Another Super High End Phono Stage! No expense spared...

[skrodahl]

I've never said that a resistance in series with the input does not contribute noise.

I'm saying that the 47k input impedance resistance is contributing to the noise, and more so with increasing frequency. That gives a resistor network like this, which is no longer two resistors in series:

I'm also saying arrivederci, at least I managed to renew my subscription this time.

 

[rothwellaudio]

I've never said that a resistance in series with the input does not contribute noise.

My mistake - I apologise.
The Douglas Self book you referenced is very good and goes into some detail about low noise designs. I don't have it in front of me at the moment so can't point you towards any particular page, but I urge you to look again to clarify how resistors affect the noise of a circuit.

 

[EngineerNate]

Reading back through this thread, the stated purpose the 2k2 or 300 ohm input resistor is to limit the current that some opamps drive back through the input load. Whether or not that is necessary with a typical MM cartridge I don't know.

That said, considering that at high frequency the cartridge impedance skyrockets to something near the level of the 47k load resistor and that at that point both the johnson noise and the current noise of the cart + 47k load resistor are significant contributors, the noise from the current limiting resistor is likely not worth discussing. And that's disregarding the fact that many well respected phono stages have similar input stages.

Just to humor the question, I've done the math for the contribution of the johnson noise for the 2k2 variant.

R_1 = 2k2 Ohms (Current limiting resistor value)
T=100F = 310 Kelvin (Resistor temp estimate for room temp + case heating)
delF = 20 Khz (Frequency range of interest)

From here: https://en.wikipedia.org/wiki/Johnson–Nyquist_noise

Using this equation but subbing in our 2k2 resistor:

1. sqrt(v_n^2) = sqrt (4*1.38e-23*310K * 2200) = 6.136 nV/sqrt(Hz) or 0.000000006136 V/sqrt(Hz)

And using our bandwidth of 20khz:

2. v_n = 6.136nV * sqrt(20khz) = 867 nV or 8.67e-7V

Assuming we've set the stage to 40dB of gain, we can expect that to be 100x higher at the output, or

3. 100*8.67e-7 = 8.67e-5V or 0.08mV.

And using the calculator here: http://www.sengpielaudio.com/calculator-gainloss.htm because I'm tired of running numbers, and assuming a 1V reference Vout from the stage, the noise from the resistor, even after amplification by both opamp stages, is at -81.93dB. Considering the absolute *best* you can hope for from the vinyl itself is something like 60dB signal to noise and the best turntables ever made have 80dB or so signal to noise ratios under laboratory conditions, this additional noise is a complete non-issue, even when looked at in isolation. Add in the electrical noise from the load resistor and cartridge and it becomes even more so.

 

[rothwellaudio]

Your calculations appear to be correct, but I'm not so sure about the conclusion that the noise contribution is completely irrelevant. It would be so easy to not have the resistor there and get lower noise (admittedly, not much lower), so why not?
BTW, I did say in post #2872 that the noise contribution isn't significant for an mm phonostage, but try doing the same calculations for an mc phonostage with 60dB or more gain and a cartridge with a 10 ohm source impedance.

 

[EngineerNate]

Your calculations appear to be correct, but I'm not so sure about the conclusion that the noise contribution is completely irrelevant. It would be so easy to not have the resistor there and get lower noise (admittedly, not much lower), so why not?
BTW, I did say in post #2872 that the noise contribution isn't significant for an mm phonostage, but try doing the same calculations for an mc phonostage with 60dB or more gain and a cartridge with a 10 ohm source impedance.

Considering my particular phono stage has roughly 1/7th the input resistor value of that calculated above which results in 89dB below 1mV reference, I'm pretty sure it's irrelevant in my case.

The real question is whether limiting the current through the cartridge is necessary. If there is no reason for the resistor to be there then it makes sense to take it out even if it isn't contributing to the signal in any meaningful way. Simply because eliminating unnecessary components is a good thing. If there's a reasonable risk to the cartridge if it is taken out, it should probably stay.

I don't think the MC argument is relevant for this particular discussion, as the extra gain would be "upstream" from this resistor if you were to use this phono stage for an MC in it's current form. Theoretically you could probably tweak the gain structure and use two opamps to get 60dB of gain but you'd likely be pushing it quite a bit at that point. As is the stage wasn't intended to be run with MC carts without an SUT or head amp... which would leave the gain/noise calcs presented above alone.

