I was reading Electronics for Vinyl (Douglas Self), and revisiting his favorite NE5534A single stage RIAA MM preamp. He mentions that the NE5534A shows a rise in distortion at low frequency due to the high closed-loop gain (meaning there's not enough open-loop gain left to keep distortion down at the device floor).
This made me revisit my RIAA design that uses separate active poles and zeros, realizing each stage only sees a max gain of 32dB or so, whereas a single stage would have about 54dB of gain at low frequency.
What's more there's no limitation on the headroom as each stage has unity gain or higher across the audio spectrum, so no stage can clip without the output clipping anyway.
The extra stage needed for correct output phase I augmented to act as a low-pass roofing filter and gain-setting stage.
A 5 pole rumble filter is available on the output if desired (why wouldn't it be?)
So why do I want to label this "blamefree"?
1) The capacitor values for each filter stage are 1nF, 10nF or 100nF, nothing awkward.
2) The nominal RIAA component values are exact - a 10nF and 31.8k make the nominal 318µs time constant for example - 31.8k can be made from two standard values in series, 30k + 1k8, and the 7.5k is already standard.
3) The input stage can benefit from the low voltage and current noise of the NE5534A without compromising on distortion by asking for too much gain
4) The 47k load resistor is synthesized to reduce its current-noise contribution - this technique is no doubt well known here, giving very good noise performance for typical MM cartridges. The aim is that no hiss is audible after a disc ends and the stylus leaves the groove - even if you've been listening at high volume (its not too difficult to be inaudible against surface noise, but once that ends I still appreciate quiet!)
5) Effective filtering of sub-20Hz noise is provided. A 5-pole Butterworth filter is not overkill here I suggest.
Here's the circuit with annotations about noise levels (in theory!).
Thoughts?
This made me revisit my RIAA design that uses separate active poles and zeros, realizing each stage only sees a max gain of 32dB or so, whereas a single stage would have about 54dB of gain at low frequency.
What's more there's no limitation on the headroom as each stage has unity gain or higher across the audio spectrum, so no stage can clip without the output clipping anyway.
The extra stage needed for correct output phase I augmented to act as a low-pass roofing filter and gain-setting stage.
A 5 pole rumble filter is available on the output if desired (why wouldn't it be?)
So why do I want to label this "blamefree"?
1) The capacitor values for each filter stage are 1nF, 10nF or 100nF, nothing awkward.
2) The nominal RIAA component values are exact - a 10nF and 31.8k make the nominal 318µs time constant for example - 31.8k can be made from two standard values in series, 30k + 1k8, and the 7.5k is already standard.
3) The input stage can benefit from the low voltage and current noise of the NE5534A without compromising on distortion by asking for too much gain
4) The 47k load resistor is synthesized to reduce its current-noise contribution - this technique is no doubt well known here, giving very good noise performance for typical MM cartridges. The aim is that no hiss is audible after a disc ends and the stylus leaves the groove - even if you've been listening at high volume (its not too difficult to be inaudible against surface noise, but once that ends I still appreciate quiet!)
5) Effective filtering of sub-20Hz noise is provided. A 5-pole Butterworth filter is not overkill here I suggest.
Here's the circuit with annotations about noise levels (in theory!).
Thoughts?
your arguments regarding your design make sense to me; so I think the designation would be fine.
but i'm biased; i'm perfectly ok with well-done op amp based phono preamps.
i'm sure the anti-op amp brigade will arrive soon to attempt to throttle you into submission.
hopefully, you have the music turned up loud enough to drown them out!
🤣
but i'm biased; i'm perfectly ok with well-done op amp based phono preamps.
i'm sure the anti-op amp brigade will arrive soon to attempt to throttle you into submission.
hopefully, you have the music turned up loud enough to drown them out!
🤣
The responses, green being the inverse-RIAA weighted input, then blue after the first stage, then red after 2nd stage, fully RIAA corrected, then cyan after the gain-setting roofing filter. Finally the rumble filter gives the purple trace that overlies the cyan one except below 20Hz...
Noise analysis, darker green is input-refered noise total, with various resistances traced (output-referenced), and the bright green output-referred total noise. First plot is with actual 47k resistor load, second is with synthesized load. The effect of R18 is clearly less in the second than R10 in the first, and the better overall noise performance can be discerned.
Note this noise analysis goes through the inverse-RIAA network as well as the preamp, and represents total noise of the recording/playback chain assuming perfect recording tech. but this implementation of a premp...
[ BTW the choice of FET opamp is arbitrary here, it was one of the one's in LTspice! Something like the OPA1642 would be suitable I think, and could be used in place of the 5532's too - only the input stage and possibly synthetic load would benefit from the NE5534A [
Last edited:
Just curious, but why didn't you put the 3180 us pole first? You have certainly outdone the old Apt-Holman preamp which only had a 3-pole rumble filter.
