is being used in place of the series dropping resistors typically found to allow for the proper voltage but without the high impedance that tens of thousands of ohms resistance between the bias rectifier and the grid would offer, yes? In this case, the plate resistance of the 12AU7 plus the DC resistance of the driver transformer secondary would be your supply impedance if I'm following you correctly.
 
In relation to the bias supply, the regulator -- in conjunction with R 25 -- forms a tapped shunt across it's output, helping to establish the correct quiescent bias voltage by creating the proper voltage drop across R25. But in this configuration, it also acts to sink any current supplied from the load as well.

Dave
 
been an excellent read so far dave.The audio gods were surely superviseing the early Fisher techs that produced these superb amplifiers.
I get goosebumpy listening to my 50a pair,dynamics,channel sepearation and eaking out all musical content available from source is what these do in spades.

hunter
 
50AZ schematic

I'm posting this info because my early group of 55As running from #s 124314 through 124616 has the same network across R9 as shown in the 50AZ schematic. C3=100uf@25v. This is not shown on the 55AZ schematic Dave posted to this thread. Also, this network is not found on my units numbered 124843 through 124899.

attachment.php


attachment.php
 
Last edited:
Roc -- Fisher -- like many companies during the 50's -- struggled with stability in their designs. During this time, as the use of increasing amounts of NFB pushed the expectations of performance and specifications ever higher -- against the limited practical experience of effectively applying it -- it resulted in a number of brands making various "adjustments" to the NFB and stability circuits of their designs (uh, dialing back performance in plain English) to achieve practical stability outside the laboratory, and in the field where their equipment was actually used. Many times, the changes were undocumented on their schematics -- or others, it wasn't even a component change at all, but a change in production lead dress.

For example, the model 80-AZ went through a number of build changes involving lead dress in Fisher's efforts to get a handle on the stability of that design. These changes to address stability basically amounted to what was called a "gimmick" in the day, and was either lauded or decried depending on your point of view.

The network you refer to on the 50-AZ, consisting of a 100 uF cap in series with a 560 ohm resistor across R9 helped (in part) to set the NFB level in these designs. By removing the network, it resulted in the amplifier operating with less NFB -- almost certainly in this case, to address low frequency stability. Of course, such a move would also increase the distortion produced, raise the damping factor, and reduce the frequency response, versus that of the original published specifications. When such adjustments were made however, the specifications RARELY ever got revised to show a reduction in performance. To do so would have been marketing suicide, since nobody else ever revised their specs either. Ultimately, it was all covered in the catchall clause of "Because Fisher Radio Corporation is always working to improve the performance of it models, it reserves the right to change....yada,yada,yada...." You get the idea.

In any event, this is why I never get too excited about models from this era actually meeting their published performance specifications. The target defining what stable, high performance really meant was moving so fast during that time that trying to keep actual performance and published specifications in agreement was quite a challenge in deed -- even at Fisher Radio Corporation.

Dave
 
Last edited:
Rich -- What is your understanding of the correct value for C1 and C2? 1 uF or .1 uF? The resolution of the schematic copies makes it tough to determine.

1 uF would be extremely large for these locations......

Dave
 
Dave; Because C1 and C2 show as "molded" on the parts list, and not electrolytic, I would presume it to be 0.1uf. I've never seen a sub 1uf lytic in a fisher prior to the Late 60's solid states, and they are rare at that.

Granted the decimal looks like it's been written in afterwards(Which I've seen in a few of them too), but look at the parts #'s for C1-2 and C6-7 They are only one # off. which would also lead me to believe it's a sub 1uf film.

Larry
 
Agreed Larry -- your reasoning makes perfect sense. The unit I have (previously worked on), has C1 removed -- which is fine -- but has a 1 uF installed at C2, which if that is the correct value, I recommend to reduce to .1 uF, which is plenty big in that location. Using a 1 uF only invites all manner of power supply undulations being added to the source signal, which likely raises IM distortion among other things.

