Solid state phono preamp design philosophy

[analog_sa]

Originally posted by gain

- active/passive equalization
passive. less active components is less possibility for noise and less complexity.

How does passive translate into less active components? Or less noise?

[x-pro]

I've done quite a few phono stages, both MM and MC, and from 1994 these always were:

single-ended amplification

passive EQ

local feedback only

mostly BJT, some with FETs on the input.

Now for MC I would probably prefer to use MM stage with a good transformer, thought in MC version for Creek (OBH-9/9SE) I've used an additional single transistor x10 stage with good results.

Alex

[SY]

http://www.diyaudio.com/forums/showt...80#post1495980



GR, I think you're a smart enough guy to make a technical point without being personally abusive.

[PMA]

So, we need some 60dB gain at 1kHz for average MC cartridge. We would probably split this gain into two stages, regardless active or passive equalization. The first amplifying stage might be linear, and MUST be very low noise. This 1st stage would be supplied from cartridge with voltage quite proportional to velocity, i.e. to frequency. In case of LP scratches, we would get fast signal with the highest voltage amplitude at the preamp input. We would probably like to have smallest overload and very fast recovery, would not we?

[gain]

Originally posted by analog_sa
How does passive translate into less active components? Or less noise?

ummm, i maybe missing something here, but to the best of my understanding, passive electronic circuits, by definition, contain only passive components (such as resistors and capacitors) and do not contain any active components (such as transistors or diodes). because of this, passive circuits can only attenuate, not amplify, the signal, and therefore the potential of introducing noise and distortion in passive circuits is less than in active circuits.

[PMA]

Well, by passive equalization we mean the inverse RIAA circuit (passive R,C elements) not included in a feedback loop. Or we may divide the inverse RIAA circuit into 2 sections and use both passive, or 1st active and 2nd passive. Active we mean in a FB loop. You can google many examples.

[myhrrhleine]

Solid state phono preamp design philosophy Post #1
Gentlemen,

I would like to open a discussion on this topic. MC phono preamp, solid-state.

- active/passive equalization
- discrete/opamp
- BJT/JFET, which position
- power supply, active PSU filter
- enclosure, shielding properties
- wiring, grounding
- SE/balanced

Hi,
my $0.02

- active/passive equalization

This is only an issue when G NFB is an issue passive means no NFB needed.


- discrete/opamp

Op-amps can do many things discretes can't. They are also simpler to build- fewer parts/less labor.


- BJT/JFET, which position

JFET (IMO) work best for inputs.

- power supply, active PSU filter

I think this depends mostly on your overall topology and PSRR.


- enclosure, shielding properties

As EM pollution increases, this becomes more of an issue...
bluetooth, wifi, cellphones...yuk!

- wiring, grounding

point-to-point and circuit boards both have their place.
PTP allows further wiring space than does pcb

- SE/balanced

I prefer balanced, but other choices (above) may dictate your options

[Onvinyl]

Hi Pavel,
good topic.
My choices and my reasoning:

first stage balanced and with good cmrr. The most elegant way to avoid hum. My main problem is, that the best sounding input stage I have tried so far isn't easily balanced (and not overly quiet).

Discrete Design: Better sound, lower noise, more open loop linearity possible.

Open loop design, means no NFB. I don't need moral wars on this issue, it's just my eyperience that t can sound great.

Passive RIAA. -> open loop possible.

BJT/FET: I would chose what sounds best...

PSU: local open loop shunt regs

Rüdiger

[lumanauw]

Example of active/passive RIAA filter can be read on AN-1651 from National Semiconductor

[maney]

Originally posted by PMA
I would like to open a discussion on this topic. MC phono preamp, solid-state.

Oooh, yes, that's a topic I've been thinking, doing some simming, and a few bits of breadboarding about, of late.

- active/passive equalization
- discrete/opamp
- SE/balanced

These three have interesting interactions with one of the notions I've found myself returning to again and again: to employ overall loop feedback, or to sweat blood to make the circuit linear enough it can live without it. If no overall feedback, then it's passive EQ with discrete gain stages - not really any choice there. And working feedback into a balanced circuit can be a bit tricky - you no longer have an input with nothing better to do than accept feedback, at least for the usual long-tailed pair design. Conversely, if you want active EQ then you're already choosing to use feedback; or if you want the convenience of ICs... well, there are ways to "misuse" some of them, such as Jan's 844-based HEC output stage, or (if I've read between the lines correctly) the use by Ayre in a DAC i-to-v stage. But for the most part using ICs means using feedback of the overall loop sort.

- BJT/JFET, which position
- power supply, active PSU filter
- enclosure, shielding properties
- wiring, grounding

I see all these as being - oh, not secondary, exactly, but more constained by the choices above, or by the requirements those choices give rise to. Which isn't to say that you might choose to fix one or more of these, and find that it forces choices in the first group, just that I'm not much inclined to work it that way. But I'm not inclined to dismiss either bipolar or FET for any circuit role, even if I do have general preferences.

So my current crazy ideas run to an undegenerated bipolar input that uses ten identical devices not for the usual brute-force NF reduction, but to linearize the open loop. Turns out that 4:1 is nearly the exact ratio for a two-pair multi-tanh cell. Did I mention that I own only low-output MC cartridges? :-) Balanced connection to the cartridge, of course - that's an excuse to rewire things. The rest of the stage would be a folded cascode loaded by a current source and passive impedances to set the gain... and one of the poles. For the second stage input I lean towards an FET pair, and of course another folded cascode with surrent source and RC network, and from there a diamond buffer to drive the outputs...

Oh, and of course an overall DC servo loop to keep things stable, though the way I'm thinking that will control the DC balances by the current source on that first stage's output.

But it's all subject to change, and most of it has changed as my thinking (and sims, and even a bit of experimentation) have pushed on it. Maybe someday I'll decide it's not going to change for a while and actually build a test specimen. I think I'm going to throw some version of Hiraga's 20W, which I either never saw before or have forogtten about, together first. It should take less time and fills a more immediately felt need.