Sonic questions about self built RIAA print

[Spooter]

Together with a friend I am busy building a phono board (he does the technical side , I do the ‘listening’ part) .

This print is built into in a control amplifier. Technical measuring of this print is quite right (if we hang it on a scoop with signal generator), but at the moment we have this board and power supply built in into the preamp, lower frequencies are more present than the rest of the sound spectrum and it sounds veiled (does not sound very open).

Other sources connected to the same control amplifier (eg CD player) and the same set of speakers sound good though. Nice open sound and a balanced sound across the whole frequency range.

The turntable connected to another (integrated) amplifier, which has a built-in phono stage, with the same set of speakers sounds fine too.

It is difficult to communicate about sound, but does anyone know what the above sound experience can explain (ie the closed sound and the emphasis on lower frequencies)?

The built phono amplifier is based on Op Amp Applications Handbook (Analog Devices Series) / Topology Considerations for RIAA Phono
Preamplifiers by Walt Jung.

http://www.analog.com/library/analogdialogue/archives/39-05/op_amp_applications_handbook.html

Section 6, pages 6-24

Also attached the technical schematics (and meausurements diagrams) of the RIAA preamp.

 

[Spooter]

Photos of the print and power print

 

[westend]

I think you need to make all things equal in your testing. Your "control amp" may have something to do with the imbalance.

 

[Spooter]

As said, other sources played through the non RIAA inputs/normal line inputs, sound okay.

what do you mean by

I think you need to make all things equal in your testing. Your "control amp" may have something to do with the imbalance.

 

[ConradH]

Go to the Hagtech site and build their inverse RIAA circuit. Use that for bench testing. You can also enter the values in Spice and feed your circuit with it, looking for flat output. At the test bench, use the inverse generator and feed the circuit with a 400 Hz square wave. If it isn't square and does funny things on the rise and fall, go fix your RIAA setting values! Also check your circuit for signal overload. IMO, evaluating RIAA circuits by gain, i.e., without an accurate inverse network, is fraught with pitfalls. Only when it passes that test can you go on to listening tests.

 

[dumptruck]

What did you do for the Rt Ct cart loading components? MC or MM? Seems like that's the prime suspect if it measures dead on.

 

[sregor]

10 nF for caps c1 and c13 seems wrong, and would (I believe) kill the treble. Try 100pF or less.
Perhaps the 10 nF would be to change a ceramic to a magnetic input, but that's just a guess.

Edit: Looked through the Jung Handbook. He uses 150 pF for magnetic cartridge, and 10nF for moving coil. (With lower resistance) So unless using a low output moving coil cartridge, need to change the 10nF to 150 pF or lower (Research cartridge loading and cable capacitance if you want to pursue this further)
Good Luck.

FWIW Walter Jung is on my short list of really respected electronics experts (for lack of a better word)

 

[ken kantor]

Together with a friend I am busy building a phono board (he does the technical side , I do the ‘listening’ part) .

This print is built into in a control amplifier. Technical measuring of this print is quite right (if we hang it on a scoop with signal generator), but at the moment we have this board and power supply built in into the preamp, lower frequencies are more present than the rest of the sound spectrum and it sounds veiled (does not sound very open).

Other sources connected to the same control amplifier (eg CD player) and the same set of speakers sound good though. Nice open sound and a balanced sound across the whole frequency range.

The turntable connected to another (integrated) amplifier, which has a built-in phono stage, with the same set of speakers sounds fine too.

It is difficult to communicate about sound, but does anyone know what the above sound experience can explain (ie the closed sound and the emphasis on lower frequencies)?

The built phono amplifier is based on Op Amp Applications Handbook (Analog Devices Series) / Topology Considerations for RIAA Phono
Preamplifiers by Walt Jung.

http://www.analog.com/library/analogdialogue/archives/39-05/op_amp_applications_handbook.html

Section 6, pages 6-24

Also attached the technical schematics (and meausurements diagrams) of the RIAA preamp.

- The 10 nF input cap is the main problem. Try 100pF to start. (The ideal value is trial and error, and depends on the cartridge and cable.)

- Some input RF supression is recommended. At least ferrite beads. Otherwise, you will get rectification problems at the first opamp.

- A polar electrolytic is used for output coupling, but it doesn't appear to be biased.

- That DC servo section could be eliminated with one cap, and might well be causing problems. Seeing as you are comfortable with C6 being in the signal path, why not stick a DC degeneration cap on U3?