Dave
 
Rich -- What is your understanding of the correct value for C1 and C2? 1 uF or .1 uF? The resolution of the schematic copies makes it tough to determine.

1 uF would be extremely large for these locations......

Dave

Dave,

Both are .1

Rich
 
Here's the network pic referenced in post #46
 

Attachments

  • 55ANetwrkPic.JPG
    55ANetwrkPic.JPG
    151 KB · Views: 71
Sidebar: Housekeeping Details

I'd like to publicly thank Rocjr for his invaluable historical information on this series of amplifiers, and how they fit together chronologically. There has been some significant backdoor communication pinpointing where the particular 55-A I have here appeared in the history of the original 50-A platform.

While I was aware of a 55-A version employing EL34 tubes, I mistakenly assumed that it appeared before the 6550 version -- this, since the EL34 tube appeared earlier than the 6550 did in history. This fact, coupled with Avery's awareness and intense desire to be the first at anything and everything in the industry, is what brought me to my understanding. However, Rocjr's information documents that in fact, it is the other way around, with the EL34 versions (there were two, the last sporting a 4 ohm output tap as well) being the final effort of the original 50-A platform.

To that point, the basic front end and Class AB2 driver platform survived to ultimately become the Fisher 200. But by that time, it was no long using a triode connected output stage, but used the same EL34 tubes now operating in pentode mode instead. Therefore, the last of the true original 50A platform stops with the final EL34 version of the 55A.

Use of the EL34 -- and at the same time, switching to the GZ34 rectifier tube with it -- solved the two biggest issues with the previous 6550 version: The high B+ surge created at initial turn on, and the strain that 6550s placed on the driver stage during Class AB2 operation. The EL34 tube is an easier tube to drive into AB2, and the GZ34 employs a cathode sleeve, producing an extended warm up time to delay the B+.

With the EL34 versions however, seeds were already being planted for what was to come: More power, more feedback, and more stability circuits. The latter two were already creeping into the EL34 55-As, with these concepts and more power only growing over time in succeeding models. The simplicity of the original 50-A design was beginning to disappear, loosing the war to sound by specification.

Of the succeeding versions of the 50-A platform then, the 6550 version of the 55-A remains most true to that of the original 50-A concept, with deviation beginning after that model. The initial steps are small in the EL34 versions, but undeniably there none the less.

I just wanted to take a moment to keep the information historically correct for any and all who own these gems. And again, thanks to Rocjr for making that possible!

Dave
 
Opportunities For Improvement

While I've published my fair share of modifications over time, all of them have been based on either demonstrable worthwhile improvements as shown in the lab, practicality of use in today's audio environment, and/or an improved listening experience. While the last item is necessarily subjective, there has been enough independent verification of most of these modifications to give validity to the statement. The bottom line is, I don't modify for the sake of modification, but rather, only when the results can produce a clear, worthwhile benefit.

In looking at this version of the 55-A under that spotlight then, there are four areas of opportunity to consider:

1. The Z-matic circuit,

2. The pre-driver stage,

3. The high voltage rectifier circuit, and

4. The bias regulator circuit.

With the exception of #3 relative to the EL34 version of the 55-A, all of the areas listed above represent opportunities for any of the amplifiers directly relating to the original 50-A platform. For this exercise then, that would specifically include anything up through and including the final EL34 version of the 55-A. But they are also relevant to the pentode output stage models following this series, that continued on with the same basic topology of driver, bias regulator, and Z-matic feedback circuits. Obviously however, any version of any of these amplifiers built without the Z-matic feature can disregard the recommendations regarding that feature.

None of the proposed modifications are overly difficult to install, and none take away from the essence of what these amplifiers represent to today's vacuum tube audio enthusiast. They simply capitalize on opportunities and observations that nearly 60 years of intervening time make possible. The following paragraphs will address each area in detail.

