Very Simple Phono Stage
Only 27 parts for the whole stereo unit.
User configurable gain 30-50dB.
Standard value components, even for the RIAA eq.
Allen Wright modded RIAA response. Accurate to +/- 0.25 dB.
Use any standard dual opamp.
Layout diagrams and photos included. Easy to build. (For those who previously visited, I have today updated the page to make it clearer and simpler to understand for the neophyte.)
-rjm
Only 27 parts for the whole stereo unit.
User configurable gain 30-50dB.
Standard value components, even for the RIAA eq.
Allen Wright modded RIAA response. Accurate to +/- 0.25 dB.
Use any standard dual opamp.
Layout diagrams and photos included. Easy to build. (For those who previously visited, I have today updated the page to make it clearer and simpler to understand for the neophyte.)
-rjm
Batteries and other PSU options.
Hi Ropie,
As noted, you could use batteries. I've never much cared for the results and that article at TNT does lend some credibility to that little prejudice of mine.
Be that as it may, as for a schematic since anything from 2 9-volt batteries on up through lead-acid cells up to caseloads of magnesium C-cells will work, I felt those who like the sound of battery PSUs are best left to implement their individual favorites.
I expect that while most people will be quite comfortable with the amp circuit, the regulation part might raise eyebrows. Pretty dinky hey? Sure, but it is "sufficient" and sounds surprisingly good - natural & musical. Please try it before experimenting with more conventional solutions.
All roads lead to Japan or some such nonsense. And I'm quite serious about those carbon resistors, too.
-rjm
Hi Ropie,
As noted, you could use batteries. I've never much cared for the results and that article at TNT does lend some credibility to that little prejudice of mine.
Be that as it may, as for a schematic since anything from 2 9-volt batteries on up through lead-acid cells up to caseloads of magnesium C-cells will work, I felt those who like the sound of battery PSUs are best left to implement their individual favorites.
I expect that while most people will be quite comfortable with the amp circuit, the regulation part might raise eyebrows. Pretty dinky hey? Sure, but it is "sufficient" and sounds surprisingly good - natural & musical. Please try it before experimenting with more conventional solutions.
All roads lead to Japan or some such nonsense. And I'm quite serious about those carbon resistors, too.
-rjm
matjans said:
I'm avoiding plunging into a description of the sonics for two reasons: the practical one is that my current system is a bit of a knock-up affair and the speakers especially not really of sufficient bandwidth to make a full evaluation. And then the more acadamic one that the designer isn't really the right person to make an objective review now is he?
For all their faults though the speakers are very coherant and clear in the midrange, with a true to life instrumental timbre even if dynamics are rather compressed.
OK? With that, and remembering that the Black Gates haven't yet been fully broken in, my "golden ears" tell me the following:
RIAA is spot on, though I swear I can hear the Allen Wright mod as a slight increase in clarity over the normal eq curve. Hum is absent, noise is low but I can't get a good feel on this. Distortion is also low. High freq oscillation - assorted opamp nastiness - is mercifully absent. That at least I can tell with certainty.
Top to bottom (remember though I don't get the true extremes with my speakers) it all seems remarkably natural. No part of the music is over-accentuated, there's no overhang or smearing. No grain, no flatness. For any LP I play - and with the VSPS hooked up I'm playing everything I can get my hands on - I run into the limitations of the speakers well before I can detect any problem with the phono stage. Note the limitations of my QED Discsaver are readily apparent under the same circumstances.
Take that as you will - I think its pretty nifty. What's musically important to me it nails authoritatively - yet its dirt cheap and a quick and easy build. Cool!
-rjm
Konnichiwa,
I have a few comments that may or MAY NOT interest you. They are meant constructive, in the interest of improving the design further....
1) For a MM sensitivity Phonostage to be used with a MM Pickup the Op-Amp's current noise dominates the noise unless the Op-Amp is very noisy. This due to the very high inductance of MM Cartidges, up to around 1H which causes material levels of impedance to appear on the input.
