Opamps fried in a Supergreen pre, (SOLVED!)

[lagoausente]

Just had an idea: what happens if you remove C13/C102/C103?

Samuel

  I´ll check it ASAP,
Just a question,  do you think that the configuration of this preamp could be more unstable then the green preamp?  Just wondering if converting it to the green would be posible, they are quite similar. Just a posible solution if finally can´t find the problem.
  I´ll check the C13/C012/C103 removement.
  Regarding to the scope, should I measure pin 1 and 7 refered to ground to see the posible oscilation?

 

[Samuel Groner]

I would be surprised if oscillation were present (the opamps sit at -13/+13 V, i.e. are clipping and hence can't really oscillate well). First check with the caps above before you get a scope.

Samuel

 

[lagoausente]

Just had an idea: what happens if you remove C13/C102/C103?

Samuel

Happens that it´s solid working. Made more than 20 swith on and off, and still working fine. 
  Man, I must to ask it,  why??  what happen?
  Tried the LM4562.
1-  On the channel that was good (before), it fumes R101.
2- On the channel that was the bad, till now, now without the caps, voltage on R101 is ok, all quiet, but no sound.
  Another thing to say, regarding caps, I added long time ago four big 4700u electrolitic caps in the battery supplie rails, two for +V, another two for -V.  If there is something wrong with this, please let me know, if have relation with the caps removed.
  In any case, would like to know about your idea, and if it´s posible to make work the Lm4562 or knowing why the hell they don´t work on this preamp.  I think others tried them on the green working fine.

 

[Samuel Groner]

There is a serious flaw in this design--let me explain: there are only two in practice relevant possibilities to kill a standard IC opamp. One is too high supply voltage (which is not the problem here), and the other is forcing the inputs beyond the supply voltage which is what happens here during startup.

Imagine at power down all opamp pins are at 0 V. Now you apply power; because of the low-pass filter R100/C102 the positive supply rail (i.e. pin 8 ) rises slowly (roughly within 100 ms) to 15 V. However, the positive inputs (pin 3 and 5) are forced immediately to 15 V, which causes large currents flowing into the opamp inputs. Every time you start this thing, the opamp gets a bit more damaged. After a couple of starts it's time to die.

This happens just in one channel probably because once there is large current flowing into one opamp (there's always one opamp first due to very minor differences in wiring resistances, parts tolerances etc.) the positive supply sags a bit and thus protecting the other channel. Now we've fixed one channel, so the other ain't protected any more.

To fix this I suggest to short out R100. I think then it should be possible to put back the caps. Likely this will also make the NSC opamps work--probably their linear common-mode voltage is exceeded because of the voltage drop across R100.

Samuel

 

[Samuel Groner]

Additional comment: this did not happen to other builders probably because they used mains powered supplies. These likely rise slow enough to not cause issues.

Samuel

 

[lagoausente]

  Thanks! Thanks! Thanks!  Solved!!

    LM4562 finally worked ok in the "bad" channel bypassing both R100 and R101.  Haven´t time today to try resoldering the caps. I´ll try at night.

This happens just in one channel probably because once there is large current flowing into one opamp (there's always one opamp first due to very minor differences in wiring resistances, parts tolerances etc.) the positive supply sags a bit and thus protecting the other channel. Now we've fixed one channel, so the other ain't protected any more.

One point here, the main swtich goes to the good channel (directly) and to the bad channel switch.  So to start the "bad" channel, by force had to switch the good channel before, but not viceversa, when switched the good channel, it was alone working.  So seems that istead of helping one to the other, maybe the result was an increased time to supply pin 8, I think.

Likely this will also make the NSC opamps work--probably their linear common-mode voltage is exceeded because of the voltage drop across R100.

Didn´t understand that, but they work now.  Will have to study.  It´s nice knowing.

 

there's not much point typing here for days

  Your efforts gived result,  and saved me hours of frustration and money of opamps dead. I would like to compensate your typing time if I can.
  I have just two doubts,  last two doubts I promise:

  1-  Regarding the not centered PGN, have any influence in the DC Bias adjustment made by RV1, RV2?  Any influence regarding sound performance, or is it ok like it is?
  2-  The extra big caps I put on the supply rails, are 4700u (like C114-C115 position) but they are "before" the power switch.  I put them because measuring with a scope the supply rails, found some AC present just when singing very hard in the mic, maybe regarding the batteries internal impedance. I thought about the batteries couldn´t be fast enough to supply the current demanded by the preamp in short times. After put the caps, the AC disappeared, and so they seem absorve that current flaws.
  Question is, should they be better after the main switch?  If so, should they make the supply rise slow enough as you told about mains powered supplies? (now are always charged, conected directly to the batts)
    In general, if my supply could be improved, I should like to know what could be done, without using regulators, what they have a lot of consumption for my batteries.

