Reverse RIAA

[Thermionic Idler]

I made up a little Reverse RIAA pcb based on this design: https://hifisonix.com/wp-content/upload ... e-RIAA.pdf - I did it for through-hole components as I'm still not into this whole SMD soldering thing. As usual with my stuff, PCB came from JLCPCB and box from Schaeffer using their U-box script that I've had previous success with.





By the way, that's not a scratch, it's a cat hair...



I used it with the Analog Discovery 2 to measure the frequency response of the Whest PS30RDT which is specced to have ±0.2dB accuracy from 18Hz to 22kHz. Interesting to see the difference in the shape of the curve between left and right channels. The left channel looks like it deviates very slightly from that, more like ±0.25dB accuracy. The right channel is within spec.

Left:


Right:


So, that seems to indicate the reverse RIAA circuit is working as it should, I'm happy with that. Of course, what I really want to measure is the output of the Pete Millett stage which seemed fairly down on treble when demoed at Owston. Problem is, I don't have an easy way to interface male XLR to female BNC at the moment.

Although thinking about it, might be able to bodge something with an oscilloscope probe...

[Nick]

https://www.canford.co.uk/XLR-ADAPTERS

[Nick]

From your link

by exploiting the fact that tolerance tightens as devices of the same value are paralleled, or added in series

Only if the distribution of the parts is normal and centered around the target value, and for that matter if both of those unproved assertions were true, you would still need to show that the parts you have have values that are distributed around the nominal values.

For example, I could take 10 1% 1k resistors, and manufacturing may have given me 10 resistors whose value is 1005 ohm to within 0.01%, doesn't matter how many of those you use they will never get closer to 1000 ohm.

[jack]

Nick's point is important and often missed. With certain devices you need to measure the resulting values to be sure - you cannot rely on bad statistics or on poor assumptions.

When dealing with known random(ish) data like noise, paralleling up generally works out of the box.

[Thermionic Idler]

Nick's point is important and often missed. With certain devices you need to measure the resulting values to be sure - you cannot rely on bad statistics or on poor assumptions.

When dealing with known random(ish) data like noise, paralleling up generally works out of the box.

Agreed, and when I was measuring the resistors, most seemed to measure the same side of 'slightly wrong' so I don't think I can claim vastly superior accuracy over the equivalent that uses only a few components. With hindsight, maybe I should have got a selection from different manufacturers to mix things up a bit, or just made a normal one with tighter tolerance components.

Still, it's good enough, and was instrumental in helping me bodge a fix to the top-end response of the Millett phono.