With FET input Op-Amp's input current noise is virtually nil, MOST BJT Op-Amp's have a rather high input current noise which will make the stage noisy with MM Pickups. For MC picups (even medium/high output types) a BJT input Op-Amp tends to be apropriate as the source impedance is low and the input noise current matters much less.
Also, MM Cartidges require a correct capacitive load (usually quite high) which forms a resonance circuit with pickups inductance to boost the treble. Usually the best course of action is to shoot for a capacitive load that places the peak at 18-20KHz and to adjust the load resistor untill a pink noise spectrum in the middle of the record shows no HF lift or droop. Then there will usually only be a small amount of HF loss in the inner groves.
2) This is a trick I learned from Allen Wright and have used many times by now. return the AC coupled Feedback loop components (R3/C1 & R4/C2) not directly to the Op-Amp output but to the output after the output coupling cap C3, follow this by the "build out" resistor R6. By insluding the output coupling into the feedback loop it's sonic influence is lowered somewhat and LF RIAA accuracy improves. Sounds notably better wherever I tried.
3) The RIAA capacitors are chose quite high in value. Remember that output stage must supply the current to drive the RIAA network and any external capacitances too. The load at high frequencies from the current values is equal to 1k in series with 3n3, this is quite a notable load which may make many an Op-Amp distort more than neccesary, especially in terms of the various non-harmonic distrotions.
I found scaling the impedances in the NFB Network up by a factor 10 (meaning all resistors * 10 and all capacitors /10), all else being equal sounded better with all the common Op-Amp's I tried. Also smaller value capacitors can often be had in better quality. For fun, try a air dielectric AM Receiver variable tuning capacitor adjusted to 330pF in the HF section of the EQ (they often have two identical sections with excellent tracking making one good for stereo). Very earopening....
The noise is usally dominated by the source and Op-Amp together, so the increase in noise from the higher feedback impedances tends to be minimal.
Combining 2 & 3 can mae a surprisingly large sonic change, which I at least consider a bona fide improvement.
Lastly, if you can find a 4-pin connector for the powersupply consider keeping the "grounds" of both supply lines seperate and only connect them together at the "power star", which would be the point where the output current loop and PSU decoupling capacitor current loop are "starred". I would recommend also seperate ground lines from there to the Input ground with R2 returned to the respective channels input "ground" point.
Anyway, just a few notes.
Sayonara
rjm said:
I have a few comments that may or MAY NOT interest you. They are meant constructive, in the interest of improving the design further....
1) For a MM sensitivity Phonostage to be used with a MM Pickup the Op-Amp's current noise dominates the noise unless the Op-Amp is very noisy. This due to the very high inductance of MM Cartidges, up to around 1H which causes material levels of impedance to appear on the input.
With FET input Op-Amp's input current noise is virtually nil, MOST BJT Op-Amp's have a rather high input current noise which will make the stage noisy with MM Pickups. For MC picups (even medium/high output types) a BJT input Op-Amp tends to be apropriate as the source impedance is low and the input noise current matters much less.
Also, MM Cartidges require a correct capacitive load (usually quite high) which forms a resonance circuit with pickups inductance to boost the treble. Usually the best course of action is to shoot for a capacitive load that places the peak at 18-20KHz and to adjust the load resistor untill a pink noise spectrum in the middle of the record shows no HF lift or droop. Then there will usually only be a small amount of HF loss in the inner groves.
2) This is a trick I learned from Allen Wright and have used many times by now. return the AC coupled Feedback loop components (R3/C1 & R4/C2) not directly to the Op-Amp output but to the output after the output coupling cap C3, follow this by the "build out" resistor R6. By insluding the output coupling into the feedback loop it's sonic influence is lowered somewhat and LF RIAA accuracy improves. Sounds notably better wherever I tried.
3) The RIAA capacitors are chose quite high in value. Remember that output stage must supply the current to drive the RIAA network and any external capacitances too. The load at high frequencies from the current values is equal to 1k in series with 3n3, this is quite a notable load which may make many an Op-Amp distort more than neccesary, especially in terms of the various non-harmonic distrotions.
