• Please note that Audiokarma will be offline briefly for updates early on Monday morning, September 28th.

Fisher's First High Power Stereo Amplifier: The SA-300

huh, well thats interesting. I have to say I've never seen that approach used before. I had to take a second look at that output stage to figure out what was going on. My first thought was that it was only going to get feedback to the upper tube, then I realized the 4 ohm tap was grounded, not the common tap.
 
Anatomy of a Modification:

Mopping Up

Details, details. Some related to the modification, some not, but all relevant to the SA-300.

1. RFI: Fisher clearly saw the SA-300 as providing monitor service in recording studios and radio stations, where RFI would run rampant. This is evident by the inclusion of C16, C18, C29, C30, C37, C39, C50,and C51 -- all designed to prevent the entrance of any RF energy inside the chassis. Unless you're living within a few miles of a commercial transmitter, these take up space and clutter the build. In the modified amplifier, they have all been removed.

2. Filtered and Unfiltered Inputs: In the original SA-300, the Filtered input provided both a high pass and low pass filter circuit that the input signal first passed through. This was stated as benefiting performance when using the amplifier with (the then new at the time) electrostatic speaker panels. Whether that's the case or not, Fisher promptly dumped the high pass portion of the filter circuit in the SA-300B, leaving only the low pass filter, so that only supersonic frequencies would be rolled off. The only other change made to the B models was the addition of a center channel output jack.

In the modified amplifier, all the original filter components have been removed, and I use the Filtered Input Jacks for a more practical purpose. The modified amplifier retains the original sensitivity of the stock design (0.90 vac rms for maximum power output), as appears at the Unfiltered Input Jack. However, this is frankly a little too sensitive for use with a typical active preamp, and certainly the 400CX-2 with which it was paired in the President console. That meant that the level controls had to be backed down, but as is so often the case, that compromises HF transient performance in the process. Accordingly, the Filtered input on the modified amplifier is now a fixed level, compensated input, requiring 1.53 vac rms to achieve maximum power output. In this way, response is not compromised, and the sensitivity of the amplifier is much more appropriate for use with the active preamp of your choice.

3. Output Tube Drive: Between the increased bias voltage the relaxed quiescent current draw requires, and the NFB provided from the partially cathode coupled configuration, the modified output stage requires basically twice the drive level that the stock output stage does, something that the stock phase inverter/driver stage was nowhere near capable of delivering, as it struggled just to deliver even that required by the stock output stage, where at 20 Hz, it basically ran out of steam before full power output was even reached. A significant portion of the success of the modified amplifier, is due to the fact that the phase inverter stage was able to be modified to provide plenty of increased low distortion drive to the output tube grids, across the full 20 Hz to 20 kHz spectrum, at full power output with both channels driven, with even a 10% reserve of drive capability remaining at 20 Hz to account for tube aging. Overall, the inverter stage becomes a much more stout driver, which is just the ticket needed for the increased drive requirements of the modified output stage. The increased drive was accomplished by raising the values of R21&R47, and R22&R48 to 43K@ 2W and 51K@2W respectively, removing and shorting out R24 and R50 (eliminating C31 as well), and replacing R64 with a 20 Hy, 20 mA Hammond choke. Finally, C23 and C44 are increased from 0.1 uF to 0.22 uF as well. Collectively, these changes allow the original 12AU7 phase inverter stage to provide all the drive necessary for the modified design. In the process, the B+ supplied to the inverter stage is not only appropriately elevated, but also filtered much more effectively as well.

Regarding the (now defunct) AC Balance Switch: The loss of this feature is not as great a loss as it might seem, as being a low level adjustment, it cannot provide for a balance adjustment made under Class B conditions, where AC balance is needed the most. As a result, the setting achieved by adjustment for minimum noise in the output rarely coincides with that produced by adjusting for minimum THD at higher power levels, which is how such adjustments are typically made. This is not to poo-poo the original scheme used, but simply to say that it's only accurate if tightly matched output tubes and phase inverter tube sections are employed. I removed the feature to maximize phase inverter/driver tube peak output capability.

