Single-stage active RIAA correction with second- or third-order Butterworth high-pass included

[arjen6t8]

I have built the design in post #158. Because of size constraints, I have recycled the PCB from a commercial Philips amplifier I had kept. The schematic resembled the design in post #158 well enough for that. I just had to piggy-back R1, R2, R10 & R11. Below is the schematic as I have implemented it, the original Philips design and a few photo's of my project. The issue is that measurements using my inverse RIAA circuit shows that the frequency response starts to rise after 10 kHz (on both channels, L + R). To rule out that this wasn't a measurement error, I've measured an old Pioneer A-110 amplifier that I have lying around. This does not show the rise in amplitude, so my thoughts are it must either have something to do with the design or the PCB layout. Since the issue is on both L & R channels, I am not thinking that it is a wiring issue (for now).

All Inverse RIAA plots have been normalized to 0 dB at 1 kHz.
The direct measurement (input to output) have not. There it can be seen that the amplification is 30 dB @ 1 kHz, which was the design criterion.

Any thoughts on this issue are welcomed.

Schematic as built

Original Philips design

PCB top side

PCB bottom side

Inverse RIAA plot normalized

Bode plot showing gain @ 1 kHz

Pioneer A-110 RIAA Amp circuit

Pioneer A-110 Inverse RIAA plot normalized

 

[MarcelvdG]

R₄ is too small, 120 ohm instead of 300 ohm...332 ohm.

The problem with single-loop active RIAA correction amplifiers with a low gain is that because the gain drops to 1 instead of 0, you get some extra treble (the ultrasonic zero, in transfer function jargon). You can correct for that with a first-order low-pass after the amplifier. The +3 dB point lies at approximately 2122 Hz times the gain at 1 kHz, about 67 kHz in your case. It's higher in your measurement because with the too low R₄, you still partly correct for it.

 

[arjen6t8]

Thanks for the response, Marcel! When we discussed R4 in a previous post, I had already soldered in a 120 Ohm resistor, and was too lazy to replace it (also because it would be a small pain to replace it). I'll make the change tomorrow! The HPF works nicely. I hope I can get away with it being single supply, as I read in an Analog Devices application note, that this set-up reduces the PSRR to 6 dB. If I pick up 50 or 100 Hz hum, I will convert both this amplifier and my turntable to a dual supply.

 

[MarcelvdG]

R1 and C1 are there to improve the supply rejection.

As R5 has ended up on the other side of R4 compared to my design, slightly reducing the resistance driving the 6.8 nF, it's probably best to use R4 = 330 ohm.

 

[arjen6t8]

I have made the change with the outcome as predicted. Next, I will hook it up to my turntable provisionally, before building it in.

Corrected schematic

Normalized Bode plot

 

[MarcelvdG]

I've written a document about designing this type of phono preamplifier, see the attachment. Section 2 details the method I have used so far, which requires fine-tuning with a pole-zero extraction program (or an ordinary simulator or measurements). A more accurate method that shouldn't require fine-tuning is described in section 3. I'm not yet convinced that the more accurate method is practically usable, but for anyone who should want to try it, see this zip file which contains the document and a spreadsheet that does the most complicated calculations: https://www.diyaudio.com/community/attachments/sriaa-zip.1452225/

 

[kgrlee]

I'm not sure MM RIAA with a single 15/18V rail is a good idea. You ju..ust have about 10dBU before O/L so it's difficult to allow for higher output cartridges which might give 10mV at 5 cm/s.

IIRC, allowing for adequate O/L at 1kHz is OK for the rest of the frequency range. Anyone have a link to the SHURE chart for max recorded velocities and their other suggested limits for vinyl?

 

[rayma]

See pages 21, 22, and 23.
https://content-files.shure.com/Kno...honograph-cartridge-technical-seminar-faq.pdf

 

[MarcelvdG]

The midband gain of the circuit of post #165 is about 30 dB, so assuming that the op-amp output can go up to 3 V from the rails, the overload margin is about the same as that of a 40 dB midband gain amplifier with a +/- 17 V or a single 34 V supply.

