Well, I'm officially done.
Some of you are aware of the trouble some folks have encountered, trying to get Heathkit AA-100 integrated amps to sound good.
It's a real puzzle, on the face of it. Good stuff in there- Mullard 5AR4/GZ34, 12AX7s and EF86s, Amperex/Mullard 7199s and Westinghouse 7591 PP outputs. One would think the amp should sound good, right?
Well... not on the face of it. At least, not sometimes. In fact, these amps have inspired a whole web-page, from a poor soul who has gone around-and-around with his... doing all sorts of stuff, to "un-constrict" the sound. Distorted at anything over medium-low levels, with questionable sonic balance too (midrangy)...
My amp fell into this category, too... compounded by the fact that obviously, something had caused some really serious trauma to one or more of the output tubes (the unmistakable signs of burnt paint on the top cover!). So, some rather serious consideration and detective work was definitely in order.
First off, a COMPLETE recap. I mean, EVERYTHING, other than a few caps in the tone controls (which never see much voltage) and one of the two can caps. Note that I HIGHLY recommend replacing the first can cap (the 60/25 @ 500v)... mine actually was ARCING internally (you could HEAR it popping!). I used a 50/50 J&J cap... seems to work just fine. Alternatively, you could use a 40/20/20 wired as 60/20 (or a 40/20/20/20 wired as 60/40) as well...
As for coupling caps- stuck with basics- Illinois poly caps from AES. Bumped everything that was 400v rated up to 630v- it's good insurance!
After reading the aforementioned web treatise, I came to the conclusion that the first agenda was output bias... or the lack of control, thereof, in the amp in stock trim. Most of that could be laid fault at the use of 470K bias-feed resistors to the output tube grids- where 300K is the maximum rated value for the 7591! So, that was immediately fixed- 220K resistors installed, check.
Next, was going around looking for any other resistors either out-of-spec or spec-out-of reality. Found that the anode resistors on the first stage of the 7199 output driver tubes were WAY out of spec- should have been 220K- one of the 220K resistors read about 800K, the other FOUR MEGAOHMS. Needless to say, those got swapped out, posthaste. Now, at least we had STABLE output, the same from both channels.
Did correct one other minor problem- some mis-matched 7199 tubes. One was exhibiting significantly higher transconductance than the other- which lead to drastically different plate voltages under load. This is a problem... due to this stage being DIRECTLY COUPLED to the grid of the phase inverter (the second side/triode of the 7199)... this leads to the phase inverter halves (the top and bottom of the triode) being biased wrong- which loses quite a bit of headroom! Juggled a couple of 7199s around- found two that read within 5% of each other on plate voltage, and close to the stock plate voltage value too... problem solved.
Well, now the amp is sounding fine... now, onto why the preamp seems to clip so much. Did some investigation- Hmmm. Plate resistors of only 100K on 12AX7s? That's fine for guitar amps... not much good for hi-fi. Also, these were set up for very little bias voltage- so, not much headroom either.
Started thinking- what's like a 12AX7, but would like a 100K plate resistor? Answer: 12AT7. Popped a pair of 6201s in there (military 12AT7s)... PROBLEM SOLVED. Without question. Twice the bias voltage- 6 dB more headroom. And only lost 1/2 dB of gain, to boot!
Now, we're getting somewhere. It sounds like a REAL amp now! But, I could still hear a little distortion on heavy bass. Time to pay more attention to balancing plate currents on the output tubes... they were a bit out of whack, with around 35ma on some tubes, while others were having a hard time hitting 25ma...
NOTE: The following includes graphic accounts of amp surgery- those with weak constitutions might want to look away...
Basically, I converted the thing over to INDIVIDUAL bias controls for each output tube. To do this, first I installed bias sense resistors on each cathode- 10 ohm 1/4 watt resistors. To do this, I mounted a little 7-pin terminal strip between the two output transformers, and mounted the four resistors on it. Then, I CUT the PC board traces to each output tube cathode, and ran individual leads to the resistors. Then, I ran a common ground lead from all four resistors, back to the power supply ground (where the cathodes grounded to before).
Next, the individual bias pots. I used 10K, 15-turn pots from Radio Shack... easy to find, worked fine. Mounted them on another 7-pin terminal board... the input and output leads to the terminal board, and the "grounds" (the leads going to the resistor to ground, that is) connected in common with an "overhead" bus wire. Here's the circuit diagram I used:
Again, I cut the PC board traces, right at the 220K resistors for each tube, and ran wires to the pots, as shown in the diagram.
Once this was done, I set all the tubes for 30ma bias. The 15-turn pots make dialing it in EASY... you can get it within a tenth-of-a-milliamp or so, pretty quickly...
WOW... what a difference. It's CLEAN. It's balanced between channels! And it sounds REALLY dynamic and neutral. In fact... dare I say it... it actually sounds like a MODERN hi-end tube amp.
Yeah, it's a bit of work... but it's the difference between an also-ran, and something pretty DAMN IMPRESSIVE, IME and IMHO...
Oh, BTW: One optional mod- I installed a 100 ohm 50 watt resistor between the output of the 5AR4 and the first can cap. This may or may not be necessary in your amp... it helped me get the voltage down to around 460V... close to stock. But, if you bias your outputs harder, you may want a smaller value (say 50 ohms) or none at all...
In all... I'm pretty darn happy now. Now, to just get a few replacement knobs, and this thing will be GOOD TO GO... maybe for another 45 years!
PS: To boot, the accompanying AJ-41 tuner is up and running too! Man, this tuner sounds GOOD! I think we may have a new contender to challenge the Eico ST97 here...
Regards,
Gordon.