4. AF Amplifier Stage: I recommend disconnecting the level controls and leaving them in place as dummy pieces. Or, if yours are sufficiently matched, then the new attenuation networks for the Filtered Inputs can still be used in conjunction with the controls, but usually, they will be insufficiently matched to allow for this. Of my controls, one measured 248K, and the other 143K. At this point, any controls I've found as suitable replacements require the mount hole to be ever so slightly enlarged, which I have resisted doing, since I don't like using level controls in the first place. If you (electrically) remove the controls, then replace them with matched 270K resistors, which will then allow the Filtered Input Jack attenuation networks to operate with a tight match as well.

Because the global NFB is now effectively being sourced from a 4Ω tap (1/2 of the original 16Ω winding used), the NFB network needs to be adjusted to maintain the same feedback factor as the original network did before the new output stage short loop was added. The new NFB network then should consist of a 3.6K resistor, and 270p cap. C28 and C49 are no longer used.

The original input grid components R12, R38, C14, C15, C35, and C36 should all be removed, and a 75K resistor connected between the input grid, and either the wiper of the level control (if retained), or the tip of the Unfiltered Input jack, where a 270K resistor to ground would also reside. If the attenuation network is desired, connect a 180K resistor in parallel with a 72 pF cap between the Filtered and Unfiltered Input Jacks.

At each 290 volt source, connect a 470K 0.5 watt resistor to ground, this to account for the higher B+ serving the phase inverter/driver stage now. These resistors also act as absolute bleeders, and are quite effective in regulating the 290 volt sources in the face of large changes in B+ voltage.

The step networks consisting of R19, R20, R45, R46, C20, C21, C41, and C42 should all be removed. They are not replaced with anything. The components of the series step networks (R17, R43, C19, C40) should remain in place as originally installed.

Odds 'n Ends:

1. The heater leads to V5 are usually not twisted. With the modification in place, these leads should be twisted to eliminate any hum from Channel B.

2. Adjust the Channel A Hum Balance control with the bottom plate in place as much as possible, lifting it on the end of the chassis where the hum control is located just enough to be able to make the adjustment with a long shaft screwdriver. If the control is adjusted without the plate in place, the adjustment made will be inaccurate once the plate is installed.

3. Power Supply Modulation: As full power is approached in both channels, power supply modulation of the output waveform ibecomes significant. This can be all but eliminated by the addition of a 64 uF @500 volt cap connected directly to the output of the rectifier tubes (Pin 8). This will raise the main B+ voltage by about 10 volts or so, which is insignificant compared to the performance improvement provided.

In the last installment, final performance of the modified amplifier will be discussed.

Dave


Below: Schematic of the "Front End". This shows the changes discussed in the text. Not shown is the 64 uF cap added to the output of the rectifier tubes.
Fisher SA-300 Front End.jpg

Below: New Phase Inverter/Driver Stage plate resistors help facilitate the increased output needed to drive the modified output stage.
SAM_2483.JPG

Below: Channel B Input Stage.
SAM_2484.JPG

Below: Channel A Input Stage.
SAM_2485.JPG

Below: New choke supplying Phase Inverter/Driver and AF Amplifier stages resides where old bias control once lived. Also, the new 64 uF 500 volt cap reduces noise under all conditions, and modulation at near full power conditions. A big benefit of this cap's installation is that now, the two 15Ω power resistors dissipate just 0.5 watt each under quiescent conditions. Along with all the other benefits of the modification then, the amount of heat generated under the hood with the modified amplifier is very small compared to the stock design.
SAM_2486.JPG

Below: The fully modified amplifier. The underside is now very clean and un-cramped in layout and presentation compared to the original build. Performance results exceed the original design in all performance categories, while the performance quirks of the original design are all gone. The heat output from the amplifier is now quite nominal in routine use.
SAM_2487.JPG
 
As Rich noted earlier, there are issues with the Sam's schematic for the SA-300. In addition to the mistake he noted (values for R69 and R70 reversed), also note:

1. The 30 volt source of AC voltage should be shown as a 55 volt source. At the Balance switch, C31 should be shown as connecting to a 55 volt source.

