GW Instek PEL-3032E Constant Resistor values seem inaccurate, is this normal?

[CatNinja]

I just got a Lecroy T3EL50015P (OEM is GW Instek PEL-3032E) Electronic Load.

The Constant Current and Voltage readings are accurate, but for the life of me, I can't seem to get an accurate Constant Resistor reading.

Input is a E3631A set at 10.00V. Here are the readings on the PEL-3032E.

500 Ohm 10.005V 0.0020A
200 Ohm 10.005V 0.0290A
100 Ohm 10.005V 0.0790A
 50 Ohm 10.005V 0.1795A
 20 Ohm 10.002V 0.4805A
 10 Ohm 10.000V 0.9815A

The readings are only reasonable at 20Ohm or less. The E3631A voltage and current readings match the PEL-3032E, so I think those are accurate.

My understanding is that the resistance is simulated, and the PEL-3032E just sinks the appropriate current to match the resistance setting. Since it measured the voltage correctly, and it knows what the resistance is set at, why is it not sinking the correct current?

Am I misunderstanding how electronic loads work? Or am I expecting too much of the accuracy of the resistance settings on this type of device?

[alm]

I don't have experience with this specific load. In general I don't expect electronic loads to be particularly accurate in constant resistance mode, although clearly the 500 Ohm current is off by an order of magnitude. Does it help if you increase the voltage? I could imagine it's having difficulty with the very low current. What does the data sheet say about constant resistance accuracy? Is there some offset off the form R * X mA to be added?

[HKJ]

Most electronic loads are not very precise when you get down to a few mA.
It may have a few mA in offset on the current measurement/setting, check the specifications.

[nctnico]

It is also very possible the DC load is oscillating because the power supply likely has a weird, complex impedance due to the output capacitor.

[Martin72]

Agree to nicos post.
We got several for the first time "mysterious" problems when connecting electrical loads to switching supplies, they interacting together in a resonant way, mostly observed by the cheaper PSUs.
On linear regulated ones we never got any issues in that direction.
At the lab we got tons of "Zentro" loads and several from Rigol, when time will allow it, I´ll do some tests with linear and switching supplies.

[alm]

Oscillation as mentioned by ntcnico is indeed another thing to check. The somewhat slow control loop of the electronic load combined with the power supply control loop can indeed easily oscillate, see for example this video:
https://youtu.be/IaWgF1SORkk?t=2057. Note that Dave calls the constant resistance mode the "power supply killer" mode

[CatNinja]

FYI, I ran some tests using a E3610A linear power supply, and got similar results.

1st voltage is from from E3610A, 2nd voltage is from the PEL-3032E. The constant resistance is set at 50 Ohms.

2.50V 2.508V 0.0260A
5.00V 4.992V 0.0765A
10.00V 9.983V 0.1790A
15.00V 15.000V 0.2815A

So it looks like the PEL-3032E is very inaccurate until several hundred milliamps or so.

What I don't understand is that the voltage and current readings seem accurate, so it should be dead simple to measure the voltage, calculate the required current from the resistance setting, and then set the sink current appropriately. Maybe I'm just thinking like a Software Engineer.

The spec say the resistance accuracy is T +/- (1% of set + 0.06S) + 0.002mS. I'm not sure how to read this. 0.06S is 16.67kOhms, so ignoring the time constant and % of set, is it really +/- 16kOhms???

Is there anything else I should check? Like hooking up a scope to see if it's oscillating?

Does anybody with a similar Electronic Load (like the Siglent DSL-1020X or Rigol DL-3021) ever use it in Constant Resistance mode, and do you see the same inaccuracies?

[nctnico]

I doubt that the current readback is that inaccurate. It would be nice if you can at least check the current with an oscilloscope. A 1 Ohm resistor will do just fine as a current shunt. I used to have some Array DC loads and these also showed bad behaviour when used in constant resistance mode.

[alm]

The spec say the resistance accuracy is T +/- (1% of set + 0.06S) + 0.002mS. I'm not sure how to read this. 0.06S is 16.67kOhms, so ignoring the time constant and % of set, is it really +/- 16kOhms???

It's a bit confusing, but they are actually specifying conductance, not resistance. So if you set it to 50 Ohm, or 0.02 S, then the actual conductance should be between 0.02 S - (0.3% * 0.02 S + 0.06 S) + 0.002 mS and 0.02 S + (0.3 % * 0.02 S + 0.06 S) + 0.002 mS, or anywhere from 12.5 Ohm to infinity. For lower resistances it's better: for 1 Ohm the uncertainty ranges from 0.94 Ohm to 1.06 Ohm. So this load really sucks for high resistances, but that's not typically what you use a load for.