I found scaling the impedances in the NFB Network up by a factor 10 (meaning all resistors * 10 and all capacitors /10), all else being equal sounded better with all the common Op-Amp's I tried. Also smaller value capacitors can often be had in better quality. For fun, try a air dielectric AM Receiver variable tuning capacitor adjusted to 330pF in the HF section of the EQ (they often have two identical sections with excellent tracking making one good for stereo). Very earopening....
The noise is usally dominated by the source and Op-Amp together, so the increase in noise from the higher feedback impedances tends to be minimal.
Combining 2 & 3 can mae a surprisingly large sonic change, which I at least consider a bona fide improvement.
Lastly, if you can find a 4-pin connector for the powersupply consider keeping the "grounds" of both supply lines seperate and only connect them together at the "power star", which would be the point where the output current loop and PSU decoupling capacitor current loop are "starred". I would recommend also seperate ground lines from there to the Input ground with R2 returned to the respective channels input "ground" point.
Anyway, just a few notes.
Sayonara
Your suggestions here are all very reasonable ones. In the spirit of discussion I’d like to respond to a couple of points.
Impedance of MM cartridges and op-amp noise.
If, indeed, the inductance of the cartridge is in the order of 1 Henry, then your suggestion that a FET input op-amp would likely have lower noise than a BJT input is at higher frequencies generally true and I urge anyone with such a cartridge to consider the OPA2134 etc. over the NE5532 in this circuit.
With a FET input op-amp it might also prove useful to scale the feedback components by 10. With larger impedance it is easier for the op-amp to drive, and with the high source impedance + low current noise op-amp there is no noise advantage to keeping it low anyway.
Since scaling the loop still keeps all the parts as standard values, its easily done.
Not all cartridges have such high impedances however. My daily-use Grado 8MZ has a resistance of 470 ohms and inductance of 45 millihenries. The more expensive models of that series have a resistance of 70 ohms and inductance of only 8 mH. Over most of the audio band the source impedance is below 1k, and best noise performance is obtained with bipolars (low voltage noise) in combination with low values for R2. This configuration will however favor op-amps with low distortion drive into 1k ohm loads - the NE5534 being one of these.
C3 in the FB loop.
That’s also an easy mod and worth trying. The only curiosity I see with doing that is the feedback at DC falls to zero. I actually prefer the sound of those nonpolar 4.7uF Black Gates to a piece of wire (I find it adds a hint of warmth with no loss of transparency) so I'm personally quite comfortable with the conventional position. The low frequency response is 3dB at 4 Hz for 10k load, or -0.25dB at 20 Hz, so likewise the RIAA error it induces isn’t something I’m going to lose sleep over.
Grounding.
Absolutely keeping the input grounds separate until the star point is the correct method. Call me unprincipled but considering the channel separation of the cartridge itself I just can’t get excited over 4 cm of shared input ground.
Separate power grounds to the transformer windings are likewise the best implementation I’ve so far seen. I would have hooked up both my Gainclone and the VSPS that way but for my lazyness: it’s a lot easier for me to come by 3 wire 16-18 gauge power cable than 4 wire.
-rjm
Impedance of MM cartridges and op-amp noise.
If, indeed, the inductance of the cartridge is in the order of 1 Henry, then your suggestion that a FET input op-amp would likely have lower noise than a BJT input is at higher frequencies generally true and I urge anyone with such a cartridge to consider the OPA2134 etc. over the NE5532 in this circuit.
With a FET input op-amp it might also prove useful to scale the feedback components by 10. With larger impedance it is easier for the op-amp to drive, and with the high source impedance + low current noise op-amp there is no noise advantage to keeping it low anyway.
Since scaling the loop still keeps all the parts as standard values, its easily done.
