measuring isolated voltages

Make your self a high speed isolator . it's better to take out the op-amp than your uC.

for a 0..10 volt output it shouldn't be to bad. use a PWM driving an opto coupler. the modulating portion maybe tricky if you attempt a 100% modulation window on the duty cycle. the power supply for the input circuit should be isolated.

After you look at it all, the $16 price tag may not be so bad.:) After all, they do price specialty parts under enough to discourage you from doing it the old way, unless you have some trick up your sleeve you're not sharing :)

If your source impedance is low enough maybe convert the voltage to a current first, then you are back to your allegro sensor again.

Robert

"Jamie Morken"

Who the hell let YOU loose on dangerous high voltage equipment ??

Someone who don't like you ?

...... Phil

Perhaps a IL300, linear opto could be used.

or maybe some other linear opto.

Cheers

"Martin Riddle" = Another PITA Top Poster

A FET op-amp differential amplifier will do the job.

Input resistors of 10 Mohms each and 100 kohms for the others.

Divides the floating DC voltage by 100 and provides good isolation.

Cost about $3.

........ Phil

It's not clear why you need or could benefit from an isolation amplifier. Is there something you haven't told us? BTW, note the painful 200uV/C drift of the TI / Burr-Brown iso122 and iso124 parts.

If this is for a larger series production I suggest to look into chopping and transfer across transformers. That's how we do that on ECG designs. Basically one RF transformer that ferries power over to the iso side at, say, 50kHz. Rectify and tap that for powering preamps and stuff, dual supplies are no problem. Tap off the 50kHz as well and use it to chop the measured signal. Send these chunks over another RF transformer and demodulate at the other side. Sounds more complicated than it is since the demodulator on the system side is just a sampler running off the same 50kHz that goes into the first transformer. Now you don't have to rely on the linearity of optocouplers or fancy top Dollar single-sourced parts anymore.

Keep the chopping frequency above the audible range of animals. Even toroids can sing. Ask the shepherd/retriever mix here in my lab, she gives me the looks and leaves when it ain't done right ;-)

Thanks that IC is a good option.

cheers, Jamie

For circuit reliability and safety I would like to isolate the high voltage power path from the measurement and control section, its a microcontroller controller SMPS.

Thanks, any ideas are appreciated!

cheers, Jamie

This sounds like a tried and tested way of doing it, I don't fully understand how the circuit works, do you have a dedicated circuit to do the chopping across the transformer or do you tap off your hbridge control signal? So chopping the signal is like using a modulator then feeding it into the RF transformer, and demodulating it on the output? If you have a circuit for this I would love to see it! :)

cheers, Jamie

Unfortunately I am not allowed to. But yes, you can use the control signal for the power driver for demod, just make sure the timing is right because on the secondary you can only use the signal on the winding itself (for the modulation part) and that transformer has limited bandwidth. So does the signal transformer. IOW both introduce a delay and you want to integrate-and-hold over the correct time frame.

He'll have a better chance with the tooth fairy, instead of the toaster fairy.

Just be careful for large series. It's over $2 and Digikey doesn't carry them. I'd consider that an "exotic" part.