[gitarretyp]

First, the background: I recently picked-up a used Antique Sound Lab AQ-1002 tube amp on ebay. It's rated at 50W x 2 @ 1% distortion with two kt88s per channel. My speakers, custom design by selah, are rated at ~85-86 db sensitivity and my current amp, ps audio hca-2 tested at 170W x 2 @ 1%, will easily drive them to levels beyond which i care to listen (>~100 db).

Now the problem: The tube amp noticeably distorts with levels that are peaking ~90-92 db. The clipping occurs in both channels with all types of music from rock to solo female vocals at roughly the same volume. From my calculations, this should really only be using ~4W or so. If it were truly outputting 50W at clipping, my 170W amp would be clipping at ~95-97db, and that simply isn't true (i would expect clipping with peaks approaching 105-110db with the ps audio, which seems roughly accurate, and ~100-105db with the asl). I've tried adjusting the bias between ~35 - 60mA with little effect on the headroom available. I've also tried different output tubes with no appreciable volume change noticed. Any thoughts on what might be the problem? Could the amp be drastically over speced? I haven't checked the internal voltages yet, but if i do, what would be the most likely culprit for decreased output? Am i simply expecting too much from the amp? The schematic is available here, if that helps

[JoshK]

Yeah something is fishy in bangkok.

If your speakers at 85db/w then you would need only 32 watts to hit 100db (64w if they are 4 ohms).  That would give you enough headroom to play at ~90db with most music.   So obviously you are not getting close to 50watts. 

I took a quick look at the schematic, and I am not going to go through the excercise to calculate the theoretical output based on the loads and biases.  However, my guess is that the second 12au7 tube is the culprit.  It likely doesn't have the drive current to drive the KT88 in UL mode to clipping and its your driver stage that is puking. 

The easy test of this would be to get your hands on a pair of 12BH7's.  Don't go spending lots of money, get a cheap pair from ebay, borrow from a friend, your own stash or alternatively EH makes new production ones that should be inexpensive.  I personally would get two pair and replace both 12AU7's, but for this test only replace the driver of each channel.

If the distortion goes down noticeably then I think we have weakly proven my guess.  Then you could try the 12bh7's in both positions and reassess. 

[woodsyi]

Shouldn't the speaker's (reactive?) impedance be a factor?

[JoshK]

Yes it should... but he seems to be off by so much, I can't imagine it is just that.

[gitarretyp]

Shouldn't the speaker's (reactive?) impedance be a factor?

Yes, of course...here's the measured impedence. Nominal ~8ohms, minimum ~5.7. I've been running off of the 8ohm tap.

[woodsyi]

That looks benign.  My Starlet integrated uses 6550 and 12au7 and gets 48watts.  12au7 may not be the cultprit.  PM Bill Baker and see if he knows more about this since he carries the brand if not the model.   

[Bill Baker]

Hello gitarretyp,

 I am not familiar with this particular model so do not know what the bias settings are for the amp. What you are experiencing is usually an indication of under biasing the tubes. There are obviously other options to consider. If you can answer a few questions, we can probably figure this out.

 How does the amp perform at "normal" listening levels?
 What bias voltage, in mV, are you reading at the cathode of the output tubes?

[gitarretyp]

Hello gitarretyp,

 I am not familiar with this particular model so do not know what the bias settings are for the amp. What you are experiencing is usually an indication of under biasing the tubes. There are obviously other options to consider. If you can answer a few questions, we can probably figure this out.

 How does the amp perform at "normal" listening levels?
 What bias voltage, in mV, are you reading at the cathode of the output tubes?

At normal (undistorted) volumes, it sounds quite nice. Warmer and rounder than my ps audio, wider soundstage, but slightly bloated bass and dynamics that aren't to the level of the ps audio.

The suggested bias point is 550mV measured across the 10 ohm cathode resistor. I've tried running them between 380mV and 600mV with little change in headroom.

[JoshK]

My rationale for the 12au7 comment was that it is known to have pretty high distortion, the 12bh7 which is a drop in replacement has less than half the distortion (at normal operating points).  So if the amp is grossly distorting because of the driver stage (my hypothesis) then this should yield an audible difference and help determine if that is the case.

Another thing to note is that triode connecting the output tube is quite a bit different load than pentode connection.  The triode connection has much higher miller capacitance thus requiring the driver stage to have much more current drive.  The 12a*7 family is not known for having good current drive abilities.  His amp is UL which is part way between triode and pentode and having part way between the drive requirements. 

As Bill was saying, one would need to reverse-engineer the operating bias and loads (I didn't see a plate load for the OPT in that schematic) to really hazard a more sophisticated guess. Bill has a lot more experience than I on these matters, my knowledge is a bit of book knowledge.

