Vinyl Lovers? Advice?

[bcarso]

Yes the old audio handbook I have from mid- '70s has a whole chapter on preamps (according to National.. but it is heavily LM381/LM387 intensive and not on their website as far as I could find..

If I was still messing with phono preamps I'd take a look at those  .3nV JFETS ( IF3602) from Interfet. A little pricey but anybody still messing with vinyl probably isn't doing it to save money.  They're just crazy quiet... Not that the preamp would be any quieter..  :

JR

I hadn't seen that part but that is a PHAT FET!  Not too bad at Ciss 300pF, and if you are using a design with heavy feedback to the source you can take most of that out.  Bootstrapping the drain will then remove most of Crss.  You will want to do that anyway because the variable and nonlinearly variable input C will give significant distortion with that relatively high Z source.

And that 300pV per square root Hz is specified at 100Hz.  Pretty nice.  A 5.4 ohm resistor's thermal noise at 300K.

EDIT:  Process information here: http://www.interfet.com/pdf/P_NJ3600L.pdf

You get rid of pretty much all of the e sub n of the preamp for any MM cartridge anyway, and i sub n is small (starting to rise at the high freq end due to coupling of thermal channel noise into the gate though).  The 47k loading R is still there unless synthetic cooled termination techniques are used.

Pretty soon you are well below self-noise of the typical cartridge, and of course far below most surface noise etc.

 

[JohnRoberts]

Yup, those are quiet enough to make a MC head amp...  Silly low noise.

JR

 

[tv]

300pF ..slightly high for an instrument input.

Your bootstraping techniques couldn't be used if simply used as a frontend to an opamp?

 

[bcarso]

Series feedback with a very small R (to not degrade noise, <5.4 ohms!) from source to ground and feedback from a considerably larger R after a lot of gain will remove most of the Cgs.  For the Cgd, you'll see it at unity multiplication having the drain going to a low Z summing node.  If you want to get rid of even more, make that node move so as to keep Vdg constant.

I note that the process data linked seems a bit more optimistic on reverse transfer capacitances than the actual part datasheet.  But pessimistic on input C: 650pF.  That's a whopper of a JFET.

The application btw is charge preamps for nuclear science detectors with high capacitance.  At mid-high frequencies the optimal noise match is where Cin of the pre is equivalent to the detector capacitance.

 

[bcarso]

It would only take 11 of these in parallel to get about a 3dB noise figure directly coupled to a 0.5 ohm ribbon.  That's nutz.

 

[JohnRoberts]

and the cost of 11 in parallel is a little nutz...

but a sweet part...

JR