Gerhard, in the article 1139 you have linked we can read that:
1)The general rule that always applies is to design for the largest possible bandwidth (i.e., maximize frequency of the dominant poles).
2)Increase in the unbypassed emitter resistance will result in reduction of PM noise.
That is exactly how I designed my new preamp. Very interesting, their conclusions are same as mine.
Also, large DC current, servo and low capacitance transistors suggested do apply.
1)The general rule that always applies is to design for the largest possible bandwidth (i.e., maximize frequency of the dominant poles).
2)Increase in the unbypassed emitter resistance will result in reduction of PM noise.
That is exactly how I designed my new preamp. Very interesting, their conclusions are same as mine.
Also, large DC current, servo and low capacitance transistors suggested do apply.
Now you know, everyone. We have known about low noise design at this level since 1968 or 40 years ago. We just had to wait for the fets to mature and reduce their 1/f noise to get them useful for audio circuitry. This happened about 1970.
PMA, good that you are still here. Jack Bybee came over this week and we did some listening tests too. Lots of fun, and my hi fi needed some TLC, as I tend to neglect it. Jack brought over a NEW item of interest, then we got into his expensive Mercedes and went to lunch at a fancy restaurant. (You see, Jack is independly wealthy from other investments, not audio, he just does audio for the love of it)
PMA said:
That paper is about designing emitter degenerated common emitter (CE) RF amplifier stages that do not employ global feedback.
The "dominant poles" discussed in that paper have nothing to do the the low frequency dominant compensation pole of a voltage feedback operational amplifier with high levels of global nfb - which has been erroneously linked to alledged problems with PIM.
Gerhard, do you really think that 1nV/rt.Hz is really super low noise design? PDF 821, and 655, are 20 years late in principle, and 2.5 times more noisy than a Vendetta input stage, that was designed 5 years before the papers were published, and uses fets also.
I'll bet the BT has more than 11pF of input capacitance and you already admit to running the FETs at high Ig so the current noise must be way over 1fA/rt-Hz. People designing a general purpose circuit for instrumentation have more than ONE trade-off.
I found my stash of TO-5 vertical MOSFETS now I could actually build a JC-80.
I found my stash of TO-5 vertical MOSFETS now I could actually build a JC-80.
Why bother with a JC-80, as it is now a 26 year old design? Why not stick to IC's, aren't they 'perfect' enough? I'm stuck with them in my Parasound phono design.
Are you going to help me 'improve' the JC-80? I would appreciate any useful input. I still have one in the back closet. You never know, it might be capable of being in the same room with a 'Blowtorch' sonically, but not at this time.
Are you going to help me 'improve' the JC-80? I would appreciate any useful input. I still have one in the back closet. You never know, it might be capable of being in the same room with a 'Blowtorch' sonically, but not at this time.
Careless, then. A T0-5 is too small to work properly in a JC-80 circuit. I know because I designed the JC-80. Either Scott is being careless, or he is trying to make the circuit work badly in order to compromise me. Either way, it is a waste of time.
Please, just kidding the JC-80 is way too complicated for a hobby circuit. I'm still working on my preamp in a pipe.
I had to borrow a digital QuanTech from the fab development guys because the old ones require a bunny suit and clean room to get at. First BF862 out of the bag measured 2.3nV at 10Hz and .6 at 100kHz @5mA.
I had to borrow a digital QuanTech from the fab development guys because the old ones require a bunny suit and clean room to get at. First BF862 out of the bag measured 2.3nV at 10Hz and .6 at 100kHz @5mA.
The problem with these is that they need to be soldered to an adaptor (SOT-23 only). I have to find a socket. I was testing about 8 different FET's for my microphone project. You were right on the 2SK222 1.3nV flat right across the band only 1.7 at 10Hz at only 1mA 9.1mS too. These Sanyo FET's were only a tiny bit worse than the Toshiba ones at half the Ciss. I wonder if they make P-channel.
scott wurcer said:
Thanks for reminding me, Scott!
I finally got the chassis made up for you for the monoblocks. How do you like the Preamp in a Pipetm logos I had engraved in the covers? Nice touch, eh? Simple, yet elegant. 😀
se
Scott - that is interesting information!
Did you try any BF861A/B/C JFETs?
If space is scarce, and lower noise is wanted, one can put several SOT-23:s on top of each other to save precious PCB space
I did not find any P-ch for the 2SK222.
P-ch devices are sooo rare.
Sigurd
Did you try any BF861A/B/C JFETs?
If space is scarce, and lower noise is wanted, one can put several SOT-23:s on top of each other to save precious PCB space
I did not find any P-ch for the 2SK222.
P-ch devices are sooo rare.
Sigurd
scott wurcer said:
Re: PIM
Hi Edmond. Yes, it's just a guess of mine that doing an AC analysis with different DC offsets would reveal PIM. I have not tried it yet. The idea is that the phase data in the AC analysis can be read with very good precision. I believe that 1/1000 of a degree of the 7 kHz sinusoid used for PIM testing corresponds to about 0.39 ns, which should be pretty good PIM resolution. That is on the order of the resolution of my PIM analyzer, IIRC.
Of course, PIM is measured with a 60 Hz large signal, so measuring it with SPICE using a varying DC offset means that we are assuming that DC behavior is similar to 60 Hz behavior.
Cheers,
Bob
Edmond Stuart said:
Hi Edmond. Yes, it's just a guess of mine that doing an AC analysis with different DC offsets would reveal PIM. I have not tried it yet. The idea is that the phase data in the AC analysis can be read with very good precision. I believe that 1/1000 of a degree of the 7 kHz sinusoid used for PIM testing corresponds to about 0.39 ns, which should be pretty good PIM resolution. That is on the order of the resolution of my PIM analyzer, IIRC.
Of course, PIM is measured with a 60 Hz large signal, so measuring it with SPICE using a varying DC offset means that we are assuming that DC behavior is similar to 60 Hz behavior.
Cheers,
Bob
scott wurcer said:
Not 40 year old technology! Those are the newest high-tech devices designed using the latest computer thermal and flow simulation software, fully profiled for thermal expansion characteristics, seals between chambers guaranteed to 2 militorr, cryogenically treated, surface finish will withstand chemically reactive environments, and all for a bargain price of only $3300.00 ; which, of course, includes performance verification and calibration for the first 9 months. After which you can purchase the factory extended support option.
Best regards...
- Status
- Not open for further replies.