1. Home
  2. analog
  3. 147043-80294

RIAA, Questions only please

I have closed the previous thread on RIAA and concluded that very few indeed understand the curves or the purpose. Here is my closing statement from that thread. For those who want to understand and have valid well stated questions I am happy to answer. 

Not wanting to leave the party without a clear and accurate statement I will say the following:

The answer to the question concerning noise reduction is that the simple filter that RIAA decided upon was to raise the high frequencies gradually by about 12 dB starting below 500 Hz, being up 3 dB at the 500 Hz pole. The circuit then cancells the pole with a zero at 2,200 Hz and there is then 3 dB of boosting left as one goes to 20 Khz. It is all done very gently with just two resistors and two capacitors.

By reversing this process on playback we get to enjoy 12 dB less noise above 500 Hz.

The RIAA part of things is the same for all cartridges. However we are accustomed to seeing RIAA combined with the 6 dB/octave compensation for a velocity cartridge. That takes off 12 dB, and along with two things that happen at the very ends of the response, brings the total EQ for a velocity cartridge to 40 dB. Next time you look at an RIAA curve ask yourself why there is that flat bench between 500 and 2,200 Hz.

An amplitude cartridge needs only the RIAA EQ of 12 dB. Which also speaks to the fact that the majority of the spectrum of a record is cut at constant amplitude. When you put a sewing needle in a paper cup and play the record you are getting amplitude playback not velocity.

I study these things because they interest me. Anyone can look up the parts values to make an RIAA filter or inverse RIAA. What interests me is that some manufacturers still get it wrong.

ramtubes

Showing 4 responses by atmasphere

atmasphere's avatar
atmasphere
7,068 posts
 
A question: are you going to apologize for mis-characterizing our preamps in public?
The statement below does not make sense. Could you clarify?

The answer to the question concerning noise reduction is that the simple filter that RIAA decided upon was to raise the high frequencies gradually by about 12 dB starting below 500 Hz, being up 3 dB at the 500 Hz pole. The circuit then cancells the pole with a zero at 2,200 Hz and there is then 3 dB of boosting left as one goes to 20 Khz. It is all done very gently with just two resistors and two capacitors.
At one point you were insisting that the cutter's constant velocity characteristic was compensated by that of the cartridge, since this isn't the case, how does that all work out?

atmasphere's avatar
atmasphere
7,068 posts
 
Actually it wasn’t Roger who asserted that. I stated that based on what was said in the following writeup:
@almarg@ramtubes

Thanks Al for that correction. My apologies Roger on this mistake.
atmasphere's avatar
atmasphere
7,068 posts
 
@solypsa mentioned cartridge design. I would like to learn more about the design of a strain gauge cartridge as it seems to me eliminating the EQ is a good thing.
A strain gauge usually employs a power supply which puts a bit of voltage across a sensor usually piezo in nature- similar to a crystal. The power supply voltage is imposed across the sensor, whose resistance changes with the strain imposed on it (in this case by the stylus). That change in resistance with a voltage across it means that the voltage is thus the audio signal.

To say that it does not need EQ is inaccurate. It is more correct to say that will only be slightly off of the required RIAA compensation. It would need less EQ than a magnetic cartridge and with the EQ will still have plenty of output so no phono preamp would be needed.
atmasphere's avatar
atmasphere
7,068 posts
 
What the article leaves me uncertain about, however, given that in modern times a cutting head is to at least a loose approximation a velocity-based transducer (meaning that it converts signal amplitude to groove velocity), is **how** the cutter is caused to cut at constant amplitude at those frequencies.
I think I can answer that. Since the cutter isn't by any means a true constant velocity device, it gets equalized so it is. The CV equalizer is a pretty complex set of time constants; I'd hate to have to work them all out, since the actual cutter response is more like a loudspeaker with some odd peaks in it (complete with a 'nose' in the high frequencies, just like any raw driver response) than anything else. 

Now the slick way to set up the whole thing would be to combine the qualities of both the RIAA curve into the CV equalizer curve so that the output was both at the same time but I suspect in order to preserve hair follicles this wasn't done and probably won't ever :)



More to discover