No it's the esotericness of the application, in other words volume of sales. The BF862 was designed for AM car radios, it beats the SK170 on every spec and costs 5 cents in quantity.
If a car maker designs a rim that takes a totally unique tire size you can bet they cost a premium.
Last edited:
@John: Yes, it's a p-channel, which is what's needed in that design. The ones I used, as I mentioned, seemed pretty quiet for MM use, quieter than the J271 they replaced. I don't claim that it's universally true for all 2N5114/16. I'd love to find a reliably quiet p-JFET for a publishable design- do you have a recommendation?
You won't get below the background noise level of the big bang. Liquid nitrogen isn't cold enough. Try liquid helium.
How fortunate microphone preamps don't have these problems. Do you think their designers know something phono cartridge preamp designers don't?
You won't get below the background noise level of the big bang. Liquid nitrogen isn't cold enough. Try liquid helium.
How fortunate microphone preamps don't have these problems. Do you think their designers know something phono cartridge preamp designers don't?
Pure physics, the brownian noise of the air on a stretched membrane is easily computed. It is rarely possible to make a useful microphone that reaches the limit of audibility, but the electronics piece of the problem is relatively easy. A picture is valuable to make this point, these are 1" B&K curves actally presented in nV for comparison.
Attachments
Last edited:
Well SY, I think that any P channel input is a poor choice. I said the same thing to Scott Wurcer, a few years ago, because he made the same 'mistake'.
Now what is the problem? In this case, ALL ELSE BEING EQUAL, P channel devices have LESS Gm and higher 1/F noise.
A perfect example is the PS geometry that SY used and its 'complement' NC geometry.
In National Semi terms, it is process 88 vs process 51.
Typically, the N channel is 3 times quieter at 100Hz than its P equivalent.
Does that mean that P channels should not be used EVER? Of course not!
As I said before, I have been using the same basic P channel part since 1973. I used them in EVERY Levinson JC-2, microphone preamp, master recorder, Vendetta Research power supply buffer, etc, etc. I have more than 1000 of them sorted to 1 ma Idss in plastic trays. Yet, I don't normally use them below line level for input stages, because they are more noisy.
Now, could you, SY, have lucked out? Yes. 1/F noise is somewhat mysterious, and is dependent on processing. WHY do you think that Linear Systems has had such a hard time getting the P channel J74 out to the public? Lack of interest? Oh no! They just have a hard time figuring how Toshiba did it as well as they did.
I realize that some scoff at innovation here, as they usually come by it, once it is established, and then it becomes OBVIOUS.
BUT, back in the old days, jfets were virtually written off as very noisy, due to what seemed to be, inherent 1/F noise, and many textbooks, even later ones, would warn against using them for ANYTHING below 100K ohm or so source impedance.
However, like the paralleling of devices was a revelation to me in 1966-7, the same journal talked about jfet noise, especially in the 2N3819, a common N channel part that can be fairly noisy. That was all that was available then. Look at the part number, they go up with time, so newer parts have higher numbers.
Guess what? Some professor, somewhere, measured 100's to 1000's of these devices and found 1 device with virtually no 1/F noise in the audio band! This was a REVELATION, because it showed that it was POSSIBLE to make a low noise 2N3819, or a similar device. You just had to get the yield up. Look what followed? NC and PS geometries, and the NI geometry, the very heart of the Levinson JC-2.
Later, Toshiba made even better parts, with higher Gm, better match, and LOWER 1/F noise.
Now, batch to batch, SOME batches of P channel devices will look similar to N channel devices, but I would not recommend it to the general public. All else being equal, if you get a design that can be 'REVERSED' in power supply voltage, then do it. Put an N channel jfet on the input and the P channel jfet on the second stage. It is just good engineering practice. It is the reverse with bipolar transistors. There, you want to use a pNp transistor at the input and an nPn transistor for the second stage. This is because the capitalized letters are the BASE MATERIAL of the bipolar transistor. Again, N is better for noise.
Now what is the problem? In this case, ALL ELSE BEING EQUAL, P channel devices have LESS Gm and higher 1/F noise.
A perfect example is the PS geometry that SY used and its 'complement' NC geometry.
In National Semi terms, it is process 88 vs process 51.
Typically, the N channel is 3 times quieter at 100Hz than its P equivalent.
Does that mean that P channels should not be used EVER? Of course not!
As I said before, I have been using the same basic P channel part since 1973. I used them in EVERY Levinson JC-2, microphone preamp, master recorder, Vendetta Research power supply buffer, etc, etc. I have more than 1000 of them sorted to 1 ma Idss in plastic trays. Yet, I don't normally use them below line level for input stages, because they are more noisy.
