Op Amp question

[Macka]

Ok, so I'm modifying a circuit that takes a DAC output with a 5V reference and runs it through some op amps and then out to a 3.5mm jack for connection to an amplifier.

One of the changes I'm making is to use a DAC with a 3.3V reference, however I want the output of the final op amp to be the same as it would have been had the DAC had a 5V reference. e.g if I put 50% into the 5V DAC I'd expect 2.5V on the output of the final op amp, if I put 50% into the 3.3V DAC, I expect 2.5V on the output of the final op amp.

Note: the op amps still use 0-5V rails for both circuits.

1) What is this circuit doing? I can see IC3A is acting as a buffer, but I don't understand the purpose of IC3B. I've tried googling various op amp configurations, but no other circuit I've come across seems to use op amps in this configuration. If I knew the name of the config, I might be able to find something on its purpose, but I'm at a loss. I understand the purpose of C5 (removal of DC offset)

2) To achieve the gain I need to make 1.65V output 2.5V and 3.3V output 5V , normally I'd add RF and RG, however in the buffer config there is no ground connection so the RF/RG ratio can't be used. If I add a ground connection and RG is that going to cause issues with IC3B? - again, not understanding how this circuit works makes it very hard for me to decide if that's the right place to put the feedback in.

If anyone can point me in the right direction to understanding this circuit, or explain what's going on here, it'd be much appreciated.

[MatthewEveritt]

The two opamps are in parallel, so it's just acting as a buffer with a bit more umph.

[MK14]

Connecting them in parallel that way (multiple op-amp outputs), MAY not be such a good idea.

https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2013/03/26/paralleling-op-amps-is-it-possible

[danadak]

This might be useful - https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2013/03/26/paralleling-op-amps-is-it-possible

Of greater concern is stability due to the 100 uF load on OpAmp output. Look at
spec sheet pulse response graph with only 50 pF load on it. Phase margin, hence ringing,
already apparent with 50 pF load.

If you want to do this right get a unity G stable OpAmp that can handle high C loads.
Or comp the OpAmp http://www.analog.com/library/analogdialogue/archives/38-06/capacitive_loading.html?doc=CN0343.pdf

Regards, Dana.

[exe]

Of greater concern is stability due to the 100 uF load on OpAmp output.

I think it's DC blocking CAP (in series with the load), it does not load the op-amp, does it?
Though I'm curious why it is there. To save some power? .

[danadak]

If there is a speaker load connected to output then the
reflected Z will still be very capacitive. Put a scope on
speaker terminals, with speaker connected, look at trans-
ient response when playing audio and not. Especially
if you see with no audio an oscillation that will indicate
there is a problem. Better yet feed a narrow pulse into
OpAmps and look at the output.

Another issue is the L load represented by speaker can
over V transient the OpAmp outputs, they should have
diode clamps to their supply rails to protect them.

Overall thats just a poor circuit for driving a speaker.

Another approach is to drive speaker in BTL mode, that doubles
V swing across speaker thereby quadrupling speaker power. And
you can eliminate large coupling cap.


https://www.maximintegrated.com/en/app-notes/index.mvp/id/1122


Regards, Dana.

[Audioguru]

Why do you think that a weak lousy old LM358 dual opamp can drive a speaker? A speaker needs POWER but a lousy old LM358 was designed to use low power so it has low output current, low idle current which produces audio crossover distortion and it has trouble with audio frequencies above only 2kHz at high levels.

The datasheet of the LM358 shows a minimum output current of only 10mA but since you use two opamps you have a peak current of 20mA. The RMS current is 14.14mA then the maximum output into an 8 ohm speaker is (14.14mA squared x 8 ohms=) 0.0016W which is almost nothing. If you use a little power amplifier like an LM386 then you will have a typical output of 0.14W into 8 ohms with a 5 supply or 0.56W when it has a 9V supply.

Other problems with your opamps are that their inputs are not biased correctly and they are followers with no voltage gain. The LM386 power amplifier already has its biasing done internally and it has an adjustable voltage gain of from 20 to 200 times.   

[bson]

1) What is this circuit doing? I can see IC3A is acting as a buffer, but I don't understand the purpose of IC3B.

Two buffers, as has been mentioned.

The confusion stems from the schematic; they should have been placed one above the other with a common input and output.  The way they're placed gives the impression (without looking more carefully) that they're cascaded, not parallel.  Signal flows left to right (or right to left in other parts of the world perhaps, not important as long as it's consistent).

