Yes, when things are broken one shot of cool spray and all hell breaks loose. Notice there is a line labeled "theory" which is not too far off, so there is a reasonable grasp on what's going on. The theory is better described in some of the papers by Lauritzen and Sah, not light reading.
Last edited:
Let me try to make things clear to the 'less technical' as to why I find jfet theory so interesting, especially with temperature.
Once, 45 years ago I had to test solid state components, mostly IC's under extreme conditions, so I had the temp extremes of -40C to +125C, or 233K to 400K to work with.
I saw how relatively easy it was to operate at lower temperatures than room temperature.
Now, in general, it is said that jfet Gm or transconductance INCREASES at lower temperatures.
Van der Ziel calculations specify the BASIC noise of a jfet is related to the inverse of 3/2[Gm]. So if the Gm increases, then the noise of a jfet should drop, and lowering the temperature should increase the Gm and therefore lower the noise. I had thoughts like this as far back as 45 years ago.
Now, what really happens? It depends! On the device, its purity, its geometry, etc, etc, and if you were to cool the jfet a little, with a refrigerator, it MIGHT get 10 dB noisier at 10Hz, or maybe some other interesting frequency compared to room temperature.
IF you further cooled it with C02 it might get quiet again, then if you get even colder, it might again get noisy. I find this fascinating, but then I have designed with low noise jfets for the last 44 years, so that might be the reason.
Once, 45 years ago I had to test solid state components, mostly IC's under extreme conditions, so I had the temp extremes of -40C to +125C, or 233K to 400K to work with.
I saw how relatively easy it was to operate at lower temperatures than room temperature.
Now, in general, it is said that jfet Gm or transconductance INCREASES at lower temperatures.
Van der Ziel calculations specify the BASIC noise of a jfet is related to the inverse of 3/2[Gm]. So if the Gm increases, then the noise of a jfet should drop, and lowering the temperature should increase the Gm and therefore lower the noise. I had thoughts like this as far back as 45 years ago.
Now, what really happens? It depends! On the device, its purity, its geometry, etc, etc, and if you were to cool the jfet a little, with a refrigerator, it MIGHT get 10 dB noisier at 10Hz, or maybe some other interesting frequency compared to room temperature.
IF you further cooled it with C02 it might get quiet again, then if you get even colder, it might again get noisy. I find this fascinating, but then I have designed with low noise jfets for the last 44 years, so that might be the reason.
Yes, stranger things have happened. But not often. Nothing against good science, only respect. My hackles are raised when the supporters of such wander off into misinformed condemnation when peering at relative unknowns. Everything we know started out as an unknown.
I thought I'd post the article link as it also plays in it's own form and connection in audio, as many of us have observed. For various reasons.
I noted, after a bit of analysis (when speaking with a given designer of circuitry), that the requirements in complex signal handling of balanced audio signals mean the perfection required goes out to about 1Mhz. Thus perfection in 3d circuit implementation and mirrored layout regarding the field considerations is paramount, regarding correctness of complex harmonics. The more perfect and mirrored the two halves are and in a RF analysis, the better it will tend to sound. Thus some of the hashy results (sonically) of some balanced designs can be partially attributed to a poor layout and other connected aspects. I treat all audio circuits that I look at or attempt as an RF interference/integration layout problem. Of course, I'm not an engineer, but I do my best. But that layout issue is an old worn out and well covered topic for this thread.
I suspect the real issue here is RF stability. All harmonics are simple: integer multiples of the fundamental. Poor RF stability can manifest itself as audio distortion, but difficult to pin down. Audio designers who don't understand RF can be flummoxed by this; fortunately the good audio designers do understand RF. Look for circuits which change their DC conditions when you touch them or wave your hand over them.KBK said:
Layout has not been examined in this thread, well at least not the interesting stuff that applies to audio frequencies in the envelope that matters. I don't think I've missed much that has been gone over here. To be fair I do understand the context of the statement though. Mike
I don't discuss layout, because I don't do it very often. I have an expert, Carl Thompson, who works with me for Vendetta, Parasound, and Constellation Audio over the years, and he does layouts for other audio manufacturers. He makes most efforts by others look amateur. As I don't do it, I leave it to him.
I personally think that very high quality circuit board layout is VERY IMPORTANT, and I have said so for the last 20 years or more. If you want to look at the original Vendetta Research design by Carl Thompson, just go the the www.Vendettaresearch.com website and look at the pictures taken by a Japanese enthusiast. It should be very revealing that these layouts were expert, and not done on an amateur computer. Check it out.
Last edited:
Yes, that is known. I've tested silicon at 77K and 4.5K, so weird is not a stranger to me..try popping 7 kiloamps into a 67mm dia slice of silicon for 4 seconds at 4.5k.
What is a fax? It that the thingy Jack Webb talked about? The fax, nuttin but the fax? Thanks for the offer John.. I really do think you should be able to discuss it clearly enough to explain though. I still see no continuity between your explanations invoking quantum mechanics, and what is being written. That's why I've asked in the first place.
Asking pertinent questions with respect to your invoking QM and weird science explanations on noise in cryo tests is not quibbling. Your descriptors did not match data presented, so you've been asked to explain. Nothing more, nothing less.
Cheers, jn
Pavel,
the Vendetta SCP is 1988.
Amateurs were using PCB design software before 1988, e.g. Orcad/EWB with student license at university, and an affordable package at home, in my case Layo1 in DOS : DOS Electronics Information - Layo1 and Orcad
(amusing bit is that the current Layo1 starter version for Windows costs €250, exact same number as it did in old currency in pre-'88)
LaYo! I wrote the User Manual and Reference Manual for that.
Bought my Audio Precision from the proceeds.
Ahhh, those were the days!
jan
- Status
- Not open for further replies.