I am currently building a conditioning circuit for a microwave probe:
https://www.diyaudio.com/community/...tyle-microwave-detecting-probes.406201/latest
Despite the on-board complexity rivalling a nuclear plant, nothing is done to correct the inherent non-linearity of the diode detector, resulting in a quadratic response.
The first and most obvious straightening measure is to apply the reciprocal function, ie square-root.
A compressor is ideally suited to the task, and I opted for the NE575, mainly because some were gathering dust in my drawers.
To use the full capability of the probe, I needed to push the circuit to the limits, ie exceeding the normal amplitude range. This is possible, however the feedback from the gain-cell resulted in a low closed-loop gain, causing instabilities and HF oscillation.
The problem is mentioned in the datasheet, along with the possible cures like a compensation caps, various degenerating resistors, but the measures are weak, suited to the "normal" working range, and they have drawbacks.
As I wanted to go almost R-T-R for the output, I needed something more radical.
This is the compressor configuration:
The solution I found was to add a noise-gain network from pin 12 to GND. 160pF + 3.3Kohm seems optimal, and cures all instabilities.
Obviously, a noise gain network has issues of its own, but there are situations like mine where this is perfectly tolerable, and it can probably be used in other cases
https://www.diyaudio.com/community/...tyle-microwave-detecting-probes.406201/latest
Despite the on-board complexity rivalling a nuclear plant, nothing is done to correct the inherent non-linearity of the diode detector, resulting in a quadratic response.
The first and most obvious straightening measure is to apply the reciprocal function, ie square-root.
A compressor is ideally suited to the task, and I opted for the NE575, mainly because some were gathering dust in my drawers.
To use the full capability of the probe, I needed to push the circuit to the limits, ie exceeding the normal amplitude range. This is possible, however the feedback from the gain-cell resulted in a low closed-loop gain, causing instabilities and HF oscillation.
The problem is mentioned in the datasheet, along with the possible cures like a compensation caps, various degenerating resistors, but the measures are weak, suited to the "normal" working range, and they have drawbacks.
As I wanted to go almost R-T-R for the output, I needed something more radical.
This is the compressor configuration:
The solution I found was to add a noise-gain network from pin 12 to GND. 160pF + 3.3Kohm seems optimal, and cures all instabilities.
Obviously, a noise gain network has issues of its own, but there are situations like mine where this is perfectly tolerable, and it can probably be used in other cases
Perhaps I don't understand? I would never "expand" a problematic signal back into unmanageable levels. I would map your compression in a (digital) table and display measurements based on a virtual expansion, not actual analog levels.
This is an interesting chip, but any custom chip it will suffer unobtanium issues. I could post a classic JFET compressor circuit, which does essentially the same thing with common parts, but I'm sure it would not be much value to you.
I have had to deal with complex analog circuits with massive "calibration" requirements in the past and I'm glad that will never happen again.
This is an interesting chip, but any custom chip it will suffer unobtanium issues. I could post a classic JFET compressor circuit, which does essentially the same thing with common parts, but I'm sure it would not be much value to you.
I have had to deal with complex analog circuits with massive "calibration" requirements in the past and I'm glad that will never happen again.
Good you got it stable.
NE / SA575 designed for low voltage
With a good note
Is beneficial for systems with 5v power
supply. Which are usually common
these days.
Another fun one, which I believe
OnSemi still makes / sells stock
is NE570
which can go up to 24 volts.
So is somewhat common to use
on 15 to 18 volt power supplies.
NE / SA575 designed for low voltage
With a good note
Is beneficial for systems with 5v power
supply. Which are usually common
these days.
Another fun one, which I believe
OnSemi still makes / sells stock
is NE570
which can go up to 24 volts.
So is somewhat common to use
on 15 to 18 volt power supplies.
I have to compress it actually, but anyway the AD conversion + the lookup table would be more complex than ~the third of a chip, 3 capacitors and 3 resistors