Building Fisher SA-100 Clone

Thorne,
Well, what you're asking about really boils down to how tubes operate. I'll PM you with some recommendations about how to get your questions answered.
Dave
 
POWER UP AND INITIAL OPERATION

Well, power up this morning went very well indeed! I found a couple of things to correct (missed a B+ connection and my input muting switch was wired incorrectly), but after than she came up like a dream on the Variac.

The amp responded very well to the controls and after roughing in the screen and EFB range voltages, I biased and balanced the new output tubes (Sovtek EL84M's--I'm a big fan of these; matched set from Mr. McShane) at 25 mA cathode current per tube and tuned the screen voltage per the formula in Dave's schematic. With the tubes drawing the recommended quiescent cathode current, I measured the voltages all around. In general, my high-end B+ voltages are about 20-25 V lower than Dave's amp, likely because I'm using a 700 VCT power transformer (201 mA) instead of the Dave's 720VCT 250mA X101ST P.T. The rectifier is a Genelex Gold Lion U77/GZ34 (a McShane recommendation), a Westinghouse (Japan) 6FQ7, and a garden-variety GE 12AX7 (for now).

With the line voltage set at 120 VAC, here's what I got (with Dave G's values in parentheses):

B+ VOLTAGES: B+ "A": 422VDC (447); B+ "B" 322VDC (339); B+ "315": 303VDC (315), B+ "225" supply: 213VDC (225)

BIAS SYSTEM: 6.3V C.T. point: 41VDC (43); "C": 2.3VDC (~2.1); "D": 13.3VDC (~15)

FILAMENTS: At P.T. Windings: 6.34 VAC (6.3); At 6FQ7: 6.28 VAC (6.3); At 12AX7 (Parallel): 6.3VAC (6.3); 5AR4: 5.15VAC (5), no tube. The 12AX7 and 6FQ7 are at the end of their respective heater strings.

EL84M VOLTAGES: Plate Voltages: 422VDC (445); Screen Voltages: 322VDC (339); Cathodes: 13.55 to 13.6 VDC (~15.25); Grids: 1.45-1.6 VDC (~1.4)

With these cathode and grid voltages, bias was right at -12 to -12.1 VDC (-13.85) across the board. In my experience, the Sovtek's are a bit cold-blooded and have to be biased pretty hot (more positive) to get the desired cathode current and this was the case here, accounting for some of the voltage differences with Dave's original set up. The fact that the EFB range control and the bias controls together can handle this end of the bias voltage range (with some room to spare) is good news indeed!

At 25 mA per tube quiescent, allowing for 3 mA for the screen current, we are running the output tubes at about 9.3W quiescent plate dissipation, about 66% of the 14W max rating for the EL84M's. Some of the magic of EFB (tm)!

The plate voltages for the 12AX7 were about 97VDC (95), the cathodes were 0.56VDC (0.65). For the 6FQ7, the plates were both at 200V (215V), the grids were at 97VDC (as expected, being direct coupled to the plates of the 12AX7), and cathodes at 100VDC (100).

I played in the amplifier for over two hours and it ran very cool. The P.T. was just warm.

OK, what about sound? Realizing that my bench speakers are expendable low-budget units, the sound was excellent with levels of detail that I've not been hearing through those speakers lately. Can't wait to get it on the 'real speakers' for a listen! With no inputs hooked up, there is a pretty good hum, but this circuit has no switching to automatically ground open inputs, but that's where my muting switch comes in handy! Dead quiet when muted and with an input plugged in and no music playing.

I put the bottom plate on it (without holes, but there seems to be very little heat build up in the chassis now, but I'll keep an eye on it) and took a couple of pics with the tubes installed and glowing away. I'm going to listen for a bit before putting it back on the bench for performance measuring in the next couple days. Very happy with the build so far!

IMG_5453.JPG

IMG_5458.JPG

IMG_5459.JPG
 
Very nice, Dave. Looks great, must sound great, runs cool, I bet you're feeling mighty handy right about now. When can we place our orders? (grin) Thorne
 
Great job, Dave! Your build stands proudly with my original! In use near daily, with the same original tubes installed as seen here, and still producing about 96% of their original new power output levels. This pic is also the pic in my identification info.

Dave

SAM_1319.jpg
 
FINISHING TOUCHES AND PERFORMANCE MEASUREMENTS

Thanks very much for the kind words, guys!

I finished measuring the new SA-100 "+" and moved it to the listening room after adding a couple of final finishing touches. I decided to go with a cover for the Hammond chassis with good open mesh for ventilation (and you can still see the tubes glowing away). I kind of always liked the look of the covers on vintage Fisher, Dynaco, and Scott power amps, so I thought I'd give it a try on this one. Having a new grandbaby around to keep out of the tubes (just about ready to crawl) is one thought as well (no cats, though!). The pilot lamp was a little too bright, so I dimmed it down with a 30 ohm 2W resistor in line with the bulb. I kind of like the amber, as opposed to the usual red or blue.

As an indulgence, I had a bronze tag made up with the Bird on it and designated the new amplifier "SA-100-DCG" to give due credit to the father of the design.