Z-MATIC CIRCUITS -- Unless you have vintage speaker systems and want to retain this feature, it should be electrically removed. The idea of increasing amplifier output impedance (which is what the Z-Matic circuit does) is an anachronism today. However, even with the Z-Matic control turned to the off position, speakers connected to the 8 Ohm tap will still be influenced by elements of this circuit.

In the off position, the effects of the circuit for speakers connected to the 16 Ohm tap are completely removed. But for speakers connected to the 8 Ohm tap, a resistance of nearly .7 Ohms is still in the circuit between the Common output terminal, and the common terminal of the OPT secondary winding. This reduces power output and increases the output impedance of the amplifier for speakers connected to this tap.

The easiest way to address this matter, is to make sure that the rear panel impedance selector slide switch is ALWAYS set to the 16 Ohm setting, regardless of which speaker tap is being used. With the switch in that position and the Z-Matic control turned off, the Common output terminal will then always directly reference ground and the common terminal of the OPT secondary winding, regardless of which output impedance tap is used.

Of course, it still accomplishes this by way of the switch contacts in the impedance selector and Z-Matic on/off switches with this approach, so the best answer is to actually remove these circuits from the amplifier. However, if you're using the 8 Ohm tap, then using the switch approach first to completely defeat the Z-matic circuit will allow you to easily judge the effects that removing the internal resistance has with your particular speakers. Then, if you want to make it a permanent change, you can do so by removing the circuit.

The Z-Matic circuit is easy enough to remove by:

1. Removing all the wiring associated with the Common output terminal, S1, S2, R32, and R34. Make sure that terminal #6 of the OPT remains grounded.

2. Connect a wire between the Common output terminal, and terminal #6 of the OPT.

3. Connect the free end of R9 to the junction of R31 and R33.

4. Connect a 1K 1/2 watt resistor from the junction of R31 and R33, to ground. Alternately, you could just replace R33 with a 240 Ohm resistor, and dispense with the added 1K resistor.

Besides limiting power output, the effects of the added internal resistance with 8 Ohm installations (as originally instructed) is to limit bass definition, which otherwise does not occur with 16 ohm installations.

PRE-DRIVER STAGE: Comprised solely of V3, this stage is required to deliver nearly 90 volts peak to each driver tube grid for the amplifier to reach full power output conditions. The 12AU7 tube used in this position is capable of producing 101 low distortion peak volts to each grid, when the B+ supplying this stage is at least 380 volts, which is the amount provided when the amplifier is producing full power output.

To provide this amount of voltage swing, each section must draw about 2.3 ma each of quiescent current in order to deliver the necessary voltage swing at low distortion. However, Fisher chose to over bias this stage, so that each section only draws no more than 1.5 ma of quiescent current in each section. This results in elevated distortion before the required drive level is reached. The fix for this is to replace the existing cathode resistor (R14) with a new value of 2200 ohms. This is less than half the value of the original R14, but causes the pre-driver stage to draw the current necessary to operate with a more linear load line, and produce the required voltage swing at the driver tube grids with minimum distortion.

In linearizing this stage however, it also produces an 11.5% increase in voltage gain by the stage. Left unchecked, this would have the effect of increasing the global NFB factor by 11.5% as well. To correct this, the value of R31 should be increased to 2500 Ohms, which corrects the feedback factor to within .6% of the original value. Alternately, a 270 Ohm resistor can simply be placed in series with R31, for nearly perfect compensation.

HIGH VOLTAGE RECTIFIERS -- The issue with the B+ surge created at turn on has been well documented. Early on with the original 50-A, Fisher used a 5V4(G) rectifier tube with its slow heating cathode to eliminate the surge. At the time however, that tube was plagued with dependability problems, so Fisher changed to conventional quick heating rectifier tubes solve the problem. However, that then created the B+ surge. The "modern" 5V4GA is a very dependable tube, and would be an easy answer, but is marginal in the 55-A chassis with respect to PIV rating, since the power transformer HV winding delivers 1.05 kV in this unit under stabilized operation.

