The difference is very small, the two resistors sample the same average current of both FET's. The multiple CMRR trims look like a pain but if you want 100dB everywhere maybe necessary.
The president of SR is a nice guy, sent me a pile of datasheets noting the parts he uses (39 - 797's in one).
Last edited:
So, that's what you meant about attenuator?But we do need that CMR.
-RM
No That was something else but it didn't work.
I think you should clearly define what you need that frontend for (or you'll get random "solutions" throwed at you). I assumed you're looking for a preamp with differential input to measure very low noise sources. If the task is to do THD measurements at high-voltage tubes you need something else, probably.
Samuel
The 7A22 manual is a free download. How things have changed, they basicly teach you in great detail how the whole thing works.
What App?
I think from reading the lit on the QA190 probe.... it is intended to measure across the ac power line (120v). Then take FFt of the signal, etc. Here though, I dont know why high voltage measurement would be needed... unless, it was tube electronics. [but low very low noise wouldnt be needed?]
Maybe for bridged power amp output measurements?
THx-RNMarsh
I think from reading the lit on the QA190 probe.... it is intended to measure across the ac power line (120v). Then take FFt of the signal, etc. Here though, I dont know why high voltage measurement would be needed... unless, it was tube electronics. [but low very low noise wouldnt be needed?]
Maybe for bridged power amp output measurements?
THx-RNMarsh
Last edited:
Yes, it's noise in HV applications. I was perplexed that a couple of folks thought that the depletion MOSFET CCS was "noisy". I don't know if there's been a lot of work done characterizing them for noise.
I've built Scott's diffamp but it keeps railing.
it sounds to me like you have the feedback polarity swapped somewhere. It can be a bear to figure out.
I have been using depletion mode Mosfets very successfully in a number of locations. I would use it to cascode a current source since the intrinsic noise is pretty high. As a cascode its gain is very low so the noise below the source dominates. However its an easy way to get 300V of compliance from a current source.
This is an important point. Jen's application is the imput of a high performance ADC for measurements. Jackinnj's application is looking inside of a tube amp. They share some aspects but have some different requirements.
To optimize the design you need to address the following:
Input noise requirements
Distortion maximum
Gain options?
Maximum common mode requirements
DC or AC coupled
Max output swing
Input impedance requirements
For jen's application I'll add some numbers to focus the task:
Input noise requirements: approx -130 dB below full output in a 20 KHz band (equal or better than the SNR of the best available ADC)
Distortion maximum: -125 dBC for any harmonics (to be better than the best available ADC)
Gain: 6, 26 and 46 dB
Maximum common mode requirements: unknown at this time? Clearly limited by the power supplies.
DC or AC coupled: Both?
Max output swing: 4V (the next stage is differential so it really should be a differential output)
Input impedance requirements: very high and very linear so the input won't introduce distortion on the output of the source if the source is not low impedance.
These are some pretty challenging design requirements.
One comment on these designs; When used with a single ended source the common mode will be 1/2 the signal level internally. This means the opamp U1a output swings at approx 1/2 the input and needs to do a pretty good job.
Yes--I have two or three 7A22s and surely a printed manual. This plugin alone makes the hunt for a 7000 mainframe (just got another 7904A in excellent condition
) worth it!If they just would have used good relays... I'm thinking of redesigning the circuit rather than fixing the pile of defective ones I've got. There are now amplifier topologies with much faster overload recovery (see Addis work for the Tek 11000 series).
Samuel
I have not studied tube circuits--I presume the expected noise voltage is one to two orders of magnitude higher than for (very) good transistor design? So I'd say a noise floor of around 10 nV/rtHz should do the trick. Also I'd suggest you stay with unity gain. This will not mess with the calibration of the AP you have, and make interpretation of the measurement data very simple.
So it looks to me as if two OPA827 connected as buffers, with a suitable AC coupling and protection network at the input, is all you need. If a bit higher noise floor is acceptable, even a TL072 will fit.
Samuel
Sorry I didn't think of it till this morning, but the common mode op-amp needs a common mode range near the plus rail. The SR circuit less so for the same result (but two more matched resistors). When I reduced the resistor values and voltage drops the op-amp is marginal unless it works .5 V from the rail, there are plenty of choices these days but it is easy to overlook.
Anybody checked out Burkhard Vogel's 'poor man's measurement amp' in Vol 7?
Fully diff, high CMR.
Jan
Attachments
I think we were looking for high input impedance. Full balance and CMR looks to need a lot of trims too.
Last edited: