OPA1656 Phono Preamp: Split from OPA1656 thread

I have prepared a real world record of the noise, maybe for @JRA (even if I doubt he is really interested in the roots of the issue), maybe for someone else who might be interested.

A simple RIAA preamp with one opamp and 40dB/1kHz gain has Shure M35X cartridge at the input. The output noise is recorded with 2 opamps, LT1028 and OPA 627. Bellow in the zip file are 2 files with recorded noise. 48kHz sampling, 32-bit float wav mono files. The FS level corresponds to 500mVrms. So, give it a chance, listen and you will know what is a "sharp" noise.
I can hear the 32bit file on my desktop system once the volume is cranked. OPA627 vs. LT1028 = rain standing inside v.s stepping outside. Big diff. You'll have me or rater my ear, glued to the tweeter. Thank you so much!
 
As I said: Cimarron is very expensive.

I want a supplier that is lower in price that will export to Brazil.

The OPA627s I have are DIP8, some OPA637s too. But I wanted to listen to the OPA1656 and OPA1642. They only come in SMD.
I did not remember you commenting on Cimarron. Sorry, I'm old. Forget 627s. The point was for $13 you can get what you want re a DIP 1656 and same with the 1642 for $13. And as I wrote they ship worldwide. What's not to like..?
 
Sorry, but what does that noise prove?
Proves that pA matter..? And now when I read that Marcel thinks one topology is inferior to another (a vs. b) b/c it's noisier I gotta wonder whether it's OCD or a real deal. I have never heard system hiss so loud that I could hear it except with my ear to the tweeter. Did you get the files to play? It was revealing. The LT1028 hiss was annoying if loud enough.
 
OPA2156, basically an OPA1656 but with very low offset (25uV, four times better than untrimmed 627), R2R input, larger input series resistors (thus a tiny bit more noise), less aggressive loop gain. Noise-wise, like 1656 it is not ideal for MM pre because of high 1/f corner but often still better than a bipolar with high current noise.
 
  • Like
Reactions: hbtaudio
If it is the total RIAA- and A-weighted noise you are interested in, there is a break-even point for white and 1/f voltage noise at about 1169 Hz. That is, when you know the density of the equivalent input voltage noise at 1169 Hz, it doesn't matter for the total RIAA- and A-weighted noise what part is white and what part 1/f. (It may sound different, though, due to the different spectral distribution.)
 
  • Like
Reactions: hbtaudio
The upper schematic comes straight from post #287. I've shown there with a couple of noise transformations that it is not a very good circuit noise-wise, mostly because of the thermal noise current that R2 injects into the signal source.

The second schematic has the exact same issue.

Yesterday evening I was a bit too tired to work out the third solution, but it actually gets worse. This circuit also has 47 kohm input impedance at high frequencies (47.6666... kohm to be precise), but it injects more thermal noise current than the others.

I've also looked at obtaining the signal for R2 from a tap on R3 (not drawn), but that doesn't help much either.
From #457:
These three schematics were meant to be 2nd order HP with direct drive from a MM cart. I don't think that schematic #3 got your full attention. It's a standard S-K but the other two have the +FB for R2 taken from R3 and that was bothersome for you b/c of yes, noise. If you have further thoughts or reminders I'm still interested in ditching a front-end S-K buffer.

From #463
a) is better than b), noise-wise. The circuit you rejected is better than both, noise-wise.
This surprised me. A subsonic filter with 6dB gain before the 40dB RIAA section (a) is better than a flat pre-filter with the 6dB gain moved into the RIAA (b). I like the part about keeping the RIAA EQ from dealing with room-slosh but then you say you liked the post-filter with gain example the best noise-wise and that threw me. Boosting the level of unwanted room-slosh garbage by upwards of +20dB (or more) before inserting that into a -12dB HPF make no sense - to me. Just seems to invite bigger evils than noise I may or may not hear from the couch. I like the "idea" of a direct MM-driven flat front filter w/47K Zin and a 46dB RIAA. The above all has to do with original phono pre c. 2021. I thought you were on to something.
 
If it is the total RIAA- and A-weighted noise you are interested in, there is a break-even point for white and 1/f voltage noise at about 1169 Hz.
Noise density with OPA627, OP27 and LT1028 (Shure M35X)

Phono1_noise_627_27_1028.png
 
  • Like
Reactions: inyashd and Drbulj
Nothing! I was just getting you to take notice of the $13 1656 and/or 1642 DIP8 shipped to your door ready to rock and roll in Brazil. You were going on about already having DIP8 627s and 637s and that wasn't the point. That's all.
Thanks, JRA!

I was already aware of what Cimarron was offering, and looking for more affordable pricing somewhere else, if available.

The DIP adapters they sell, which would be necessary to try and compare different dual SOIC types, but they do offer pre-assembled versions with different opamps.
 
From #457:
These three schematics were meant to be 2nd order HP with direct drive from a MM cart. I don't think that schematic #3 got your full attention. It's a standard S-K but the other two have the +FB for R2 taken from R3 and that was bothersome for you b/c of yes, noise. If you have further thoughts or reminders I'm still interested in ditching a front-end S-K buffer.

The problem with all three circuits of post #457 is that both R1 and R2 contribute to the equivalent input noise current. The lower the parallel value of R1 and R2, the higher their contribution to the equivalent input noise current density √(4kT(1/R1 + 1/R2) becomes. I didn't bother calculating the value for the third circuit, as the parallel value of R1 and R2 is obviously much lower than for the first two.

From #463

This surprised me. A subsonic filter with 6dB gain before the 40dB RIAA section (a) is better than a flat pre-filter with the 6dB gain moved into the RIAA (b). I like the part about keeping the RIAA EQ from dealing with room-slosh but then you say you liked the post-filter with gain example the best noise-wise and that threw me. Boosting the level of unwanted room-slosh garbage by upwards of +20dB (or more) before inserting that into a -12dB HPF make no sense - to me. Just seems to invite bigger evils than noise I may or may not hear from the couch. I like the "idea" of a direct MM-driven flat front filter w/47K Zin and a 46dB RIAA. The above all has to do with original phono pre c. 2021. I thought you were on to something.

In general, the more voltage gain up front, the less the effect of the voltage noise of the second stage. As the voltage noise of the second stage is not very large anyway, the difference is only small.

I still think my single-stage circuit is the most suitable, by the way.
 
  • Like
Reactions: JRA
I don't see the choices as all that difficult. At the input have as little as possible between the source + its load and the dominant noise generating stage and have all large signal overhead issues resolved as soon as possible. Optimally this requires a single stage that does all of the joujou, including a good subsonic filter. Every other stage added diminishes either noise performance or signal overhead capacity. If the subsonic HPF is too early, noise increases. If too late, overhead decreases.

All good fortune,
Chris
 
  • Like
Reactions: MarcelvdG