A couple of my recent proposals involved products that have an external resistor connecting a pin to ground. My test system has to be able to test different resistance values. I thought I'd talk through a few options.
The most obvious option is to get some resistors and switch them in. You might need a small circuit board to hold the resistors and maybe you add a transistor or a relay for switching. Alternatively, you can go off the shelf for a sinking digital output or a relay. In all cases, check the on resistance for your transistor/relay. Sometimes, the cheaper ones are rated for an on resistance of >10 ohms, which may not be accurate enough for your test. In general, for the resistor accuracy, I buy the best possible. For instance, I can buy an 0.01% 1k resistor for $2.05, which I think would be worth it.
If you go this route, I believe there should be some sort of regular maintenance on the fixture with your resistance. Maybe annually or so, just doublecheck that your reference resistor is in line with your expectation.
If you have a spare voltage measurement channel in your system, I would definitely use it to measure the voltage across resistor as that can help diagnose what may be wrong in the product if there's a failure. For example, if the voltage is not the expected value, then maybe there's extra resistance in the product; if it not the expected value, maybe there's an issue with a voltage reference or some other component.
If you only have to test one or two values, I think this is the most cost effective way to go. I would just keep in mind that it's not really flexible so if version B has different specifications (or maybe an improved accuracy requirement), changes to the test system hardware may be necessary.
Another option is to get a programmable resistor. I have seen these with accuracy less than 0.1% and greater than 5%. This is definitely a good option if you want more flexibility in your system. Also, most programmable resistor boards have at least 4 channels, so that might be a good fit depending on how many measurements are needed.
It's possible that a programmable load might work for your application, but in general, those are designed for high power applications and if your resistance measurement need is for, say, a thermistor, a programmable load is unlikely to have the necessary accuracy.
My last idea here is a source measure unit, which might be the most expensive option (at least $3k). In the majority of designs that I've seen, the resistance input completes a resistor divider circuit and the voltage is measured. So one could use the SMU to sink current and then you can use the voltage sense as a sanity check. The obvious advantage is that (most) SMUs are very accurate (sub 1uA or sometimes 1nA accuracy). Another advantage is that you don't really have to worry about series resistance in your cabling. Some disadvantages are SMUs usually have just 1 or 2 channels and, of course, you're likely to be around $4-5k in cost.
Hopefully, this will help spur some thoughts the next time you have this kind of measurement.
We just sent you an email. Please click the link in the email to confirm your subscription!
OKSubscriptions powered by Strikingly