2. The -25 volt source at the wiper of the bias control should be shown as a -52 volt source. At the DC Balance Controls, the wipers of these controls should be shown as connecting to a -52 volt source.

3. The 8.5 volt indication shown at pin 8 of the EL34 tubes in Channel A should be 7.8 vdc. Ideally, this would also be the voltage indication at the same point in Channel B.

4. The unmarked terminals on the Speaker Terminal Boards should be labeled as "GND".

Dave
 
A spectacular contribution, much of which I even understood
index.php


The undersides of these units were always a bit intimidating to me.
Now, slightly less so. This thread alone justifies the entire website imho!
 
One Little Detail --

I keep forgetting to mention. When replacing the main B+ can (C1) with an original type replacement can, be sure to replace these washers in the same position as shown here (top of the picture is towards the chassis innards):
SAM_2497.JPG

Now I could go on at great length about how this was Fisher's latest cutting edge technology at using ferrite beads to further enhance the rejection of RFI in the SA-300, but of course, that would be a bunch of hooey. The fiber washers are actually a clever answer to what must of been a last minute "oops" moment: This can is located so close to the edge of the chassis that without the washers, the cover would bulge out at that point as the bottom lip of the cover would hit the can. With the three washers installed as they are however, there is a slight gap formed between the top of the chassis, and the bottom skirt of the can, allowing the bottom lip of the cover to slide into the gap formed (at the bottom of this pic), and eliminate the bulge. Crisis avoided!

Dave
 
Wrap Up

In the Lab:

The performance improvements delivered by the modified amplifier are significant. In the first place, the amplifier will now easily deliver 35 watts RMS from each channel with both channels driven, and do so from 20 Hz to 20 kHz with low distortion. The stock amplifier could only be rated as delivering 30 watts RMS under the same conditions, so there is a slight uptick in power output. What is more notable however, is that it delivers this power increase with significantly less distortion than the stock design delivered at lower power output. Under the conditions given, the modified amplifier delivers:

1. @ 20 Hz = 38.44 watts RMS with 1.80% THD. (Compares with 30.3 watts @ 2.4% THD stock)

2. @ 1 kHz = 42.24 watts RMS with 0.05% THD. (Compares with 37.5 watts @ 0.60% THD stock)

3. @ 20 kHz = 36.91 watts RMS with 1.0% THD. (Compares with 33.1 watts @ 3.75% THD stock)

4. IM Distortion @ 35 watts RMS equivalent power output = 0.10% IMD. (Compares with 1.0% @ 31.5 watts equivalent stock)

The distortion figures change very little whether each channel is driven individually, or together. Driven separately, each channel can deliver as much as 49 watts RMS now throughout the majority of the audio bandwidth. An accurate rating of the amplifier however would now honestly have it rated as a 70 watts RMS basic stereo amplifier.

Being a vacuum tube amplifier with an output transformer, IM Distortion becomes important. The low IM distortion of the modified amplifier then is particularly noteworthy, again shining a large spot light on the quality of the output transformers employed.

The performance match between the two channels is such that posting the individual results of each channel would result in changes to the average performance results given here that would only affect the second digit to the right of the decimal point. The performance match between the two channels is very tight in deed. Other than (almost surely) the SA-1000, no other piece of Fisher stereo gear will deliver it's full rated power output from 20 Hz to 20 kHz with both channels driven and with low distortion. This is not to poo-poo the other pieces, but simply to say that best of the rest is generally only good down to 25 Hz or so. With a full 35 watts RMS from 20 Hz to 20 kHz available in both channels at the same time, the modified SA-300 is now a notable standout piece in the Fisher stable of stereo products.

Frequency response of the modified amplifier is very flat, being +/- 0 db from 15 Hz to 20 kHz, and down 1 db at 40 kHz.

