how to get isolated inputs with adjustable Vih/vil thresholds and adjustable hysteresis

Hi - I'm trying to do something that may be... odd. I'm working on a logic analyzerish sort of project. I'd like to have 4 sets of 8 inputs. Each set would be completely isolated (say 500V isolation, though I could be convinced to go lower if needs be). I would think my max VIn would be about 5V or so. I would like to have pretty good control over what constitutes a low and high signal (as this would be used on many different systems with different voltage levels). Ideally I'd like to be able to be able to set the input low voltage or the input high voltage, and also set the hysteresis. I'd want these settings to be the same for all 8 inputs, but individually configurable for the 4 different sets. Everything will be getting controlled by an FPGA. So - this is how I'm imagining each set of 8 inputs:

Ideally, I'd find an octal comparator that had an adjustable hysteresis. I'd wire all of the inverting inputs together. I'd have a DAC with two outputs - one to the hysteresis pin, the other to the inverting inputs of the comparators. Then - if I wanted to set VIL I'd program in that voltage into the DAC and presto - it'd work. If I wanted to set the VIH I'd program in that voltage into the DAC and it'd work, but the output would be inverted (which would obviously be easy to correct for in the FPGA). This all seems fairly simple to me - though I'm sure I'm forgetting all sorts of important things. Like I wonder if any comparators like what I'm looking for even exist... Haven't found one yet - but I bet it's out there.

So - then I'd just have a bunch of digital inputs and outputs for this set of inputs. Getting isolated power to it isn't too bad - but what about getting the digital signals isolated? I'm hoping to hit 100MHz+ bandwidth. Ideally faster - but 100MHz I would be pleased with. The DAC signals will be plenty slow (I don't plan on changing them often)

- but the comparator outputs - those are the ones I'm worried about. A quick search on Digi-Key finds the fastest opto-couplers to be 50MHz. I'd ideally also like to keep size and cost down as much as possible. Size moreso than cost. I think I'd be looking at probably somewhere along the lines of 12 isolated signals/channel (the 8 inputs + 4 for the SPI DAC, assuming I use an SPI DAC) - so finding small optos would be awesome.

Is there a nice way around this problem? Does it sound like I'm attacking this from the right angle? Or am I completely off?



Reply to
Loading thread data ...

you could always roll your own with Gas photodiodes 1 ns risetime part should do the job about $40 each.


Reply to

fiber optic ethernet transceivers? and a short fiber.


Reply to

Try a transmission line transformer driving a fast comparator with a bit of hysterisis.

I'd wrap a bit of minature coax around a toroidal core or a small RM core. If you wanted to go for broke, you could use RG405 semi-rigid coax or Quickform86, but VMTX 50MTX or twisted enamelled wire would probably be good enough.

The low frequency cut-off frequency wil be fairly high, but you can use plenty of hysterisis on the receiving comparator to hold the output logic level between transitions.

With the right comparator you could get close to a GHz.

-- Bill Sloman, Nijmegen

Reply to

Firstly, consider putting a microprocessor on each of the 'sets' to hold and update all the signal gain and hysteresis or dead-band info, the things that are only going to change slowly. The communication to that microprocessor can be a simple optoisolated serial channel. If you use addressing in the serial commands, a single channel can drive all the sets.

The best high-speed isolation is still a transformer, and you might be able to drive with a flip/flop and receive with a Schmitt trigger so that only the fast rise/fall of a true transition changes the receiver state. It won't pass DC, but after a couple of transitions, the output will stay true to the input, in the logic sense. It might be useful to XOR the inputs to a test pulse generator, to ensure agreement as part of a reset process.

Also, there are very fast opto devices, in the multigigahertz range, if you go to fiber optics catalogs; sometimes you can serialize/ deserialize fast enough... look into AMD's 'Taxi' chips, for instance.

Reply to

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.