Cheers,
Nathan

 

[rothwellaudio]

The real question is whether limiting the current through the cartridge is necessary.

Yes, I agree, that's a valid question. If there was a DC blocking capacitor at the input there would be no chance of bias current flowing through the cartridge (barring capacitor leakage current), but this design eliminates capacitors where possible so bias currents should beconsidered. I don't think bias current through the cartridge would damage it but it might cause the stylus to be off-set from its natural rest position. Would that affect the sound? I wouldn't like to say one way or the other.

I don't think the MC argument is relevant for this particular discussion, as the extra gain would be "upstream" from this resistor if you were to use this phono stage for an MC in it's current form. Theoretically you could probably tweak the gain structure and use two opamps to get 60dB of gain but you'd likely be pushing it quite a bit at that point. As is the stage wasn't intended to be run with MC carts without an SUT or head amp... which would leave the gain/noise calcs presented above alone.

Two op-gain stages (each high gain) is not an uncommon approach for mc phonostages and they look basically the same as this one - they simply have more gain. Whatever Johnson noise the resistor makes gets an extra 20dB gain (or more) for an mc phonostage and it does make a difference.

 

[EngineerNate]

Agreed on all points.

Regarding MC stages, I probably spoke too generally, but my point is that without other modifications the CNC and it's derivatives aren't suitable for direct MC input. Concerns about the input resistance in this particular case based on the requirements of MC carts aren't overly well supported. The stage isn't designed for MC carts.

If someone wants to modify the stage to produced 60dB of gain there is much more than just R1 to be changed, so changing R1 to be suitable for MC carts is part of a much bigger modification and shouldn't be looked at in isolation.

 

[Hamish119]

Reading back through this thread, the stated purpose the 2k2 or 300 ohm input resistor is to limit the current that some opamps drive back through the input load.

You may have read that (erroneous) conclusion here. It appears to be 'guesswork' (it's a reasonable guess, I guess).
The only circumstance where the noise resistor would limit current is under severe fault conditions: i.e. the Op-Amp has suffered a catastrophic internal failure.
Dave explains why...., but knowledge is required.

There were also some raised eyebrows here (post #70).

The noise source modification error appears to have crept in on page 2 of this thread.

...this additional noise is a complete non-issue, ...

"Okay I have been thinking about this while I was away, I noticed in my Rod Elliot P06 and our current CNC that there is a slight rustling sound when you put your ear near the speakers,..."

Could it be Johnson noise generated by a 2k2 resistor?

Post #93 also hints at the source for the deviant noise modification.

Under normal operating circumstances, the series noise generating resistor generates additional/unnecessary noise only. It does not limit current!

Cumulative Noise Contributor.

 

[EngineerNate]

Hamish, do you normally listen to your records with your ear pressed to the speaker? Doing that for very long will very quickly result in you not listening to anything at all. The noise of an input resistor will be especially irrelevant then.

 

[rothwellaudio]

I have no idea why there appears to be so much love for this resistor, so many posts defending it, and such insistence that its noise contribution is somehow benign.

 

[EngineerNate]

No one here is arguing that the input resistor isn't contributing in some way to the noise performance of the phono stage. But at -89dB in my particular case (300 ohms) I'm personally completely unworried about it.

I'll go watch Dave's video at lunch today. He's someone I respect in the greater field of electrical engineering. Maybe the resistor is completely unnecessary. I don't pretend to be the final authority on any of this, but I do contend that noise at -80dB or less is not worth the amount of derision being thrown around in this thread. Neither I nor @skrodahl made the call to add it and in his kits Skrodahl reduced the value to 300 ohms in direct response to noise concerns. I don't know what more can be asked for on that front.

In response to Rothwell's question, I'll just say that I believe it's likely more of a response to the general timbre of the comments being made rather than any particular love for the resistor. It shouldn't be surprising that people hit back when other individuals come into threads swinging.

Numbers have been presented that convinced me (why I ran them in the first place...) that the noise contribution, while present, was minimal. I questioned the current limiting hypothesis as well. I'm personally completely happy to be shown that it's not necessary, but bold fonts and clever abuse of acronyms aren't exactly convincing arguments.

Cheers,
Nathan

 

[rothwellaudio]

I'm personally completely happy to be shown that it's not necessary...

Bias current flows into the op-amp's +ve and -ve inputs. It flows through through the cartridge and through the resistor in question (if it's there). The bias current isn't reduced by the presence of the resistor, so it makes no difference to how much current flows through the cartridge.

