Op amp supply Quiz - what sounds best ?

[March Audio]

The problem being sought as being solved, is in one word , mediocrity

Sorry but that statement is utterly meaningless.

Can you outline your technical reasons for thinking a high impedance supply will improve sound quality? Im serious, I just want to hear your theories. There may be basis to them.

Also what tests have you performed to quantify and demonstrate this assertion and demonstrate this apparent "mediocrity"?

If you dont have any I think its pretty clear you have just jumped to some conclusions about the circuit design and cause and effect.

 

[House de Kris]

The thing is, this is not a high impedance supply, but rather a low impedance supply. You need to analyze the circuit from the opamp's point of view. With the 100-220uF cap right on the pins, it is a low impedance source. It is isolated from the system supply by the 100 ohm resistor. The local cap acts as a quiet battery for the opamp, and is trickle charged by the series resistor to the system supply. The opamp's current draw is tiny, compared to the amount of stored energy in that huge cap. It won't waver from current demand changes by the opamp. This is standard operating practices in circuit design. I've even used inductors instead of resistors with a cap before and after the inductor. This is called a pi-filter and has a long history. Move along, nothing new here.

 

[March Audio]

The thing is, this is not a high impedance supply, but rather a low impedance supply. You need to analyze the circuit from the opamp's point of view. With the 100-220uF cap right on the pins, it is a low impedance source. It is isolated from the system supply by the 100 ohm resistor. The local cap acts as a quiet battery for the opamp, and is trickle charged by the series resistor to the system supply. The opamp's current draw is tiny, compared to the amount of stored energy in that huge cap. It won't waver from current demand changes by the opamp. This is standard operating practices in circuit design. I've even used inductors instead of resistors with a cap before and after the inductor. This is called a pi-filter and has a long history. Move along, nothing new here.

Correct, you do need to look at this from the op amps POV.

Its typical practice to use something like a 10uF and 0.1uF cap near as possible to the supply pins. A 100 to 200uF cap will have high ESR in comparison. If you have a quiet supply in the first place (as you should) a resistor is not necessary to isolate. In fact it only provides its own source of voltage modulation due to the varying current flow charging the caps, small though that may be.

My issue with whats going on here is the premise of the original question and the assumptions that went with it. The assumption that what is there is "mediocre" or that doing something different will lead to an audible improvement. Its a bit like op amp rolling.

 

[House de Kris]

In fact it only provides its own source of voltage modulation due to the varying current flow charging the caps, small though that may be.

I apologize for not writing clearly, but there isn't voltage modulation of the supply rails on the device because the caps are so huge. The device plus cap current demand is smoothed out to a near constant level that only has minor fluctuations way below the usable pass band of this application.

 

[JohnYang1997]

Hmmm measurements?

 

[March Audio]

I apologize for not writing clearly, but there isn't voltage modulation of the supply rails on the device because the caps are so huge. The device plus cap current demand is smoothed out to a near constant level that only has minor fluctuations way below the usable pass band of this application.

Size isnt everything . What about their ESR? How close can you get those large caps to the supply pins? What about the inductance/ impedance of the trace from those large caps to the op amp pins?

 

[JohnYang1997]

You only need two 1uF 0603 caps for decoupling and Boom! Under -130dB THD there is!

 

[March Audio]

Hmmm measurements?

Indeed.

 

[March Audio]

You only need two 1uF 0603 caps for decoupling and Boom! Under -130dB THD there is!

Precisely. This is a non problem.

 

[stereo coffee]

Sorry but that statement is utterly meaningless.

Can you outline your technical reasons for thinking a high impedance supply will improve sound quality? Im serious, I just want to hear your theories. There may be basis to them.

Also what tests have you performed to quantify and demonstrate this assertion and demonstrate this apparent "mediocrity"?

If you dont have any I think its pretty clear you have just jumped to some conclusions about the circuit design and cause and effect.

A LM833 as an example has a maximum of 8ma - it is not necessary to supply it with any more.
Hence advantage can be made to supply to its actual requirements as you indicated earlier with a regulator

Try low noise voltage regulator, ie one that also has good rejection at high frequencies (simple ones dont) and follow the op amp manufacturers recommendations WRT decoupling/bypass caps

Which needs to be a Vref/ R type ie a current regulator, rather than a voltage implementation.

However back to the Quiz I will accept your answer, of lets say a LM317 and resistor and a LM337 and a resistor, as being quite a good answer
if implemented as Vref/R , which admirably demonstrates its better than the manufacturers 100 ohm effort ( pioneer PDS507 ) , because with Vref/R we can tailor to the op amps actual requirements ie 8ma - and there is then two bandgap devices, replacing passive resistance in our improved circuit. .

 

[BDWoody]

which admirably demonstrates its better than the manufacturers 100 ohm effort ( pioneer PDS507 ) , because with Vref/R we can tailor to the op amps actual requirements ie 8ma - and there is then two bandgap devices, replacing passive resistance in our improved circuit. .

Which improves what in terms of sound...?

 

[JohnYang1997]

A LM833 as an example has a maximum of 8ma - it is not necessary to supply it with any more.
Hence advantage can be made to supply to its actual requirements as you indicated earlier with a regulator



Which needs to be a Vref/ R type ie a current regulator, rather than a voltage implementation.

