EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: idpromnut on October 19, 2014, 03:34:44 am
-
So I don't have details yet (since I'm still working on them!), but the basic issue is that I need to have both of my power channels in my supply isolated, and I decided that I would put the micro-controller in the "non-isolated" section (i.e. not in either of the two isolated power channels).
What I need is some way of measuring the output voltage. I seem to think the typical way of doing this is to put an ADC in the isolated area and opto-couple the SPI/I2C interface back to the micro. Is there any other way of doing this, perhaps with an ADC that offers internally isolated inputs? I was thinking an instrumentation op-amp, but I don't think the inputs are isolated.
-
A2D on the isolated side with opto couplers (or a dedicated isolation chip like Maxim do) across the barrier would be the way I'd go.
-
Hello,
for SPI you will need 2 optocouplers.
I would use a microcontroller with integrated ADC (e.g. PIC16F683 with 10 BIT ADC or PIC24F16KA301 with 12 BIT ADC) and output on asynchronous serial comms line to the central controller for the display.
So you will need only 1 optocoupler.
But you could also use the ADUM-Devices (ADUM1201) for isolation.
There are also devices which transfer some power to the isolated section (ADUM5241)
With best regards
Andreas
-
Not measuring per se, but controlling can be done isolated. With an optoisolator in the feedback loop of a regulator (usually a TL(V)431 reference and error amp, and an SMPS controller on the primary side), the gain error of the opto goes away and you get desired output voltage.
If primary side control and monitoring is required, yeah, it's probably something harder. There never were any good analog isolation methods; some of the more expensive all-in-one chips used pulse or modulation coupling methods (e.g., PWM, FM).
Tim
-
I would use local feedback so the isolated links between the converters and microcontroller will not be part of the feedback loop. That relaxes their speed requirements considerably. How quickly do you need to update the voltage and current limits and read back the output values?
If your accuracy requirements are not severe, they do make linear optocouplers. For more stringent requirements, isolation amplifiers can also be used.
-
Thanks guys!
@David: the channel will be fed by a PWM signal through an opto-coupler with all the required feedback and control systems in the isolated section (i.e. no active feedback will be required by the micro). However, I want to be able to read back what is happening at the outputs (either direct or via a sense connection) and that's where the above need came from.
I think I'm going to put an ADC in the isolated section for each channel with the SPI/I2C digital lines fed back through an opto-coupler. Thanks for the feedback!
-
@David: the channel will be fed by a PWM signal through an opto-coupler with all the required feedback and control systems in the isolated section (i.e. no active feedback will be required by the micro). However, I want to be able to read back what is happening at the outputs (either direct or via a sense connection) and that's where the above need came from.
I understand. The different rise and fall times of the optocoupler solution will tweak the duty cycle a little bit but you can calibrate that out.
I think I'm going to put an ADC in the isolated section for each channel with the SPI/I2C digital lines fed back through an opto-coupler. Thanks for the feedback!
This is the common solution these days. In the past, they often used a voltage to frequency ADC which produces two pulse trains with the ratio representing the input. This easily works up to 12 bits and beyond.