Need help with an automated DAQ system that would test 100 fuses

[billthebuilder]

Hi, I am trying to characterize the IRL performance of a SMT fuse. I want to design a PCB that would allow me to do Kelvin (4-wire) resistance measurements on a quantity of fuses (say 100). Ideally, it would test each one over a range of current and over a range of temperature and record that data through the DAQ. My trouble is imagining a way that I can do a 4-wire measurement on many fuses. Either I have one 4-wire setup per each fuse or I have one 4-wire setup that repeats the test on each fuse independently.  Any ideas on how I could do that?

Also, any recommendations on an affordable DAQ system?

I will add more info on the current and temperature range and the specific fuse to be tested soon.

Thank for your help in advance.

[Vgkid]

Are you going to test them until the blow, and you want to characterize them that way, or just measure the resistance?

[branadic]

Sounds like the classical use of a PXI system, unless you have a digital multimeter with multiplexer, that switches the 4 wires all the way around the devices under test.

[Vgkid]

Now that I have better read your OP, it looks like brandiacs method will be best. I would have put about 10 in series with a constant current source, and measured the voltage. Then computed those to find the resistance.

[ap]

If you want to go cheap but dont want to build something on your own, you could use 3499s as the switch (they can be bought at very low prices) for both voltage and current (if you go higher current, use power relais driven by the 3499).

[daqq]

Well, you don't need to do a full 4 wire measurement for a 100 fuses. You only need one current source, connect them in series and then measure the voltage on them. That'll limit the amount of wiring needed. It's a 4-ish wire measurement. So, 200 sense signals + 2 force signals ought to do it. Just a large mux should do the trick.

[billthebuilder]

Are you going to test them until the blow, and you want to characterize them that way, or just measure the resistance?

I want to measure the resistance, don't need to test them until they blow. Our input voltage will be ~12 - 17 V and I want to test them at 1 A, 50 A and 120 A over a temperature range of 70 °C - 100 °C. The fuse is expected to have a resistance of ~0.9 - 1.3 mOhm, so we were thinking we had to do 4-wire measurement to accurately measure the fuse's resistance. Better ideas?

Also, we want to measure the temperature of each fuse during the test.

[billthebuilder]

Sounds like the classical use of a PXI system, unless you have a digital multimeter with multiplexer, that switches the 4 wires all the way around the devices under test.

Thanks. A PXI system was something I was considering. Although, I want to measure the fuse's resistance, which is very small. In this case, the fuse's datasheet states that is expected to have a resistance of ~0.9 - 1.3 mOhm, so we were thinking we had to do 4-wire measurement to accurately measure the fuse's resistance because it is so small, this would eliminate the error introduced by the resistance of the voltmeter leads, had we just put a current through it and tried to calculate resistance that way. Our input voltage will be ~12 - 17 V and I want to test them at 1 A, 50 A and 120 A over a temperature range of 70 °C - 100 °C.  So if the input current was 1 A and the fuse resistance is ~1 mOhm, then I would need an ADC with a resolution to measure well below 1 mV. Better ideas?

Also, we want to measure the temperature of each fuse during the test.

[Marco]

If it's just 100 fuses you won't be able to earn back much engineering time over doing it with some ugly hacky test bench setup and manpower.

[billthebuilder]

Well, you don't need to do a full 4 wire measurement for a 100 fuses. You only need one current source, connect them in series and then measure the voltage on them. That'll limit the amount of wiring needed. It's a 4-ish wire measurement. So, 200 sense signals + 2 force signals ought to do it. Just a large mux should do the trick.

The reason we concluded we needed to do 4-wire measurement was because the fuse's resistance is very small. In this case, the fuse's datasheet states that is expected to have a resistance of ~0.9 - 1.3 mOhm, so we were thinking we had to do 4-wire measurement to accurately measure the fuse's resistance because it is so small, this would eliminate the error introduced by the resistance of the voltmeter and ammeter leads, had we just put a current through it and tried to calculate resistance that way. Our input voltage will be ~12 - 17 V and I want to test them at 1 A, 50 A and 120 A over a temperature range of 70 °C - 100 °C.  So if the input current was 1 A and the fuse resistance is ~1 mOhm, then I would need a DAQ with an ADC resolution to measure well below 1 mV. Better ideas?

Also, we want to measure the temperature of each fuse during the test.

[billthebuilder]

If it's just 100 fuses you won't be able to earn back much engineering time over doing it with some ugly hacky test bench setup and manpower.

Marco, this would be 100 fuses at a time but I think the goal is that this test would be repeated over the course of production to ensure that our fuses are meeting a crucial specification. Do you mean to say, buying a system would be better than building one? What's the solution you would suggest?

