EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: gaddra on January 03, 2015, 07:20:35 pm
-
I'm currently working on a MEMS driver with an output of 0-230V on 128 channels (using the HV256). The high level system looks like this:
(http://i.imgur.com/4qudvgD.png)
Each XMP has 64x 0V to 4.096V DAC channels connected to the HV amplifier.
The HV amplifier and each XMP will have isolated 12V power inputs, however the amplifier is not isolated from the XMP's.
I was wondering what the best way to protect the XMP's in case of mishap on the HV amplifier- it's much more preferable to damage the amplifier than the XMP's.
-
I'm guessing you mean this HV256 (http://www.microchip.com/wwwproducts/Devices.aspx?product=HV256), but what are these XMPs? Are they going to be on the same board as the HV256, or separate modules?
Presumably the worst case failure would be the high voltage supply being applied to one of the DAC channels, right? If you're going to use the full current output capability of all 128 channels of the HV256s your HV supply needs to be capable of around 100mA, right? you could protect each DAC channel with a pair of rail clamp diodes and a series resistor (standard overvoltage protection). Except that to keep the dissipation in the resistor below, saw, 1/4W, with the output shorted to 230V, you need a resistance of 225K to limit the current through the resistor to 1mA. What does that do to your signal integrity in normal operation? You'd have to balance the amount of power dissipation you're willing to tolerate/able to fit on the PCB against your output impedance requirements. You could put a fuse in series with a lower value resistor, but the lowest current commonly available SMD fuse AFAIK is 50mA, which means briefly dissipating 11.5W.
Another option would be to detect when one of the DAC channels encounters an overvoltage fault and shut down the HV supply to limit the duration of the fault.
-
Don't clamp to the power supply! Most regulators cannot sink current. Use (low leakage) transzorbs or zeners instead.
-
I prefer to clamp to the supply, and use a big TVS there instead. If nothing else, all inputs will be clamped to the same level. I like consistency. I suppose this will save cost as well, since clamp diodes are cheaper.
Tim
-
you need a resistance of 225K to limit the current through the resistor to 1mA.
A depletion mode MOSFET could do it with a couple of K.