 

[ghazzer]

I'd like to remind everyone that Jim's tongue was practically poking thru his cheek when he penned the title for this thread.

This is a simple, inexpensive, versatile, great sounding phono preamp that almost anyone can assemble and enjoy. I think that it is astounding how many AKers (and others)

have built and enjoy our own version(s). At less than a Grant for the basic build it has opened up new vistas and listening pleasure for so many. My hearing has really gone downhill over time but I can still hear significant improvement between a basic Sansui 5000A vs the 5000A with the CNC.

I think that EngineerNate is spot on about how far down in the mud some of these noise contributions really are. The good news is that I will definitely never be able to discern the differences, even with some very hi end headphones.

So, please enjoy picking your nits, but stop throwing dispersions about things that most of the community are completely happy about.

Enjoy - - -​

 

[rothwellaudio]

I think that EngineerNate is spot on about how far down in the mud some of these noise contributions really are. The good news is that I will definitely never be able to discern the differences, even with some very hi end headphones.

So, please enjoy picking your nits, but stop throwing dispersions about things that most of the community are completely happy about.​

​

I apologise if my comments have been taken as "dispersions". I merely pointed out that the noise performance could be improved very easily and for absolutely no money by linking out one resistor. The thread turned into a bit of a bun fight because at least one poster took offence at that and flatly denied the truth of it.

 

[Llewellyn]

Hello, I have only just joined, also recently taught myself how to use eagle CAD.
This is my attempt so far at making the CNC phono stage.
Would love to hear what you guys say about the schematic and if i should change any core components.
I have not yet made my silkscreen layer for all values.
Should i leave the input resistor in and bypass it if i like, or is it completely unnecessary?
I have made all 150R gain resistors 100R, will this be ok gain wise?
I am using this as practical diy experience and also need a phono stage for myself.
I will have the pcb professionally made by a china manufacturer that made some Nelson Pass bride of Zen preamp boards in Cascode configuration for me, that i also pcb designed myself. (Still in the mail)
This design has not been finalized yet and i am open to ideas about the circuit. Thank you very much for having me.

 

[EngineerNate]

Are you adding so much local power filtering because of your chosen power supply setup?

As a note, I spoke with a well respected designer regarding input resistors on opamp phono stages and he told me that they were the simplest way to prevent RF interference and had fewer downsides than more elaborate solutions. He was also of the belief that their contribution to the noise floor was well below that of typical vinyl surface noise.

Cheers,
Nathan

 

[skrodahl]

I'd invert the traces, placing the ground fill on the top layer. Less thermal mass on the soldering side makes it easier to solder.

Also, move the 100 nF ceramics as close to each power pin as possible. Keep the two electrolytic caps closest to the power inputs, skip the rest.

 

[Llewellyn]

Thanks Nathan,
I have many filtering caps because i had the space since i'll be using 1/2 watt dale resistors, I can't see the harm of having more rather than less. I've not decided on the power supply yet but i think i'll use a small teriodal with a bridge rectifier, and make a simple Transistor Series Voltage Regulator for it. The electrolytics are 25v versions, should i go for high voltage and less capacitance?

Thank you Skrodahl, i can see how the 100nf poly caps ill be using are best for noise suppression as close to the power pins as possible.
Is there no benefit to having extra on board capacitance, or is it a draw back, or does it not matter?
As for the ground plane i've not had problems soldering because of it before, except for components coupling to it need extra heat. i have good temp controlled solder stations that make it easy.
I will play around with inversing the circuit also.

 

[EngineerNate]

For your 100nf local decoupling caps I'd go with some C0G (NP0) ceramics.

I don't think there's any harm in adding the extra capacitance if you'll have the room. Capacitors are cheap. But it's likely not buying you any audible improvement.

 

[rothwellaudio]

I've not decided on the power supply yet but i think i'll use a small teriodal with a bridge rectifier, and make a simple Transistor Series Voltage Regulator for it. The electrolytics are 25v versions, should i go for high voltage and less capacitance?

A high value of capacitance may cause the regulators to struggle - but probably not. There's no advantage in having having electrolytics with a much higher voltage rating than is required. About 10% over is plenty in my opinion.
I see lots of phonostages where "the audio circuit" is on one PCB and "the power supply" is on another, with the voltage regulators on the power supply PCB. To me it makes more sense to have the regulators on the audio PCB so they are close to the circuit it they're meant to be supplying regulated voltage to.