the legendary tg12412

[warpie]

Hi,

After some time, I now got the chance to experiment again with this project.
I'm testing my proto boards and I noticed a couple of interesting things, so (as always) your input is more than welcome 

Both of the issues I noticed are DC-leakage related.

1) When I power up the unit, the audio output (after C101) reads about 400-500mV! This values decreases (painfully slow) and after 25-30 minutes it settles down to 20-30mA. Is this normal?

2) Also, I noticed that on the filter sections (points 22,25,23 on the schematic) I measure about 550mV DC and this doesn't decreases with time.  Again, I don't think that's normal behaviour as there are blocking caps, but please correct me if I'm wrong.

The schematic can be found here

[Michael Tibes]

1 is kinda strange, do you have any load? if you just measure dc without any other load than the dmm things might take a while to settle? Otherwise a leaky cap?

2 might be ok at first glance, as long as all 3 points are at he same dc level.

Did you hear the beast with music?

Michael

[warpie]

1 is kinda strange, do you have any load? if you just measure dc without any other load than the dmm things might take a while to settle? Otherwise a leaky cap?

Aha- indeed!  I had connected only my DMM and the scope. Now (with load) it settles at 55-60mV in a couple of seconds.
However, I still believe 55-60mV is quite high? Also, I'm a bit concerned about connecting the transformer because maybe it'll magnetise it.

I swapped a couple of caps (the 470uF one) and they all show the same value.
(I must say that the output, from the soundcard to the EQ, reads only 7mV)

2 might be ok at first glance, as long as all 3 points are at he same dc level.

All 3 points read 560mV. Is there any way of improving it? I guess such amount of DC will cause 'scratces' and 'clicks' on the pot/switches...

Did you hear the beast with music?

Very roughly. I liked what I heard but unfortunately I haven't enjoy it yet. I'm still troubleshooting...

thanks for your input!

EDIT: mA/mV confusion 

[Michael Tibes]

regarding 2: is there any crackle? If I get it right, then things will be fine when all 3 points are at the same dc level because then there is actually no dc across the pot. It would only be a problem if there was dc between these points, but measuring from 22 to 23 or 25 should show no dc. The voltage between the pot and gnd is irrelevant in terms of crackling in this case. So if I get things right there might be nothing to improve, getting rid of the voltage between the pot and gnd might even affect the transistor bias?

What are you measuring at the output, mA - not mV? Maybe just connect 47k from out to gnd and see what happens?

Michael

[warpie]

regarding 2: is there any crackle? If I get it right, then things will be fine when all 3 points are at the same dc level because then there is actually no dc across the pot. It would only be a problem if there was dc between these points, but measuring from 22 to 23 or 25 should show no dc. The voltage between the pot and gnd is irrelevant in terms of crackling in this case. So if I get things right there might be nothing to improve, getting rid of the voltage between the pot and gnd might even affect the transistor bias?

You're right. I was measuring with reference the gnd. There's no DC when measuring between pins. There's no crackle but I was thinking that it might develop after a period of time.

 

What are you measuring at the output, mA - not mV? Maybe just connect 47k from out to gnd and see what happens?

Sorry, I meant mV and not mA (I edited my post). Thanks for pointing it out. A 47K resistor improves the DC by 5-6mV so I now read around 50-55mV. What amount of DC is acceptable at the output and moreover, is there's any danger of damaging the output transformer?

[Michael Tibes]

I'd replace the output cap with the biggest foil cap at hand and measure again, keeping the 47k. That would be in order to make sure you haven't bought a batch of leaky 470uF caps. If I'm not overlooking something then there should be no dc. I am lost as well in terms of 'how much dc is too much', that should also depend on the transformer.

Michael

[warpie]

Thanks Michael, I'll try a bigger lytic although the 470uFs I tried are from different batch and different brand.
FWIW, Maybe I should mention that instead of the original 2n4898 and 2n4910 transistors, I'm using NTE218 and NTE175,
which are supposed to be modern replacement.

Also, one more thing if you don't mind. Can you please elaborate a bit about the 47k resistor? I'm experimenting with different values and I noticed that the lower the value, the lower the DC. Is there a limit as to how low the resistor should be?

[Michael Tibes]

There isn't much to elaborate, it's just a path to ground to discharge the output cap. 47k is just a rough guess, not too low to present a load but low enough to show some discahrging effect after a little while... And it's mostly available 

Lowering the resistor wouldn't be a cure but just a cover up, if you want to get rid of all the dc you gotta find out where it comes from. I just glanced at the schematic and it seems like nothing else was connected to the output after the cap, so I believe theoretically it could only be leakage. That's why I would try a film cap, afaik they shouldn't have any leakage. I usually have something like 10uF lying around, that should do to see what difference it makes.

Michael

[warpie]

Thanks Michael, I have some 10uF Wimas that I'll try (although I'd expect that such a small value will mess with my frequency response).
But I'll try it anyway

EDIT: I tried a 10uF film and it looks to work OK. DC offset is around 3mV and the frequency response at the bottom end is as flat as with the
470uF. I tried lots of lytics (different values and brands) and oddly enough they all appear to produce 50-60mV DC offset so I tend to believe that this is how the original circuit behaves (or that the problem is laying somehere esle...)

[Michael Tibes]

I'm not good at mathematics, but I'm sure the 470uF are on the very safe side not to affect the lowend - with the lowest possible load. Often a lot smaller values are used. 10uF might be ok with a not too high load, that's probably why it mesures ok. If the offset is gone with the film cap it points to leaky caps in my opinion.

Michael

[warpie]

Thanks for all you help Michael. Really appreciated! 

[leadbreath]

hey warpie, did you finally finish this project?

[warpie]

hi leadbreath,

I've put it aside at the moment as I'm very busy with other things but I'm getting there slowly slowly..

[etheory]

I could be completely wrong here, but here's a hypothesis.

Film caps that I am aware of are non-polarized, but electros (at least most of the ones I've come across recently) are polarized.

So, I noticed recently in one of my projects that if you insert an AC coupling cap in an audio path backwards, such that the difference in voltage across it is opposite to it's polarity, then it maintains a DC offset.

However, simply reversing the cap, fixes this completely. The dc offset falls to 0 with a time constant relative to RC (where R is the load impedance and C is the capacitor value itself).

Is it simply possible that the schematic you are using has the polarity of the AC output cap wrong?

This is conjecture, but I have seen the same symptoms before in different circuits.