Not all cartridges have such high impedances however. My daily-use Grado 8MZ has a resistance of 470 ohms and inductance of 45 millihenries. The more expensive models of that series have a resistance of 70 ohms and inductance of only 8 mH. Over most of the audio band the source impedance is below 1k, and best noise performance is obtained with bipolars (low voltage noise) in combination with low values for R2. This configuration will however favor op-amps with low distortion drive into 1k ohm loads - the NE5534 being one of these.
C3 in the FB loop.
That’s also an easy mod and worth trying. The only curiosity I see with doing that is the feedback at DC falls to zero. I actually prefer the sound of those nonpolar 4.7uF Black Gates to a piece of wire (I find it adds a hint of warmth with no loss of transparency) so I'm personally quite comfortable with the conventional position. The low frequency response is 3dB at 4 Hz for 10k load, or -0.25dB at 20 Hz, so likewise the RIAA error it induces isn’t something I’m going to lose sleep over.
Grounding.
Absolutely keeping the input grounds separate until the star point is the correct method. Call me unprincipled but considering the channel separation of the cartridge itself I just can’t get excited over 4 cm of shared input ground.
Separate power grounds to the transformer windings are likewise the best implementation I’ve so far seen. I would have hooked up both my Gainclone and the VSPS that way but for my lazyness: it’s a lot easier for me to come by 3 wire 16-18 gauge power cable than 4 wire.
-rjm
Konnichiwa,
Yes, Grado's use a different technology to Moving Magnet Cartridges (they are a "third kind" of cartridge outside the MM/MC division with Decca's being the "fourth kind"). The much more typhical Shure V-15 clocks in at 680mH or 0.68H....
Well, that is a personal choice.
True. Another Mod I had not mentioned but actually highly usefull and somewhat original would be to make the EQ curves adjustble to match also the DECCA & COLUMBIA Stereo EQ curves and possibly CCIR (Europe mainly eastern)....
I understand that, channel separation was never the consideration, rather current loops per se are.
Sayonara
rjm said:
Yes, Grado's use a different technology to Moving Magnet Cartridges (they are a "third kind" of cartridge outside the MM/MC division with Decca's being the "fourth kind"). The much more typhical Shure V-15 clocks in at 680mH or 0.68H....
rjm said:
Well, that is a personal choice.
rjm said:
True. Another Mod I had not mentioned but actually highly usefull and somewhat original would be to make the EQ curves adjustble to match also the DECCA & COLUMBIA Stereo EQ curves and possibly CCIR (Europe mainly eastern)....
rjm said:
I understand that, channel separation was never the consideration, rather current loops per se are.
Sayonara
Checked it out and noticed the trace from the top of R5 to the blue trace connecting R2 to the IC isn't visible. It's "under" the resistor - there are only a couple of pixels peeking out! I'll fix that shortly.
Thinking about T's comments I'll branch the component list early on into a Version A (low impedance, bipolar-op-amp, for Grado-type carts) and a Version B (high impedance, FET op-amp, "typical" MM carts) with the difference being the 10x scaling of the feedback loop.
Since I have to fix the layout for the invisible trace I migth as well modify the grounding too.
-rjm
Thinking about T's comments I'll branch the component list early on into a Version A (low impedance, bipolar-op-amp, for Grado-type carts) and a Version B (high impedance, FET op-amp, "typical" MM carts) with the difference being the 10x scaling of the feedback loop.
Since I have to fix the layout for the invisible trace I migth as well modify the grounding too.
-rjm
Moving iron
Moin T,
there is only one group between MC and MM: the MI = moving iron type. grado and decca share the same principles (some BangOlufsen too); of course decca is mechanical more complicicated. Please check this link for further information (sorry folks, german only and pdf size is >3mb!)
http://mitglied.lycos.de/carawu/hifiscene/Artikel_RLA.pdf
Carsten
Moin T,
there is only one group between MC and MM: the MI = moving iron type. grado and decca share the same principles (some BangOlufsen too); of course decca is mechanical more complicicated. Please check this link for further information (sorry folks, german only and pdf size is >3mb!)
