Weird LM3886T problem

[Piisami]

Hi,

I'm having a weird problem with LM3886T Gainclone. The problem is that, I'm getting -25.0V DC at output at all times . It doesn't matter if something is connected at input or not. The resistor Ri (see the link for the schematic) is heating like ****. What could be the problem.

I'm using this connection: Mick Feuerbacher Audio Projects

And dual mono PSUs:
http://myweb.tiscali.co.uk/nuukspot/...PSU_MkIVSE.png

AMP is the same as presented except there is no pot at input (Rin). PSU Is the same except I have much more filtering and there is a ground loop breaker. On both channels, the PSU and AMP are in different chassis. The PSUs have proven to be fully working in different projects.

This is my 1st Gainclone, I'm more used to tube amps...

[pacificblue]

Strange that Ri is getting hot. 25 V at the output should only lead to an insiginificant heat dissipation of less than 1 mW there. Try a 150 µF capacitor in series to Ri to block any DC current.

If that does not help, check for shorts between pins 3 and 4 or replace the LM3886.

And choose the muting resistor according to the datasheet's recommendations. 10k is much too small.

[wintermute]

I think 10K is fine for the Mute resistor at 10K the current will be approx 2.3mA well within the range shown on the current vs Mute graph which shows up to 10mA

formula is RM is calculated using: RM ≤ (|VEE| − 2.6V)/I8 where I8 ≥ 0.5 mA

note the ≤ for RM and the ≥ 0.5 mA The biggest value you could get away with with 27V rails is about 47K but that leaves no margin for error, a smaller value will ensure that the chip is fully un-muted.

In addition to what pacific blue has suggested, I'd check the voltage on Pin 9 and double check that you have Ri connected to pin 9 and not somewhere else edit: and also check the value of Rf and Ri that they are what you think they are

Tony.

[Piisami]

I found the origin of the problem! I'm ashamed to admit and I'm laughing at myself

There was sort between the pins 1 and 9 I just don't know what was i thinking when soldering that wire..

Anyway, now i have 91.9mV and 101.9mV DC offset in the two amps. I assume that should be ok, since in my Classe Seventy there is 150mV D offset and no speaker has blown yet. The preamp's D offset at output is 0 - 0.1mV.

-Sami

[AndrewT]

modify your circuits to reduce output offset to <=20mV and preferably <10mV.

[Piisami]

I don't like idea of adding Ci. Is there any other possibility to reduce the DC offset?

-sami

[AndrewT]

Hi,
yes, match the input resistances of both inputs (R+IN = R-IN).

This applies to the total parallel resistance of all connections between input pins and ground. There may be a small residual output offset after this is done. This is due to slight input offset currents in the chipamp. This can be eliminated by trimming the input that benefits from a slightly lower input pin resistance. There is no way to tell which pin will require the lower resistance without actually powering up and testing.

BUT,
this will demand a DC coupling across both the feedback route and also a DC coupling across the input route.
If you change source from a DC coupled to an AC coupled, your output offset reverts to the poor result you have now.

This can only be overcome safely (for your speakers) by AC coupling both the input AND the NFB. The AC coupling of the input can be at EITHER the source or at the power amp, but not both.
This can be very easily solved by having two power amp inputs, one with the DC blocking cap and another that bypasses the DC blocking cap.

[Atilla]

Coupling your feedback with a properly-sized cap will eliminate any DC gain in the circuit, so in the case of a minimal DC offset coming from the source, your output should have only the same value.

Full safety is achieved trough a good-quality cap at the input. Do not be afraid of using one, the values that you need are quite minimal, given the input impedance of the amp. With a good cap you literally won't know the difference.

My chipamp has a fairly standard cap in the feedback path. Unless all the rest of your equipment is divine and your ears golden - you won't be able to notice a significant change after the caps burn in a little.

At any rate, as with any opamp, matching the impedances for both inputs will reduce your DC offset. My chipamp has lower than 12mV offset on both channels, with different sources connected. Try reaching a lower value.

A switch for plain/DC-coupled input is a very convenient thing to have. It's not a choice you'd regret.

A 150mV is not *that* scary value (am I right here guys?), but it does rob you from some speaker dynamics and with modern circuits it should definitely be lower, unless something is not right. My own DC-protection circuit would trigger very close to that value, just for example.

[AndrewT]

Originally Posted by Atilla

Coupling your feedback with a properly-sized cap will eliminate any DC gain in the circuit, so in the case of a minimal DC offset coming from the source, your output should have only the same value.

no!
if the input offset is zero then there is a residual output offset. This is the input offset current times Rin times stage gain.
If there is an input offset, the residual offset and the input offset add (algebraically) together.

[AndrewT]

Originally Posted by Atilla

A switch for plain/DC-coupled input is a very convenient thing to have. It's not a choice you'd regret.

yes. A very good option, or if you have space add a DC coupled input socket.