See :
http://www.nanovolt.ch/resources/low_distortion_oscillators/pdf/low_distortion_oscillator_design.pdf
Fig. A8, A12.
I cannot measure than low, so I cannot verify.
But it is still fun to try those MELFs out.
Patrick
http://www.nanovolt.ch/resources/low_distortion_oscillators/pdf/low_distortion_oscillator_design.pdf
Fig. A8, A12.
I cannot measure than low, so I cannot verify.
But it is still fun to try those MELFs out.
Patrick
Those numbers are at +20dBu since they go typically as V squared I still don't see it in a phono or line stage.
The problems with the Vishay metal foil resistors happen where the thermal time constants of the foil and substrate diverge the most, low audio frequencies. The good tempco of the s102 is from the balance between the substrate and the foil but wherenthe substrate doesn't track the foil fast enough you get the nonlinearity.
My guess is that there are some through hole resistors (rn60?) that may be as good in distortion but poorer in tempco. And way cheaper than s102's.
Sent from my LG-H811 using Tapatalk
My guess is that there are some through hole resistors (rn60?) that may be as good in distortion but poorer in tempco. And way cheaper than s102's.
Sent from my LG-H811 using Tapatalk
OK, that clarifies it. My only application is quasi-DC test heads, the last one we did used >$2000 in .005% resistors.
The MELFs have a much better pulse power rating per board area
because they have much more "active" surface than 0805 & friends.
I do not like them simply because they roll away.
BUT:
Has anybody here managed to buy or even to test these JFETs?
I have their catalog in .pdf, but on their website they must have
hidden them carefully.
because they have much more "active" surface than 0805 & friends.
I do not like them simply because they roll away.
BUT:
Has anybody here managed to buy or even to test these JFETs?
I have their catalog in .pdf, but on their website they must have
hidden them carefully.
Attachments
We bought some die but the MEMS mic group was sold before we got to play with them, they were VERY expensive IIRC.
gerhard,
Just as a follow up, the context was that one of the MEMS guys wanted to mock up a naked mic element and JFET charge amplifier. Those FET's in die form had a feedback capacitor for this purpose included and their gm/C FOM was quite good.
it was simply a quick way to evaluate a topology. This would have been one way to increase dynamic range from the simple back to back diodes on a MOSFET gate. I honestly remember the samples were in the $200 each range.
Just as a follow up, the context was that one of the MEMS guys wanted to mock up a naked mic element and JFET charge amplifier. Those FET's in die form had a feedback capacitor for this purpose included and their gm/C FOM was quite good.
it was simply a quick way to evaluate a topology. This would have been one way to increase dynamic range from the simple back to back diodes on a MOSFET gate. I honestly remember the samples were in the $200 each range.
OUCH!
Selecting somewhat similar IF3602s already hurts enough!
Do you think that the Microchip LND150N8G would make a good cascode
for the IF3602 ?
The MMBFJ310 seems to work nicely for BF862-class, but I'd like
to run the IF3602 at more current. They may sweat a little bit for their money.
But when I see the improvement at -100°C on the picture above, it might
make more sense to use the current for a Peltier cooler. If it wasn't such
a mess with the condensing water!
Selecting somewhat similar IF3602s already hurts enough!
Do you think that the Microchip LND150N8G would make a good cascode
for the IF3602 ?
The MMBFJ310 seems to work nicely for BF862-class, but I'd like
to run the IF3602 at more current. They may sweat a little bit for their money.
But when I see the improvement at -100°C on the picture above, it might
make more sense to use the current for a Peltier cooler. If it wasn't such
a mess with the condensing water!
The low C's are attractive, the low gm would give some Miller component to input C I would think if you are running lots of Id on the IF3602. Have you explored the tradeoffs of using one of those medium power ultra low noise bipolars as a cascode? The escaping Ib is added noise but not sure it is a factor in the total budget.
They were die in tiny quantities and for all I know there was some minimum charges involved. The company seemed like a nuclear physics lab spin off of a university and very small scale (at the time at least).
I had a bipolar in the cascode, but that makes bootstrapping of the cascode
clumsy. Even if the source of the input FET has only 200 mOhm to GND,
that does not mean that the FET runs in pure CS. By action of the 40 dB
feedback divider, the FET sees a replica of its AC input at the source, just
as if it was a follower.
So, when I connect the gate of the cascode to the source (and not GND),
I get the bootstrapping for free; no level shifter or gate driver is required.
I also think that this quasi-follower behaviour is the real reason for the
negative input impedance we see on all those circuits.
BTW, the cascode transistor does not need to be extra low noise;
it adds very close to nothing to the eqiv. input noise.
I have made an opamp-less version of the amplifier with just a
pnp differential stage between the active load and the output
to bring the dc level back to zero. That delivers still more than
enough loop gain; the extra 100 dB of an op amp in the loop
are simply excessive after the high gm into the active load.
The active load is required for PSSR.