Fisher answered this problem in succeeding versions and models by specifying a GZ34 rectifier tube, which does eliminate the surge -- but at the same time, also produces 20 more volts of B+ at the output of the filter choke in this particular 55-A chassis. That does address the surge problem, but requires increased negative grid bias voltage to maintain the quiescent bias set point on the Power Monitor -- which in turn just piles on to the driver section requirements. That problem was addressed by specifying EL34 power tubes when the change was made to GZ34 rectifier tubes, which have reduced drive requirements over that of the 6550.

So these tube changes that Fisher made do resolve the B+ surge problem, and do so without introducing any new ones. These tube changes could be retrofitted into this version of the 55-A that I have here. But doing that starts to really change the character of what this version of the 55-A originally represented. Therefore, it has been decided to go a different route.

By installing an appropriate timing module and relay, the heater voltage to the big 5AW4 rectifier tubes can be delayed by about 30-40 seconds, to completely eliminate the surge, while still using quick heating rectifier tubes. The relay employed however should have 15-20 Amp contacts as rated for resistive loads, as these rectifier tubes employ 3.7 Amp heaters each. The contacts need to be hefty to have good durability over time. When the contacts close, the rectifier tubes heat, and B+ ramps up with a nice smooth rise, completely unlike B+ on/off switches that guitar amps often use for standby operation.

An appropriate relay and module has been found for the job. And while the cost plus shipping for these items will be around $60, this approach has one advantage that no rectifier tube can provide: Upon the slightest loss of AC power, the timing module resets and starts its timing sequence all over again. This means that any damage from short AC power cycling pulses can forever be eliminated, so that the 5AW4 rectifiers in this model should last a lifetime. When these components arrive, I'll post a pic of their installation.

One way or another however, this is a concern that must be addressed in these units. With typical line voltages today, the surge rises to just over 600 volts, and this gets applied to many of the filter cap sections before the audio tubes warm up. With the main filter cap only being rated for 500 volts, and the succeeding caps only rated for 450 volts, they get beat up pretty bad with each power on cycle. The older caps had poor ESR ratings compared to modern caps, but they were better at handling surge voltages than the equivalent component of today is.

That brings us to the the bias regulator circuit, which will be addressed in the next post.

Dave
 
The Bias Regulator, Part I

If the driver section is the heart of a class AB2 amplifier design, then the bias regulator must surely be the liver if transformer coupling is used. As explained earlier, during Class AB2 mode, you can either pull the grids positive from a resting voltage source that is more positive than the grid, or you can push it positive from a resting source that is more negative. But either way, stability of the voltage source used to provide the grid current is imperative if distortion is to be minimized, and driver effectiveness is to be maximized. In the case of these amplifiers, the grid is both pulled positive by the C10/C11, and pushed positive by the transformer.

Now the positive voltage source (the B+ supply to V4 & V5) is of little concern in this design, largely because the driver stage itself is a push-pull Class A design. Therefore, the quiescent current drawn by each tube is enough to cover the total current delivered by each tube. More specifically, while only one tube conducts current through C10/C11 at a time, it is still a symmetrical action, such that with both tubes contributing to the signal developed in the transformer secondary all the time, the total current drawn by these tubes then is largely and inherently constant. As a result, a simple dropping resistor can be used to supply the stage, with little change in B+ voltage supplied to the driver stage during Class AB2 mode, save that produced from drop in the main supply itself.

On the negative source side however (the bias supply), the picture is not so pretty. Here, because the output tubes effectively represent a full wave rectifier configuration in so far as the current developed in the transformer secondary is concerned, the grid current developed on each side of the push-pull signal does not cancel each other out as with the positive source powering the driver stage, but rather, is effectively added together in a very non-asymmetrical way, pushing current into the bias supply, and causing the bias voltage to actually increase. Once again, this is why the bias regulator tube (V1-B) is NOT designed to stabilize the basic source of bias voltage, but to provide a sinking action to any current applied to it.

So how effective is this regulator? As a basic point, obviously good enough to enough to allow Class AB2 operation. But it is far from perfect, as will be shown.