[Zero999]

Why do you think that a weak lousy old LM358 dual opamp can drive a speaker? A speaker needs POWER but a lousy old LM358 was designed to use low power so it has low output current, low idle current which produces audio crossover distortion and it has trouble with audio frequencies above only 2kHz at high levels.

The datasheet of the LM358 shows a minimum output current of only 10mA but since you use two opamps you have a peak current of 20mA. The RMS current is 14.14mA then the maximum output into an 8 ohm speaker is (14.14mA squared x 8 ohms=) 0.0016W which is almost nothing. If you use a little power amplifier like an LM386 then you will have a typical output of 0.14W into 8 ohms with a 5 supply or 0.56W when it has a 9V supply.

To be fair, it doesn't say what the load is. For all we know it could be some 600Ohm headphones.

However, you're right about the low bandwidth and crossover distortion. the NE5532 would be a better choice but we don't know whether it's suitable here because the power supply rails aren't shown. If it's 5V then a CMOS op-amp such as the MCP602 would be better.

[Macka]

The two opamps are in parallel, so it's just acting as a buffer with a bit more umph.

ahh, right, I see, it's actually like this attachment.

If there is a speaker load connected to output then the
reflected Z will still be very capacitive. Put a scope on
speaker terminals, with speaker connected, look at trans-
ient response when playing audio and not. Especially
if you see with no audio an oscillation that will indicate
there is a problem. Better yet feed a narrow pulse into
OpAmps and look at the output.

Another issue is the L load represented by speaker can
over V transient the OpAmp outputs, they should have
diode clamps to their supply rails to protect them.

Overall thats just a poor circuit for driving a speaker.

Ok, I've modified the original post to clear it up, the intention is this will be connected to an audio amplifier, not directly to speakers. Will this circuit still have issues with that (it seems many people have built this circuit in the past with no issues)?

If I need to, I don't mind picking another op amp, because I could use IC3B for another purpose rather than putting 2 dual op amps on the PCB I could then get away with a single dual op amp IC

Why do you think that a weak lousy old LM358 dual opamp can drive a speaker? A speaker needs POWER but a lousy old LM358 was designed to use low power so it has low output current, low idle current which produces audio crossover distortion and it has trouble with audio frequencies above only 2kHz at high levels.

The datasheet of the LM358 shows a minimum output current of only 10mA but since you use two opamps you have a peak current of 20mA. The RMS current is 14.14mA then the maximum output into an 8 ohm speaker is (14.14mA squared x 8 ohms=) 0.0016W which is almost nothing. If you use a little power amplifier like an LM386 then you will have a typical output of 0.14W into 8 ohms with a 5 supply or 0.56W when it has a 9V supply.

Other problems with your opamps are that their inputs are not biased correctly and they are followers with no voltage gain. The LM386 power amplifier already has its biasing done internally and it has an adjustable voltage gain of from 20 to 200 times.   

Sorry, no I don't think it could, to be honest I hadn't considered it. This circuit comes from another person's design that I'm modifying to increase functionality. I've modified the original post, the intention is to drive an audio amplifier, rather than speakers directly.

Based on what you and others have said, I think I might be better off with a slightly more powerful op amp. What I would like to do is have this and a separate a mono headphone output. So I'll try to find a dual op amp that can meet my requirements - I'm not sure if the LM386 is suitable, need to look into it further.

P.S. are you the "Audioguru" from electro-tech-online?

To be fair, it doesn't say what the load is. For all we know it could be some 600Ohm headphones.

However, you're right about the low bandwidth and crossover distortion. the NE5532 would be a better choice but we don't know whether it's suitable here because the power supply rails aren't shown. If it's 5V then a CMOS op-amp such as the MCP602 would be better.

Load is an amplifier, however I also plan to have a separate headphone output. power supply is 0 - 5V.

[Macka]

Alright, I've had a bit of a hunt around and I'm looking at the following IC TSV324IPT (data sheet http://www.st.com/web/en/resource/technical/document/datasheet/CD00063292.pdf).

Again, what I'm trying to achieve is a single output designed for an audio amplifier (PA system/PC speakers/other) and a single headphone output.

Since my DAC is 3.3V ref, I have worked out my feedback resistors as Rf= 10kOhm and Rg = 19.6kOhm (see attached schematic).

If all this is ok, then I just need to work out the headphone amp circuit.