Misterious Cap, noise issue on NOVANEX AUTOMATIC 6 amp

[razabri]

Alright, got the 1st mod done and tested. The voltages are exactly as you predicted and no change at R21 and R22 - 1.32v at diodes, 0.07v at R21 and 0.52v at R22. No heat, no noise, no problems - only transistors are tiny bit warm, not even hot.

Also, it sound to me as if there is an improvement regarding that noise at low volume - it's not so consistent and it clears off quicker than before.

I can rig those diodes together with sockets to try out best resistor value and if you feel this is not quite safe, I agree on using some of the suggested protective methods - for the sensing diode, I'm not quite sure where it would go in the circuit (drilling the heatsink and securing it is not an issue). Also, 2W resistor(s) sound good to me too and I'm not sure which method would be less invasive, or simply better to choose.

[Rob Strand]

The good news out of that is nothing broke which gives some confidence about moving to the next step.  A small improvement is also good news.   The bad news is those simple mods don't go far enough.

So as far as improving the sound the next step is to add a resistor in series with the diodes.  A value will end-up in the 100 ohms to 220ohm region - fine adjustment is required.    The thing is without adding some protective measures I can pretty much guarantee at some point the amp will overheat by itself and the output transistors will eventually fail.  It might start off Ok but after using it for a period it will get thermal runaway and cook, you might even smell something getting hot.

It's obviously possible to do a complete set of mods adding resistors and diodes like a "normal" amplifier.   However, that's going to be a big butchered mess.   I'm thinking we should add a small valued emitter resistor to Q9, say only 0.47 ohm for insurance.  It can be added by cutting tracks or by lifting the emitter lead of Q9 from the board (probably need to remove Q9 and straighten the emitter) then adding the series resistor.  That's relatively easy to do.   Next add one diode to the heatsink.    The issue here is if you want to poke the diode through the heatsink and the PCB board you have a find a place to put the diode without interfering with the transistors on the top and the tracks on the bottom.   Placing the diode close to the transistor is better in this case.  You can sense either Q9 or Q10, both are in series so they will both heat up.

Just to be clear the mod replaces R19 with two diodes in series with a small valued resistor and one of those diodes is on the sink.

It's a funky set-up.   You definitely won't find any amps design from scratch like this.


Another mod that seems worthwhile is to change R14 from 18k to 5.6k.  You could even try it with the two diodes as they are now.

[razabri]

Great, thanks for reply - I'll try and hook it up like so, perhaps using a trim pot to vary the resistance and fine tune it... Will do the 5k6 resistor placement first and see how it goes. Cheers!

[razabri]

So as far as improving the sound the next step is to add a resistor in series with the diodes.  A value will end-up in the 100 ohms to 220ohm region - fine adjustment is required.

In this setup, I'm not sure if there's any difference in what goes first - diode, diode, resistor or other way around, or if there's any difference with which diode would be the sensing one...

Another mod that seems worthwhile is to change R14 from 18k to 5.6k.  You could even try it with the two diodes as they are now.

Did this one - I'm not sure if there's a bit of an improvement with sound, as I didn't make any recording of original state, but there may have been. Think I notice that cap releasing pop when switching the amp off comes a bit later now and it may or may not be a bit louder...

[Rob Strand]

In this setup, I'm not sure if there's any difference in what goes first - diode, diode, resistor or other way around, or if there's any difference with which diode would be the sensing one...

As a first order approximation it doesn't matter.   Unfortunately when you start running wires around it can matter.  For example one magic combination doesn't oscillate and another does.    If you get trouble with oscillations we will have to deal with it.

Did this one - I'm not sure if there's a bit of an improvement with sound, as I didn't make any recording of original state, but there may have been. Think I notice that cap releasing pop when switching the amp off comes a bit later now and it may or may not be a bit louder...

Again the good thing is no problems.

The "problem" with the amp is it has quite a lot of DC offset.  That offset causes a reasonable amount of DC current through the speaker when there's no signal.   The current is only supplied in one direction and by one output transistor.   It's enough current that is messes with the bias currents when the amp is sitting there with no signal.   While that in itself might not make a large impact of the sound of the crossover distortion it may promote weird thermal problems when we start trying to set the bias to a minimum level.  If the DC current through the speaker is more than the bias current, one output transistor might get hotter than the other during idle which could cause things to drift warm-up.   It's something to take note of and we will see how things end-up.

