Screenshot
I usually don't bother with the T connection, from now on I'm going to have Jan redraw all my schematics.
This was intended originally as an improvement (I hope) on the front half of the pre-amp in the Levinzon IEEE paper. With his current-in second stage it might work for MM. Just playing around.
Cool circuit!
I verified the circuit in my simulator. Well, I got 0.5nV/SQRT(Hz) at 20KHz, and a total noise of 80nV from 20-20kHz.
Can it be made complementary so to get rid of the cap?
I verified the circuit in my simulator. Well, I got 0.5nV/SQRT(Hz) at 20KHz, and a total noise of 80nV from 20-20kHz.
Can it be made complementary so to get rid of the cap?
Unfortunately the FET's would need heroic matching, better to think servo but have not considered it. I probably ignored the parasitic Rs and Rd of a few Ohms each. As it is couldn't be more Zen.
I'm not up on battery noise, but there is some PSRR (25dB or so) but if the battery was horrible you might want to bypass it well.
EDIT AudioXpress found 9V alkaline noise "unmeasurable" and the series resistance is 1-10 Ohms.
Last edited:
What the battery? That's great, RTI no contribution at all. The FET 17.5 actually why quibble use 2 or more current, I misread Lorizen's numbers, that was for the relatively unavailable Interfet part.
EDIT - 2 at 8mA total gives 72mS gm or ~10 Ohms noise resistance. My Quan-Tech measurements showed the Toshiba Fets to be right on the 1/.7gm as the equivalent noise.
Last edited:
Sorry I said I misread Lorizen's data. A 2SK369 has ~36 mS at 4mA so two (2) in parallel @ 8mA will have 72mS or 9.72 Ohms (+-) noise equivalent resistance. Syn08 did a little better with BF862's but he used a lot of them.
Attachments
The point here I thought, was DIY. Anyone has access to this and so what that it's not perfect.
It is important, because you were stating openly, how EASY it was to make 0.4nV/rt Hz, or 10 ohms equivalent noise, and how my efforts were so easy for those engaged in the craft, from the get-go. No, it was not easy, 40 years ago, and it is still difficult with a single jfet. Just like you would have to parallel 5-10 AD797 devices to make the spec., you need two 2SK369's, or 2SK146's, or four 2SK170's to make the spec. This is without ANY excess noise, or ballast resistors, so match them well.
To make what is state of the art, appear easy, would be great, BUT you must achieve your claims.
To make what is state of the art, appear easy, would be great, BUT you must achieve your claims.
Hi Scott,
It is better to re-arrange this for a simple common source stage (as has been tradition since the 80's), DC couple the input to the cartridge and as the output carries DC in either case use only one output Cap (or better use a common mode DC derived from the output to the rest.
For more fun, we can use an active load with another 2SK369 (with a 300K gate leak resistor) and use the drain load of the lower 2SK369 for the complete RIAA (e.g. with the usual 40mA/V we get a gain of 1000 (60dB) in the midband using 10n in series with 30K with a parallel 3n Cap. For even more fun add a third 2SK369 ac cascode and a fourth as follower to isolate the RIAA.
Now we have an all j-fet, MC capable, low noise Creek OBH -8 Clone which is as "Zen" as they come (what is the sound of one 2SK369 clapping?)...
My only worry is linearity in this 4 * 2SK369 J-Fet RIAA and the fact that it inverts polarity, which will bother some.
Ciao T
It is better to re-arrange this for a simple common source stage (as has been tradition since the 80's), DC couple the input to the cartridge and as the output carries DC in either case use only one output Cap (or better use a common mode DC derived from the output to the rest.
For more fun, we can use an active load with another 2SK369 (with a 300K gate leak resistor) and use the drain load of the lower 2SK369 for the complete RIAA (e.g. with the usual 40mA/V we get a gain of 1000 (60dB) in the midband using 10n in series with 30K with a parallel 3n Cap. For even more fun add a third 2SK369 ac cascode and a fourth as follower to isolate the RIAA.
Now we have an all j-fet, MC capable, low noise Creek OBH -8 Clone which is as "Zen" as they come (what is the sound of one 2SK369 clapping?)...
My only worry is linearity in this 4 * 2SK369 J-Fet RIAA and the fact that it inverts polarity, which will bother some.
Ciao T
Heroic matching is my middle name Heck, this is DIY, so no problemo.
Please show us a complementary version without input caps.
Batteries surely need to be bypassed.
Please show us a complementary version without input caps.
Batteries surely need to be bypassed.
Adding another 2SK369 with my values as in the attached schematic gives about 0,531nV/SQRT(HZ) at 1kHz, and a total noise 20-20k of 65nV.
Attachments
Hi,
Good question. I do detect difference with polarity, but cannot reliably decide on a preference (unlike some of the self proclaimed polarity gurus), but as most gear is non-inverting, inverted sound mildly bothers me...
For yourself, just ask yourself, does it make a difference if the pretty girl singer you are listening to (possibly Ruby Rhod from the 5th Element Style) sucks or blows?
Does It?
Ciao T
Good question. I do detect difference with polarity, but cannot reliably decide on a preference (unlike some of the self proclaimed polarity gurus), but as most gear is non-inverting, inverted sound mildly bothers me...
For yourself, just ask yourself, does it make a difference if the pretty girl singer you are listening to (possibly Ruby Rhod from the 5th Element Style) sucks or blows?
Does It?
Ciao T
Yup, that's where this came from. As a demo I built a Chinese large capsule mike that ran off of harvested room light. I used one of those Japanese waving toy cells and 6 or 8 of those Cherry Semi micro PV arrays to bias the capsule. You still get pretty low noise out of a 2SK170 at 25uA.
Thorsten I said it was just a fun circuit. I was thinking of something I might put in a headshell and phantom power without rewiring. It might not work out but it did lighten things up so to speak.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.