Hello, I'm a newbie when it comes to "designing" my own circuits, so I would like some input on the attached circuit.
The goal is to have a capacitor coupled opamp that will serve as a buffer for a 25K volume pot. Source impedance will be less than 1.3k ohms, I would like the circuit to be able to play with amplifiers that have an input impedance as low as 30K. The opamp will be the LME47910 or LME47920/LM4562. Power supply pins will be decoupled with multiple low value ceramic capacitors.
What are your thoughts on the resistor and capacitor values?
Would you add or remove components to improve stability?
Any input would be greatly appreciated.
Thank you
The goal is to have a capacitor coupled opamp that will serve as a buffer for a 25K volume pot. Source impedance will be less than 1.3k ohms, I would like the circuit to be able to play with amplifiers that have an input impedance as low as 30K. The opamp will be the LME47910 or LME47920/LM4562. Power supply pins will be decoupled with multiple low value ceramic capacitors.
What are your thoughts on the resistor and capacitor values?
Would you add or remove components to improve stability?
Any input would be greatly appreciated.
Thank you
Attachments
Its basically fine. R2 and R3 are not needed. R4 is to large, it need only be perhaps 33 ohm, just enough to isolate the opamp from cable capacitance effects.
(Although this is AC coupled you do realise it still needs a spilt supply as drawn)
(Although this is AC coupled you do realise it still needs a spilt supply as drawn)
When the pot cursor is at ground, it happens sometimes that the circuit oscillates.
I always keep R2.
I have encountered the problem a few times.
There is always some parasitic capacitance at the input of an amplifying device and some parasitic inductance in the wiring, so a small value resistor will tame the tendency of the LC circuit to resonate.
This may look like an excessive precaution but I think it is often used, it is innocuous . It is sure that Blehmbo has already seen it somewhere.
There was a paper by Erik Margan on the subject at the input in an old Electronics World issue.
There is always some parasitic capacitance at the input of an amplifying device and some parasitic inductance in the wiring, so a small value resistor will tame the tendency of the LC circuit to resonate.
This may look like an excessive precaution but I think it is often used, it is innocuous . It is sure that Blehmbo has already seen it somewhere.
There was a paper by Erik Margan on the subject at the input in an old Electronics World issue.
Thanks forr. Its obviously a problem you have encountered and know the solution that works to prevent it.
Very low-noise opamps are particularly susceptible to this, as their Rbb' inevitably is low and does little to isolate input capacitance (or even negative input impedance).
That's also why a direct connection from out to -in is recommended against with the AD797. Other, less "hot-rodded" parts can take that just fine (Yamaha must use NJM2068s like that by the bucketload, I think 5532s don't mind either, nor do NJM4556As - note how none of these are remotely close to 0.9 nV/sqrt(Hz) in terms of e_n).
The only common audio part that potentially needs a 470 ohm output isolation resistor is a TL072 though. Most any other part should be perfectly happy with 47-100 ohms. Something like an LM4562 should also drive loads down to less than 600 ohms with ease.
R3 can also be used to reduce impedance imbalance between -in and +i if the noise penalty isn't an issue but low distortion is. The source impedance presented by a 25k pot fed by a low-impedance source can be up to 6.25k after all, with a minimum of near-zero ohms. It's not a big issue for the LM4562 (LME49860 was found to exhibit about 0.04% of input impedance distortion at 100k imbalance, 10 kHz @ 7.75 Vrms, unity gain, so at 6k25 it would be like 0.0025%, proportionally less at lower levels - so at 1.5 Vrms it would be down to 0.0005%), but a lot of FET input parts are about one order of magnitude more critical.
That's also why a direct connection from out to -in is recommended against with the AD797. Other, less "hot-rodded" parts can take that just fine (Yamaha must use NJM2068s like that by the bucketload, I think 5532s don't mind either, nor do NJM4556As - note how none of these are remotely close to 0.9 nV/sqrt(Hz) in terms of e_n).
The only common audio part that potentially needs a 470 ohm output isolation resistor is a TL072 though. Most any other part should be perfectly happy with 47-100 ohms. Something like an LM4562 should also drive loads down to less than 600 ohms with ease.
