• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Spec 4 with Problems

OK, two things to do, watch the + 68v(pin 2) and -68v(pin 6) power supply voltages over the drift period to see if the power supply is inconstant.

Also I would watch the output offset that is drifting to see if there are correlations between the drifts in the two channels - like the right channel rises 60 millivolts while the left rises how much, and is it repeatable?


I would also tie both R6's to ground instead of the wiper of VR1. That gives us a rock solid offset input (not something dependent upon the stability of the supply voltages), even if the value isn't zero at the output. The voltage readings at the bases of Q1a & Q1b did stay stable, implying the Q1 stage is doing it's job.

If that doesn't turn up anything, then we do the gain of one thing with either a capacitor or bypassing the R22 feedback resistor.

I feel that anything misbehaving after Q1 in the circuit by drifting would also affect the idle currents' reading.
 
The power supply is relatively constant. Pin 2 started at 63.6 v, pin 6 at -67v. After 30 min, pin 2 was at 64.5v and pin 6 at -67.4v. Change was very slow and in one direction over the 30 min.

The drift on the right channel, with R6 grounded and C7 removed was about 15mv.

The drift on the left channel, with no changes to the circuitry was about 50mv.

The channels moved independently, i.e. one would go up with no changes in the other and vice versa.
 
Either C7 is leaking DC, or the drift is coming from the DC offset adjusting circuit or the power supplies for it. The 15mV drift you are getting now is acceptable, so now lets remove the short on R6, adjust for 0V on the output, and see how far the output drifts with C7 still disconnected.
 
Back to square one.

I left the unit on for an hour and monitored the voltage at the speaker terminals for the right channel, the one with the resistor grounded and C7 removed. Earlier, the drift was about 15mv. Now it is 70mv, about the same variation as in the left channel, with the intact circuitry.

The power on pins 2 & 8 is relatively constant as indicated in previous posts.
 
Without reading the whole thread and you have goods experienced guys working with you. I think you have disconnected the input ac coupling capacitor to make sure there is no leak current coming into the input differential pair.
I am guessing you have a thermal change with time across the board and affecting the transistor operating points to drift out of step over time. This is not a problem with chip amp. Discrete amps are prone to thermal difference between transistors and diodes. Amps use feedback to get stability. Discrete need it more than chip amps.
Try taking a drinking straw and blow air on one transistor at a time while watching the offset. One lung full slowly (30 seconds) each should do it. You don't want any freeze mist as that will cause all to step forward and you won't find the real one. Starting from the input differential and work back. Most likey culpit is within the first 3 stages and that is why some amps have the diffential pair in a single chip/can. No need to go into the output power trans. If trans don't do it, try the diodes. They use diodes to compensate trans bias, but being discrete they don't lock step thermally so you can have drift.
 
Last edited:
Leave everything the way it is, but short across R22, NOT to ground. Solder a wire across it's 2 terminals, don't use a clip lead, any intermittent contact will cloud the issue.
 
Shorted R22, and R6 is still grounded and C7 removed.

On start up, had a reading of 50mv. Over 40 min, it slowly dropped to 28 mv and stayed in a range of 28 to 30 mv after that.
 
miklemke said:
Shorted R22, and R6 is still grounded and C7 removed.

On start up, had a reading of 50mv. Over 40 min, it slowly dropped to 28 mv and stayed in a range of 28 to 30 mv after that.
Click to expand...

What about the other channel? This was the channel that was drifting the most?
 
From these results, I think the drift is being caused by something closer to the output, maybe those diodes that caused more drift when you sprayed them with freeze spray. If the diff amp were at fault, I don't think reducing the gain would have improved the drift.
 
If the last resistor shorted is tied to one the diff amp trans base (was that the feed back resistor?). Then my vote is the diff amp trans. The equivalent resistance from voltage source to base was changed in one leg. The base current would drop some volts through the equivalent resistor. Which is in effect like applying a small volt to one diff amp input and cause the amp output to shift. Smaller resistor with base current makes a smaller signal to the diff amp and output shift.
I think the 2 diff trans are likely:
1. they were not gain and maybe Vbe matched. (There is a reason to match them or use the ones with 2 trans in the same chip/can)
2. Some small temp diff would caue small operating point shift (leakage, gain-->base current) between them even if they were identical trans. There is a reason to put them in the same chip. If the trans are fets then base current is of no concern, but leakage and gain are still there.
3. the collector current is much different between the 2 trans. This has to do with how much feedback to get 0.0 volt difference between the 2 diff inputs, the common emitter current source and the collector resistor. Diff Ic automatically makes them operating in a different point (base current). Temp change will cause the (leakage, gain-->base current) to change relative to each other which cause the amp output to shift.

my 2 cents
 
Q1, Q2 & Q3 are very sensitive to temperature changes. I blew on them with a straw as Solderiron recommended and saw significant and immediate changes in the mV readings.
 
>Q1, Q2 & Q3 are very sensitive to temperature changes

Single chip amp is so much better than discret in thermal stability and controllable overall performance. Some discret application in the old days had a portion of the circuit in a small regulated temperature oven/chamber. You can't do that here.

If Q1 and Q2 are the differential trans (Emiters connected together). You can try to get a matched pair (same DC gain) at the operating collector current. Unfortunately you have to hand select them. The ideal operating point for the differential pair is the collector currents should be equal and the base to source equivalent resistance should also be the same. Layout wise the diff trans should be as close as possible to get temp tracking (if they ase not in the same package).

Q3 is then likely the current source to the differential pair emitters or the voltage gain and level shift stage (base tied to one of the differential collector). Better designs have a temp compensating diode tied to the base of Q3. Not much you can do beyound that in discrete design.
 
Q1 is the differential transistor. It is an SA798. Basically two transistors in one package with the emitters tied together internally.

Because of the issues I've been having, I picked up another SPEC 4. This one has issues also, but most importantly to me it has one good amp. This amp has stable voltages. There are some differences in circuitry, but most interesting, it has a cover over Q1. It also has the extra diodes connected to Q1 that I mentioned in previous posts. (I guess that indicates these diodes were some production modification.)

I put this good amp in the SPEC 4 with the drifting voltages and it was stable. To me that leaves out the power supply and anything else on the meter amp board as a source of the drifting problem. It looks like the problem is related to the amp board itself. The quest goes on.
 
Since the diff amp is in a single package, the 2 transistors should be matched. But if they are old, the gains of each could now be different. Try covering Q1 with heatshrink tubing to insulate it a bit, see if that improves the drift.
 
I did cover Q1 with shrink tube and it did improve the drifting from the standpoint that it isn't as fast a change. It still drifts though.
 
You don't have a fan on over your workspace, or an air conditioner vent circulating air in the room do you?

I have learned to turn off all air currents when working on these things. I see it affecting bias typically, but you could rule it out as a contributing factor.
 
>I put this good amp in the SPEC 4 with the drifting voltages and it was stable

Would you please run a straw air flow test on Q1, Q2, Q3, Q4, Q5, and D1 on this board? Curious mind wants to know :)
 
You must log in or register to reply here.
Back
Top Bottom