  Best regards,

 

[Samuel Groner]

Bypassing both R100 and R101.

I don't think it will be necessary to bypass R101 too (doesn't really hurt either though). In any case you should modify both channels just to be sure...

Didn't understand that, but they work now. Will have to study.

It's not hard to understand; in the LM4562 datasheet on page 3 there is a specification called Common-Mode Input Voltage Range. It says that they guarantee a properly working opamp only if the inputs are 2 V below the positive supply (and 2 V above the negative, but that doesn't matter here). R19/R23 set the common-mode input voltage to +12.8 V; with the NE5532 the positive supply (directly at pin 8 ) is +14.65 V, so the difference is just 1.85 V where we need at least 2 V for the LM4562. And the LM4562 draws more current, so the voltage drop across R100 is larger, lowering the positive supply even further. Bypassing R100 increases the positive supply to perhaps +15 V--gives 2.2 V difference which makes the LM4562 happy.

Note that the guaranteed common-mode input voltage range of the NE5532 is actually even larger--3 V. So it was just good luck that the preamp worked at all.

Regarding the not centered PGN, have any influence in the DC Bias adjustment made by RV1, RV2?

Didn't understand the question--what's "PGN"?

The extra big caps I put on the supply rails are "before" the power switch.

One disadvantage is that due to leakage some small current will flow even if you switch the preamp off. Don't know how much this will shorten their life.

Should they be better after the main switch?

I think I'd put them after the switch, but the difference ain't large. Make sure to use a robust switch as relatively large currents flow when the caps charge.

Samuel

 

[lagoausente]

Quote
Bypassing both R100 and R101.

I don't think it will be necessary to bypass R101 too (doesn't really hurt either though). In any case you should modify both channels just to be sure...

  On my first try with R100 bypassed only, LM4562 didn´t start, but I´m not sure if I wait enough seconds. After bypassing R101 it started after 1 second or so.  Anyway, if doesn´t hurt, just can let it bypassed.

It's not hard to understand; in the LM4562 datasheet on page 3 there is a specification called Common-Mode Input Voltage Range. It says that they guarantee a properly working opamp only if the inputs are 2 V below the positive supply (and 2 V above the negative, but that doesn't matter here). R19/R23 set the common-mode input voltage to +12.8 V; with the NE5532 the positive supply (directly at pin 8 ) is +14.65 V, so the difference is just 1.85 V where we need at least 2 V for the LM4562. And the LM4562 draws more current, so the voltage drop across R100 is larger, lowering the positive supply even further. Bypassing R100 increases the positive supply to perhaps +15 V--gives 2.2 V difference which makes the LM4562 happy.

Note that the guaranteed common-mode input voltage range of the NE5532 is actually even larger--3 V. So it was just good luck that the preamp worked at all.

  Understood, 

Quote
Regarding the not centered PGN, have any influence in the DC Bias adjustment made by RV1, RV2?

Didn't understand the question--what's "PGN"?

  Sorry, it was PGND,  Power GND.
  Repeat the question,  all GND are joined, Audio GND, phamphom and power supply GND.
  So I have two 8,4V batteries for +16.8V another two for -V. So have four 8.4V batteries in series, and take the Power GND from the middle point.
  So, when discharge is not 100% equal in all batteries, I can get diferent +V then -V in reference to ground.  Since all grounds are conected, don´t know if having this unbalancing between ground, +V and -V can affect the DC bias that I adjusted with RV1 and RV2. (Pin 3 of Ic3, and Pin3 or Ic4), or even don´t know if T1 and T2 are afected through R9, R8..,
  The shem of this preamp is beyond my actual knowledge really. 

 

[Samuel Groner]

With RV1 and RV2 you trim the CMRR of the mic preamp, not any DC conditions... There is no significant performance problem with unequal power supplies.

Samuel

 

[lagoausente]

With RV1 and RV2 you trim the CMRR of the mic preamp, not any DC conditions... There is no significant performance problem with unequal power supplies.

Samuel

Fine,  you were right about R101 not needed, it really was R100 that made LM4562 work, was I was confused, have rechecked that.

  Best regards,  I owe you one  (few really).