Finally, the stability of the amplifier remains absolute, with no amount of capacitance only loading causing any tendency towards oscillation. Pulse signals produce rapid settling regardless of load condition.

A few scope pics will help detail this performance. Each waveform is labeled, and so are self explanatory:
SAM_2491.JPG

SAM_2492.JPG

Below: At 20 kHz, this is particularly noteworthy performance as many amplifiers present a full power 20 kHz sine wave that -- besides typically being reduced in power output -- also becomes notably phase distorted as the output tubes try to deal with the extra load of the winding capacitance that larger OPTs invariably have. The stock SA-300 is better than many in this regard, but does still display notable distortion at this frequency due to the partial cathode bias used to help curtail the side effects of the partial direct coupling scheme used. The modified design removes both of these design elements, with 20 kHz performance being one of the huge beneficiaries as a result.

SAM_2493.JPG

SAM_2494.JPG

SAM_2495.JPG

SAM_2496.JPG

Finally, after discovering how to best adjust the Channel A hum balance control (described earlier), the hum and noise performance figures in both channels (-101 db below 35 watts) now exceed that of the stock design, thanks to the improved filtering provided by the additional cap at the output of the rectifier tubes, the EFB Screen and Control Grid Regulators, and the choke now powering the Phase Inverter/Driver and AF Amplifier stages.


In the Listening Room

In my insulated 12' X 16' listening room, the modified SA-300 is used with my Fisher 400CX-2 (President) preamp, Cornwall II speakers (101 db), and Teac P-650 CD Player. Using the new 1.53 volt inputs, there is no audible noise of any type anywhere in the listening room except right at the speaker boards under quiescent conditions, with the volume control set to a normal listening level.

As for listening material, I enjoy a wide variety of music from The Rippingtons, to Bach and Beethoven, to Patsy Cline, and 38 Special. From big band, to rock, to orchestral, and pipe organ. The modified SA-300 got subjected to all of it and more.

On "I am a Town", Mary Chapin Carpenter's voice is startling when it first appears, as if she is in the room above the center of the speakers. Very detailed, with no sibilance at all. Very natural sounding through horns, which is no easy task.

On any one of a number of Rippington tracks, the sound is seamless, energetic, and detailed, with no one instrument lost to another. Everything is there, with no muddiness at any volume level, soft or loud. Horns can sometimes put electronic instruments in your face and mar the presentation, but with this amplifier, everything stays in place for a very balanced presentation.

One of the most impressive aspects of the modified amplifier however is in the bass register. On Saint-Saëns Symphony No 3 (The Organ Symphony), you feel the bottom register of the organ notes as much as hear them -- just as you would if there in person.

This amplifier is one that brings a smile to your face in the first few moments you hear it. For those who critically listen, it doesn't take long to know if everything is right. This thing is just fun to listen to. Classically Fisher like in that it sits there so unimposing and unassuming, never drawing attention to itself -- until you hit "play", and then the fun begins.

I certainly always admit to the possibility of bias, as I have participated in more well run double blind listening test events than I care to remember (on both sides of such events), and have seen such bias first hand -- those who were absolutely convinced of a position going in, but unable to substantiate it going out. As a result, my absolute fall back position is that the modified amplifier sounds at least as good as the stock unit, and in my opinion, far better, being simply more lively, energetic, and effortless in its task. It never gives a sense of limitation. Of equal importance to me, is that on top of the great listening experience and unmistakable Fisher presence in the room, it is very satisfying to know that I can have that experience -- extended as long as I like -- knowing that even with the cage in place, the amplifier is providing all of that without cooking itself into oblivion.

As a result, this amplifier has moved from a position of one AC prong already out the door, to one of setting a standard for others to rise to. I have a number of very good lower powered amplifiers which are very enjoyable to listen to. But having the great sound and extra brawn of this amplifier -- on top of its really excellent lab performance -- makes this amplifier a standard for others to rise to, and take notice of. If you have one of these amplifiers, and have the concerns I did regarding the stock design, then I encourage you to give this modification a try -- I don't think you'll be disappointed!