http://mitglied.lycos.de/carawu/hifiscene/Artikel_RLA.pdf
Carsten
VSPS
i have completed your VSPS yersterday and i must say that i am very happy with the results , the sound is very detailed, a bit on the bright side...
i haven't yet decided how to box the unit and whether to leave PSU separated or try to put it in the same box, also is it absolutely neccesary to put it in a box made of metal (to provide case ground) or could the box somehow be made of wood?
btw. the parts have costed me about 15 euro!
once again thanks for the very simple (cheap) project
i have completed your VSPS yersterday and i must say that i am very happy with the results , the sound is very detailed, a bit on the bright side...
i haven't yet decided how to box the unit and whether to leave PSU separated or try to put it in the same box, also is it absolutely neccesary to put it in a box made of metal (to provide case ground) or could the box somehow be made of wood?
btw. the parts have costed me about 15 euro!
once again thanks for the very simple (cheap) project
Usually the turntable has to be connected to the circuit common. This is normally done at the phono pre through attaching to a metal case which is itself connected to the circuit common.
But you can attach the TT ground wire to the amplifier ground if you prefer, or directly to the circuit common of the VSPS. So, no, you don't have to have a metal case though I would recommend it for the shielding it provides.
For the brightness, there are two points: 1) the Allen Wright RIAA curve is "lighter" (brighter?) than the traditional RIAA. 2) the Black Gate output caps (if you are using them) take about 3 weeks to settle in. For the first few days they sound hyper-loud and flashy (bright?), then they turn abruptly stuffy/cloudy before they finally begin to clear up and the bass fills in. Week 3-4 is the best because each day the system sounds noticeably better than the day before.
I've been through this about three times now and it is highly unsettling each time. I should really break in the caps before listening to them but each time I tell myself I can't be bothered...
So I'd give it about a month before passing a final verdict. You can in the meantime try shorting over R3 with a paper clip or suchlike to defeat the Allen Wright treble rise.
If you build the "HiZ" version its very easy to just wire over R3.
If with the LoZ version you short over R3 maybe the brightness gets worse, this is actually even better result because it proves that its the NE5532 driving low impedance load rather than the AW curve thats at fault. Solution is to switch to HiZ feedback loop.
I'm still investigating the sound I'm getting "as built" and I haven't got around yet to tweaking even the obvious things like switching opamps.
Hopefully you'll keep us updated as well.
-rjm
But you can attach the TT ground wire to the amplifier ground if you prefer, or directly to the circuit common of the VSPS. So, no, you don't have to have a metal case though I would recommend it for the shielding it provides.
For the brightness, there are two points: 1) the Allen Wright RIAA curve is "lighter" (brighter?) than the traditional RIAA. 2) the Black Gate output caps (if you are using them) take about 3 weeks to settle in. For the first few days they sound hyper-loud and flashy (bright?), then they turn abruptly stuffy/cloudy before they finally begin to clear up and the bass fills in. Week 3-4 is the best because each day the system sounds noticeably better than the day before.
I've been through this about three times now and it is highly unsettling each time. I should really break in the caps before listening to them but each time I tell myself I can't be bothered...
So I'd give it about a month before passing a final verdict. You can in the meantime try shorting over R3 with a paper clip or suchlike to defeat the Allen Wright treble rise.
If you build the "HiZ" version its very easy to just wire over R3.
If with the LoZ version you short over R3 maybe the brightness gets worse, this is actually even better result because it proves that its the NE5532 driving low impedance load rather than the AW curve thats at fault. Solution is to switch to HiZ feedback loop.
I'm still investigating the sound I'm getting "as built" and I haven't got around yet to tweaking even the obvious things like switching opamps.
Hopefully you'll keep us updated as well.
-rjm
The 50kHz Neumann correction (I don't call it AW as AW, AFAIK, did not 'invent' it) gives a 0.5dB rise at 20kHz. Lower than the frequency response aberations in almost any MM, MC, or dare I say it, curring setup. This is not the cause of brightness.