[gitarretyp]

My rationale for the 12au7 comment was that it is known to have pretty high distortion, the 12bh7 which is a drop in replacement has less than half the distortion (at normal operating points).  So if the amp is grossly distorting because of the driver stage (my hypothesis) then this should yield an audible difference and help determine if that is the case.
...

The dirver stage distorting makes some sense to me, but the only other preamp tubes i have are 12ax7 (10 times the gain of 12au7, if i remember correctly). I could try a 12ax7, but i'm not sure the effect would be desirable.

[JoshK]

12ax7 wouldn't work in this case.  I'd try to hunt down some dirt cheap but working 12bh7's (just a pair).

I don't have a pair or I'd loan them to you.  You can see if tubesandmore.com has a used pair, which are usually a lot cheaper.   If it doesn't help, flip them on ebay for little or no loss.

[gitarretyp]

12ax7 wouldn't work in this case.  I'd try to hunt down some dirt cheap but working 12bh7's (just a pair).

I don't have a pair or I'd loan them to you.  You can see if tubesandmore.com has a used pair, which are usually a lot cheaper.   If it doesn't help, flip them on ebay for little or no loss.

I just realized that i have a single 12at7 in the reverb stage of one of my guitar amps. Would it be suitable to try in one channel?

[JoshK]

I wouldn't.  The 12at7 requires different operating points iirc, and has higher gain.  It is more linear however than the 12au7.

[FredT300B]

This is a long shot, but are you sure you're reading your voltmeter correctly? My first tube amp, a PP Assemblage EL34, called for a 40mV bias setting. I set the bias so my digital auto-ranging meter read 40, and the amp clipped at relatively low volume levels. Then I discovered I was setting the bias to 4mV instead of 40. Turning the bias pots up until the meter read 400 solved the problem. I had been misreading the range indication on the meter. 

[gitarretyp]

This is a long shot, but are you sure you're reading your voltmeter correctly? My first tube amp, a PP Assemblage EL34, called for a 40mV bias setting. I set the bias so my digital auto-ranging meter read 40, and the amp clipped at relatively low volume levels. Then I discovered I was setting the bias to 4mV instead of 40. Turning the bias pots up until the meter read 400 solved the problem. I had been misreading the range indication on the meter. 

Interesting suggestion, but i am reading the meter correctly .

[JoshK]

So the two suspects are bias and driver.  Even if 50w is highly optimistic from kt88 PP (which shouldn't be the case in AB1), I think there is something not right here. 

Do you know the plate to plate load of the OPT?  You can measure the resistance of the primary (far leg to leg and then center to leg) to determine the load. 

[gitarretyp]

While we're discussing the input tubes, could old or failing preamp tubes cause the problem? The seller wasn't sure how many hours were on the 5814a's. He thought maybe upwards of 1000 or so.

Also, what would happen if the input 12au7s were biased incorrectly? They have pots on the cathode that may be incorrectly adjusted.

By the way, thanks for everyone's help so far.

[JoshK]

Bill will have to chime in on how it would react if the balance were incorrect.  That is what the first pot on the first gain stage does.  It is to set the balance between the two legs of the differential amp.  I imagine it could cause distortion since the pushing and pulling won't be balanced.   I don't know the best way to measure balance....I'd naively try to measure the current through each side of the nulling pot.

I'd also try to see if all the resistors are soldered down well, especially ones to ground.  Do this with the power off and the caps drained!

[Bill Baker]

Hi Guys,
 First things first, do not use any tube other than a 12AU7 in this circuit. I am unable to print the schematic from the computer I am working from so I have to go back and forth. I must admit that it is unusual to see a bias put on the input tubes.

Where are you taking your bias measurements from? I do not see any bias pots for the output tubes on the schematic. If this circuit diagram is correct, they are indicating that the bias is read at points "A" and "B" (on the schematic) or at least that is what they usually indicate when found on ASL's schamatics. These measurements should be different from what you would read at the cathode resistors.

  I don't want to get into further details not knowing how familiar you are with probing around inside a tube amp. There are both AC and DC voltages that you may have to measure to be sure this amp is running properly. If not comfortable or capable of doing this on your own, please bring in some help by someone familiar with such procedures.

[Bill Baker]

 I wanted to add a few things here before going forward with adjusting the bias based on the "A" and "B" test points. I would do this very gradually. By this I mean taking measurements here to see what you have. If they are much lower than you recommended bias point, you could adjust them slightly till you start to see a rise in the volatge. At this point, I would go back and check the bias at the cathode (10 ohm) resistor to make sure it is not going through the roof.
 While you are making adjustments, be sure to keep an eye on the plates of the output tubes. If you see them starting to develop an orange glow....STOP.
 If you can continue to increase the voltage at the A & B points with no ill effects at the plates, I would stop at about 10 - 20mV below the recommended setting and see how the amp sounds at that point. If things start to get much better, you can proceed further.
 If this doesn't work, you need to get someone to take a look at the amp.