Sound check into my Klipsch KG4's is fantastic, with the kind of low end punch, outstanding mids, and airy, detailed highs Dave described in his original build. I can see why this amp would be a daily driver; its really easy to listen to, to say the least.

I'll get to the measurement stuff in the next post, but first here are a couple shots of the finished product taking it's place in the stack.

IMG_5472.JPG

IMG_5490.jpg



IMG_5475.jpg
 
SA-100-DCG MEASURED PERFORMANCE

First, a note on the B+ voltages I measured in a post above. I noted how they were running a bit lower than Dave's original build and described the 20 VAC difference (lower in mine) in the HV output of the PT as a difference. Also worth noting is that the Hammond 193L 5H 300 mA choke I used has a higher DC resistance than the 2.6 H 193K in Dave's build: 57 ohms vs. 26 ohms.

I measured the voltage drop across the 30 ohm 10W resistor just off the rectifier before the choke with no signal and it was about 4 volts (didn't write it down!), indicating a quiescent current flow of 130-140 mA, which seems about right for four EL84M tubes biased at 25 mA cathode current per tube, a few mA for the 12AX7 and 6FQ7 plates, and the rest of the circuit current draw for the regulators, etc. The Hammond 273BX PT is rated at 201 mA at 700 VAC CT and indeed, after three hours operation at pretty fair volume, was only mildly warm.

Frequency response and distortion first, then the square waves:

MEASURED PERFORMANCE SUMMARY--In short, outstanding frequency response and very low distortion indeed; rated power as expected:

Output Power (measured at first sign of clipping): 21.1 WRMS per channel, 1 kHz, single channel driven. 18.3 WRMS per channel, both channels driven. Full power was maintained 20-20 kHz except at 20 Hz, where output was 18W RMS for each channel, single channel driven. Clipping was gradual and perfectly symmetrical; recovery was quick.

Signal input voltage to obtain full output power was a little above 1.3 VRMS.

I set the AC balance controls for both sides for lowest 2nd harmonic distortion (corresponding also to lowest overall distortion) via FFT at 3/4 rated output power before measuring distortion and frequency response levels.

FREQUENCY RESPONSE dead flat from 20-40 kHz at 1W, and the same at 1 dB below max power, except for very slight (0.5 dB) roll off right at 40 kHz.

THD+N vs Frequency: Extraordinarily low at 1W: <0.2% 20-20kHz (<0.1% across much of the spectrum); slight rise above 20 kHz R channel compared to very little L channel; At 1dB below max power, THD+N <0.5% for most of audible spectrum, rising at very low and high frequencies, L channel somewhat lower than R in this regard.

THD+N vs Power @ 1 kHz: Below 0.2% up to max power; L channel very slightly lower across the bandwidth compared to R channel.

NOTE on Channel differences: Recall that I installed the step network values that Dave evaluated as best for his pair of AA-100 output transformers. I just duplicated these for my build up to this point. With respect to distortion, both channels perform admirably compared to pretty much everything I've measured lately, but the L channel is just a bit better than the right.

QUESTION: would changing the step network values for the R channel to be more like the left even out these slight differences in performance?

Editorial comment on performance:
I just rebuilt a pair of McIntosh MC-30 monoblocks for a guy and this little power amplifier compares very favorably to the legendary Mac's, even besting them on flat frequency response to 40 kHz and 1W THD+N vs. frequency levels. At 1 dB below max power, though, the Mac's with their legendary transformers had significantly lower distortion at the low and high ends (very low!), but not so much in the middle of the spectrum. Remarkable!

Performance plots in next post, titles self-explanatory.
Dave
 
Last edited:
Frequency response and Total Harmonic Distortion + noise scans, performed on HP 8903B audio analyzer:

Dave
 

Attachments

  • SA-100C Freq Resp 1W L copy.jpg
    SA-100C Freq Resp 1W L copy.jpg
    59.8 KB · Views: 49
  • SA-100C Freq Resp 1W R copy.jpg
    SA-100C Freq Resp 1W R copy.jpg
    60.2 KB · Views: 44
  • SA-100C Freq Resp 16.8W L copy.jpg
    SA-100C Freq Resp 16.8W L copy.jpg
    61.2 KB · Views: 47
  • SA-100C Freq Resp 16.8W R copy.jpg
    SA-100C Freq Resp 16.8W R copy.jpg
    60.4 KB · Views: 44
  • SA-100C THD+N v Freq 1W L copy.jpg
    SA-100C THD+N v Freq 1W L copy.jpg
    62.2 KB · Views: 46
  • SA-100C THD+N v Freq 1W R.jpg
    SA-100C THD+N v Freq 1W R.jpg
    61 KB · Views: 46
  • SA-100C THD+N v Freq 16.8W L copy.jpg
    SA-100C THD+N v Freq 16.8W L copy.jpg
    62 KB · Views: 46
  • SA-100C THD+N v Freq 16.8W R.jpg
    SA-100C THD+N v Freq 16.8W R.jpg
    61.6 KB · Views: 46
  • SA-100C THD+N v Power L.jpg
    SA-100C THD+N v Power L.jpg
    54.3 KB · Views: 47
  • SA-100C THD+N v Power R.jpg
    SA-100C THD+N v Power R.jpg
    53.8 KB · Views: 44
SQUARE WAVE TESTS--SA-100-DCG

Here are 10 kHz square waves for L and R channels. The shots show good rise time with the expected/desired slight initial overshoot, with what looks to me like good damping and generally quite good form for both. Note that the left channel rise is slightly less than the right channel--see above comments about the different step networks in the channels. Could this also be descriptive of the slightly better performance on distortion tests of the L channel, or just differences in the output transformers? Opinions and comments on all of the above, but particularly the square waves most welcome!