- What does your second "Figure 1" depict?

- And some editorial.... it's not clear to me why some designers like to take the RIAA circuit out of the feedback loop. In practice, this makes the opamp work harder, reduces headroom and increases noise. And for zero benefit. But, such is audiophilia...

-k

 

[Spooter]

Some more information about the built print used for a MM cartridge:

http://www.analog.com/library/analogdialogue/archives/39-05/op_amp_applications_handbook.html

R1 and R19 are 47k, as befits a MM cartridge
C1 and C13 are not posted.
The values ​​shown are for a MC element . The print is made ​​so that by placing, and the value of these components can be selected for an MC or MM phono stage.

The RIAA filter comes from reliable source ( Jung ), you can (if you want) calculate it using the formulas from http://www.aes.org/e-lib/browse.cfm?elib=3197
Because capacitors always have a fairly high inaccuracy in value (better than 10 % is hard).
I have recalculated with the values ​​that are in now :
C19 + C27 + C30 = 97nF 3x (Polypropylene , remeasured with LRC meter )
C31 + C33 = 98nF (polypropylene , remeasured with LRC meter , so there is still about 1 % error here)
On that basis, the resistors are slightly adjusted relative ' Jung ' to get the RIAA right again
R26 = 10k
R23 = 750 +47 = 797
This can be further optimized , but it is still an order of magnitude 1 %

C6 and C22 , of course, be omitted in favor of the DC servo. Again, the choice of the " end user " left.

 

[ConradH]

- .... And some editorial.... it's not clear to me why some designers like to take the RIAA circuit out of the feedback loop. In practice, this makes the opamp work harder, reduces headroom and increases noise. And for zero benefit. But, such is audiophilia...

-k

OK, a bit OT, but my feelings exactly, or maybe we're both crazy?

 

[Spooter]

@ken kantor
("And some editorial.... it's not clear to me why some designers like to take the RIAA circuit out of the feedback loop. In practice, this makes the opamp work harder, reduces headroom and increases noise. And for zero benefit. But, such is audiophilia..."):

You are not alone, and I agree this circuit has the disadvantage of requiring a 2nd opamp, but there are some valid points to make for passive RIAA equalization:
- the RIAA correction is correct, there is no need to compensate the high frequency zero which results from the minimum gain of 1 of the non-inverting op amp. This would be easy to do though.
- Does passive RIAA compensation really make the op amp work harder? The op amp load is minimum 10k for this passive circuit. When placing the RIAA circuit inside the feedback loop, the load impedance above the 2nd pole is about 75nF (in series with a low ohmic resistor, e.g. 100 Ohm, from the negative input to ground) , which is only 106ohm at 20kHz and falling...... There are few opamps that can deliver rail to rail output for such a load at 20kHz, so there goes your claimed voltage headroom! The point is: current headroom/load impedance is important too.
- the reduced voltage headroom is no real problem. The first stage can take over 400mV for a 10V output. With a 6.8mV Rega Bias cartridge, I do not expect this to happen with any record. All the sources I can find say 10x the nominal velocity is the maximum to be expected.

Besides, I do not understand why the circuit would need 20dB or more voltage headroom.

 

[ken kantor]

OK, a bit OT, but my feelings exactly, or maybe we're both crazy?

Well, we are probably more crazy for hanging out in DIY than for our opinions on RIAA circuits....

Frankly speaking, I think this is one of those areas where the bulk of professional opinion lines up on one side of the issue, while many in the audiophile community have come to believe that an alternative approach has advantages, without being able to say exactly what those advantages are. This leaves the disadvantages clear and well-documented, while the advantages are vague and hard to understand. Of course, there is the usual chorus of, "just listen." But, the suggested listening tests never seem to isolate the circuits in question as the only, or even the primary, variable.

