Dear audio enthusiasts,
I have read lots of op-amp datasheets to date.
While I miss a bit of knowledge in electronics, this was pretty educational to know (at least roughly
) how generally those chips behave and should be used.
But I would like to know something more precisely and maybe you can help me with that:
Usually, distortion rises with higher gains. This is obvious for high gain preamps, thd curves for 40dB or 60dB of gain are always much higher than thd at unity gain.
Also the output impedance rises with the gain and the higher distortion may arise from there (?), at least in high current amps driving a low impedance.
But then in some datasheets there were curves or declarations, which suggested to me that the lowest THD would not necessarily be at unity gain, but could be at gains e.g. between 1 and 10.
Well, I'm sure many of the knowledgeble persons here have the answer. And I'd be very glad to hear it!
Background: My intent is to use an op-amp before a chip amp, and I'd like to use the chip-amp at the gain it works best (above all distortion-wise)
, because it has the hard work to do with driving the speakers! And the op-amp in front of it has an easy job anyway, so it's no problem to increase the gain on it.
The question is intended to be more universal, but if you really want to know what I plan to do: I think of using the preamp op-amp at a gain of about 10 (or a bit more), and an OPA549 at a gain of only 2, because I imagine that it would be more comfortable then with driving the heavy speaker load.
Please, please, enlighten me!
Dominique
I have read lots of op-amp datasheets to date.
While I miss a bit of knowledge in electronics, this was pretty educational to know (at least roughly
) how generally those chips behave and should be used.But I would like to know something more precisely and maybe you can help me with that:
Usually, distortion rises with higher gains. This is obvious for high gain preamps, thd curves for 40dB or 60dB of gain are always much higher than thd at unity gain.
Also the output impedance rises with the gain and the higher distortion may arise from there (?), at least in high current amps driving a low impedance.
But then in some datasheets there were curves or declarations, which suggested to me that the lowest THD would not necessarily be at unity gain, but could be at gains e.g. between 1 and 10.
Well, I'm sure many of the knowledgeble persons here have the answer. And I'd be very glad to hear it!
Background: My intent is to use an op-amp before a chip amp, and I'd like to use the chip-amp at the gain it works best (above all distortion-wise)
, because it has the hard work to do with driving the speakers! And the op-amp in front of it has an easy job anyway, so it's no problem to increase the gain on it.
The question is intended to be more universal, but if you really want to know what I plan to do: I think of using the preamp op-amp at a gain of about 10 (or a bit more), and an OPA549 at a gain of only 2, because I imagine that it would be more comfortable then with driving the heavy speaker load.
Please, please, enlighten me!
Dominique
Hi,
Using an op-amp compensated for unity gain operation
at high gain gives you high distortion at high frequencies.
Using an op-amp compensated for high gain operation
at high gain gives you low distortion at high frequencies.
Using an op-amp compensated for high gain operation
at low gain gives you oscillation, lots of it and loads of
distortion if it doesn't expire due to overheating.
Basically you should not change your power amplifiers gain
margin, the amount of gain reduction that causes oscillation.
If you increase gain, you increase the gain margin, you can
reduce compensation. If you decrease gain, you need to
increase compensation.
/sreten.
Using an op-amp compensated for unity gain operation
at high gain gives you high distortion at high frequencies.
Using an op-amp compensated for high gain operation
at high gain gives you low distortion at high frequencies.
Using an op-amp compensated for high gain operation
at low gain gives you oscillation, lots of it and loads of
distortion if it doesn't expire due to overheating.
Basically you should not change your power amplifiers gain
margin, the amount of gain reduction that causes oscillation.
If you increase gain, you increase the gain margin, you can
reduce compensation. If you decrease gain, you need to
increase compensation.
/sreten.
Hello,
Op-Amps provide huge gains, often 1,000,000 times.
They achieve low distortion by using most of the gain for negative feedback (NFB).
Thus a gain of 10x would leave 100,000x for the NFB and reduce distortion quite a bit.
A gain of 100 would leave 10,000x for NFB and distortion would be higher because less NFB is available.
A gain of 1,000 would leave 1,000x for NFB and distortion would be higher because less NFB is available.
A gain of 1 would leave 100,000x NFB and distortion would be lower because more NFB is available.
Output impedence is lowered by NFB too.
But the output impedence is usually low enough that it isn't a factor.
Personally, I'd set the OPA549 for unity gain and get all the voltage gain from the pre-amp.
Maybe a 5534, or OPA134 type device.
Op-Amps provide huge gains, often 1,000,000 times.
They achieve low distortion by using most of the gain for negative feedback (NFB).
Thus a gain of 10x would leave 100,000x for the NFB and reduce distortion quite a bit.
A gain of 100 would leave 10,000x for NFB and distortion would be higher because less NFB is available.
A gain of 1,000 would leave 1,000x for NFB and distortion would be higher because less NFB is available.
A gain of 1 would leave 100,000x NFB and distortion would be lower because more NFB is available.
Output impedence is lowered by NFB too.
But the output impedence is usually low enough that it isn't a factor.
Personally, I'd set the OPA549 for unity gain and get all the voltage gain from the pre-amp.
Maybe a 5534, or OPA134 type device.
Thank you for helping!
So, if I understood correctly,
while the amp is assured to work as it should,
the lowest distortion will appear at unity gain.
Except of course if the input signal will be too high and cause distortion, or the amp is fast and easily starts oscillating at low gains.
Hey, maybe I should test the distortion of the op-amp pre and the power amp together in chain, at various gains, so they both together will have the wished gain of... 24 or so...
pre x24 / power x1
pre x12 / power x2
pre x6 / power x4
pre x3 / power x8
OK, I know, I' talk to much, I should diy more...
Thanks again!
