surface noise and cartridge/ phono pre questions

Jonathon, I'm guessing that you've not heard one of our preamps in probably 15 years! Time to hear one again :)

I agree that its possible to use feedback without surface noise issues, but- its a lot harder to do. In addition, I have found that universally the use of feedback will cause the circuit to take on a hardness or brightness that is not part of the original signal. Of course there can be a lot of variables in any design; I am stating this out of working with many circuits over a period of decades- there are always circuits out there that are exceptional.

With regards to feedback I have yet to hear one that really does it right (no excess 'surface' noise, no compression of dynamics, no brightness), although IMO this is a subject for another thread.

Ralph: The last time I could do a one-on-one comparison was more like 5 years back. And since then, I have also learned how to get the perceived s/n ratio down by a clearly perceivable margin, through more work on power supplies, parasitics optimization, circuit stability, more comprehensive analyses and so on (smile).

>Time to hear one again :)

Probably true for both of us (smile).

>I have found that universally the use of feedback will cause the circuit to take on a hardness or brightness that is not part of the original signal.

My findings are different. My findings are that there is an appropriate amount of global NFB for a given circuit, and this amount of NFB is best dialed-in by listening. FWIW, I think that 0dB generally sounds OK, but increasing the NFB up to about 20dB does not sound so OK. With more NFB than 20dB, things start sounding OK again, but there will be a point above at which the sound starts worsening again. Your ears will tell you where that point is.

The general trend that I have found is that the more intrinsically linear and stable the circuit is, the more global NFB can applied without damaging the sound. Also, the parasitics arising from the physical construction of the circuit are just as important as the topology.

I should add that I do use non-NFB topologies for certain sections or functions in my designs, and these sections have a clear effect on the sound. It's not like I am anti-zero NFB (grin).

Rather, I don't have any particular feelings regarding NFB, either for or against. I consider it as just another tool. If global NFB makes sense in the context of the topology and it gives better results, I use it. If it doesn't, I don't.

I agree with you that it is easy to make a global NFB circuit that sounds hard, has an unnatural "sheen", polite dynamics, compressed front-to-back depth and so on.

But it doesn't need to be so.

Thank you for a polite, sensible discussion.

kind regards, jonathan

Static charge on records is also a very strong candidate, especially if your in house conditions include lack of humidity. First things first.

Jcarr, I think you may find this article by Nelson Pass to be interesting:

https://passlabs.com/articles/audio-distortion-and-feedback

It seems to re-enforce some of your comments above. Norman Crowhurst wrote about some of this as well, although I'm don't have a link handy, I suspect I could find the article on Pete Millet's website given enough time :)

Essentially though, the issue as I see it is that as you add feedback, the problem/solution is in the way the 5th 7th and 9th harmonics are handled, but you also have concerns with intermodulation at the feedback node. This can result in a harmonic and *inharmonic* noise floor, with harmonics up to the 81st (although the circuit may lack the bandwidth for that).

The concern I have has to do with human hearing rules. The 5th, 7th and 9th are measured by the ear brain system to determine how loud the sound actually is; so if they are altered even in very tiny amounts the sound will be artificially louder and brighter than it really is.

There is also the issue of detail- due to the ear's masking rule, louder sounds can block the presence of quieter sounds. If the louder sounds are distortion then you will have less detail. It turns out that the one exception to the masking rule is the ability to hear into a noise floor composed of hiss; the ear can hear about 20 db into such a noise floor (something usually ignored by digital advocates). I suspect that this may be because analog hiss has a lot in common with wind noise, something with which our ear/brain system is very familiar. But if the noise floor is composed of harmonic and inharmonic noise, the detail below it will be lost as the ear cannot hear into that kind of noise floor.

If you want a link to Crowhurst's comments regarding this feedback phenomena, I can find it but may not be able to take the time until after CES.

Have a good show!

Ralph: I agree with everything that you have written above.

All distortion is not equal to the ears, and higher odd-order distortion sounds increasingly obnoxious.

Also, the character of the noise floor has a significant bearing on far into it the ear can hear. Completely random "soft" white noise, and the ear can hear deep into the noise floor. 20dB sounds about right, and correlates to what I have been told by Keith Johnson. But if there are harmonics, supply or signal-related rectification artifacts along for the ride, the ear's ability to hear into the noise floor stops much earlier.

If the designer doesn't pay attention to these areas, low-level detail is impaired while IMD or inharmonics typically creep into the noise and distortion products, and the result is a circuit that invariably does not sound natural or "believable".

Excellent post! And I hope that you have a good show, too.

kind regards, jonathan