-k

 

[ken kantor]

@ken kantor
("And some editorial.... it's not clear to me why some designers like to take the RIAA circuit out of the feedback loop. In practice, this makes the opamp work harder, reduces headroom and increases noise. And for zero benefit. But, such is audiophilia..."):

You are not alone, and I agree this circuit has the disadvantage of requiring a 2nd opamp, but there are some valid points to make for passive RIAA equalization:
- the RIAA correction is correct, there is no need to compensate the high frequency zero which results from the minimum gain of 1 of the non-inverting op amp. This would be easy to do though.
- Does passive RIAA compensation really make the op amp work harder? The op amp load is minimum 10k for this passive circuit. When placing the RIAA circuit inside the feedback loop, the load impedance above the 2nd pole is about 75nF (in series with a low ohmic resistor, e.g. 100 Ohm, from the negative input to ground) , which is only 106ohm at 20kHz and falling...... There are few opamps that can deliver rail to rail output for such a load at 20kHz, so there goes your claimed voltage headroom! The point is: current headroom/load impedance is important too.
- the reduced voltage headroom is no real problem. The first stage can take over 400mV for a 10V output. With a 6.8mV Rega Bias cartridge, I do not expect this to happen with any record. All the sources I can find say 10x the nominal velocity is the maximum to be expected.

Besides, I do not understand why the circuit would need 20dB or more voltage headroom.

The reason the opamp is "working harder" has to do entirely with loop gain. I wasn't considering impedance, actually. Also, that HF zero problen is not inherent in the topology. Rather, it arises from an improper choice of gain stage specs.

When I was younger, I did quite a bit of RIAA design work. At one point, I flew over to England to assist Stan Curtis with the design of the preamp in the original 640. Apparently, dealers and customers were complaining about distortion problems. It didn't take long for Stan and I to discover that the amp driving the passive network was being over-driven despite careful attention to gain staging. This very much corrolated with my own experience that the frequency response resonances of the cartridge and front-end network, combined with the tendency to over-cut many LP's, led to much higher than expected signal levels in that stage.

I suggested, of course, a new topology. However, Stan was convinced of the benefits of the "passive" approach to EQ, and I was convinced of the wisdom of Stan. So we left things as they were, with the solution being a reduction in the first stage gain. The cost, of course, was noise performance. Anyway, this was many decades ago, so I can't remember specifically how the gain staging was done.

As an aside, my personal favorite design was one I did for NAD in the early 80's that split the various corners between the two gain stages, to buy a bit of optimization. Of course, all this is a bit silly in the big picture, given the gross frequency inaccuracies in the overall cartridge/arm/cable/pre system.

 

[Spooter]

@ken kantor: Thank you for your reaction, I'll forward it to the designer of the print.

And by the original design of the 640 do you mean the Cambridge 640? What it is the difference with the 640 which came on the market?

 

[Spooter]

Reaction from the designer and builder:

"Also, that HF zero problem is not inherent in the topology. Rather, it arises from an improper choice of gain stage specs."
What 'gain stage specs' would you suggest which do not have the HF zero then? An inverting op amp is not an option noise-wise.

Do you think the >60dB open loop gain at 20kHz of the op37 amp is insufficient? The loop gain at 20kHz is >1000 /25 = 40.

More to the point: all this does not explain the observed bass boost.

 

[HypnoToad]

Something to think about, with modern op amps you don't need a complex design, in fact at times the simpler the better.

 

[sregor]

You never stated what the actual input is. Unless it was a low output moving coil cartridge, you will get a major roll off in the highs, or a bassy sound. Even a signal generator with a 600 ohm output would roll of the high end.

 

[HypnoToad]

You never stated what the actual input is. Unless it was a low output moving coil cartridge, you will get a major roll off in the highs, or a bassy sound. Even a signal generator with a 600 ohm output would roll of the high end.

Yes, if his input impedance is indeed 49.9ohms it's only suitable for a LOMC cart.

 

[ken kantor]

@ken kantor: Thank you for your reaction, I'll forward it to the designer of the print.

And by the original design of the 640 do you mean the Cambridge 640? What it is the difference with the 640 which came on the market?

Yes, the Cambridge 640. Not sure what your question is, can you run it by me again? Thanks!

-k

 

[ken kantor]

Reaction from the designer and builder:

"Also, that HF zero problem is not inherent in the topology. Rather, it arises from an improper choice of gain stage specs."
What 'gain stage specs' would you suggest which do not have the HF zero then? An inverting op amp is not an option noise-wise.

Do you think the >60dB open loop gain at 20kHz of the op37 amp is insufficient? The loop gain at 20kHz is >1000 /25 = 40.

More to the point: all this does not explain the observed bass boost.

This is spelled out pretty clearly in Lipschitz's paper. Check it out, and let me know if you want to discuss it further.

To my thinking, for a really world-class, reference-grade preamp, 40 is a bit on the low side for loop gain. I'd like to see at least 40 dB.

I thought the bass problem was traceable to the input loading error, or am I confused?

-k