high end version of low end phono stage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am wondering if anyone has any thoughts on using this basic phono stage design (http://www.geocities.com/rjm003.geo/rjmaudio/diy_pho5.html) but using something like the Borbely discrete opamp. Pricewise, to do this would cost about $150 for a pair of opamps, + at least another $150 for caps worthy of the exercise and a pair of power supplies. Would this design be up to the task, or, for the same money, should I look at something like a bottlehead seduction, or Cary PH-1 kit? (those are both tube designs, I am not sure of other ss phono tages in this price range.)
 
AX tech editor
Joined 2002
Paid Member
This single stage design is inherently limited in concept, and no substitution of parts will get past it. Better designs in terms of distributing the correction tasks over two gain blocks are available and provide lower noise, improved RIAA conformity and lower interaction between xover points. There are many posts on that here on the forum if you are willing to search for them. Walt Jung has also published an article on this issue complete with good arguments for the distributed topology.

Jan Didden
 
i'm not sure you will find agreement with all of those statements jan

noise and dynamic range are "best" with the single stage feedback eq - as long as you're looking at op amp style gain blocks

see thorsten's "el-cheapo" for a finely tuned implementation - even justiifes the high freq zero as being desirable

if you want to correct for the high freq zero then a extra buffer stage may be good - but i would not distribute the riaa constants over the 2 stages
 
Borbely phono

You might consider ordering the 419 phono kit. Mine was about 300.00 USD delivered. It has the discrete gain stages, active hybrid or passive RIAA, and low noise discrete regulators.
The circuit board comes with fairly high grade parts to finish. You still need a raw 29 -31 volt dual rail supply, and chassis.
The gain stages are nice, use 2SK170 and 2SJ74 transistors. There are also servos to keep the dc drift down.
I have not built mine yet. Need to build a box to put it in first. From the layout and design I expect it to be excellent.

George
 
janneman said:
This single stage design is inherently limited in concept, and no substitution of parts will get past it. Better designs in terms of distributing the correction tasks over two gain blocks are available and provide lower noise, improved RIAA conformity and lower interaction between xover points. There are many posts on that here on the forum if you are willing to search for them. Walt Jung has also published an article on this issue complete with good arguments for the distributed topology.


I can't find an article referring to distributed topology in Walt's publication list. In page 28 of Op Amp Applications Ch. 6 he refers to passive EQ having better RIAA curve conformance but I see this as academic given the variability in cartridges and source materials.

I agree with jcx on active EQ (theoretical) superiority for noise and overload performance, though I believe in practice that component and layout choices are more important. As for RIAA conformance, I have verified +/- 0.1dB on a single stage active RIAA with 72dB Avol using single-stage inverse RIAA.

Of course xover points are interactive in single-stage topologies but this does not preclude close conformance if some experimental adjustment is possible, OTOH I don't see where distributed constants are detrimental except perhaps for overload performance when using a first stage with flat gain vs f.

I believe the main advantage of distributed topologies is in cases with low gain stage Avol and high output impedance (resistive loaded discrete), and in convenient calculation of RIAA network values for close conformance.
 
I believe the main advantage of distributed topologies is

Hi Nuvistor

Have you actually built one? It goes against the minimalist approach (unless you integrate the line stage) and is difficult to make fair comparisons but the split riaa sounds significantly better to my ears. Not to mention that it allows to use really high quality caps without having to resort to paralleling.
 
analog_sa said:


Hi Nuvistor

Have you actually built one? It goes against the minimalist approach (unless you integrate the line stage) and is difficult to make fair comparisons but the split riaa sounds significantly better to my ears. Not to mention that it allows to use really high quality caps without having to resort to paralleling.

Fair question. I was responding to janneman in a technical context only.

My only experience with distributed was a version of the Nat Semi AN-346 using 5534s with JFET first stage, built in the 80's, haven't used it in years, 3180/318us active followed by 75us passive. Recent experience has been only with single stage EQ, passive with op amps, active with CCS-loaded tubes, for MM cartridge, my personal preference is the tube stage. I have wanted to build a EAR-834 like topology (flat gain, 75us passive, 3180/318us active) for awhile but have not gotten around to it, I assume this is what you are thinking of as "split RIAA".
 