Other loads have better specifications, but often under the condition of some maximum voltage or current. For example this Rigol load specifies in a footnote that for CR mode the input voltage shall not be lower than 10% of full scale (which might be something like 0.4 A and 1.5 V depending on how they define this). This Siglent load has the same condition. They specify the accuracy similar to the PEL-3032E, but explain it better (see the note under [1] in the datasheet).

[CatNinja]

Thanks for the explanation. This makes a lot more sense now. Math was never my strong suit. Basically, I can't just use +/- the error, but have to convert to conductance using 1/R, then calculate the error, then convert back to resistance using 1/S.

So, it looks like this class of load doesn't really have that useful of a constant resistance setting. Resistance has to be fairly low to be accurate (about 1 Ohm according to your calculation), and current and voltage has to be > 10% full range (looks like 1.5A and 50V?), which matches my observations above, where it got more accurate as the resistance got lower, and the current got higher. I just didn't expect it to be that far off at low values.

I still think that they could do much better in software. For example, I can manually simulate a resister by looking at the voltage reading, then calculating what the current should be based on my desired resistance, then set that value in constant current mode, and be within a percent as the CC mode is pretty accurate. But that's just the Software Engineer in me speaking. I'm guessing that errors in the hardware control loop don't simply add up this way.

[alm]

The 10% of full scale condition was for Rigol and Siglent loads. I didn't notice it in the GW-Instek data sheet. But it wouldn't surprise me if this load is also more accurate for higher voltages and currents.

Naively you'd think it's the multiple of voltage read back error and current programming accuracy. Plus some sort of leakage term at high resistances (I think that's what the + 002 mS is about). I haven't done the math to see if that matches.

Note that these should be the worst case specs. You're unit might well perform better. Still, I think it's odd that they offer resistance in the kiloOhms with such barndoor-wide specs. Maybe send Lecroy or GW-Instek an email to ask them to check. I wonder if they slipped a decimal somewhere in the data sheet.

[HKJ]

Note that these should be the worst case specs. You're unit might well perform better. Still, I think it's odd that they offer resistance in the kiloOhms with such barndoor-wide specs. Maybe send Lecroy or GW-Instek an email to ask them to check. I wonder if they slipped a decimal somewhere in the data sheet.

As I wrote before the issue is probably with the current. Use a higher voltage and it will handle higher resistance, because the current is higher.

[alm]

I agree, but that doesn't make the specifications that say the unit has resistance ranges up to 50 kOhm, but there might be 16 Ohm in parallel with that 50 kOhm, less ridiculous. Maybe they missed a prefix and in should be 0.06 mS?

[Hexley]

The data in the table below was taken from another GW Instek product, the GPP-4323 power supply/active load. While this 32V/3.2A load is not in the same class as the PEL-3032E, perhaps the data may be of interest to some.

Channel 1 was set to Load mode, constant resistance, and the load was stepped from 500 ohms to 10 ohms.
An external voltage source was applied, and the voltage at the load terminals was measured with a 5.5 digit voltmeter.
The current flowing into the load was measured with a 6.5 digit milliammeter.
The equivalent load resistance was off by just under 4% at 500 ohms, and was much less than 1% for loads <50 ohms.

GW Instek GPP-4323 Load Behavior (Channel 1) CR Vload Iload Req. Err 500 9.9535 19.17 519 3.84% 200 9.8758 48.31 204 2.21% 100 9.7474 96.42 101 1.09% 50 9.8934 197.2 50 0.34% 20 9.7299 486.6 20 -0.02% 10 9.4674 947.9 10 -0.12%

[Hydron]

I'll have a quick go with my cheap AliExpress load later tonight, see what results I get out of that (also a 500V load, but nowhere near as nice as the Lecroy one, kinda regretting not grabbing one of those when I had the chance).

Btw don't try and use the GPP-4323 load in constant resistance mode for anything dynamic - the resistance simulation is only done at 10Hz or something (on the earth referenced processor, not the floating one doing the fast current/voltage control loop) so it's ugly and ~unusable. Constant current/voltage don't have the issue. Load mode is basically just a nice bonus on the PSU, it's not really a first class feature.

@CatNinja: I also invite you to open up the Lecroy load and see what's going on in there - it's certainly not built down to a price!

Edit:
Checked my DCL6104A @ 500 ohms 10Vin, it says 20mA but actually draws ~17mA, with a fairly large ripple (a few mA RMS with spikes of a little under 10mA, this load is not great at low currents!). -3mA offset is the same at any resistance or in current mode though, so it's just a calibration issue, nothing fundamental. For some reason the offset seems to change to +7mA when I use 4-wire mode (a little hacked addition I made to the 6104A), which makes less sense. The load also disagrees with being driven by my SMU when in CR mode, but that's not a big surprise (SMU is fussy).

[CatNinja]

Thanks to everyone for all your help. I have opened a ticket with Lecroy support on this issue, and I will report back on what they say.