I'd build a composite amp. The Neurochrome design looks well done unless it has some flaw I'm not aware of. I doubt I could do better.
My experience with power amp ICs consisted mostly of repairing blown low end guitar amps. Quit that as an unrewarding pastime, but I suspect most failed because of abuse or bad design, not because of some inherent problem with the chips, in fact I'm surprised some survived as well as they did. OTOH, the old STK modules I don't have much good to say about, and the "new" ones are worse. Wonder if problems with well known power ICs could be traced to counterfeits? Long ago I used a lot of very well known high voltage power ICs. They had a high failure rate until the manufacturer realized they had an internal tin whisker problem. They replaced every one they ever shipped- that's how you can tell a good company!
Also following the other thread, and sometimes confuse it with this one! Did you notice my recent PM?
My experience with power amp ICs consisted mostly of repairing blown low end guitar amps. Quit that as an unrewarding pastime, but I suspect most failed because of abuse or bad design, not because of some inherent problem with the chips, in fact I'm surprised some survived as well as they did. OTOH, the old STK modules I don't have much good to say about, and the "new" ones are worse. Wonder if problems with well known power ICs could be traced to counterfeits? Long ago I used a lot of very well known high voltage power ICs. They had a high failure rate until the manufacturer realized they had an internal tin whisker problem. They replaced every one they ever shipped- that's how you can tell a good company!
Also following the other thread, and sometimes confuse it with this one! Did you notice my recent PM?
That Passive filtering comes with more noise and less headroom has to be taken with a big corn of salt.
When taking the circuits F and G from posting #23, but adjusted in such a way that both have the same feedback resistors over the opamps, but with unaltered gain, headroom from 20Hz to 20 Khz is exactly the same 25dB for both versions.
The A weighted noise measured with LTSpice with a Cart connected (600 Ohm, 300 mH and 120 pF for cable plus input cap) is 86 dBA for the active version and 85 dBA for the passive one, when measured from 20Hz - 20Khz ref 5mV@1Khz.
Since these opamps have input current cancellation, one should correct these figures with -2dB resulting in 84dBA and 83dBA resp.
Given the fact that 65 dBA is about the point where noise from the Phono Amp first starts to becomes noticeable at normal listening levels, but only with your ears on the speaker grill, 84 dBA and 83 dBA are ridiculously good and way beyond the point that still makes sense.
To conclude, yes there is 1 dB more noise, but at a level far beyond what is needed, and headroom is exactly the same for both.
When simulating the frequency response for the two circuits however, one can see a nasty 10dB peak around 1Mhz for the active version, with a much slower roll off as for the passive version, that does not show this peak at all.
Two 100pF caps on C1 and C2 will make things quite bit better, but it becomes not as smooth as the passive version.
It is obvious that this is far in the supersonic region, but the circuit seems to be more susceptible to HF oscillations and for RF signals entering the Phono Preamp.
So you might have to be more careful when designing an active filter Phono preamp.
In the passive version, Cap values needed for the RIAA are much larger, which makes C17 very large with 4uF in this very case.
When going to 1uF for C17 and making the surrounding timing resistors 4 times as large, you will lose 2 dB SNR, bringing the A weighted noise with Cart connected to 81 dBA, which is still more than excellent and way beyond what is practically needed.
Hans
When taking the circuits F and G from posting #23, but adjusted in such a way that both have the same feedback resistors over the opamps, but with unaltered gain, headroom from 20Hz to 20 Khz is exactly the same 25dB for both versions.
The A weighted noise measured with LTSpice with a Cart connected (600 Ohm, 300 mH and 120 pF for cable plus input cap) is 86 dBA for the active version and 85 dBA for the passive one, when measured from 20Hz - 20Khz ref 5mV@1Khz.
Since these opamps have input current cancellation, one should correct these figures with -2dB resulting in 84dBA and 83dBA resp.
Given the fact that 65 dBA is about the point where noise from the Phono Amp first starts to becomes noticeable at normal listening levels, but only with your ears on the speaker grill, 84 dBA and 83 dBA are ridiculously good and way beyond the point that still makes sense.
To conclude, yes there is 1 dB more noise, but at a level far beyond what is needed, and headroom is exactly the same for both.
