Troubleshooting help for 12AQ5 amp

[Carlp]

I just recently re-built my El Cheapo (a PP 12AQ5 amp designed by Eli Duttman over on Audiophiletalk.com) because I had been having problems with my earlier build. Circuit is in the attached image below.

Note I've used a "booster iron to bring up the B+ by 12v above this schematic (per recommendations in the long thread) so the PS looks slightly different. After wiring I checked B+, B- and heater voltages, plugged in tubes, set the CCS in the tails of the AT7s to 0.7v (per the design spec, giving 3.5mA per side of the 12AT7 splitter/driver) and gave it a test run. Not bad sound quality but sound level was definitely not close to the 8 wpc it should be. My 1.8 wpc SE pentode amp (another Eli D amp) puts out more SPL through an otherwise identical setup.

Back to the bench to measure voltages. Here's what I have:

B+ = 344
B2+ (for 12AT7s)
R = 344
L = 345
B- = -153

These PS levels are all ballpark good.

B2+ at 12AT7 Anodes
R (inverting) = 174
R (non-inv) = 169
L (inverting) = 106 (too low)
L (non-inv) = 78 (too low)

B- at end of CCS
R = -66
L = -99 (off)
B- at cathodes
R = 2.25
L = 0.35 (too low

The outputs all appear to be fine, per the voltages below
B+ at 12AQ5 anodes
R (inv) = 341
R (non-Inv) = 339
L (inv) = 342
L (non-inv) = 340

12AQ5 Cathodes
R = 23
L = 24

B+ and B2+ were as expected. B- was about what I'd expect too based on voltages posted by other builders. Overall, the R side seems to be close to what I'd expect, though I thought I'd see closer to 200v on the AT7 anodes. The 50K load resistors run 3.4mA or so, half the CCS current per side of the AT7s. That's OK.

L side has problems. AT7 anodes are low (burning too much B+, one side running at 4.8mA, the other at 5.3mA or so). Oddly, while the CCS output voltage is too low (-99v instead of -66v), current seems to still be OK at 7mA. 7mA through the CCS into the cathodes is good, but the two halves of the AT7 are running at 4.8mA and 5.3mA through the anodes. That's a total of 10.1mA for the tube, even though the CCS shows 7mA into the cathodes. Is that possible? I rechecked the L side CCS and it's still at 0.7v (7mA). Could there be a leak to ground somewhere? Could grid current account for those readings?

I appreciate any thoughts on how to troubleshoot this. Oh, and I've checked the heater voltages, and they appear good, perhaps a little high.

[Carlp]

Just FYI, tonight I tried removing the CCS and using a resistor to bias the AT7 cathodes for testing purposes. I added 220 ohm 1w resistor in line with the existing cathode resistor (with a bypass cap) hoping to bias at about 3.5mA (it was MUCH easier just to add the resistor, but I forgot that the existing resistor is 10 K ohm, not 10 ohm). As a result, I biased with 10.22K ohm instead of 220 +/- for the dual cathodes and got almost 9v on the cathodes. That's only about 1mA I think.

Still, both channels were even at 9v (dual cathodes were tied together, so same on each) and the B2+ at the anodes was the same both sides, suggesting maybe a CCS problem? Ideas?

[wrenchone]

Since you have the CCS out of the circuit, why not bias it up with 20V or so and see if it really is a CCS?

[Carlp]

Thanks for the suggestion. I've wondered if there is a way to test this CCS (in or out of circuit). Can you give me more details on how to do that?

Right now I know the CCS holds current steady in circuit. But it seems to burn too little negative voltage on the L side (dropping from -150 to -100 instead of -150 to -65. What I don't understand is whether this is due to a failure in the CCS or some other factor.

The 100k CCS resistor drops .7v for both the L and R channels, meaning it is (and presumably the entire CCS are) passing 7mA current. That suggests there's less resistance in the L side CCS or the mosfets are doing something wrong. Unless something else could be going on with the pathway to ground?

The B- leaves the CCS, goes through a single 10k resistor, then enters the cathodes of the AT7. I've confirmed it is a 10k resistor, but the voltage drop problem occurs before that resistor anyway.

Seems I'm convincing myself it's probably a CCS problem and if I can confirm the problem I'll probably rebuild it soon. Any more ideas though?

Thanks.

[wrenchone]

If you have a bench supply (you should, if you're serious about working with electronics), use it to place ~20V across the CCS with a DVM in series to measure current. You should be able to use the pot to adjust current to that specified for the amp. One adjusted, the current should stay constant if you adjust the supply voltage upwards. If not, it's not a CCS and one or both of the fets are zorched. Eli can comment on this, but I think one DN2540 fet is sufficient for a relatively undemanding application like a LTP tail current source - cascoding the CCS is really gilding the lily.

[Carlp]

Perfect, thanks Wrenchone. I can't speak to why the CCS is in the LTP but you've helped me think through how to approach the next step. From what I've read in the build thread, the DN2540 is a bit finicky - fragile, actually. So I wouldn't doubt it's fried, but I'd rather do a test before assuming so.