In fact I have a prototype on the bench, including 50kHz correction, measuring almost ruler flat up to 20kHz, and yet I can turn it from lush into steely merely by swapping one capacitor type for another.
I would rather look into the compound response of the opamp + RIAA network, calculated/simulated with the ACTUAL measured R and C values.
I would also ensure that the opamp is well-fed and decoupled resistively from any capacitive load, including the output cable.
In fact I have a prototype on the bench, including 50kHz correction, measuring almost ruler flat up to 20kHz, and yet I can turn it from lush into steely merely by swapping one capacitor type for another.
I would rather look into the compound response of the opamp + RIAA network, calculated/simulated with the ACTUAL measured R and C values.
I would also ensure that the opamp is well-fed and decoupled resistively from any capacitive load, including the output cable.
thanks for your replies!
i forgot to mention in the earlier post that i built the loZ version, with NE5532, also i couldn't use Black Gate's since they are not avaliable in Croatia.
one more thing that bothers me is that the amount of hum slightly increased as the unit is breaking in, the hum in inaudible at normal listening levels, but when i turn up the volume it is there and it's annoying....
the cartridge is Rega bias
i forgot to mention in the earlier post that i built the loZ version, with NE5532, also i couldn't use Black Gate's since they are not avaliable in Croatia.
one more thing that bothers me is that the amount of hum slightly increased as the unit is breaking in, the hum in inaudible at normal listening levels, but when i turn up the volume it is there and it's annoying....
the cartridge is Rega bias
noise
Ran through some analysis for ya -
Rs is the source impedance (cartridge) contribution
Vn is the op-amps voltage noise, 5 nV/rtHz
In is the product of the current noise and input impedances
Rf is the resistor noise of the feedback arm (R2 and Rf combined)
The plot shows various contributions to the total noise, for a NN5532 / 1Khz. The resistors are as per a VSPS in the hiZ configuration: R2 2.2k, Rf nominal@1kHz (220k). That's a hypothetical worst case scenario, just to see if hanging big resistors off a NE5532 makes a significant difference. The answer is: no, not a big difference. Switching to a OPA2134 won't change the total noise all that much either.
Remember that input-referred noise performance is not exactly the be-all and end-all of a phono preamp design. I just thought a quick calculation might be informative, that's all.
-rjm
Ran through some analysis for ya -
Rs is the source impedance (cartridge) contribution
Vn is the op-amps voltage noise, 5 nV/rtHz
In is the product of the current noise and input impedances
Rf is the resistor noise of the feedback arm (R2 and Rf combined)
The plot shows various contributions to the total noise, for a NN5532 / 1Khz. The resistors are as per a VSPS in the hiZ configuration: R2 2.2k, Rf nominal@1kHz (220k). That's a hypothetical worst case scenario, just to see if hanging big resistors off a NE5532 makes a significant difference. The answer is: no, not a big difference. Switching to a OPA2134 won't change the total noise all that much either.
Remember that input-referred noise performance is not exactly the be-all and end-all of a phono preamp design. I just thought a quick calculation might be informative, that's all.
-rjm
Attachments
Re: noise
Konnichiwa,
Hmm, lets a Shure V15 (680mH/500R IIRC) with a 47k Parallel resistor.
The Impedance of this is around 910 Ohm @ 100Hz, 4k3 @ 1KHz and 22k5 @ 10KHz.
The Philips Datasheet lists the NE5532 as having an Ein of 5nV|/Hz at 1KHz and the Iin as 0.7pA|/Hz.
This current noise translates in a voltage noise derived from the impedance and the noise current as 0.6nV|/Hz @ 100Hz, 3nV|/Hz @ 1KHz and nearly 16nV|/Hz @ 10KHz. Or in other words, the noise from the source will be 10db higher at high frequencies than that from the source and in the "midrange" it will be more less the same as the self noise.
In other words, the self noise of the Phono Amp changes from "pink" to "white" with an emphasis on high frequencies and reduced S/N ratio. Given that the NE5532 is a little marginal for phono applications to start with, using a low noise FET type will eliminate the noise rise with frequency and keep the noise "pink" and low, with traditional MM Pickups.