Now, could you, SY, have lucked out? Yes. 1/F noise is somewhat mysterious, and is dependent on processing. WHY do you think that Linear Systems has had such a hard time getting the P channel J74 out to the public? Lack of interest? Oh no! They just have a hard time figuring how Toshiba did it as well as they did.
I realize that some scoff at innovation here, as they usually come by it, once it is established, and then it becomes OBVIOUS.
BUT, back in the old days, jfets were virtually written off as very noisy, due to what seemed to be, inherent 1/F noise, and many textbooks, even later ones, would warn against using them for ANYTHING below 100K ohm or so source impedance.
However, like the paralleling of devices was a revelation to me in 1966-7, the same journal talked about jfet noise, especially in the 2N3819, a common N channel part that can be fairly noisy. That was all that was available then. Look at the part number, they go up with time, so newer parts have higher numbers.
Guess what? Some professor, somewhere, measured 100's to 1000's of these devices and found 1 device with virtually no 1/F noise in the audio band! This was a REVELATION, because it showed that it was POSSIBLE to make a low noise 2N3819, or a similar device. You just had to get the yield up. Look what followed? NC and PS geometries, and the NI geometry, the very heart of the Levinson JC-2.
Later, Toshiba made even better parts, with higher Gm, better match, and LOWER 1/F noise.
Now, batch to batch, SOME batches of P channel devices will look similar to N channel devices, but I would not recommend it to the general public. All else being equal, if you get a design that can be 'REVERSED' in power supply voltage, then do it. Put an N channel jfet on the input and the P channel jfet on the second stage. It is just good engineering practice. It is the reverse with bipolar transistors. There, you want to use a pNp transistor at the input and an nPn transistor for the second stage. This is because the capitalized letters are the BASE MATERIAL of the bipolar transistor. Again, N is better for noise.
How fortunate that when you listen to the .0000000000001% of all the phonograph records out there made by WA and CCR your phono playback preamp won't be the limiting factor for noise. But for all the rest of them that had the misfortune to be recorded using brand X microphone preamps, well we'll just have to live with their hssssssssssssssss, won't we. BTW, living in SF the way you do, what is the background street noise level you have in your house? If you keep the windows closed what is the background noise level of the air conditioning system, heating system, or the neighbor's TV set where you listen? Me, I live out in the country where the air is clean and clear, the water pure, and the only sounds besides the birds twittering...is all of the systems like the forced air heating and air conditioning equipment that makes my house liveable. Now of course you're going to tell me that even though the preamp noise is below the background noise floor, it's effect is subliminal? There's a good ad slogan for you; "my preamp is so good you can't hear how good it is."
Yes, Scott, and if that graph is a very OLD one, then I improved the S/N of the electronics by as much as 10dB and INSISTED that B&K follow my example. They did, and all the new electronics is 10dB quieter. It was not an 'easy' sell, but their OWN independent measurements proved me right, and they quietly upgraded their preamp the 2619, and I have the B&K measurements that prove it.
Who says that you have to live with mediocrity?
Who says that you have to live with mediocrity?
Yes, well, some people like performance for performance sake. Some people push transformers when they are relatively expensive and distortion prone, to reduce noise, and gloat about it when they get a dB or 2 better than someone else, without a transformer. Yet when it comes to fundamental design choice of an N or P device, where the P device is inherently more expensive, and noisy, they recommend the P device, because that is what they found in their 'junkbox'. Heck, yesterday I found a J111 (N channel NC geometry) on the rug next to me. Should I make a project with it? Of course, I have a 1000 more in the next room, and hundreds of other devices that might be more suitable, but that was what I found on the rug! ;-)
I miss that Pieter. I used to live in Glion, above Montreux for a couple of years. Wonderful, quiet, and quite a change, from SF. Still, much of the HISS that we hear, is higher frequency, and many people use headphones that screens out the outside noise. Much of the low frequency noise is ignored, by both our ears and our brains. If not, analog tape would never would have been successful, vinyl recording would never have worked, and condenser microphones would only be used for measurement.
These are the same kind of people who would have you believe that the golden spike is the only part of the Transcontinental Railroad that matters. And in the next breath they tell you that they are engineers. Just like trash collectors in NYC who are called "sanitary engineers" or high school janitors who are called "stationary engineers." That's how they justify their high fees for services they perform too.
Nope, please read the plot and do the math 82nV at 50Hz is 50pF and 10G Ohms. So please present your data. A 1" capsule does 5-12dB SPL self noise there is no way to improve it. There are dozens of mikes at the limit of the capsule thermal noise.
Last edited:
- Status
- Not open for further replies.