[Marco]

Marco, this would be 100 fuses at a time but I think the goal is that this test would be repeated over the course of production to ensure that our fuses are meeting a crucial specification. Do you mean to say, buying a system would be better than building one? What's the solution you would suggest?

Okay, so more than 100 fuses. I was suggesting just bodging together a couple of fuse holders with measurement wires coming out, taping thermocouples to the fuses, putting it in a toaster oven and running all the tests manually But obviously you need something slightly more sophisticated.

How cost sensitive is this? Do you have a Labview license available? How reliable does it have to be? (ie. can you use second hand NI hardware which might not be readily replaceable if it breaks, requiring the attention of an engineer to get it running again with something more up to date.)

The NI DAQ cards tend to have differential inputs BTW.

[billthebuilder]

Marco, this would be 100 fuses at a time but I think the goal is that this test would be repeated over the course of production to ensure that our fuses are meeting a crucial specification. Do you mean to say, buying a system would be better than building one? What's the solution you would suggest?

Okay, so more than 100 fuses. I was suggesting just bodging together a couple of fuse holders with measurement wires coming out, taping thermocouples to the fuses, putting it in a toaster oven and running all the tests manually But obviously you need something slightly more sophisticated.

How cost sensitive is this? Do you have a Labview license available? How reliable does it have to be? (ie. can you use second hand NI hardware which might not be readily replaceable if it breaks, requiring the attention of an engineer to get it running again with something more up to date.)

The NI DAQ cards tend to have differential inputs BTW.

We can spend some money on it, probably could get Labview. Would like it to be reliable enough that we could send it to our manufacturer after and they could run the tests alongside production. I'm figuring I will need to go with a NI system but trying to learn more about DAQ options to figure out what I need.

[alm]

The reason we concluded we needed to do 4-wire measurement was because the fuse's resistance is very small. In this case, the fuse's datasheet states that is expected to have a resistance of ~0.9 - 1.3 mOhm, so we were thinking we had to do 4-wire measurement to accurately measure the fuse's resistance because it is so small, this would eliminate the error introduced by the resistance of the voltmeter and ammeter leads, had we just put a current through it and tried to calculate resistance that way. Our input voltage will be ~12 - 17 V and I want to test them at 1 A, 50 A and 120 A over a temperature range of 70 °C - 100 °C.  So if the input current was 1 A and the fuse resistance is ~1 mOhm, then I would need a DAQ with an ADC resolution to measure well below 1 mV. Better ideas?

Using a current source plus voltmeter is exactly what a 4-wire measurement from a DMM does. If you wire multiple fuses in series, the current will be equal for every fuse as long as the input impedance of your voltmeter is much larger than the impedance of the fuse (should not be difficult) (KCL). Series resistance in the wires to the current source are irrelevant assuming it has sufficient compliance to overcome the resistance. If you use separate wires/traces for the voltmeter (Kelvin connections), series resistance in the connection to the voltmeter should not matter either (again, assuming sufficiently high input resistance). I agree with daqq that this could significantly reduce the switching complexity (since you only need to switch two terminals, and each terminal can be used for two measurements).

If you would wire five fuses in series and connect a current source across the two ends:

Code: [Select]

--[---]---[---]---[---]---[---]---[---]--
|       |       |       |       |       |
A       B       C       D       E       F
Then terminal 1 of your (differential) voltmeter would only have to connect to nodes A, C and E. Terminal 2 would have to connect to terminals B, D and F. So two three-pole single-throw switches (or six channels in a relay matrix) would be sufficient in this example, instead of a five-pole quadruple-throw switch (or 20 channels in a relay matrix).

As for the low voltage drop, this problem exists regardless of how you source the current (the only way to change it is to increase the current). So you need a sufficiently sensitive voltmeter regardless (be it a DMM with external mux or DAQ, possibly with an external pre-amplifier).

[ap]

If you want to achieve decent accuracy, you need a nanovoltmeter in conjunction with a decent current source and any suitable mux (be it NI or Keithley 7001 or Keyight 3499 or whatever). Measureming miliohms with a standard precision meter (even 3458A) is not recommendable. Of course if your uncertainty requirements are low, maybe something else will do, but does not sound that way, you should do your math to determine that.

[Kleinstein]

If it is about a lot of fuses, the fuse holders also have to be reliable to stand changing the fuses often and fast. For highest accuracy one might to get a 4 wire contact directly to the fuse and thus might not be able to use standard fuse holders.

Not sure one should really have 100 channels, unless you really have a shit lot of parts to test. Even than a board with easier to change fuses and less channels might be more useful than a large number of fuses in the board. The more normal scanners are more like 16 or 32 channels. With the low voltages and low source impedance normal CMOS switches / MUX should be good enough, just in case you want the switches as part of the board.