variable gain setting for LM3886 based power amp to eliminate need for pre-amp

Dear Gainclone enthusiasts,

many schematics of LM3886 based audio amplifiers were published and I'm really surprised by the
number of possible configurations (non-inverting, inverting, DC-servo, brigded and so on).
But I was not able to find an answer to my idea / question. Perhaps somebody here is able to
to comment on.
The underlying idea / goal for my project is to have the complete DA conversion and amplification
steps in an loudspeaker having active crossovers. The speaker is only fed with a digital signal
supplied by a TOSLINK connection. Until now it is not completely clear whether analog of FIR
crossovers will be used. But that is not for importance for the discussion.
In traditional audio implementions everytime the signal chain is as follows: source, input selection,
volume control, power amplification. The drawback according to my understanding is, that the power
amplifier usually uses a fixed gain setting and volume control is done in between pre- and power
amplifier. The result is, that the signal is strongly attenuated and than amplified again.
The question now is, whether it is possible / beneficial to do the volume control in the power amplifier
to have full scale signals as long as possible. The LM3886 itself is an operational amplifier.
It is possible to use it in non-inverting and inverting configuration. In both cases the gain is
defined by only 2 feedback resistors. In the inverting configuration the gain can be set in between
-infinite to xxdB. Why nobody has done this until now? Is there any drawback (stability ...) which
I do not see at the moment?
I'm really looking forward to your response.
Best regards so far...

Yours Stephan
 
Op amps have certain gain limit where they are stable, and also function optimally. As billshurv said, LM3886 has to have certain gain.

I thought about similar idea also, to control gain rather than attenuating input signal, but I realized that controlling gain is not as simple as attenuating the input signal. But I saw somewhere that there is such amplifier, I don't remember what it was.
 

billshurv

Member
Paid Member
2014-03-01 11:53 pm
It's the way all amplifiers work. You aim to have your gain structure such that you don't have too much excess gain but as different music has different levels of compression you need a minimum of 20dB of adjustment, which you could do digitally.

So start with how efficient your speakers are and how loud you listen and work from there!
 
look at the datasheet.
The stability margins can be predicted from the graph of Figure 49.
At a gain of 10times (+20dB) the phase margin is 65° (180-115)
This is OK for driving a non reactive load, but can usefully be made a little bit higher.
For driving a reactive load it is far too low. Aim for around 80° to drive a speaker. This requires the gain to be set to +26dB (20times) to +28dB (28times). This is the "sweet spot" mentioned by a few Builders.

You can apply external components to change the stability margins. eg, you can get good reactive load stability @ a gain of +20dB with a couple of small capacitors and a resistor.

But, you can't just change the gain on the fly.

The gain and the stability must be set to match each other and then left alone. Volume changes are then done via a conventional attenuator.
 
Thanks a lot for the numerous replies. In deed, figure 49 of the data sheet is completely saying the truth. A gain setting below 20dB causes instability. It was not present to me, that the phase margin is so little in these power opamp chips.
There are still 2 opportunities left. Take the 16Bit input signal, scale it to 24Bit and use these additional 8Bit for volume control. The other way is to do the classical way maybe by using an digital potentiometer. I'm not sure which one is better. I guess DAC THD figures and so on are given at full scale signals and are much worse when using only the lower bits.
Best regards so far...

Yours Stephan
 

Arty

Member
2011-02-24 5:04 pm
to some extent it is sortha possible with unity gain stable opamps, but..
it has its quirks, and matching of channels would end up interesting too.
but, even if it was really possible without issues i mentioned, say.. you anter the feedback network to use a potmeter. and then hen you turn the pot, and it looses contact fora tiny amount of time.. nah,....
its not that good trust us.
you get no gain in quality to be fair.
it may seem like an interesting solution, but then again, it ain't.
 
Besides the chipamp only being stable at a high gain.
How would you like to adjust the resistor settings?

Let say you manage to get some volume information trough the SPDIF interface.
SPDIF should support 6channel 24Bit so if you use 1 channel for volume you could simply use a PGA2311(2ch) or PGA4311(4ch) to adjust the gain of the dac(s) signal(s)

These chips outperform the LM3886 (based on specs)

If you want some gain control just put a resistor in front of the input pin.
(if you have a resistor between in+ and in-(gnd))
I don't mean the chip in+ and In- inputs but the inputs of the amplifier board!