[PRR]

Rob: I doubt you need more tools, but this begs for attention:

https://www.diyaudio.com/community/threads/bob-cordells-power-amplifier-book.171159/post-6886335
"Bobs book has info on simulating thermal circuits for power amps ... and thermal bias stability with emitter resistors ... For those who would like to do more electrothermal simulations I have made available 5 subcircuits "widgets" that convert the following models to be electrothermal: BJT, MOSFET, JFET, Diode, and IGBT."

[Rob Strand]

Rob: I doubt you need more tools, but this begs for attention:

https://www.diyaudio.com/community/threads/bob-cordells-power-amplifier-book.171159/post-6886335
"Bobs book has info on simulating thermal circuits for power amps ... and thermal bias stability with emitter resistors ... For those who would like to do more electrothermal simulations I have made available 5 subcircuits "widgets" that convert the following models to be electrothermal: BJT, MOSFET, JFET, Diode, and IGBT."

Yes, it's all relevant.  I don't have Bob's book but I'm aware of it.    Self has some stuff on it as well and in fact Ed Cherry's 1967 (1968?) book.

I'm approaching these mods as a delicate balancing act.

Up front I'd say the biggest decision point is to use one sensing diode on one sink or two sensing diode, one on each sink.    Life is made a little more difficult in that choice because of the quasi-complementary output stage.    The quasi-complementary only has one junction in the Vbias loop.   The drivers are off the main sink.  If the single sense diode is well coupled to the transistor it should track reasonably well.   The current source diodes help a little in thermal tracking when we add the series resistor to the bias network, so too the other bias diode .  Since the transistors are in series they should both have about the same temperature rise.  Two diodes might over do it, it could back-off the bias current and go fuzzy (due to crossover distortion) when warm, especially if we target minimum bias.     I know of reliable amps which use a single diode which *isn't* thermally coupled so well to the heat sink.  Some amps have diodes with no thermal coupling and manage to survive.     I've gone for the single well coupled diode to simplify the mods.  Worst comes to worse we can add another sense diode.   The emitter resistor is kind of a back-up.   The heatsinks are small and I'm only suggesting a low valued emitter resistor to reduce power loss, so it's only a weak back-up.     If we can avoid asymmetries like the DC offset current I think we should be able bias this thing out of the crossover distortion zone.

The 220pF compensation cap is actually pretty hefty and doesn't help hide the crossover distortion (not to mention the low diff pair current and slew-rate issues).    However changing this is going to open-up a whole heap of risks for oscillation.   Even the small changes and re-biasing could cause issues.   No Zobel and no output inductor means we don't want to push our luck by tweaking the compensation cap.

Anyway, the idea is to tread carefully.

[razabri]

Success!

Well, it took me some time and effort to get around and complete the mod you suggested, Rob, but now I can report that it has been well worth it.

I followed your guide to the letter and have placed one diode through the heat sink, fixing it in place, and on other end there's the other diode with a 220 Ohms trim pot - it was set at about 100 Ohms to begin with and it pretty much stayed like so (now it's somewhere at 115-125 Ohms), but once switched on, the amp is working perfectly and without any issues, and most importantly, that slight fizz, distortion on low volume is completely gone! I could maybe go all the way to 220 Ohms, but I don't see a reason to do it.

I did add that 0.47 Ohm, 2W resistor on one of the output transistors emitter, as suggested too.

Here's the pic - it may not be most elegant solution, but that's as far as I can go, for now.

The amp is not heating at all, even at high volume, but I guess that it's thermal stability can be confirmed once it has been through more trial (I have played it for some time now, no issues).

So, thanks man, this has been an interesting one to work on and you and few other guys here have made an effort and I appreciate it very much - cheers to all!

[Rob Strand]

Success!

That's great news.

Well, it took me some time and effort to get around and complete the mod you suggested, Rob, but now I can report that it has been well worth it.

Here's the pic - it may not be most elegant solution, but that's as far as I can go, for now.

The mod looks neat to me, you did a really good job.

I followed your guide to the letter and have placed one diode through the heat sink, fixing it in place, and on other end there's the other diode with a 220 Ohms trim pot - it was set at about 100 Ohms to begin with and it pretty much stayed like so (now it's somewhere at 115-125 Ohms), but once switched on, the amp is working perfectly and without any issues, and most importantly, that slight fizz, distortion on low volume is completely gone! I could maybe go all the way to 220 Ohms, but I don't see a reason to do it.

