Hi Gary, thanks for the input. Do you know Ed Oxner? He was my original source for explaining the problem, when I was confronted with it in the 1970's. You know, this year they are OK (mostly), then (magically) they virtually all go noisy. Whatever the problem, a change in process seems to make things worse, and the product essentially useless for low noise inputs.
John,
No, I don't know Ed other than through your references to him over the years. He's from a slightly earlier generation than me - I only started working on this stuff seriously in the 80s.
Regarding things changing randomly over time - well you get what you measure. If things change over the course of a few years, it just means that the fabs don't have transistor noise as one of their control parameters. The fabs are incredibly tightly controlled these days, so things shouldn't shift randomly. There are regular process tweaks to "improve" things - raising yield or reducing costs are often drivers.
Others have pointed out that the market for discrete transistors is pretty small and most manufacturers don't worry too much about them. They put their best engineers on the development of the high volume digital stuff. In the world of ICs, analog transistors are often neglected when developing process technology despite the fact that analog performance is getting more and more important for mobile devices. People still believe that digital rules the world and the performance of analog can be an afterthought. I'm sure the folks at Analog Devices or TI push their process suppliers so that they don't ignore analog but it's often not the primary thing in people's mind in the fabs.
---Gary
You guys are embarrassingly out of your element pointlessly making no point. Frankly you don’t know what you are talking about. Please go back to worrying about silver wire, etc. For folks here wanting to learn about what’s going on in the IC business it is best to look elsewhere.
Scott, Kirkwood makes fets for a living. What is wrong with this situation?
I am not talking about IC's, I am talking about jfets, like the ones that I discovered back in 1973 that were exceptionally low noise and relatively low cost, like the E110, that I used for the phono input stage of the Levinson JC-2. Like owning a QuanTech noise analyzer, that cost me a month's pay, used, back in 1975. Could YOU have made such a sacrifice for a piece of test equipment? I have evaluated jfet and bipolar transistor noise with a QuanTech noise analyzer since 1967, when I worked at Ampex. I have many evaluation samples from that era and from then on. I am the one who discovered the low noise of the E110, and Ed Oxner put the noise graph in the Siliconix 'FET Data Book' 1977, NIP geometry, p 4-27,through my input. And later, when the devices started to become noisy, he told me that changes had been made in the processing, and this was permanent. I then started using 2SK146 ultra noise jfets in 1979. I started using them immediately in the reproduce stage of the Mobile Fidelity Supermaster 30 ips recorder that very year. Did you even know about them?
An associate and I are attempting to match YOUR test results between the 2SK170, 2SK389, and the BF862. Interested in the results? We go lower in frequency, calibrated in nV/rt Hz and get comparable but different results.
I am not talking about IC's, I am talking about jfets, like the ones that I discovered back in 1973 that were exceptionally low noise and relatively low cost, like the E110, that I used for the phono input stage of the Levinson JC-2. Like owning a QuanTech noise analyzer, that cost me a month's pay, used, back in 1975. Could YOU have made such a sacrifice for a piece of test equipment? I have evaluated jfet and bipolar transistor noise with a QuanTech noise analyzer since 1967, when I worked at Ampex. I have many evaluation samples from that era and from then on. I am the one who discovered the low noise of the E110, and Ed Oxner put the noise graph in the Siliconix 'FET Data Book' 1977, NIP geometry, p 4-27,through my input. And later, when the devices started to become noisy, he told me that changes had been made in the processing, and this was permanent. I then started using 2SK146 ultra noise jfets in 1979. I started using them immediately in the reproduce stage of the Mobile Fidelity Supermaster 30 ips recorder that very year. Did you even know about them?
An associate and I are attempting to match YOUR test results between the 2SK170, 2SK389, and the BF862. Interested in the results? We go lower in frequency, calibrated in nV/rt Hz and get comparable but different results.
Hmm... Maybe there's a connection here. I imagine oven-heating *could* act as a crude annealing process, although microwave would not be my first choice.