In the original 50-A, maximum power output is a (measured) 40 watts RMS. Placing a matched pair of 6L6GC tubes in this chassis confirms this, as did measurements by the Navel Research Center. At 125 mA of quiescent current draw, this required a bias voltage of -40 volts to achieve. With nearly 88 peak volts applied to each grid, this then produced ~ 17 ma of transformer generated grid current during Class AB2 operation at 40 watts output. As explained earlier, more grid current is actually drawn than this due to the action of C10/C11. But the bias regulator only has to sink the current generated by the transformer secondary, so that is the figure relevant to this discussion. At 40 watts power output, the voltage across the bias regulator had risen from -40 vdc, to -52 vdc. This represents 30% regulation, which is very poor. Of course, without the regulator, it would be much worse, and prevent the amplifier from even entering Class AB2 mode.

In producing the first version of the 55-A, 6550 tubes replaced the 6L6 class tubes used in the 50-A. Quiescent current rose to 180 mA from a bias voltage of -42 vdc. With 82 peak volts drive to each grid, a maximum power output of 50 watts RMS is produced, with the transformer generated grid current reaching 20 ma. Obviously, bigger grids produced more grid current at less peak drive. In this scenario, the voltage across the bias regulator has now risen to -59 vdc, and regulation has deteriorated to > 40%.

With more grid current being developed when driving 6550 tubes, it means that the load that the driver stage operates into has been numerically reduced (harder to drive), causing the driver stage and the bias regulator to both work harder. For a driver section that was already nearly maxed out, and a bias regulator that was already in over its head with 6L6 class tubes, then upping the output stage to 6550s is just piling on, adding insult to injury. As a result, while power output increased 10 watts with the change in tubes, distortion did not, with both tubes producing ~ 1.8% THD at 1 kHz, at the onset of clipping for their respective power output levels. The vast majority of this is generated in the maxed out driver section.

Against this performance, an obvious point of modification for improved performance then would target the driver stage. But that presents a significant effort to redesign that stage for greater undistorted output swing, that would also change the character of the original product. So can any improvement be had within the confines of the existing design?

In fact there is, but to achieve it, it is important to understand this salient point: ANY loss of regulation across the output of the bias regulator during class AB2 operation represents needless drive that the driver stage has to provide a signal swing against to counter act. With the poor regulation displayed then from the stock regulator design, this wasted swing is significant -- particularly when judged against the swing of the driver section all but being maxed out under existing full power conditions. The best attack to improve performance then is not to modify the driver stage for greater output, but modify the bias regulator for less loss.

Fisher came to realize this as well, and in future Class AB2 platforms, doubled up both sections of a 12AU7 to lower the impedance of the regulator. Ultimately however, this just improves the situation from grossly poor, to very poor. So what can be done?

This is a problem begging for a SS solution, and fortunately, a very simple one exists. Now before you all go and get up in arms, follow this through. What is needed is a very low impedance to sink the Class B2 current into when that mode commences. Low impedance is exactly what SS components are all about. Could you do it with tubes? Yeah, but why? it would take more and bigger tubes than the regulator currently consists of, and STILL likely not approach the performance that using SS devices can deliver.

And then there is this, too. Class AB2 -- and therefore the grid current associated with it -- doesn't even kick in until about 23-24 watts RMS, so at power levels under that point, the impact of the circuit change is Zero, Zip, Nada. During Class A and AB1 conditions, the impedance of the stock regulator circuit is set by C15-C. All by itself, it provides a plenty low enough impedance during these modes of operation to make the regulator invisible. It would continue to do so with the proposed modification. So the modification would only present itself sonically above 24 watts of power output. At that point, the effect of the modification would be the same as removing a resistor in your speaker leads. Would you intentionally install a resistor at that location? I don't think even the most diehard trendy audiophile would do such a thing. Yet that is exactly what the stock regulator represents to the Class B audio signal.

