In this circuit R3 is split in two. Can someone tell me what is R3a doing? I don't see its equivalent in other RIAA implementations. Thank You!!
Website of Wayne Stegall - Low-noise Hybrid Op Amp/Discrete Phono Preamp
Website of Wayne Stegall - Low-noise Hybrid Op Amp/Discrete Phono Preamp
Imagine that C1 and C2 have zero impedance at very high frequencies.
You now have the gain setting resistors of R3a (~32r) and R3b (10r)
That sets the very high frequency gain to ~4.2 times (+12.46dB)
the HF response is falling at 6dB/octave (the pole of the filter) due to the RIAA curve until the effect of R3a steps in to flatten the curve. This is the zero that Ica mentions.
But the high frequency zero is placed in there as an attempt to cancel the pole of the recording chain.
There is a strong argument that the recoding chain does not place a single pole at the high end around 50kHz to 100kHz.
But places a complex set of poles around that frequency range.
You then follow that with, what sounds better:
a.) Ignore the extra recording pole
b.) add a simple zero that creates a different error, since it cannot match the recoding complexity
c.) try to second guess what the recording complex poles were for each manufacturer and have an adjustable complex set of poles and zeroes to correct for the recording.
We seem to be agreed that option c.) is not worth the effort.
There seems to be fairly equal arguments for adopting a.), or b.).
They both work, neither are absolutely accurate. Both sound good when the rest of the RIAA is accurately reproduced.
You now have the gain setting resistors of R3a (~32r) and R3b (10r)
That sets the very high frequency gain to ~4.2 times (+12.46dB)
the HF response is falling at 6dB/octave (the pole of the filter) due to the RIAA curve until the effect of R3a steps in to flatten the curve. This is the zero that Ica mentions.
But the high frequency zero is placed in there as an attempt to cancel the pole of the recording chain.
There is a strong argument that the recoding chain does not place a single pole at the high end around 50kHz to 100kHz.
But places a complex set of poles around that frequency range.
You then follow that with, what sounds better:
a.) Ignore the extra recording pole
b.) add a simple zero that creates a different error, since it cannot match the recoding complexity
c.) try to second guess what the recording complex poles were for each manufacturer and have an adjustable complex set of poles and zeroes to correct for the recording.
We seem to be agreed that option c.) is not worth the effort.
There seems to be fairly equal arguments for adopting a.), or b.).
They both work, neither are absolutely accurate. Both sound good when the rest of the RIAA is accurately reproduced.
Last edited:
Ok. This makes sense, I am surprised I do not see it used all the time. So it causes the gain to trend to some fixed value at "infinity" . It seems that the people that do not insert it are going route a.). I think it be best to ignore it in the feedback network and put it all in a passive 3.18 E-06 low pass on the output.
My designs are all passive and I habitually leave that zero out of the equation, it seems to have arisen at some point as compensation for the behavior of a particular Neumann cutter head. It's outside of the range of human hearing, and might only be an issue with cartridges with very extended ultrasonic response. Incorrectly implemented it may cause measurable droop in HF response.
It's one of many things people do including sometimes attempting to correct for the HF response anomalies of a very specific cartridge stylus combination.
In the few demonstrations I have heard I have preferred the standard approach, and the network with the best overall frequency response conformance to the RIAA standard. Always disappointing when the trendy thing is technically and sonically inferior to stolid, solid engineering based on long understood principles. 😛
It's one of many things people do including sometimes attempting to correct for the HF response anomalies of a very specific cartridge stylus combination.
In the few demonstrations I have heard I have preferred the standard approach, and the network with the best overall frequency response conformance to the RIAA standard. Always disappointing when the trendy thing is technically and sonically inferior to stolid, solid engineering based on long understood principles. 😛
- Status
- Not open for further replies.