My task in life has been to identify low noise components and to design topologies that use them best, for both low distortion and low noise. In the beginning of silicon transistor design, base resistivity was an almost neglected component in audio design. A few people knew that there was a difference between NPN and PNP, but there was little more really known, EXCEPT what the QuanTech noise analyzer could tell you. With such an instrument, it is relatively easy to separate the low rbb' devices from the high rbb' and this was not seriously considered by designers until several years, after I got started with it. What it took, was to virtually measure every seeming candidate, and this is where the 2N4401-2N4403 low rbb' combination came from, back in the 1960's. The people who actually took this seriously were the Japanese, and by 1978, about 10 years after I originally found a low noise solution to the MC problem, they started making some extraordinary devices both bipolar and jfet. They just blew the American products out of the water, so to speak, in almost every way: F(t) low rbb', complementary devices, etc. In any case, I have followed low noise devices for decades, and have asked 'jfet experts', some who have written textbooks on the subject like Ed Oxner, what was going on, especially when something went 'wrong' and the devices were not performing properly. I have given my 'explanations' here from what THEY told me, and what I am pretty sure that they stated, but I am no device designer, and I ONLY measure them, not build them. Some of my terminology could be 'quaint' to an expert in the field, or to someone who studied at MIT, where they apparently instill 'proper definitions' in early, much like upper class British schools teach 'correct speaking' that later separates them from the average person. And so it goes.
Mr. Curl
I find it a bit of strange.
I often have a hard time comprehending Mr. Scott Wuscer’s posts, mostly due to his choice for not using 'correct speaking' in his writing form.
He is the last person contributing here that I would name him an upper class snobbist.
But then, I am an outsider
Regards
George
Absolutely.
Technical definitions have to be precise for to count as definitions.
(If I use the term “resistance” in place of “reactance” I only confuse thinks.)
Someone who insists on the use of proper technical definitions then, while using plain language for to convey his thoughts, should not be coloured as elitist (or implied as such).
And since You brought up the case of the Man, the articles of Bob Pease - intended for the technical masses out there – are not in the direction you are pointing
Regards
George
I can vouch for the quality of Ed Oxner's work. His Prentice Hall textbook "Power FETs and their Applications" of 1983 was a wonderfully clear and thorough treatment, and I persuaded two of our University departments to buy copies. It included useful sections on Audio, and featured some audio applications of the Japanese power-JFET of the time [the Static Induction Transistor, commercialised later by TOKIN for ultrasonic fish-finding(!)]. His application notes were useful too. And he gave generously to me in 1983, spending time attending to my student correspondence, on the subject of power MOS applications, in audio and RF.
I have the 1977 NIP data, too, so here's the E110 and NIP general characterisation data.
Attachments
Last edited:
Ohh, N-I-P (aka Negative Intrinsic Positive)
Blimey, what a well tutored expression.
Ed Oxner is a great guy, I don't think he would advocate going back to 2 inch wafers and all diffused processes. Field plating, gettering, and annealing were all figured out years ago IC JFETS are well below 100Hz noise corners and GR noise is absent.
Gpapag, if I added spell checking and editing for grammar and removed the obsure sarcasm it would be all work and no play. When I went to MIT you could still submit term papers in longhand, pity those teachers.
Gpapag, if I added spell checking and editing for grammar and removed the obsure sarcasm it would be all work and no play.
When I went to MIT you could still submit term papers in longhand, pity those teachers.
Hi Joshua_G. Look I'm sorry but I don't know how to post a link. I am just not very knowledgeable about that sort of thing. I am pretty sure the details are correct. As PMA says the search engine is good. That's how I tracked it down yesterday having recalled reading it back in May. Probably the easiest way is looking up Russellc in the members list search facility and checking out the "threads started by him".
Sorry can't be more helpful. But it is a v.good thread and put my mind at rest over some purchases I made................but will also create some anxiety for people not so lucky!!
EDIT. To bring you up to speed on that thread it is "currently" on page five of the PASS LABS forum (that will of course change). Last post 21st May. Or better still go to the member list, found under "Community" and use the 'members search' to find Russellc.
Cheers Jonathan
Sorry can't be more helpful. But it is a v.good thread and put my mind at rest over some purchases I made................but will also create some anxiety for people not so lucky!!
EDIT. To bring you up to speed on that thread it is "currently" on page five of the PASS LABS forum (that will of course change). Last post 21st May. Or better still go to the member list, found under "Community" and use the 'members search' to find Russellc.
Cheers Jonathan
Last edited:
- Status
- Not open for further replies.