riaa circuit

[lt cdr data]

I hope this has worked ok for all.

A very good sounding riaa, I think I have the values right, but if not, please let me know, thanks

[Netlist]

I could offer this spreadsheet, perhaps you know of its existence.

/Hugo

[lt cdr data]

another one!!



The output is actually taken bottom right via a series 51 ohm resistor, not shown

[jcx]

your output offset servo destroys the gain matching condition for the output active compensation - R14 is too small and should be moved, why not 1 meg into U2 +in, 1 Meg hardly effects component tol error at U2 +in


not that I think active compensation is particularly useful in any event, Jung's multiloop composites are likely to be more useful in this location as the phono pre out may drive cable C and variable impedance - which also detracts from the active compensation

your images are so large so you can't tell but jpeg really sucks for denser drawings with small text - use .gif

[andy_c]

I've got some info on my web page about one popular active RIAA equalization circuit, including getting the circuit to match the Neumann time constant, computation of RIAA equalization errors using SPICE, as well as separating out the op-amp induced portion of the error. The link can be accessed via the WWW button below this post.

[Leolabs]

How long u have been working on this???

[lt cdr data]

they aren't my designs, but they both work and sound v good. They are both well known circuits of items.
Its my first attempt at uploading, so be patient...any tips gratefully received

Any comments about the top circuit (should I post in the vinyl section?), it uses single supply and is my fave phonostage to date.

[Graham Maynard]

Hi Andy,

Excellent investigation.

My best sounding phono pre-amp split the RIAA process into two so that one IC was not working over such a wide gain/frequency range.

Following a separate buffer with pre-setable variable R + C cartridge matching, the first active IC created the low frequency lift to 20Hz ( not 50Hz ) plus the gain necessary to acheive the required midband output level.
There was separately switchable 'rumble' filtering for discs where the extended low frequency response shook the room.
All ICs were run flat gained through mid to hf, so no equalisation could go wrong at any stage, just plenty of headroom via the psu rails.

This was followed by precision resistor-capacitor passive filtering for the upper RIAA characteristic, and of course it simultaneously cut high frequency noise and hf distortion products from the first two stages.

Finally there was an IC output plus transistor buffer, which could now cover the Neumann time constant.

It sounded superior, no matter what the dB figures suggested, and where every dB of signal to noise ratio is so important it was fractionally quieter too.
It's low impedance output was totally isolated from the RIAA equalisation IC/circuitry so the sound could not be 'sucked' or dynamically challenged by power amplifier or cable impedances.

I had bought our home and little feet had arrived so I needed money; I sold it for £100 about 25 years ago. £36 of that was for the ICs alone, and this was a fair expense in those days, but I'm blowed if I can remember their make or number now; they were the latest high spec of their day. The guy who bought it kept coming back to tell me how he was amazed by the open-ness, clean-ness of attack, stability of sound and sheer clarity.
Aspects not heard via conventional 'all-in-one' RIAA equalisation stages, no matter how good the base amplifier.

I would love to see your type of investigation cover something like that, maybe including noise performance as well as amplitude/frequency.


Cheers ......... Graham.

[tlf9999]

As to the newman constant, has it been proven that we can actually hear the difference between two circuits where one is corrected for it and another not?

Some spice programs (later OrCAD versions for example) come with optimizers and one may be able to just use that function to let the computer find optimal values for the equalization network. Not sure if that has been done by anyone.

[andy_c]

Yes, that's been done by Jim Hagerman. There's a link to his analysis in my article, but for convenience I've duplicated it here. http://www.hagtech.com/pdf/riaa.pdf

Closed-form solutions are nice because you can just take the formulas and put them into Excel, MathCad or whatever. One of the issues with optimization is that there's not always a way to constrain the optimizer so that it is forced to pick R and C values from a list of standard values. Often the deviation of the calculated value from the standard value can be much more than the tolerance of the component itself.