Its the only way to avoid headroom compromise - that pole has the biggest variation in gain across the audio spectrum, about 40dB, so it has to come first.
The other pole and zero combine nicely in one stage to only have a 20dB variation
[ Ah, I think I'm wrong about this, the intention is also to have eough gain in the first stage to mean the later stages don't contribute to the noise performance - yes that's it, the HF pole allows the first stage to have a largish gain across most of the spectrum so it can climb out of the noise floor of the first stage... The LF pole wouldn't have gain across the whole spectrum and the zero would clip before the output I think. ]
The other pole and zero combine nicely in one stage to only have a 20dB variation
[ Ah, I think I'm wrong about this, the intention is also to have eough gain in the first stage to mean the later stages don't contribute to the noise performance - yes that's it, the HF pole allows the first stage to have a largish gain across most of the spectrum so it can climb out of the noise floor of the first stage... The LF pole wouldn't have gain across the whole spectrum and the zero would clip before the output I think. ]
Last edited:
I was not aware of the virtual ground technique or separate networks for the 318us zero and 3180us pole. Kudos for that.
My experience is that the 47K noise becomes small once in parallel with a cartridge.
I can blame the first stage for having the +1 error... still not as accurate as passive EQ. 😉
Ed
My experience is that the 47K noise becomes small once in parallel with a cartridge.
I can blame the first stage for having the +1 error... still not as accurate as passive EQ. 😉
Ed
Previous discussions about active cooling and rumble filters.
https://www.diyaudio.com/community/threads/deluxe-phonostage-with-vertical-rumble-filter.290723/
https://www.diyaudio.com/community/threads/deluxe-phonostage-with-vertical-rumble-filter.290723/
Its shunted by the cartridge, yes, but at higher frequencies the cartridge impedance rises a lot, exposing the 47k more and more. For instance the model of a 350mH cartridge will be about 22k reactance at 10kHz, 44k at 20kHz. So most of the high frequency noise is from the input transistors and the 47k resistor.
BTW I've run the noise analysis bypassing the inverse-RIAA step (which I think makes more sense really), plots below.
This represents the noise you'll hear with the stylus out of the groove just sitting there, rather than the noise relative to the programme material. The steep rise with frequency of the hiss is clear.
I replaced the initial NE5534A with a noiseless opamp:
And redid the noise plots with the NE5534A current- and voltage-noise out of the equation.
The two plots, real 47k on top, synthesized below:
The noise hump from the 47k resistor from 4kHz up is more prominent.
And redid the noise plots with the NE5534A current- and voltage-noise out of the equation.
The two plots, real 47k on top, synthesized below:
The noise hump from the 47k resistor from 4kHz up is more prominent.
Okay, that matches my experience. LF noise is greatly reduced with the cartridge in parallel, whereas HF noise is small to begin with due to the EQ.
I see that I need to re-learn Spice. My pre-amp may be the last to have been designed using the noise contours that used to be included on transistor datasheets. 😉
Ed
U3 should oscillate without external compensation capacitor.
You could consider using an ultralow voltage noise, ultrahigh current noise op-amp for U6.
Cartridges tend to have an ESR that increases with frequency.
You could consider using an ultralow voltage noise, ultrahigh current noise op-amp for U6.
Cartridges tend to have an ESR that increases with frequency.
Ah, that's a good call, its not unity gain stable. Well my LTSpice model doesn't spot this, but then again it doesn't let you add an external cap either!
U6 is indeed better with lower voltage noise, so AD797 I guess...
This is a bit of a trip down memory lane for me! I became interested in phono preamps when I discovered my flatmates' moving coil cartridge setup in heavy tonearm sounded so much cleaner than my "State of the Art" lightweight moving magnet Shure V15 IV / damped SME tonearm system.
I tried complicated balanced line input with NE5534AN opamps, the NE5534 being considered a good low level opamp, especially suited to microphones.
I tried varying the input load capacitance and resistance.
I never fixed it. Everything I did made it better in some ways and worse in others.
Just to remind myself here is a simple Rotel RA931 phono and buffer stage, which is a mature design:
The NE5532 doesn't need a compensation capacitor at unity gain.
The WORST problem with moving magnet cartridges is the high inductance. Adding, say, 200pF capacitance to the 47K load creates a resonant circuit that in theory flattens the frequency response. This would not matter with a microphone IMO, but with a stylus a resonance gets the stylus moving in an opposite direction to the applied groove force. Thus you get a sort of mistracking that you can hear as a fuzziness at the very top end. I could never fix it without wrecking the noise performance.
A few MM cartridges like Grado had much lower inductance, and sounded much better at the top, but then needed heavier tonearms and higher tracking force to behave, and thus were nearer moving coil ideas.
One handy tip I did discover is to keep the turntable isolated and well away from loudspeakers. The stylus in the groove is a microphone and picks up sound from the loudspeakers, especially bass, and thus clutters and muddies the sound.