Happy listening!

Dave
 
Last edited:
Very nice. You really cleaned things up and minimized the number of components.
When I get around to rebuilding my (spare) SA300, I'll be using your mod.


"One AC Prong Already Out The Door"? Funny line.
 
I am so envious at everything you do, Dave. I wish I had one tenth the understanding you have with which to explore and test different designs, and measure the results. It took me 56 years to become the master guitarist (and that's being overly egotistical) I am. I'd need another lifetime and start at age 9 to get to the knowledge you have. I am in awe. I wish I could just apply some of your skill to my 500-C. Sigh.
 
Great write up Dave. As always I understood about 30% of the terminology but that thing sure is a beauty. High praise from "the master". I'll bet with those stats, the Cornwalls are just trembling!
 
I always learn something (and usually a lot) reading Dave's posts, excellent information - and presentation.......
 
Late to the party, but that's a fantastic result! I kinda expected EFB to be part of the mix. I appreciate the detailed history lesson at the beginning of the thread. But I don't know who ran marketing at Fisher, or why it was important to mention it????
 
Mr. Avery Fisher was the acknowledged marketing guru at Fisher Radio Corporation, having a very heavy hand in the look and advertising of the various products produced, and also made sure all the milestones of Fisher Radio were marketed and capitalized on to bolster the image of the company beyond the high water that the quality of the products alone achieved.

The importance of the history lesson is that the design is so clearly a square peg into a round hole effort (very un-Fisher like in that regard), because there were other far easier ways to achieve the performance results they did for the power class chosen. That Fisher was so obviously determined to accomplish it with EL34s however -- and accomplish it the way they did with EL34s -- strongly suggests that the design was specifically driven by the use of that particular tube (the marketing angle), and that they did not want to follow the crowd and have to pay any homage or royalties (there's the bean counters) for any technology that others could lay claim to. The SA-300 is the result of such thinking. That Fisher achieved the results they did in spite of it all is to their eternal credit, but man-oh-man! Talk about going around the world backwards to get across the street!

Dave
 
Last edited:
3. Push-pull partial cathode coupling of the output stage.
Re-reading this post where this "partial cathode coupling" was discussed. Wow. Just wow. Petty sure I would never have thought of applying push-pull local loop feedback from secondary of OPT to cathodes of output stage in this way.

But now where do I connect the speaker leads with the 4 ohm secondary lead grounded? Between 4 and 16 ohm outputs, which effectively gives me a 4 ohm output?
 
The speakers remain connected as they always were -- between the Common, and whatever impedance tap they should connect to. Yes, in uniquely the case of 4Ω speakers, they could be connected to either the Common and 4Ω tap, or the 4Ω and 16Ω taps -- this because the 4Ω tap is the CT of the winding -- but this was always the case even with the original design. The speakers -- of any impedance then -- still connect as they always did. It's just that the Common terminal no longer operates at ground level. The 4Ω terminal does so that an equal NFB signal is applied to both sides of the push-pull connection. Test equipment then -- where multiple ground paths are commonly in place -- now need to be connected between the 4Ω and 16Ω connection, so as to eliminate any shorts or ground loops. Of course, any voltage readings or wave form levels observed with this type of testing scenario should be treated as 4Ω signal values.

Partial cathode coupling of the push-pull connection has received only light press over the years, because even with though it has great potential (reducing distortion by the square of the FB factor), a great many applications can't make much use of it due to the topology of the output stage employed. That is, lower Gm tubes like that of the 6L6 or 6V6 family, or output stages operating with a UL connection, are effectively hamstrung relative to this type of connection since the relative gain of such stages and tubes is reduced which then minimizes its effectiveness. As a result, more often, when cathode coupling was used, it was typically executed by way of a tertiary winding that could supply plenty of NFB in spite of the output tubes or output stage topology used. Think Fisher 70 through 100 amplifiers, which all used this approach (the 70 had a UL and cathode tertiary winding, while the 80 and 100 amplifiers only had the tertiary winding). Because these amplifiers all used the 6L6 family of tubes however, they had to employ a separate tertiary winding to achieve any effectiveness from the connection.