However back to the Quiz I will accept your answer, of lets say a LM317 and resistor and a LM337 and a resistor, as being quite a good answer
if implemented as Vref/R , which admirably demonstrates its better than the manufacturers 100 ohm effort ( pioneer PDS507 ) , because with Vref/R we can tailor to the op amps actual requirements ie 8ma - and there is then two bandgap devices, replacing passive resistance in our improved circuit. .

So the power supply is not regulated in the first place? Oh well.

 

[March Audio]

A LM833 as an example has a maximum of 8ma - it is not necessary to supply it with any more.
Hence advantage can be made to supply to its actual requirements as you indicated earlier with a regulator



Which needs to be a Vref/ R type ie a current regulator, rather than a voltage implementation.

However back to the Quiz I will accept your answer, of lets say a LM317 and resistor and a LM337 and a resistor, as being quite a good answer
if implemented as Vref/R , which admirably demonstrates its better than the manufacturers 100 ohm effort ( pioneer PDS507 ) , because with Vref/R we can tailor to the op amps actual requirements ie 8ma - and there is then two bandgap devices, replacing passive resistance in our improved circuit. .

OK, now calculate the voltage drop across a 100 ohm resistor with 8mA flowing through it.

V=IR

0.008 * 100 = 0.8 Volts

So assuming your op amps current draw is varying with the output music signal and actually reaching 8mA (it wont in reality as the load impedance is far higher than implied here) you will have a supply varying 0.8 volts with the music. Whilst a cap will smooth this out to a degree, do you think it is a good starting place to have a wibbly wobbly supply??

 

[stereo coffee]

So the power supply is not regulated in the first place? Oh well.

Not locally to the op amp devices as a needed current regulation - which their current maximum dictates from data sheets as being needed which I am suggesting should be catered for, or at least examined.

 

[March Audio]

However back to the Quiz I will accept your answer, of lets say a LM317 and resistor and a LM337 and a resistor, as being quite a good answer
if implemented as Vref/R , which admirably demonstrates its better than the manufacturers 100 ohm effort ( pioneer PDS507 ) , because with Vref/R we can tailor to the op amps actual requirements ie 8ma - and there is then two bandgap devices, replacing passive resistance in our improved circuit. .

yep, you mean use a regulated supply. I dont think there is anything new here.

 

[stereo coffee]

OK, now calculate the voltage drop across a 100 ohm resistor with 8mA flowing through it.

V=IR

0.008 * 100 = 0.8 Volts

So assuming your op amps current draw is varying with the output music signal and actually reaching 8mA (it wont in reality as the load impedance is far higher than implied here) you will have a supply varying 0.8 volts. Whilst a cap will smooth this out to a degree, do you think it is a good starting place to have a wibbly wobbly supply??

Once again quite the opposite a vref/r is needed to stabilise and set current maximum. You are presently at the top of the list in the Quiz in suggesting (with quite a bit of help from me ) a Vref/R implementation in both rails.

 

[March Audio]

However back to the Quiz I will accept your answer, of lets say a LM317 and resistor and a LM337 and a resistor, as being quite a good answer
.

However this does not automatically equate to an audible improvement in sound quality. Op amps will have a PSRR which varies from design to design but may typically be 80dB or more. This will reduce the effect of any PSU fluctuation. So, back to my earlier questions, have you quantified the problem at hand? What is the effect on the output of the op amp of the allegedly "mediocre" resistor/cap input?

 

[March Audio]

Once again quite the opposite a vref/r is needed to stabilise and set current maximum. You are presently at the top of the list in the Quiz in suggesting (with quite a bit of help from me ) a Vref/R implementation in both rails.

You have lost me. Why do you want to limit current?

 

[House de Kris]

Size isnt everything . What about their ESR? How close can you get those large caps to the supply pins? What about the inductance/ impedance of the trace from those large caps to the op amp pins?

All valid comments. 100-220uF is truly huge in value. Personally, if I was using such a large value, I'd put 1uF cap in parallel with it. But, remember, this is Marantz and Pioneer designs. Pretty much the kings of "good enough" circuit design. ESR is also a concern, but why assume it's excessively large? I seem to recall low-ESR caps in the 220 or 330uF range available in the SMT D size package. You could place this right next to the part, have only a couple millimeters of fat trace, with a 1uF on the back side right on the pins. Nothing is impossible, nor hard in this case. Just good circuit design practice. And yes, I have measured improvements in performance from this age old technique. Parts aren't just thrown into a design willie-nillie, everything must be justified by measurable improvements.

 

[March Audio]

All valid comments. 100-220uF is truly huge in value. Personally, if I was using such a large value, I'd put 1uF cap in parallel with it. But, remember, this is Marantz and Pioneer designs. Pretty much the kings of "good enough" circuit design. ESR is also a concern, but why assume it's excessively large? I seem to recall low-ESR caps in the 220 or 330uF range available in the SMT D size package. You could place this right next to the part, have only a couple millimeters of fat trace, with a 1uF on the back side right on the pins. Nothing is impossible, nor hard in this case. Just good circuit design practice. And yes, I have measured improvements in performance from this age old technique. Parts aren't just thrown into a design willie-nillie, everything must be justified by measurable improvements.

Its excessively large because its just not required in the context of this circuit. 10 + 0.1 uF in 0603 packages is all thats required in this context. These will work better as they will be lower ESR and you can get them right next to the pins. This is pretty standard advice from op amp manufacturers.