Consider that when the driver has a full head of steam behind it (producing 20 ma of transformer grid current), the stock bias regulator appears as a 2950 Ohm impedance between the the driver transformer's secondary CT, and ground. The bias voltage has risen an addition -17 volts to -59 vdc in producing Class B operation, that the driver must first overcome before Class B operation even commences.

If we take V1B, and replace it with an N-Channel power MOSFET and a single resistor, this picture improves unbelievably. Such a device typically has an "on" resistance of well under 1 ohm. Now, that -42 volts of bias under quiescent conditions? -- REMAINS at -42 volts at 50 watts of Class AB2 power output! And that 82 volts of peak signal drive per grid that the driver had to develop before? -- is now only 65 peak volts per grid. This gives the driver section some considerable breathing room then under Class B2 conditions. It still takes 23 volts of peak positive drive in Class B2 mode to produce 50 watts of power output, but now it is no longer "on top of" -59 volts of static bias, but just -42. Now, the driver section -- with its improved output and linearity from the pre-driver -- looks like a real he-man to the 6550 grids.

So what's the results of such a move?

That's in Part II.

Dave
 
The Bias Regulator, Part II

With the driver able to fully drive the output tubes in Class B now, a number of things happen: All of a sudden:

1. The 55-A can actually produce 55 watts RMS at mid-band frequencies.

2. The power bandwidth is flattened to produce 50 Watts RMS across the entire 20 Hz to 20 kHz audio bandwidth. Before, distortion was severe at 20 Hz over 40 watts RMS, and the unit really couldn't even produce 40 watts at 20 kHz.

3. Distortion drop significantly. Whereas before, it was either 1% at 1 kHz at 47 watts, or 1.8% at 50 watts, now it is just .5% at 55 watts.

4. And then there's this: Now, minimum low level THD (1 watt) and minimum high power THD at the onset of clipping both occur at the same quiescent output tube current level of just 130 ma -- not the 180 ma of the stock design. Distortion at 55 watts is as indicated above, while at 1 watt, it clocks in at just .08%. Both of these figures are lower than the stock design produces at these power levels, while the THD at 23 watts (the transition power from Class AB1 to AB2) remains unchanged. Clearly, the 180 ma quiescent current requirement of the stock design for 6550 output tubes was for the benefit of the driver section, by reducing the static bias voltage as low as possible to help the driver section out. Now however, the output stage quiescent current level can be set for the most linear point of it's own operation to minimize distortion even further.

5. Finally, with the reduced output stage quiescent current, both output and rectifier tube life are impacted in a very positive way.

So here is a list of modifications that can have a meaningful, positive impact on the performance of any of the versions of this amplifier that sprung from the original 50-A platform. Component life is clearly improved by delaying rectifier tube warming. Rectifier and output tube life are clearly extended with reduced quiescent current flow. Power and bass definition is improved for speakers connected to the 8 Ohm tap, and overall power, power bandwidth, and distortion are all significantly improved through improved pre-driver and bias regulator performance. And best of all, it will all be about as sonically transparent as it gets when power levels below 20 watts are used.

Rocjr has indicated that performance is maximized when quiescent current is set over 200 ma, using space heater output tubes (KT90s). I have no doubt that this is true, as in essence, what the higher idle current was actually doing was giving the driver some extra breathing room when driving medium efficiency speakers with demanding material. Individual preference may still prefer a higher quiescent setting. But with the modification to the the pre-driver and bias regulator circuits, it is no longer necessary for the purpose of coming to the driver circuit's aid, and distortion is actually increased when using such an elevated setting.

Next time, I'll glue this all together with some performance pics and final discussion.

Dave
 
Dave,

That Mosfet replacement is a nice touch.

Looking forward to the wrap up and hopefully modified schematic ???

Frannie
 
Indeed. SS parts do regulate way better than tubes can. Any chance the MOSFET can be rigged in to plug into the original tube socket for a no-modification improvement?

I don't own one of these, just thinking out loud here.
 
Back
Top Bottom