I guess the best way to state the aim is to set that trimpot to the lowest resistance possible but still remove the fizz.   The higher the value the more risk of the amp heating-up uncontrollably.  (There's a remote chance of setting it too low where the fizz comes back when the amp is warm or hot.  It depends on how the sensing diode tracks the temperature of the output transistors as it can drive the output transistors either on or off depending on the tracking.)

If it works with 115-125 ohm that's fine, leave it there.   The values I gave are only estimates, not something to follow rigidly.  In fact that resistor would be the hardest value to estimate in the whole amplifier.  It depends on fine details of the output transistors and the diodes all of which we know nothing about.

I did add that 0.47 Ohm, 2W resistor on one of the output transistors emitter, as suggested too.

I'm pretty sure you put the transistor on the emitter of Q10 and not Q9.   The goodness it adds is very much reduced on Q10.

The amp is not heating at all, even at high volume, but I guess that it's thermal stability can be confirmed once it has been through more trial (I have played it for some time now, no issues).

That's a good sign.  Having the second diode close to the heatsink like in your pic might actually help push things towards the safer side.

So, thanks man, this has been an interesting one to work on and you and few other guys here have made an effort and I appreciate it very much - cheers to all!

You did great.  I'm really happy it turned out.

It all seems to be working fine and it might be best to leave things alone and just monitor it externally.

Measuring the bias current is a good way to judge where you stand regarding overheating.
It's not hard to measure but  there's always risk of slipping and blowing something up when you
do the measurements!   Since it's working you don't need to act but by measuring
the current and at least you know where you stand.

I'll let you decide.


What you do is disconnect the speaker and measure the voltage across the 0.47ohm emitter resistor.   You need to set the multimeter to 200mVDC.    Measure the voltage across the resistor then calculate Ibias = Vmeasured / 0.47 ohm.

Rough idea of voltage

Ibias [mA]   Vmeas  [mV]
0.1                   0.047
0.5                   0.235
1                      0.47
5                      2.35
10                    4.7

Some cautious evaluations would be:
- less than 1mA the risk of the amp heating up due unforeseen events is quite low.
- around 5mA OK but personally I would be keeping an eye on things for a while.
  I'd perhaps even get the amp hot and see how high the bias current got.
- Around 10mA  I would either try to back the trimpot off to minimum point by ear, or
  I would be doing a lot of testing heating the amp up and checking bias current didn't
  even get over 10mA.

If the minimum current to get rid of the fizz is more than 10mA I'd be thinking about putting
the second diode on the other heatsink, or even adding the second emitter resistor to Q9.

A bias current of 10mA or more is very normal but you can only set it to these currents if the diode sensing
is known to track well and there are emitter resistors on both transistors.

[Rob Strand]

Perhaps to give you some confidence in your current set-up:   Based on the 120 ohms, some of your previous measurements, and some measurements I've got on my computer, it's looking like the bias current is well under 1mA.    So it's very likely you are in the safe zone.   I'm not surprised there's no audible crossover distortion with such a low bias current since I've seen other amps work similarly.

[razabri]

Thanks Rob! I did try and set things neatly - that heat sink was a bit harder to drill than I assumed 
You're right about that resistor - I kind of assumed that it makes no much difference where it is from previous discussion, plus it was a bit easier to approach, but I might as well move it to the right spot at Q9...
I'll try and get the measurement as suggested and calculate the bias the way you described - interesting thing is, the amp is not getting hot at all, playing loud or at low volume, even with bass...
Only thing that's a bit troubling now is that volume pot, as it still gets scratchy, and I'm not a fan of the idea of replacing it - perhaps I can try to save it with a better, dry spray thing, as I've already soaked it with that other one I have.

[Rob Strand]

You're right about that resistor - I kind of assumed that it makes no much difference where it is from previous discussion, plus it was
a bit easier to approach, but I might as well move it to the right spot at Q9...