 

[kgrlee]

See pages 21, 22, and 23.
https://content-files.shure.com/Kno...honograph-cartridge-technical-seminar-faq.pdf

Thanks for this rayma. The SHURE chart shows a max peak velocity of ju ... ust below 70 cm/s which is 50 cm/s rms. The most sensitive cartridges are about 2mV per cm/s which would give 100mV. A gain of 30dB gives 3.162V rms which is 4.47Vp. Adding 3V gives 7.47V which is ju.uust under 15V single supply.

Hoevever the peak 70 cm/s is around 7kHz where RIAA EQ is some 10dB below 1kHz.. But da Golden Pinnae designs with passive EQ should beware.

So Marcel is right that a single 15V supply is ample (10dB margin) for '30dB gain'. I am suitably reprimanded. My excuse is designing RIAA preamps with 2 or 3 BJTs in da Jurassic times when THD would be excessive well before the preamp clipped.

 

[arjen6t8]

@MarcelvdG

If I want to change my amp's roll-off to 3rd order, are the below values for R0, R1, R2, R6, R7 and C2 correct?

I followed the formulae in your document.

Rcart = 1k3 (my carts spec)

Fsub = 16 Hz, C5 = 10 nF -> L = 1/(4pi^2 * Fsub^2 * C5) = 9895,2 H

R7 = SQR(L/C5) = 994 k -> 1 Mohm

C2 = (1/2*pi*Fsub) / (R6+Rcart//R0) -> R6 = (1/2*pi*Fsub)/C2 - Rcart//R0
Since R0 >> Rcart -> Rcart//R0 = Rcart

Take C2 = 150 nF -> R6 = (1/2*pi*16)/150nF - 1k3 = 65 kohm

R6 consists of R6//(R1+R2) ->
R6 = 130 kohm
R2 = 124 kohm
R1 = 6k04 ohm

 

[MarcelvdG]

In your last equation, R6//(R1+R2) should be R6//R2, as R1 is practically shorted by C1 for frequencies of the order of 16 Hz. The equations are only approximate, so I will run the values through LINDA.

The PSRR at 100 Hz would become somewhat worse with the smaller C2, but you made the time constant R1 C1 larger, which improves it again.

Did you manage to solve your muting problem? Do you have problems with subsonics with the second-order version?

 

[arjen6t8]

Thanks, I understand. The muting problem is largely fixed. I have incorporated a relay at the output which shorts the signal at power-down and has a delay when powering up. I can only hear a slight click at power down of the turntable at high volume levels. But it is one of the reasons for the following:
I don't have subsonic issues, but I might build a second version, and since it does not require any more parts, I thought, why not. I am planning on removing the RIAA pre-amp parts from my integrated amp and replacing all parts with your design, so I will have a fully integrated amp again with your RIAA pre-amp instead of the factory design (which only has a first order HPF at some very low frequency). I am also planning to put a provision for C4 externally at the back of the amp, so I can play with the termination. I am hoping that I will get rid of some of the 50 Hz noise I am still seeing with my current build and implementation. The new design will have 2 x 15,5 V~ supply, no single supply, so that would make R6 = 64k9 Ohm (and R0 = 169 kOhm). I was a little lazy, so I modified my existing schematic, instead of taking one of the dual supply ones.

 

[arjen6t8]

PS - I have attenuated the gain of the RIAA preamp with 11 dB with a resistive divider at the input jacks of the integrated amp, because I still found it too sensitive compared to my CD and tuner input channels. What would be better noise wise, reducing the gain of the RIAA amp by 11 dB to around 20 dB or keeping the gain of the RIAA amp at 30 dB and attenuating 10/11 dB at the output? R12 would increase. C5 would become unpractically large, for one, at 100 nF (it would require 10 x 10 nF 5% caps for accuracy, or a hand selected 100 nF 5% part). The other values would require me going through it first. Perhaps there would be other issues.

 

[MarcelvdG]

Thanks, I understand. The muting problem is largely fixed. I have incorporated a relay at the output which shorts the signal at power-down and has a delay when powering up. I can only hear a slight click at power down of the turntable at high volume levels.

I meant the problem that one channel was muted permanently due to some malfunction, like you wrote on another thread a couple of weeks ago.

But it is one of the reasons for the following:
I don't have subsonic issues, but I might build a second version, and since it does not require any more parts, I thought, why not.

The higher the order of a Butterworth subsonic filter, the more suppression of subsonics and the flatter the response in the passband, but also the more phase shift you get.