For Grados NE5532 is not a bad choice, though other op-amp's are being made, with potentially superior sound.
Sayonara
Konnichiwa,
rjm said:
Hmm, lets a Shure V15 (680mH/500R IIRC) with a 47k Parallel resistor.
The Impedance of this is around 910 Ohm @ 100Hz, 4k3 @ 1KHz and 22k5 @ 10KHz.
The Philips Datasheet lists the NE5532 as having an Ein of 5nV|/Hz at 1KHz and the Iin as 0.7pA|/Hz.
This current noise translates in a voltage noise derived from the impedance and the noise current as 0.6nV|/Hz @ 100Hz, 3nV|/Hz @ 1KHz and nearly 16nV|/Hz @ 10KHz. Or in other words, the noise from the source will be 10db higher at high frequencies than that from the source and in the "midrange" it will be more less the same as the self noise.
In other words, the self noise of the Phono Amp changes from "pink" to "white" with an emphasis on high frequencies and reduced S/N ratio. Given that the NE5532 is a little marginal for phono applications to start with, using a low noise FET type will eliminate the noise rise with frequency and keep the noise "pink" and low, with traditional MM Pickups.
For Grados NE5532 is not a bad choice, though other op-amp's are being made, with potentially superior sound.
Sayonara
Your numbers look right to me.
But .. you'll notice its the source noise 4kT Rs rather than the current noise term which dominates even for the NE5532 -at least below 10 kohm. So no benefit to further lowering the current noise by going with a FET input stage except for the 20kHz case.
(For those (trying to) follow - Rs is a function of just the input impedance and isn't influenced by the opamp at all. Current noise voltage is a product (loosely) of the source impedance and the opamp current noise.)
I'm not saying no benefit to a FET input - there is a benefit (the high frequency situation you mentioned) but it is more marginal than I originally thought - I had assumed current noise would dominate even at moderate frequencies and source impedances.)
and I'll tack this update on to avoid posting twice:
I tried shorting over R3 to see if I could hear the effect of the treble lift induced by the AW mod.
Well, I can plainly hear it ... more properly I can hear the effect of shorting over R3 in the VSPS circuit. Besides decreasing the high frequency response, the feedback impedance at high frequency is also decreased. What I hear could be due to the altered phase/frequency, impedance, or level of feedback.
Nonetheless it plays like a straight high-frequency lift: without R3 the sound is slightly soft. Putting R3 back in adds a more "live" feel to the music. Individual notes seem more precisely defined. Percussion has more transient snap.
In my system I preferred the eq. "with R3" by a wide margin - though it had me scurrying back to re-adjust the stylus azimuth. I wouldn't call the circuit bright, not at all in fact - but it will reproduce any upstream crappiness with pristine fidelity.
YMMV
rjm
But .. you'll notice its the source noise 4kT Rs rather than the current noise term which dominates even for the NE5532 -at least below 10 kohm. So no benefit to further lowering the current noise by going with a FET input stage except for the 20kHz case.
(For those (trying to) follow - Rs is a function of just the input impedance and isn't influenced by the opamp at all. Current noise voltage is a product (loosely) of the source impedance and the opamp current noise.)
I'm not saying no benefit to a FET input - there is a benefit (the high frequency situation you mentioned) but it is more marginal than I originally thought - I had assumed current noise would dominate even at moderate frequencies and source impedances.)
and I'll tack this update on to avoid posting twice:
I tried shorting over R3 to see if I could hear the effect of the treble lift induced by the AW mod.
Well, I can plainly hear it ... more properly I can hear the effect of shorting over R3 in the VSPS circuit. Besides decreasing the high frequency response, the feedback impedance at high frequency is also decreased. What I hear could be due to the altered phase/frequency, impedance, or level of feedback.
