By the way, the pole-zero extraction program I use is LINDA, written by Frank Verwaal when he still worked for the late Catena Microelectronics in Delft. It runs quite well in a DOSBOX under Linux. I'll post the results in the other thread.
Double-check against a simulator. This is not a case of adding pure real and imaginary numbers.
Ed
Ed
You have to take into account the + 1 term when dimensioning the RIAA correction network, such that the extra zero around -400π krad/s is the only error it causes.
When there is an extra zero at -400π krad/s, corner frequency 200 kHz, the error it causes at 20 kHz is 20 dB • 10log(√(1 + (20 kHz/200 kHz)2)) = 10 dB • 10log(1 + (20 kHz/200 kHz)2), which is about +0.04 dB. It follows straight from Pythagoras's theorem, you don't need a damn simulator for that.
I can look up the precise zero location this evening; I think it is a bit more negative than -400π krad/s, but not much. -400π krad/s must be about right, as the corner frequency is roughly the mid-band gain times the corner frequency of the highest RIAA pole, so 100 times 2122 Hz.
When there is an extra zero at -400π krad/s, corner frequency 200 kHz, the error it causes at 20 kHz is 20 dB • 10log(√(1 + (20 kHz/200 kHz)2)) = 10 dB • 10log(1 + (20 kHz/200 kHz)2), which is about +0.04 dB. It follows straight from Pythagoras's theorem, you don't need a damn simulator for that.
I can look up the precise zero location this evening; I think it is a bit more negative than -400π krad/s, but not much. -400π krad/s must be about right, as the corner frequency is roughly the mid-band gain times the corner frequency of the highest RIAA pole, so 100 times 2122 Hz.
Last edited:
The reason why I personally prefer to avoid the extra supersonic noise from the 1+ effect is shown in the 1st attachment.
I have added a noiseless 100R resistor at the output, alternatively coupled to a grounded 1pF and 7.64nF cap for getting the 208Khz pole, and of course also with a virtual MM cart connected to it's input.
This shows the difference in noise output beyond 200Khz with extra filter in blue and without +1 filter in red.
Further have I simulated the A-Weighted noise in the audio range from 20Hz to 20Khz with a 500mH and 600R MM cart attached
At the output this adds up to 45.5uV noise, or with a gain of 39.66dB into 474nV equivalent input noise.
When using a MM with 5mV@1Khz at 5cm/sec, this results in a fabulous S/N of 80.5dBA, thanks to the very good OPA1656.
Hans
I have added a noiseless 100R resistor at the output, alternatively coupled to a grounded 1pF and 7.64nF cap for getting the 208Khz pole, and of course also with a virtual MM cart connected to it's input.
This shows the difference in noise output beyond 200Khz with extra filter in blue and without +1 filter in red.
Further have I simulated the A-Weighted noise in the audio range from 20Hz to 20Khz with a 500mH and 600R MM cart attached
At the output this adds up to 45.5uV noise, or with a gain of 39.66dB into 474nV equivalent input noise.
When using a MM with 5mV@1Khz at 5cm/sec, this results in a fabulous S/N of 80.5dBA, thanks to the very good OPA1656.
Hans
Attachments
Yeah, now I see, thx.
Will see what difference it makes in the simulation.
Hans
Very minor improvement in low-frequency noise and a better subsonic filter response (closer to maximally flat).
I forgot one thing: a very small noise density increase around 450 Hz because the series capacitor doesn't tune out the 500 mH cartridge reactance there anymore.
I assume it is a noise with a MM cartridge related to 5mV/1kHz input. Is SNR 80.5dBA "fabulous", or just as good as to be expected?
It is very good, I agree. Below the result with cartridge included, real world measurement (OPA627, LT1010, servo OPA134).
Very good, however not fabulous. Just what you get with the best JFET opamps. With 2SK170/2SJ74, it can be done even better.