I am planning on removing the RIAA pre-amp parts from my integrated amp and replacing all parts with your design, so I will have a fully integrated amp again with your RIAA pre-amp instead of the factory design (which only has a first order HPF at some very low frequency). I am also planning to put a provision for C4 externally at the back of the amp, so I can play with the termination. I am hoping that I will get rid of some of the 50 Hz noise I am still seeing with my current build and implementation. The new design will have 2 x 15,5 V~ supply, no single supply, so that would make R6 = 64k9 Ohm (and R0 = 169 kOhm). I was a little lazy, so I modified my existing schematic, instead of taking one of the dual supply ones.

I've run your values through LINDA. My simplified equations always underestimate the required R₇, as I neglect a bunch of damping terms. R₇ = 1.1 Mohm works better than 1 Mohm.

Besides, if you want to be perfectionistic about it, C₃ can best be reduced to 10 uF and R₅ changed to whatever value results in 7334.6 ohm when you connect the amplifier input impedance in parallel with it. The time constant of the output AC coupling then becomes 76.646 ms, which is what you need to compensate for a zero that should be in the origin but isn't. (In the second-order version, that's done with the input AC coupling.)

 

[MarcelvdG]

PS - I have attenuated the gain of the RIAA preamp with 11 dB with a resistive divider at the input jacks of the integrated amp, because I still found it too sensitive compared to my CD and tuner input channels.

How much do you attenuate those channels? I'm used to the phono input sounding softer than CD even with 42 dB of midband gain, when the CD is not attenuated.

What would be better noise wise, reducing the gain of the RIAA amp by 11 dB to around 20 dB or keeping the gain of the RIAA amp at 30 dB and attenuating 10/11 dB at the output? R12 would increase.

Attenuating at the output, but I'm surprised you need such low gain. I've tried to keep R₁₂ below 1 kohm to keep its noise negligible, when you can't do that anymore, the noise will get a bit worse.

 

[arjen6t8]

Thanks for your feedback!

The muting issue I solved. Too much solder on a joint creating a short circuit on the output of my muting relay.

The input sensitivity of my line inputs of my amp is 100 mV. I have attennuated my CD input by 20 dB. I have a few albums both on CD and LP. So I made the record sound as loud as the CD. This is what it turned out to. For me this gives a sensible sensitivity on the volume control.

 

[arjen6t8]

@MarcelvdG
At the moment the RIAA amp (2nd order HPF) is installed in the turntable. As I've mentioned before, I have attenuated the output by -11 dB in the RCA plug at the power amp's input. See the first schematic of what the complete set-up looks like.

Now I am looking at integrating the RIAA amp (as 3rd order HPF) into my integrated amp. I would like to use R4/R5 to drop the sensitivity by 11 dB. See the second schematic of what this would look like. The order of C3, R4, R5 and C9 is as they are on the PCB of my integrated amp.

I could use R4 = 22 k and R5 = 11 k again to drop the sensitivity, but then I run into issues with correcting for the zero at 2122 Hz * A (@1 kHz) = 2122 Hz * 33.1 = 70.2 kHz. C9 would become 100 pF (1/(2*pi*f*R4)) and that is hard to manage since I don't know C in2 of my amp.

I could lower the combination of R4/R5. If I would drop the total load seen by the Opamp to around 6 kOhm (10 times the minimum load of 600 Ohm),
then R4 = 4k2 and R5 = 1k8. Then C9 would become 540 pF. If I allow around 50-100 pF for C in2, then I could use 470 pF for C6 and expect some accuracy for correcting the zero at 70.2 kHz.

My question is what I can (or should) do with C3. It has to do with correcting the HPF around 1 Hz and I have read it is not critical. Since I have C in1 in my amp, can I not just remove it? What is wise?

Current situation. RIAA amp in turntable. Attenuation in RCA plug

Proposed new situation. RIAA amp integrated in amp.

 

[Nick Sukhov]

Since I have C in1 in my amp, can I not just remove it? What is wise?

delete electrolytics C3 + Cin1 and use DCservo @ your poweramp input

 

[MarcelvdG]

You can remove C3.

Regarding the treble correction, according to post #150, in Linda, the extra zero ends up at -442140 rad/s, so the extra first-order low-pass filter has to have a time constant of 2.2617 us.