Nonetheless it plays like a straight high-frequency lift: without R3 the sound is slightly soft. Putting R3 back in adds a more "live" feel to the music. Individual notes seem more precisely defined. Percussion has more transient snap.
In my system I preferred the eq. "with R3" by a wide margin - though it had me scurrying back to re-adjust the stylus azimuth. I wouldn't call the circuit bright, not at all in fact - but it will reproduce any upstream crappiness with pristine fidelity.
YMMV
rjm
Pick and op-amp, any op-amp!
For fun I reworked the sim to include the frequency dependence of an inductive source and a complex feedback impedance.
Now we have frequency on the X-axis, with noise calculated for a Shure V15VMR (1kOhm + 425 mH). VSPS circuit, NE5532, LoZ version shown.
1/f low frequency flicker noise is not modelled, so I haven't shown below 100 Hz. Opamp noise taken as the 1Khz datasheet values.
The blue line is the source noise. That's fixed by the cartridge and cartridge alone. The game is to try keep everything else below this as much as possible.
The horizontal brown trace is the opamp voltage noise. It starts to dominate at low frequencies.
The pink trace is the current noise. Somewhere above 12kHz, it overtakes source noise. (For the Grado cartridge it doesn't do this until above 100 KHz.) With a FET input opamp the current noise drops off the graph, no longer important.
The light blue trace is the noise of the feedback loop. It decreases at higher frequencies because of the RIAA filter. In the LoZ version shown, it doesn't play any significant part, but with the HiZ version its about 8 nV/Hz, about the same as the voltage noise of the OPA2134, and slightly higher than an NE5532 (5nV).
So it all confirms what we've been discussing:
FETs get you lower noise at high frequencies.
Bipolars get you better noise at low frequencies, unless the feedback impedence is high.
Somewhere in the middle it doesn't matter since source noise dominates.
And its all dependent on which cartridge you use.
-rjm
For fun I reworked the sim to include the frequency dependence of an inductive source and a complex feedback impedance.
Now we have frequency on the X-axis, with noise calculated for a Shure V15VMR (1kOhm + 425 mH). VSPS circuit, NE5532, LoZ version shown.
1/f low frequency flicker noise is not modelled, so I haven't shown below 100 Hz. Opamp noise taken as the 1Khz datasheet values.
The blue line is the source noise. That's fixed by the cartridge and cartridge alone. The game is to try keep everything else below this as much as possible.
The horizontal brown trace is the opamp voltage noise. It starts to dominate at low frequencies.
The pink trace is the current noise. Somewhere above 12kHz, it overtakes source noise. (For the Grado cartridge it doesn't do this until above 100 KHz.) With a FET input opamp the current noise drops off the graph, no longer important.
The light blue trace is the noise of the feedback loop. It decreases at higher frequencies because of the RIAA filter. In the LoZ version shown, it doesn't play any significant part, but with the HiZ version its about 8 nV/Hz, about the same as the voltage noise of the OPA2134, and slightly higher than an NE5532 (5nV).
So it all confirms what we've been discussing:
FETs get you lower noise at high frequencies.
Bipolars get you better noise at low frequencies, unless the feedback impedence is high.
Somewhere in the middle it doesn't matter since source noise dominates.
And its all dependent on which cartridge you use.
-rjm
Attachments
Same data as above, but referred to the output by multiplying through by the gain and displayed in dB.
The RIAA eq. means lower noise frequencies are amplified more than higher noise frequencies.
If anyone can suggest a better opamp to use for MM carts than the OPA2134 let us know! The AD825 has a good reputation, though you'll have to mess around with SOIC packages. It's noiser than the OPA2134 however...
Likewise a better op-amp than the NE5532 for Grado MI types.
The RIAA eq. means lower noise frequencies are amplified more than higher noise frequencies.
If anyone can suggest a better opamp to use for MM carts than the OPA2134 let us know! The AD825 has a good reputation, though you'll have to mess around with SOIC packages. It's noiser than the OPA2134 however...
Likewise a better op-amp than the NE5532 for Grado MI types.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.