When simulating the frequency response for the two circuits however, one can see a nasty 10dB peak around 1Mhz for the active version, with a much slower roll off as for the passive version, that does not show this peak at all.
Two 100pF caps on C1 and C2 will make things quite bit better, but it becomes not as smooth as the passive version.
It is obvious that this is far in the supersonic region, but the circuit seems to be more susceptible to HF oscillations and for RF signals entering the Phono Preamp.
So you might have to be more careful when designing an active filter Phono preamp.
In the passive version, Cap values needed for the RIAA are much larger, which makes C17 very large with 4uF in this very case.
When going to 1uF for C17 and making the surrounding timing resistors 4 times as large, you will lose 2 dB SNR, bringing the A weighted noise with Cart connected to 81 dBA, which is still more than excellent and way beyond what is practically needed.
Hans
Last edited:
If you have a list of the inductance of 'reputable brands' I would appreciate a PM.
Err.rh! ALL good modern cartridges use the 'cantilever resonance to prop up the falling HF response'.
The last of the SHURE papers on cartridge design will give you an idea of what the SOTA is.
Not only did I insert a 300uH Cart coil, I also forgot the 47K Cart loading.
I should have stayed in bed instead of simulating in the middle of the night.
So forget the silly graph in my previous posting, it does not show the correct behaviour.
With the right values now, 600Ohm, 300mH , 47K and 65pf cable/input capacitance I recalculated the SNR.
I also replace the OP37 with the 5534, so I don't have to correct for the input current cancellation that the OP37 has.
For the active version with Cart connected, A weighted SNR is now 81.8dBA.
For the passive version with 4uF for C17, SNR is 81.1dBA, only 0.7dB lower as it's active brother, all measured from 20Hz to 20Khz, ref 5mv@1Khz.
The Headroom for both circuits is the same 25dB over the whole audio range from 20Hz to 20Khz.
And when using the same resistor values for both versions, noise is only 0.7dB worse.
So to say that you lose on headroom and noise with a passive filter, has to be taken with a large grain of salt.
A disadvantage of the passive version is that you need much larger caps for the RIAA filters.
Especially C17, as in postings #42 circuit diagram, is quite big with 4uF.
When going to 1uF, SNR worsens a little bit to 80.3dBA.
I can only repeat that these are all excellent figures, way beyond what is practically needed.
I am not advocating one system in favour of the other, but to say that a passive filter has more noise and less headroom is not creating a correct image.
My personal preference is in using passive filters, but that's because I prefer to use fully balanced RIAA preamps, having a fantastic CMRR, resulting in less susceptibility for all sorts of disturbances.
And Fully balanced Riaa Preamps and passive filtering goes hand in hand.
Hans
I should have stayed in bed instead of simulating in the middle of the night.
So forget the silly graph in my previous posting, it does not show the correct behaviour.
With the right values now, 600Ohm, 300mH , 47K and 65pf cable/input capacitance I recalculated the SNR.
I also replace the OP37 with the 5534, so I don't have to correct for the input current cancellation that the OP37 has.
For the active version with Cart connected, A weighted SNR is now 81.8dBA.
For the passive version with 4uF for C17, SNR is 81.1dBA, only 0.7dB lower as it's active brother, all measured from 20Hz to 20Khz, ref 5mv@1Khz.
The Headroom for both circuits is the same 25dB over the whole audio range from 20Hz to 20Khz.
And when using the same resistor values for both versions, noise is only 0.7dB worse.
So to say that you lose on headroom and noise with a passive filter, has to be taken with a large grain of salt.
A disadvantage of the passive version is that you need much larger caps for the RIAA filters.
Especially C17, as in postings #42 circuit diagram, is quite big with 4uF.
When going to 1uF, SNR worsens a little bit to 80.3dBA.
I can only repeat that these are all excellent figures, way beyond what is practically needed.
I am not advocating one system in favour of the other, but to say that a passive filter has more noise and less headroom is not creating a correct image.
My personal preference is in using passive filters, but that's because I prefer to use fully balanced RIAA preamps, having a fantastic CMRR, resulting in less susceptibility for all sorts of disturbances.