Alas, I forgot to run the 0.25 uF stability test this run, but I would expect the same level of stability that Dave got in his original build. Work for next time it goes on the bench!

Thanks
Dave

10 kHz square wave, LEFT Channel:

IMG_5469.JPG

10 kHz square wave, RIGHT Channel:

IMG_5470.JPG
 
Super job Dave!! And thanks for sharing your measured data.

The supersonic differences you note between the two channels is no doubt due to slight differences between the two output transformers -- a much more common situation than not. Your right channel 10 kHz square wave looks ideal. You might try increasing the value of the cap in the step network for the left channel slightly, and using a pot for the resistor to adjust the waveform for best appearance. It may raise the 20 kHz THD figure slightly -- but it's better to have the smoother square wave appearance, as typically, not even a full watt of power is developed at 20 kHz -- even when playing aggressive music material at maximum power output -- so the slight change in full power 20 kHz THD is meaningless.

Again, a very very nice job. I have no doubt that you'll definitely enjoy the fruits of your labor!

Dave
 
Thanks much, Dave. I'll do exactly that when I have it on the bench next and will report on the adjustment of the L channel step network. I'm already enjoying the amp very much; it's opening up even more as things play in.

I appreciated a kward thread that featured you in conversation with him on interpreting square wave quality and tuning step networks in amplifiers from 2013 and would offer it again to anyone with interest in this science/art:

http://audiokarma.org/forums/index.php?threads/how-i-tuned-my-push-pull-feedback-amp.539036/

Also, I'm thinking of building a pre-amp to go with it (I'm using a 'passive pre' input selector/level control right now to good effect). If anyone has recommendations on a favorite circuit I'd love to hear them!

Dave
 
Thanks, Tim. Lotta work, but nice when it's done and you get it working well, then you can kick back and just listen!
 
Thanks! The sound is excellent; the bass is tight and deep, mids outstanding with great clarity and detail on the highs. I think it matches Dave's description of his original build: kind of an EL84 amp on steroids (paraphrase). Musical, but with the kind of kick you get from a 7591 amp (I have several, with my X-200 being a great example). The amp sounds particularly good through my Klipsch KG4's with a very nice sound stage. I'm not an 'audio poet' in describing these matters, so that's about the best I've got. Let's just say I'm enjoying listening to it a lot as it plays in without any hint of fatigue.

Ultimately, I'll be comparing it to my Scott 299B (a tough test, as the 299B is outstanding to my ears), but it's early for A/B comparison in the process.

Dave
 
Couple things: First, the amp just sounds better and better. I am really, really happy with the way the build turned out.

Second, about the pre-amp I mentioned, above: the 'itch to build' has been scratched pretty well building the SA-100-DCG, but I have been looking at many circuits and approaches to building a nice pre-amp for the new amp. After much study, I had pretty much decided to build a version of the Scott LC-21 stereo pre-amp circuit, being very fond of H H Scott gear and their approach. I know it may be heresy, but I really want tone controls with my pre-amp. I use them on all my integrated amps to compensate for recording and amp/speaker combo variations and Scott's tone control circuits seem to me to work very well.

I had actually been looking for an LC-21 or a Scott 130 to buy, but nothing had appeared in quite a while and they come up only rarely. After actually ordering a cabinet for the build, I found a dark-face LC-21 that had just come up on eBay and bought it. Looks to be in pretty fair shape, lacking a couple of knobs and a little interesting work evident under the chassis. Supposed to have a bad power switch as well. In any case, I'll give it a good going over and see how it mates up with the Fisher clone.
 
I got the LC-21 in and it is indeed an interesting case. Will start a thread on it in the Tube Audio forum and then report on how it matches with the new amp!
Dave
 
I decided to build on a Hammond black powder-coated steel chassis and have been busy planning, measuring, and drilling and cutting (metal bits everywhere!). The parts have been attached now and wiring has commenced.
…..

I worried a little about eddy currents in the steel chassis (Dave used aluminum, I believe), but I wanted the stiffness of a steel chassis. I did not raise the PT off the chassis as a preventive, but if this turns out to be an issue I'll use phenolic to offset the PT from the chassis.

Dave

Hi Dave,

what size Hammond chassis did you use? You may have mentioned it somewhere but I can’t find it. Did you have any issues with the steel chassis, other than the mentioned weight? If you were to build the clone again would you try for a smaller chassis?
 
Back
Top Bottom