Just my 2 cents.
I tried complicated balanced line input with NE5534AN opamps, the NE5534 being considered a good low level opamp, especially suited to microphones.
I tried varying the input load capacitance and resistance.
I never fixed it. Everything I did made it better in some ways and worse in others.
Just to remind myself here is a simple Rotel RA931 phono and buffer stage, which is a mature design:
The NE5532 doesn't need a compensation capacitor at unity gain.
The WORST problem with moving magnet cartridges is the high inductance. Adding, say, 200pF capacitance to the 47K load creates a resonant circuit that in theory flattens the frequency response. This would not matter with a microphone IMO, but with a stylus a resonance gets the stylus moving in an opposite direction to the applied groove force. Thus you get a sort of mistracking that you can hear as a fuzziness at the very top end. I could never fix it without wrecking the noise performance.
A few MM cartridges like Grado had much lower inductance, and sounded much better at the top, but then needed heavier tonearms and higher tracking force to behave, and thus were nearer moving coil ideas.
One handy tip I did discover is to keep the turntable isolated and well away from loudspeakers. The stylus in the groove is a microphone and picks up sound from the loudspeakers, especially bass, and thus clutters and muddies the sound.
Just my 2 cents.
Phono cartridges have such weak coupling through the mechanical-to-electrical conversion that electrical loading can't really effect the moving parts of the cartridge. High inductance cartridges are of course still very sensitive to capacitive loading, which combines with the moving parts' resonances (especially the vinyl compliance / stylus effective moving mass resonance) to give wildly varying results. Fortunately we grow old and can't hear that stuff anymore, given time.
All good fortune,
Chris
All good fortune,
Chris
This design will make it impossible to upgrade . The inability to apply any other phono preamp ( without changing the design of the turntable ) will be a big disadvantage of this method.
This method increases the effective mass of the tonearm . And it will create additional mechanical resonances of the tonearm . And again - it will be impossible to apply another phono preamp . Universality will be lost .
I mounted the phono preamplifier in the turntable cabinet when I was experimenting, a Thorens TD160 IIRC.
I wasn't too worried about the thin twisted pairs that ran down the tonearm, My telecoms background told me they pretty much work as advertised for a 1 foot run and would be about 300 ohms impedance..
RCA phono leads (around 100 ohms impedance) are slightly more problematic over a metre or so. Impedance is SQRT L/C as it goes, so the inductance is a tiny 1uH if I have done the sums right.
The theory then says they are about 100pF capacitance when terminated in 47K. This disappears when terminated around 100 ohms as in usual Moving Coil practice, which means the load does not reflect the incoming signal.
I had a good time with Grado cartridges, although they need a heavy tonearm to be stable, being low compliance, and are not the greatest trackers.
I have dug up a comparison of a high inductance AT VM 520EB cartridge at 800R DC and 460mH, and a Grado Prestige 3 at 475R and 45mH:
Only watch if you are interested! This is not an advert for Grado cartridges!
Here's the specs, and notice the Grado unsurprisingly has greater bandwidth because it doesn't use the normal 500mH/200pF/47k circuit to get FR about right:
https://www.audio-technica.com/en-gb/vm520eb
https://gradolabs.com/cartridges/prestige-series-2
My interest is that not all high output cartridges are the same on electrical output parameters, and this is reflected in how you get the best out of them.
Just looking at it, I would say the Grado is going to work well with no load capacitance whatsoever! Which makes a case for a turntable mounted preamp.
Sorry for the long post. 😳
I wasn't too worried about the thin twisted pairs that ran down the tonearm, My telecoms background told me they pretty much work as advertised for a 1 foot run and would be about 300 ohms impedance..
RCA phono leads (around 100 ohms impedance) are slightly more problematic over a metre or so. Impedance is SQRT L/C as it goes, so the inductance is a tiny 1uH if I have done the sums right.
The theory then says they are about 100pF capacitance when terminated in 47K. This disappears when terminated around 100 ohms as in usual Moving Coil practice, which means the load does not reflect the incoming signal.
I had a good time with Grado cartridges, although they need a heavy tonearm to be stable, being low compliance, and are not the greatest trackers.
I have dug up a comparison of a high inductance AT VM 520EB cartridge at 800R DC and 460mH, and a Grado Prestige 3 at 475R and 45mH:
Only watch if you are interested! This is not an advert for Grado cartridges!
Here's the specs, and notice the Grado unsurprisingly has greater bandwidth because it doesn't use the normal 500mH/200pF/47k circuit to get FR about right:
https://www.audio-technica.com/en-gb/vm520eb
https://gradolabs.com/cartridges/prestige-series-2
My interest is that not all high output cartridges are the same on electrical output parameters, and this is reflected in how you get the best out of them.
Just looking at it, I would say the Grado is going to work well with no load capacitance whatsoever! Which makes a case for a turntable mounted preamp.
Sorry for the long post. 😳