Conversion from 0-10V to resistance - Methodology

Question

Summary

Finding the most appropriate methodology for converting a variable voltage output to a variable resistance to mimick/replace a thermistor based sensor.

Use case/Example

There are two equally compelling and frequent use cases:

1. Providing several devices (air handling units, heat pumps, boilers, district heating exchangers, ...) within a building or building complex with the same outside air temperature (oat) information. (No deviation, no excess physical sensors and wiring.)

2. Manipulating the oat sensor value that is sent to a given device, as a means to control such a device from a supervisory system, if the device does not offer any other (better) methodology for achieving external control.

Problem

I have analog outputs (0-10 V DC +/- 0.5%, 12 bit or PWM at 0.01 Hz, 0.1 Hz, 1 Hz, 10 Hz, 100 Hz speeds) or digital outputs available.

I'd like to connect one of these outputs to the input on the device to be controlled, where an outside air temperature sensor would otherwise be installed.

My goal is to control the supply temperature setpoint for the device indirectly by feeding a "fabricated" oat temperature to the device. This will work because the device is oat compensated, meaning it determines the current supply temp setpoint based on the current outside air temperature.

For example, if the device is setup to deliver a supply temp of 50 degree C at an outside air temperature of -10 degree C, the device will do exactly that when the oat sensor input shows a resistance of approximately 25.8 kohm. So regardless of the actual current outside air temperature, if I generate a resistance on the sensor input of 25.8k, the device will deliver 50 degree C supply. Thus, rather than scaling my output into ohms or equivalent oat, I can simply scale it to the desired supply temperature.

The device in question expects an NTC 20 sensor to be connected to this oat input. Other details, like current, isn't published. But this is one specific example and one concrete solution I need, but I am also looking for a mostly universal solution, although I get the scaling needs to be adapted.

An NTC 20 used as an oat sensor means I'd need a range of 3-33 kohm approximately, and a resolution of at least around 100 ohm.

Changes do not need to happen fast, so solutions with capacitors are ok. From change to settled can be eg. 5 min.

The 0 V on both my supervisory system and the device to be controlled are connected to ground/PE/protective earth.

What is the best methodology for achieving this in a simple and accurate way?

Solutions considered

A. Digital potentiometer - Doesn't seem to like such a broad range, and also concerned with heat dissipation and drift? It also doesn't readily accept a 0-10 V input, and as such would need a custom circuit with fx a microcontroller and Modbus communication to the supervisory system.

B. DIY digital potentiometer - a binary array of 8 bit (128 for the least significant value) would give me the range and accuracy needed (32640...128 ohm at 128 ohm resolution, would equate +100...-14 degree C at <1 degree C accuracy). Seems over engineered and like a misguided solution though (compared to eg C or D below).

C. PWM + "some circuit" - I understand some combination of a capacitor, a resistor and a PWM output can give me what I need, but I can't find further information on it, probably because I don't know the name of this circuit?

D. 0-10 V or PWM directly - With common 0 V rails, I might be able to calculate the voltage the input on the controlled device expects at various temperatures and simply provide that voltage directly to the input?

E. LED/LVN (I think?) - Seems a possible solution, but I understand there are issues with long term drift. It also - like the digipot - requires custom circuitry. I don't like "maintenance-required-solutions", and with this I understand the LEDs may loose intensity etc over the years.

F. Servo/motorized pot - I've seen this suggestion a great number of times, but it seems equally silly as to build a temperature controlled box for the oat sensor of the device to be controlled to be placed in. I'd like to steer clear of "solutions" like these.

So for now, what I'm hoping for is clarification of the best methodology to do what I'm asking in a simple and economical way.

English is not my native language, so my suspicion is that the simple solution eludes me simply because I don't know the name of it.

Answer 1

If you have PWM, you can be synthesizing a resistance directly. All you need is an analog switch, like a 74HC4066 or CD4006. Parallel all switches in the package, and add an external resistance that adds to the switch resistance to yield the lowest resistance you want to synthesize.

The PWM will be multiplying the conductance of that resistor, i.e. PWM will be proportional to the reciprocal of resistance multiplier.

For example, if the "static ON" resistance of the switch+resistor is 100 ohms, then at 50% it will be 200 ohms, at 10% it will be 1kOhm, etc.

This works well.

Another approach is a PWM-controlled resistance multiplier: put a known reference resistance \$R_{ref}\$ in series with your synthesized pass element (e.g. a mosfet), and let an op-amp regulate the voltage drop across the pass element to be a multiple of that on the known resistance. Set the gain \$A_V\$ on the op-amp to be equal to the ratio of maximum resistance to the reference resistance. Then the PWM is used to chop the reference voltage to to the op-amp. The PWM 0-100% will map from \$1 \times R_{ref}\$ to \$(1+A_V) \times R_{ref}\$.