Konnichiwa,

janneman said:
This single stage design is inherently limited in concept, and no substitution of parts will get past it. Better designs in terms of distributing the correction tasks over two gain blocks are available and provide lower noise, improved RIAA conformity and lower interaction between xover points.

That is one view.

With my rather ancient mid 90's "Analogue Addicts" Phono I set out to demonstrate the opposite.

Okay, step one. Let us consider a MM Phono using "Op-Amp" structures. (we don't do MC here, a seperate stepup optimised for the job is a better choice).

Is it possible to attain better than 0.1db RIAA accuracy with off the shelf parts, reliably, in a single stage? Of course, easy peasy.

Is there any specific advante to the active single stage design?

Well, most Op-Amp's open loop gain rolls off as 1st order lowpass somewhere between 10 - 100Hz. The RIAA may be seen simplified as 50Hz lowpass (ignoring the midrange breakpoints), so the RIAA EQ around a single stage keeps actually the degree of negative feedback around the stage the same within a few db regardless of frequency.

Is there any specific disadvantage to 2-Stage topologies?

Of course, ignoring the extra parts etc., the 2 Stage topology invariably incurs a severe headroom penalty, a larger one if passive EQ is used.

Oh, it sounds much worse too.

I find that FOR OP-AMP structures ONLY, the single stage design as is the AA/ElCheapo and the more recent RJM is a better choice, objectively and subjectively.

I'm normally all down with Walt, but with solid state, limited rail voltages and all else a 46db gain single stage RIAA is the best choice. For MC only the split topology would seem to make more sense but suffers the same disadvantages, except the two stages are now essential ever which way, a good open loop MC Pre-Pre or a good MC Transformer usually outperforms other solutions though.

Sayonara
 
Konnichiwa,

nuvistor said:
I have wanted to build a EAR-834 like topology (flat gain, 75us passive, 3180/318us active) for awhile but have not gotten around to it

Just a note, the EAR is not like you describe, it has a "flat" gainstage outside the NFB loop and a fully active EQ (all three time constants) wrapped around a single ECC83 stage with cathode follower.

Sayonara
 
Konnichiwa,

nuvistor said:
So what are your opinions on the topology I incorrectly described as EAR 834P, specifically, flat gainstage with no NFB, followed by 75us passive, followed by 3180us/318us active?

It depends. When using stuff like ECC83's (poor linearity, poor bandwidth, very high gain) often an active EQ around it is a better choice.

Otherwise I am mainly in favour of split stage of single stage "type b"passive equalisation in valve circuitry, if linearity, output impedance etc. do not cause concern.

I'm not sure if I where to make the 3180/318uS EQ passive or the 75uS, probably the prior, not that latter and probably by varying load with transconductance input stage (no headroom penalty). But if I was I'd probably throw the 75uS back into the mix and be done.

Sayonara
 
Kuei Yang Wang said:
I find that FOR OP-AMP structures ONLY, the single stage design as is the AA/ElCheapo and the more recent RJM is a better choice, objectively and subjectively.

How about a single stage active EQ around two single-ended discrete gain stages? Here is a simple example with resistive loaded stages (L1, R9 represent MM cartridge), mu stages would provide more open-loop gain.
 
Kuei Yang Wang said:
Otherwise I am mainly in favour of split stage of single stage "type b"passive equalisation in valve circuitry, if linearity, output impedance etc. do not cause concern.

I'm not sure if I where to make the 3180/318uS EQ passive or the 75uS, probably the prior, not that latter and probably by varying load with transconductance input stage (no headroom penalty). But if I was I'd probably throw the 75uS back into the mix and be done.

"type b" refers to Jung's 1980 topology paper?

Not quite sure what you mean by the last statement, I take it you prefer 3180/318us passive as part of 1st valve load, followed by 2nd valve with 75us passive?
 
Konnichiwa,

nuvistor said:
"type b" refers to Jung's 1980 topology paper?