Unfortunately out of the three possible positions that resistor position is the only one that doesn't offer much insurance against overheating.  Back in reply #78 I posted a schematic which shows the correct way to do it with two resistors.  In order to keep the mods to a minimum we only use one resistor.  Position R25 is the best but not practical to mod, R24 not bad.  Then there's the third position in emitter of Q14 which is least effective because it's outside of that middle "loop". 
https://1.bp.blogspot.com/_YpJ002RhDIU/TOariOVWrYI/AAAAAAAAAJA/3-9S7R4uQcs/s1600/Quasi+circuit.JPG
In our case the equivalent to the R24 position is the emitter of Q9

It's up to you to re-mod it.  In fact instead of removing the emitter resistor on Q10 you could just add another resistor to Q9.

I'll try and get the measurement as suggested and calculate the bias the way you described - interesting thing is, the amp is not getting hot at all, playing loud or at low volume, even with bass...

I think it all worked out well.  If you set the trimpot to a low value like you have now it's unlikely to overheat.  Adding the emitter resistor is more like insurance against unforeseen events.

Only thing that's a bit troubling now is that volume pot, as it still gets scratchy, and I'm not a fan of the idea of replacing it - perhaps I can try to save it with a better, dry spray thing, as I've already soaked it with that other one I have.

Scratchy pots are annoying and the sprays don't work in bad cases.  In some cases you can disassemble the pot and clean it out then spray some lubricating cleaner in there after.   You can also bend out the spring a *little* to increase the track force.  None of that seems possible with the style of pots on the amp as they are riveted together.   A sneaky trick would be to swap the Volume and Treble pots, assuming they are the same volume and taper.

Another possibility is it might not be the pot at all.   The amp design promotes scratchiness because the pot wiper is DC coupled to R11.  If you put a cap between the pot wiper and R11 it might stop the scratchiness altogether.   The cap could be non-polar 100nF to 220nF.
***  I'll need to check if that has any significant effect the amps DC offset.
[I checked.  There is a change and there's a small improvement changing R14 from 5k6 to 6k8 to compensate.  There's other ways to deal with it but the other ways need a larger input cap.  We are talking small tweaks here.]

[razabri]

Hey Rob, thanks for another great feedback and I wish you a very happy New Year!

Thanks to your guidance, I'm now enjoying this little amp every day, practicing and experimenting with effects.

Sure, I can add another resistor as described and just leave that one already installed in place.

For the pot, I could try with a cap, but I'm afraid that it's rather some sort of bad contact inside, as it's mostly scratchy on starting point and at the end of the turn - guess some debris and dust could be the issue - this is not much of a problem, but I'd like to get it solved, so I guess I'll try a better spray and then go for adding a cap.

[razabri]

Oh, and I had another post here that didn't get far - I tried to figure out this clipping setting of an effect, so I'd appreciate it if you could have a quick look at it - https://www.diystompboxes.com/smfforum/index.php?topic=128377.msg1235841#msg1235841

[Rob Strand]

Hey Rob, thanks for another great feedback and I wish you a very happy New Year!

Thanks, happy new year to you too!

For the pot, I could try with a cap, but I'm afraid that it's rather some sort of bad contact inside, as it's mostly scratchy on starting point and at the end of the turn - guess some debris and dust could be the issue - this is not much of a problem, but I'd like to get it solved, so I guess I'll try a better spray and then go for adding a cap.

These problems take time to weed out.   There's no doubt the pot has issues but adding the cap can often reduce the scratchlyness quite significantly.   There's other tricks like soldering a high value resistor across pins 1 and 2, or pins 2 and 3, of the pot.   In the case of your amp I'm thinking the cap might be more effective.  Of course the pot might to totally beyond help.

If you took it to a repair shop they would probably try cleaning the pot and if that didn't work they would just replace it.   Repair guys can only spend so much time on the problem.   They want the problem fixed and they definitely don't want it coming back.

[razabri]

Thanks Rob, I'll definitely try the cap then - sounds simple enough and it may be beneficial. Another thing I'm thinking about is to somehow reduce it's SNR, as we have got rid of that distoriton, I get the wish for it to be even cleaner, if possible. It's not something of an issue, but it would be a nice thing to do.

[Rob Strand]

. Another thing I'm thinking about is to somehow reduce it's SNR, as we have got rid of that distoriton, I get the wish for it to be even cleaner, if possible. It's not something of an issue, but it would be a nice thing to do

Way back in reply #34 I suggest a few resistor changes to the preamp.  Given the post-preamp volume control it's a good idea to bias the preamp for maximum headroom if you want a clean tone.  So maybe you should try those mods.    I'm assuming the lack of clean is caused by the preamp.