Controlled Resistor emulator / process meter

Wow. I woke up at 3AM with a new concept for an electronic load box. We have orders for 16 and no design, actually no specs! It would have to be high power and needs to behave well with PWM drive. It would look like a solenoid or a torque motor or a servo valve to a controller.

I have a different simple circuit for a fake resistor, using an opamp and an MDAC, which I could share. But you may as well use a DPOT.

Yes, I was thinking about a DPOT. Maybe just an eval board, with a USB to SPI converter, and then it's easy and fast

MDAC could also be an option

Now you say it, I have a Siglent scanning multimeter, so could arrange the relays in R2R fashion. That would be fast and sure to work, with no capacitance at all and floating. Nice hint :-)

Could just buy some cheap relay modules from Aliexpress...

Very nice, thanks Lasse

I found this on a Danish site:

During the search I also found a USB controlled 8 channel Digital potentiometer. Quite nice:

SPI interface evaluation module:

If you know it's not gonna take a lot of current, a resistor is just a current controlled by a voltage; LM13700 or any equivalent OTA has that basic function. You have to identify one terminal of the resistance that is not grounded nor driven (that's where the current source/sinks from) and connect the differential input (with suitable attenuation) to both terminals. And, there's the small matter of nonlinearity for large voltages on the differential inputs.

The upside: it's cheap. About a dollar. The downsides: you gotta power it, and it has compliance limited by its power rails, and the bias current has to be supplied to a negative-rail current mirror input. If the thermistor is balancing a bridge, that bias current controls the resistance (actually trans-resistance, before you connected the output to one voltage-sense input pin).

It depends on how the measurement is made. The NTC could be fed from a standard pullup scenario, then a current source would work. It could be multiplexed, it might not work. It could be a slope converter type, would not work. Could be a bridge configuration, higher voltage, would probably not work. The resistor method with relay is simple and guaranteed to work at almost any voltage and drive method.

Vactrol? (Though I doubt they're still made.)

Of course you could just put two NTCs on a small Peltier (TEC), close a feedback loop on one, and connect the other to the heat pump.

I usually do temperature control loops by putting a copper pour next to the TEC (with HS grease), with one side hanging out a bit, and soldering one end of an 0603 NTC to it, right next to the TEC.

One of those on each side, with an aluminum spreader to make sure the two sides track each other, would be easy to hack together in an afternoon, including the FB loop.

Cheers

Phil Hobbs

Yes, that also crossed my mind. Never tried that before, but could be a fun exercise.

Could do it mechanically, a servo motor on a dual gang potentiometer. One potentiometer connected to the regulation loop, the other to the heat pump.

motorized potentiometers are available off the shelf

you can still get them,

Could be easy to compensate though, just record what the heat pump reports the outside temperature is, and make a table lookup

IIRC you used to be able to get two in a package - one LED with two LDRs

- and you could wrap one LDR around an op-amp to linearise the other one.

Why would multiplexing be a problem? There's a bit of imbalance if the current has no where to go, but you could clamp it somewhere safe.

Would work fine, as far as I can tell. It might have a current-limit issue if the slope is very fast (high current, low duty cycle doesn't self-heat the original thermistor), though. My proposal is NOT a fixed current source, it uses the differential input terminals to make a resistance-like element, so if the original thermistor made a R-C exponential decay, so will the OTA circuit,

If the thermistor is 1k to 15k, and it doesn't self-heat horribly, it's NOT driven by 'higher voltage'. That's only a problem if your power supply for the chip isn't ground-referenced at the appropriate level.

For a slope converter you have a rising voltage and falling voltage, or it could be a more advanced method where a fixed resistance is used for reference, and only the time deviation is observed to calculate the resistance. Such a converter could be fast, so a emulated resistor needs to have be faster with no overshoots or other settle times.

Your right it could work, it might not work.

The temperature circuits I have seen/have designed, are all driven from voltages below 5V, so yes, only a slight risk it would not work.

is it a one-of ? it shouldn't take many minutes with a scope and a few resistors to figure out if it's just a pullup and you can just inject a voltage

They may move to another heat pump, so I need it to work nomatter how the temperature measurement is done. But, your right, could be interesting to see how they do it. One always learns from other designs :-)