Nope, not really. Jungs Type A & Type B are from my view the same, both subject to identical problems and hence lumped by me as "type a passive EQ" (in other words generic since the 1940's).

My "type B" (which in the context of Jungs paper would be "type C") looks like this here:

attachment.php


It frees you from the tyranny of Lipshitz et al and sets the EQ free, at the cost of one extra resistor....

nuvistor said:
Not quite sure what you mean by the last statement, I take it you prefer 3180/318us passive as part of 1st valve load, followed by 2nd valve with 75us passive?

I mean i would likely lump the 3180/318/75uS EQ directly into the anode/drain/collector load of a pure transconductance stage, tus obtaining a naturally equalised output which is "flat" after the first stage.

Sayonara
 
Konnichiwa,

nuvistor said:
How about a single stage active EQ around two single-ended discrete gain stages? Here is a simple example with resistive loaded stages

In such cases it is very much a tossup. What does the interstage slewing look like. Is the open loop bandwidth sufficiently stable with signal, etc. Some stages require looped feedback to operate well, others don't.

By STRICT GUT FEELING I would opt for passive eq in that specific case, but I'd have to do a more complete analysis to be more certain.

Sayonara
 
KYW -

Regarding your "type B", I hadn't really noticed the extra resistor (I presume 48k7 in your schematic), I can see how this would reduce breakpoint interaction. Do you have any maths to share correlating time constants to R/C values? If not I can derive them from your example.

The example I gave with resistive loading has low open loop gain and a bit high output impedance so I agree that passive is a better practical choice. The version I use has about 72dB open loop gain (mu of 80 ^ 2, using 5755 with CCS/mu follower) so it needs loop feedback with frequency compensation. The 2nd stage Cdom Miller capacitance has the slewing burden, slew rates with only the music signal is of no consequence but scratch or pop signals could challenge slew rate. I think that "excessive" slew rate minimizes pop/scratch noise, mine clocks in at about 40V/us, should be plenty.

My generalizing opinion is that passive EQ is better with linear low gain (30 - 40 dB) stages of known and not necessarily low output impedance, active EQ is better with high gain (>60 db) stages of lesser linearity but require low network and gain stage output impedance for best performance, the final choice depends on your tastes and implementation, there is no one best solution for all cases.

I believe that active EQ is regarded as inferior because of implementations with high network and drive impedances and too low quiescent current, Dyna PAS for example. Graham Slee and Nelson Pass probably do it right given their reputation, they use active EQ.
 
Konnichiwa,

nuvistor said:
Regarding your "type B", I hadn't really noticed the extra resistor (I presume 48k7 in your schematic), I can see how this would reduce breakpoint interaction. Do you have any maths to share correlating time constants to R/C values? If not I can derive them from your example.

Well, 31K6//316K = 28k7

28K7 + 48K7 = 77k4

77K4 * 1nF = 77.4uS

The rest is the difference between theory and practice... ;-)

nuvistor said:
My generalizing opinion is that passive EQ is better with linear low gain (30 - 40 dB) stages of known and not necessarily low output impedance, active EQ is better with high gain (>60 db) stages of lesser linearity but require low network and gain stage output impedance for best performance, the final choice depends on your tastes and implementation, there is no one best solution for all cases.

I believe that active EQ is regarded as inferior because of implementations with high network and drive impedances and too low quiescent current, Dyna PAS for example. Graham Slee and Nelson Pass probably do it right given their reputation, they use active EQ.

Yeah, I'm more or less in agreement, high network impedances are not per se bad though.... It's all in the implementation.

Ciao T
 
Konnichiwa,

carlmart said:
What parts could work well in order to get a silicon discrete or IC version to replace the tubes?

Non, most solid state stuff lacks the headroom to use passive EQ well.

Easy calculation, LP nominal = 5cm/S, LP standard maximum = 25cm/S (RMS), so expect normal 14db more output from the cartridge than nominal 5cm/S rating. If you have enough gain your headroom goes unless you go active EQ or push the supply voltages into Valve territorry, where you may as well use valves.

Sayonara
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.