Dominique
So, if I understood correctly,
while the amp is assured to work as it should,
the lowest distortion will appear at unity gain.
Except of course if the input signal will be too high and cause distortion, or the amp is fast and easily starts oscillating at low gains.
Hey, maybe I should test the distortion of the op-amp pre and the power amp together in chain, at various gains, so they both together will have the wished gain of... 24 or so...
pre x24 / power x1
pre x12 / power x2
pre x6 / power x4
pre x3 / power x8
OK, I know, I' talk to much, I should diy more...
Thanks again!
Dominique
Dominique said:
Hi,
You cannot do this. The power amplifier will expire rapidly.
(Nevermind the preamplifier is unlikely to have enough
voltage swing for the unity gain power amplifier case)
The power amplifier must be used at its designed gain (or greater).
you have been warned..........
/sreten.at any gain a OPA549 power "chip amp" will always have more distortion than any decent quality audio op amp used at gains up to +24 so the question is moot, other factors will determine what you can do
cascading separate feedback amplifiers is in any event a poor way to reduce distortion, distortion is reduced by high loop gain so using the same gain stages within a single feedback loop gives lower distortion at the same total gain than any partitioning of the gain between separate feedback amplifiers using the same op amps
the entertaining part comes from trying compensate a feedback loop with 2 op amps for stable operation and still get the distortion reduction advantages of their combined open loop gain
Walt Jung shows some multiloop feedback circuits but he uses fast current feedback output op amps and the slower input op amp supplies the dominant pole compensation
http://www.analog.com/library/analogDialogue/archives/39-05/op_amp_applications_handbook.html - section 6
an example with a fast output amp:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=45794&highlight=
with the 900 KHz unity gain bandwidth of the OPA549 you have to really slow down the input op amp - and lose most of its added loop gain if you use these compensation schemes
an alternative is to let the OPA549 supply the dominant pole and add a very fast input op amp with local feedback that controls its added phase shift at 900 KHz to sufficiently low levels;
an option might be a LM6171 input op amp with local modified integrator feedback with a high frequency gain of Av +24 -- this gives ~ 3-4 MHz pole from the input gain stage which adds ~ 10-15 degrees excess phase at 900 KHz with the global closed loop gain set to +24 (a lead capacitor in the global feedback network can cancel this added phase and improve the phase margin even further)
the added gain of +24 improves the composite amplifier’s distortion at all frequencies by a similar factor vs the OPA549 supplying all of the gain, even more of the LM6171’s gain can be made available to reduce audio frequency distortion by the extra pole of the modified integrator local feedback adding gain below ~ 100 KHz so it doesn’t add too much more excess phase at 900 KHz
cascading separate feedback amplifiers is in any event a poor way to reduce distortion, distortion is reduced by high loop gain so using the same gain stages within a single feedback loop gives lower distortion at the same total gain than any partitioning of the gain between separate feedback amplifiers using the same op amps
the entertaining part comes from trying compensate a feedback loop with 2 op amps for stable operation and still get the distortion reduction advantages of their combined open loop gain
Walt Jung shows some multiloop feedback circuits but he uses fast current feedback output op amps and the slower input op amp supplies the dominant pole compensation
http://www.analog.com/library/analogDialogue/archives/39-05/op_amp_applications_handbook.html - section 6
an example with a fast output amp:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=45794&highlight=
with the 900 KHz unity gain bandwidth of the OPA549 you have to really slow down the input op amp - and lose most of its added loop gain if you use these compensation schemes
an alternative is to let the OPA549 supply the dominant pole and add a very fast input op amp with local feedback that controls its added phase shift at 900 KHz to sufficiently low levels;
an option might be a LM6171 input op amp with local modified integrator feedback with a high frequency gain of Av +24 -- this gives ~ 3-4 MHz pole from the input gain stage which adds ~ 10-15 degrees excess phase at 900 KHz with the global closed loop gain set to +24 (a lead capacitor in the global feedback network can cancel this added phase and improve the phase margin even further)
the added gain of +24 improves the composite amplifier’s distortion at all frequencies by a similar factor vs the OPA549 supplying all of the gain, even more of the LM6171’s gain can be made available to reduce audio frequency distortion by the extra pole of the modified integrator local feedback adding gain below ~ 100 KHz so it doesn’t add too much more excess phase at 900 KHz
Hi,
thank you all fro participating! That's a very interesting subject to me!
@sreten
The opa549 doesn't have to be used at higher gains, but you're right and I forgot: Most op-amps for the first stage won't have enough voltage swing!
@jcx
that sounds very interesting! I only wish I'd understand all of you've said! I've read about those poles, but didn't understand that well to date.
But e.g. if I'd simply use a global feedback and an opa2132 first and the opa549 then, both with a gain of about 5, overall about 25, will the circuit be easily unstable?
From your description it sounds a bit difficult to get such a circuit stable.
And yes, from the THD graph, the opa549 doesn't look that fine, but I think it will perform a bit better with 8ohm speakers, or equally with two chips in parallel to drive 4ohm speakers.
Cheers,
Dominique
thank you all fro participating!
That's a very interesting subject to me!@sreten
The opa549 doesn't have to be used at higher gains, but you're right and I forgot: Most op-amps for the first stage won't have enough voltage swing!
@jcx
that sounds very interesting! I only wish I'd understand all of you've said!
I've read about those poles, but didn't understand that well to date.But e.g. if I'd simply use a global feedback and an opa2132 first and the opa549 then, both with a gain of about 5, overall about 25, will the circuit be easily unstable?
From your description it sounds a bit difficult to get such a circuit stable.
And yes, from the THD graph, the opa549 doesn't look that fine, but I think it will perform a bit better with 8ohm speakers, or equally with two chips in parallel to drive 4ohm speakers.
Cheers,
Dominique
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.