R3 can also be used to reduce impedance imbalance between -in and +i if the noise penalty isn't an issue but low distortion is. The source impedance presented by a 25k pot fed by a low-impedance source can be up to 6.25k after all, with a minimum of near-zero ohms. It's not a big issue for the LM4562 (LME49860 was found to exhibit about 0.04% of input impedance distortion at 100k imbalance, 10 kHz @ 7.75 Vrms, unity gain, so at 6k25 it would be like 0.0025%, proportionally less at lower levels - so at 1.5 Vrms it would be down to 0.0005%), but a lot of FET input parts are about one order of magnitude more critical.
It's a solution I am aware of, but rarely implement.
So far it has not bitten me.
Many discrete amplifiers show a low value resistor just before the base of the LTP pair, like a base stopper.
I have recommended R3 to many of our Members.
Some datasheet demand R3.
Last edited:
Thank you guys for all of your input and I will make changes accordingly. I do have one more question regarding the distance between the components.
Should I expect detrimental effects if I were to mount the Opamp, opamp decoupling capacitors, R2*, R3*, and R4 on a separate board that is placed a couple inches away from the rest of the circuit using twisted pairs of wires (dual opamps per channel for a differential signal)? star grounding would be implemented.
Thanks again!
Should I expect detrimental effects if I were to mount the Opamp, opamp decoupling capacitors, R2*, R3*, and R4 on a separate board that is placed a couple inches away from the rest of the circuit using twisted pairs of wires (dual opamps per channel for a differential signal)? star grounding would be implemented.
Thanks again!
When datasheets claim that their chip is "unity gain stable" does that mean that R3 is not required?
That should be fine. The point most susceptible to noise pickup is the high impedance opamp + input. If fed from a low impedance source then there is little problem. The output is at low impedance and wont pick up noise.
Why P = 25 kOhm ? All modern sources can handle much lower loads.
I suggest 10 kOhm (a good ALPS) or less (4.7 kOhm, 2.2 kOhm) but log pots with such values are less common.
I suggest 10 kOhm (a good ALPS) or less (4.7 kOhm, 2.2 kOhm) but log pots with such values are less common.
No. 'Unity gain stable' means the chip itself has sufficient HF rolloff that an ideal unity gain circuit will be stable. We don't know how to build an ideal circuit, so R3 may still be required. I suppose that under some circumstances R3 could make things worse so be prepared to put a wire link in.shredhead said:
Last edited:
unity gain stable means it does not oscillate when loaded as described by the manufacturer.
But, look at the gain/phase plot.
We find that the stability margins change a lot from unity gain to gains far in excess of 3.
It is the stability margins that determine the "shape" of the output when various loads are applied. That "shape" is heard as unreal or real sound quality.
A phase margin of 45degrees sounds quite different from a phase margin of 80degrees and different again if a little bit of parasitic capacitance is added to the loading. Try 100pF, or 1nF. The manufacturer might show some of these effects in their datasheet.
Last edited:
Since you posted the above I've read it about 40 times and still can't figure it out. Would you please explain further? Especially, "The source impedance presented by a 25k pot fed by a low-impedance source can be up to 6.25k after all, with a minimum of near-zero ohms," which defeats me entirely.
For quick reference I've re-posted the Original Poster's circuit below. After that is posted a generic circuit of mine that I think might be applicable somehow, or maybe not...but it seems to relate somehow...or not.
.
Attachments
Last edited:
Wire a pot up like this and measure the resistance between the wiper and either point A or point B.
What is the lowest value or R in relation to pot value ?
What is the highest value of R in relation to pot value ?
What is the lowest value or R in relation to pot value ?
What is the highest value of R in relation to pot value ?
May be some confusion between load impedance of a volume pot and source impedance of a volume pot? It is the source impedance which might create common-mode distortion problems in some opamps. Bentsnake may be thinking of load impedance?
Thanks, Mooley. Doing as told, after a bit.
DF96, thanks for commenting, but I can't make head or tail of the situation either-which way. But working on that.
.
What I understand he is saying is that some opamps can add extra distortion due to the source impedance of the signal, especially if the two input pins see different impedances. A volume pot gives a source impedance of between zero and a quarter of the pot track resistance, hence there may need to be a roughly matching impedance on the other pin.
- Status
- Not open for further replies.