Well, back to business. It is much more practical to make something like this design of Scott's with stock Toshiba 2sk170 devices. The newer IF3601-2 devices will have more 1/f noise, because they changed the process over the years. This is a tip from Ed Oxner, still a technical consultant for INTERFET.
Ed must be getting on now. I remember the Silicinix (IIRC) fet hand book. He wrote most of the app notes. That was in the late 70's when I was still young.
This means that the 2008 parts I have should have higher 1/f noise the the '94 parts? How much? Is the noise at 1 KHz worse? (And 1/f noise with a corner at 100 Hz would need to be pretty bad to be audible given the LF rolloff of your hearing at low level). Are they still made in the 2" line? I have some 2008 IF3602 and some 2004 IF3602 as well as some custom parts made from that die for me from 1994. Were there better years? Vintages? How important was the storage?
Demian, you have a Quantek, you should measure them. They changed the process to ion implantation, and usually this makes things noisier.
Hi,
Ed's Book remains available via Amazon and is seriously suggested reading:
Ciao T
I have it - unfortunately in storage in Europe. I may just buy it again since getting it out is near impossible.
I suspect this is overplayed, 1/f is 1/f, popcorn would be another thing. At a 100Hz corner any of the perceptual weightings give just about nothing for 1/f noise contribution.
The problems with ion-implanted JFET's have been solved, I'll even tip my hat to a competitor. Plots of noise for OPA827....
Attachments
Scott,
At one time 1/f was also called popcorn noise! This changed when device defects started really creating what we now call popcorn noise. So there can be some confusion. (I have a nice 1928 book that refers to popcorn noise!)
ES
But popcorn noise does not have a 1/f spectrum and 1/f noise does not sound like discrete level changes. As I said our understanding has improved. BTW I apologize for the Ohm's law comment. I should have just pointed out that it is not a physical law like conservation of charge. We like to refer to non-ohmic contacts as those with barriers or semiconducting junctions. Ohm's law is useful in setting up state variable equations which soon turn into differential equations and dI/dV is what matters.
Whilom, as olde stories tellen us,
Ther was a duc that highte Theseus;
Of Atthenes he was lord and governour,
And in his tyme swich a conquerour,
Language does change with time. So it is important to keep the usage in context. For example the 0 db level for acoustic measurements actually has changed. (The current reference is 10 db lower than the old US one!)
Yes popcorn noise as we use the term today is quite different than the oldest uses I know of. Of course even actual popcorn has changed since I was young! So the noise it makes has also changed a bit.
I think that this 'parsing' of words with regard to 'popcorn noise' just confuses the issue. Both Scott and Ed are correct from THEIR reference point, but what matters is what 'popcorn noise' is defined as, TODAY, not 80 years ago.
The reason this is important is that '1/f noise' has a rise of 3dB/octave, and 'popcorn noise' has a rise of 6 dB/octave, and they are caused by somewhat different mechanisms. In truth, we hardly know how '1/f noise' is actually created. Of course Scott might tell us that he knows, and I am sure he knows more than me, but I will go to 'Noise in Electronic Devices and Systems' by M.J. Buckingham for my reference.
This is, of course, part of the 'black magic' that fet designers covet. For example, Toshiba never released specifically how they made the 2S389, j109, etc, etc, and they just closed shop. Other manufacturers have to 'cut and try' to get similar specs. and this is why there are delays in P channel parts availability, today.
The reason this is important is that '1/f noise' has a rise of 3dB/octave, and 'popcorn noise' has a rise of 6 dB/octave, and they are caused by somewhat different mechanisms. In truth, we hardly know how '1/f noise' is actually created. Of course Scott might tell us that he knows, and I am sure he knows more than me, but I will go to 'Noise in Electronic Devices and Systems' by M.J. Buckingham for my reference.
This is, of course, part of the 'black magic' that fet designers covet. For example, Toshiba never released specifically how they made the 2S389, j109, etc, etc, and they just closed shop. Other manufacturers have to 'cut and try' to get similar specs. and this is why there are delays in P channel parts availability, today.
Hi John,
Did not know another year already passed, TEMPVS FVGIT.
Many happy returns.
Ciao T
Did not know another year already passed, TEMPVS FVGIT.
Many happy returns.
Ciao T
- Status
- Not open for further replies.