And Fully balanced Riaa Preamps and passive filtering goes hand in hand.
Hans
Link?
I have to admit that its hard to separate the peaking caused by the cartridge/preamp circuit and that caused by the resonances, but if you compare say a shure VST-V or AT150 with lightweight cantilevers against cartridges with somewhat more massive assemblies it does seem you get the flattest reponse from old skool lightness.
Jose,
Here it is.
With 10 Ohm connected to Out accuracy from 10Hz to 100Khz is 0.02dB
With 5 Ohm accuracy is 0.005dB from 10HZ to 100Khz.
View attachment AntiRiaa_2.asc
Hans
Why wouldn't the SPICE model account for the current noise when only one input needs to be accounted for. BTW the noise correlation is not well presented anywhere, if you look at the data sheet it is very different at 1kHz than at 10Hz to me this indicates that the mechanism might even vanish at high frequencies. Base current is from recombination in the base so the noise from the input pair itself can not correlate at all. Take a look at the LT1028 datasheet and see how small the effect is by 1kHz. Ignore the mistake in the unbalanced graph, the base line is from Rs not 2Rs (every data sheet/app note I looked at today had mistakes, TI, LT, and ours).
Last edited:
I've dabbled in vinyl cartridge design but was never as successful as with my speaker & amp efforts and, unlike them, never for commercial production.
My mentors were Tony Emerson and the polymath Dr. Don Barlow, who used to do this for LEAK.
I should clarify that the resonance I'm referring to is that of tip mass/vinyl compliance/stylus compliance.
Chapter & verse on the subject is in various SHURE JAES papers but I've been a beach bum for nearly 2 decades so you'll have to search JAES yourself. Please let us know what you find.
Actually it's dead easy. The electrical peaking is simple 2nd order & set exactly by the inductance, the capacitance and the resistances. If you parse these out, the remaining effects are 'mechanical'.
This ISN'T simple as there are at least 2 interacting resonant systems. It gives rise to non-intuitive practical results ... eg making the 'stylus compliance' stiffer often gives 'better' HF response & trackability.
IIRC, there are separate SHURE papers detailing both the mechanical & electrical developments for each V15 from Type II to V. There's also a separate paper on the M24H surround sound cartridge IIRC which has important insights on 'tuning'.
A LoZ MC cartridge doesn't have the extra (2) degree(s) of freedom given by the electrical bits so often, the tip mass & HF trackability has to be worse than the MM 'version' for flat response. The clearest illustration of this is the MM Technics 205 which I believe has the lowest tip mass ever achieved and its MC 305 sister ... both excellent cartridges BTW.
V15 type V (VST?) may have even lower tip mass but it came out while I was in the bush.
_________________
I assume you are not referring to 'cantilever beam resonances' as these are always EVIL and avoided. Again the SHURE papers have full details.
To Jose and Hans
There are quite a few RIAA encoders out there.
The one I used in msg #23 - well I don't recall where I found it, but you say, Jose, that the response is different from Rod Elliot's, so I divised a comparison between the 2 (see attachment). I've also included an encoder, which IMHO is accurate to within +/- 0db (!).
When running them, it's apparent that they only differ for (very) high frq, where my encoder keeps rising with 6dB/octave as it should - in theory.
There are quite a few RIAA encoders out there.
The one I used in msg #23 - well I don't recall where I found it, but you say, Jose, that the response is different from Rod Elliot's, so I divised a comparison between the 2 (see attachment). I've also included an encoder, which IMHO is accurate to within +/- 0db (!).
When running them, it's apparent that they only differ for (very) high frq, where my encoder keeps rising with 6dB/octave as it should - in theory.
Attachments
This is mathematically the most accurate way to produce Riaa and Anti_Riaa curves.
In the frequency domain the results are perfect, but in the time domain they produce strange phase and amplitude errors and shouldn't be used.
View attachment Riaa-Inverse_Riaa.asc
Hans
In the frequency domain the results are perfect, but in the time domain they produce strange phase and amplitude errors and shouldn't be used.
View attachment Riaa-Inverse_Riaa.asc
Hans
How so, RIAA is defined by a minimum phase polynomial in s, the time and frequency domain behaviors are not independent?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.