The downside of the high FB level tertiary approach -- and particularly when used with lower Gm tubes, is that is asks for tremendous output from the driver stage to overcome the NFB generated by the cathode coupled connection. Therefore, unless designed carefully, advances in distortion reduction made in the output stage can be lost in the driver stage. And there's another thing to consider. While the connection does act to reduce distortion, it also acts to reduce power output if taken too far, since the NFB voltage applied to the cathode also acts to effectively reduce the screen grid voltage as well -- and particularly so if a UL winding is used in conjunction with a tertiary cathode winding.

Of course, this all gave rise to the famous McIntosh Unity Coupled output stage, where a cathode coupled tertiary winding is employed to maximum benefit (representing 50% of the primary winding), while cross coupling of the plate and screen connections effectively holds the screen grid voltage constant, so that maximum power is still developed, and at minimum distortion as well thanks to the cathode coupled connection. But -- it also explains why all the shenanigans are required of the driver circuit to drive the McIntosh output stage to maximum power output (requiring hundreds of volts p-p). To their everlasting credit however, Frank and company developed a wonderful R/C coupled driver system for their output stages, and the rest as they say, is history.

Now back to the way I employed it in the modified SA-300, the output stage in this unit is a perfect candidate for the partial cathode coupled connection used, as the output stage employs high Gm tubes operated in pentode mode, meaning that output stage gain is relatively high compared to other typical tube and output stage scenarios. This means that maximum distortion reduction will be achieved for the amount of NFB available, and because that FB is limited by that which can be supplied by the output winding, it's hardly enough to impact power output -- and particularly so because the EFB™ screen grid voltage was adjusted to account for the partial cathode coupled connection. Also keep in mind that because both the EFB Screen and Control Grid Regulators act to maintain ideal output stage operating conditions under all manner of dynamic conditions, in this case then, they are also seeking to maintain the effectiveness of the partial cathode coupled NFB connection as well!

The partial cathode coupled output stage connection as executed in the modified SA-300 -- and even with the relatively limited FB it provides in this application -- still effectively cuts the post EFB distortion in half, or basically down to just 0.5% mid-band at full power output before any other NFB is applied. And importantly, it is being applied directly to the stage generating the most distortion within the amplifier. With the driver stage modified to then be able to drive the modified output stage with plenty of low distortion drive, the relaxed global loop can then easily bring that distortion level down to the (typically) 0.05% THD measured at full power. Of course, the original goal was to in fact relax the global loop, so as to be able to eliminate the partially direct coupled networks and the collateral damage they bring to the party, and all without causing any increase in distortion.

In the end then, the amount of NFB applied to the modified SA-300 is exactly the same as that used in the original design, but being spread over two separate loops, it is now had with greater stability, and without the side effects of the original design, with EFB then providing a X12 reduction in overall distortion. The real beauty is that now, all of these systems are working together in harmony to maximum benefit of each other, with significant proof available in both the measurable and audible pudding. This amplifier so modified, has now risen to a seriously high level on my enjoyment scale!

I hope this helps!

Dave
 
Last edited:
I meant to add also that likely the most commercially successful use of partially cathode coupled push-pull NFB can be found in most ARC (Audio Research Corporation) amplifiers. In his designs, Bill Johnson implemented it using 6550 tubes. To get the most out of the modification, he also operated the tubes in pentode mode, and because of the capability of these tubes, he could operate them at a reduced screen grid voltage (compared to what the EL34s requires for a similar power output at typical plate voltages). Operated this way, the tubes would still easily develop 50 watts RMS of usable power output (and more with higher plate voltage), but ended of with the effective gain of an EL34 output stage, so that in his design as well, the connection effectively cut output stage distortion in half, before the application of any outside loops.

Let me again also stress that it takes a well designed output transformer for the effective implementation of this type of NFB.

Dave
 
Back
Top Bottom