Comparator with open collector "precise " hysteresis

A good thing to do is to use a comparator that has built-in hysteresis, and stop playing games with external resistors. For example, I just used Microchip's MCP65R41 in a design. This comparator includes a built-in voltage reference, and yet it comes in a sot-23 package and only draws 2.5uA. The MCP65R41 has push-pull output, and the MCP65R46 version is open-collector. Both versions have 3.3mV of hysteresis. What's more, they only cost 50 cents.

Open _drain_ ;-) ...Jim Thompson

Precise hysteresis can be done with two comparators and an r-s flop.

That's how I do it in all my custom chip designs that need precise trip points.

In fact, if you dissect the venerable 555 timer, that's exactly the trip-point architecture you will find.

Compared to analog-based Schmitt's, that approach is much more tolerant of noisy transitions. For example, see NoiseBlank.pdf on the S.E.D/Schematics Page of my website. ...Jim Thompson

In a built-up circuit, you still need something to prevent the comparators from oscillating on slow transitions.

Cheers

Phil Hobbs

In a full-custom approach the R-S FF is actually a part of the comparators, the R-S FF blanks the appropriate comparator. ...Jim Thompson

Do you? Just let them oscillate!

"John Larkin" ha scritto nel messaggio news: snipped-for-privacy@4ax.com...

Thanks for all the good ideas and hints

If you want a more 'perfect' square wave, for instance as an element of a square/triangle generator, you can employ MOS switches (analog switch, such as CD4016) with symmetric power supplies. That, or simply using a CMOS comparator, or gate, works with light loads.

Or drive, through a full-wave diode bridge, a current regulator (jFET with gate/source shorted together) . The other AC terminal of the diode bridge goes into a load resistor to ground. V across the load resistor will be symmetric about ground. That second solution is a good reason to want a two-terminal current regulator (as discussed here recently).

Sure. Some latching comparators have the latch function in the input structure, so that you can drive it from the comparator output without the thing going nuts. Lots of others will misbehave badly if you try this.

If you have one of the right ones (the MAX900 used to be a fave of mine), you can prevent oscillation by AC coupling from the output to the latch pin. A sufficiently quick RS FF could do that too, but getting it right is a bit harder than just putting a large resistor from output to noninverting input.

I can see the attractions of the two-comparator + FF approach--it doesn't push so much crap backwards out the inputs, for one thing. That's one of the main attractions of the AC-coupled latch approach as well.

Cheers

Phil Hobbs

You can do that, assuming that the period is slow compared with the setup and hold times of the FF. I'd expect that it would spray around a certain amount of EMI that may get into the other channel, especially if you're using a dual package to reduce the board space penalty. If the desired hysteresis is small, you could have both sides going nuts together, no? A decent latch function with a bit of AC speedup is a win.

Cheers

Phil Hobbs

Hey, Jim, it's a wonder I didn't revert to open-plate!

I haven't done any tube designs in years, but I could easily revert ;-) ...Jim Thompson

In my youth...

...Jim Thompson

Another approach that I don't see mentioned here is to feed the comparator output (with a pullup resistor, obviously) into the control input of an analog multiplexer (eg. 1/3 of a 4053) and use that to select a comparator input voltage from a voltage divider off a reference.

For example if you want the comparator thresholds to be 1.00V and

So your input goes into one comparator input The common on the mux goes to the other input

You can reverse the output by swapping inputs to the comparator *and* the multiplexer (feedback must always be positive though).

The advantage is that the comparison voltages don't depend on the the supply voltage or saturation output voltage of the comparator, provided you get a valid logic level out of it.

Best regards, Spehro Pefhany, Trexon Inc.

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Get...with the times? Erm...

Tim

...Jim Thompson

ll.

Seems I tested some of that 900 series a while back with an HP synth ultra- slow 5 minute input transition, monitoring the output with a LeCroy Gsps tr ansient capture- there was nothing- just an ultra clean switch. IIRC minimu m internal guaranteed hysteresis was 8mV. Imbeciles diss Maxim but the performance of their product line has always b een excellent.

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Here's mine,

They're just sitting around (the nuvistors) because I don't happen to have any sockets. :(

Tim