ARRAY electronic loads

[dolomike]

It would be more for analyzing performance of some of the power components and PCB in a design and not for testing closed loop control although there isn't much to the loop that would be affected anyway.

We do have switched resistance banks but would like to test with a constant current load ideally.

Mike

[rx8pilot]

I'm not quite sure a standard DC load will do what you want because the control loop will always lag behind and could mess with the control loop of the system you are testing. For this kind of tests I designed & build a switched resistor DC load myself. I can set it to a specific resistance and because its purely resistive it will keep that setting no matter what appears at the input.

That is what I did for my setup. Resistors and capacitors are dumb and predictable. I have an ARRAY DC load, but I am looking at a Chroma mainframe so that I can automate with LABView.

[LabSpokane]

Well, we normally run DC motors in automotive and heavy duty applications so this isn't a lightweight system. I think we would normally run CC or CR mode for most tests but some could be upwards of 800W. I believe this can be achieved by paralleling the devices in CC or CR mode. Some re-brands state they can be paralleled but I didn't recall seeing anything in the Array manual about it.

Voltage: 12V/24V nominal but some tests can be at 28V.
Current: ~10-15A but some tests may require almost 30A
Wattage: Most tests are done at 12V so we could do most of it with ~400W but if the parallel combos work, then we can get up to the ~850W for the extreme tests.
PWM Frequency: 5kHz-20kHz
Filtering: What type of filtering are you asking about?

I guess a lot depends on the current slew rate but the manual states 1mA/us~4A/us on CCH mode which is quite a range so not sure what the driving factor will be.

The transient generator can handle up to 50kHz so I would expect it could handle a PWM source upwards of that frequency as well.

Thanks for the feedback.

Mike

Could you or someone else explain how a constant resistance / current can simulate a motor load?  Not that a electronic load could do much better of a job, but unless you get some sort of a custom wind, typically a power resistor is wound to be non-inductive.

[dolomike]

We've looked at the Chroma stuff as well but I don't want to spend over $10k for a 1200W system and probably not use half the features. We aren't looking for pinpoint accuracy since we are using 10's of amps and not worried about mA resolution/accuracy.

To help simulate a motor, you do have to add in a comparable inductance. Although you won't ever get the same behaviour as a motor at least you can get it close enough to validate some aspects of the design. The nice thing about the DC load is the transient generator which could help with ripple testing to a certain extent.

Another way we have done it is to drive a motor coupled to another and connect that to a load bank.

I'd like to have some finer grained control over the load that a switched resistor bank can't easily do especially when operating at varying voltages.

Mike

[nctnico]

I'm not quite sure a standard DC load will do what you want because the control loop will always lag behind and could mess with the control loop of the system you are testing. For this kind of tests I designed & build a switched resistor DC load myself. I can set it to a specific resistance and because its purely resistive it will keep that setting no matter what appears at the input.

That is what I did for my setup. Resistors and capacitors are dumb and predictable. I have an ARRAY DC load, but I am looking at a Chroma mainframe so that I can automate with LABView.

That is rather crude. I came up with this (neatly controlled with a microcontroller):

[rx8pilot]

Definitely crude, but I needed it in one day and had to build it from parts on hand. Serves the purpose while being the ugliest piece of gear on the bench by a long shot.

Yours looks like its nearly ready for a production run. Nice.

Sent from my SAMSUNG-SM-G890A using Tapatalk

[forrestc]

Well, we normally run DC motors in automotive and heavy duty applications so this isn't a lightweight system. I think we would normally run CC or CR mode for most tests but some could be upwards of 800W. I believe this can be achieved by paralleling the devices in CC or CR mode. Some re-brands state they can be paralleled but I didn't recall seeing anything in the Array manual about it.
......
Filtering: What type of filtering are you asking about?

In relation to the filtering:  I guess I should have asked "are you using PWM to derive DC, or are you using it to drive a motor".

My experience is that a DC load doesn't really like a dynamic power source.  For instance, to illustrate I hooked my Array load up to my Rigol Power Supply, and set the load to constant resistance mode, and set the rigol to deliver at most 1A at 24V.   See the video below to see what happens.....  in short the load can't adjust correctly with this type of power source - it keeps trying but it doesn't succeed.



(Sorry about the shaky handheld Cell-phone video).

[dolomike]

We are using PWM to drive a motor - we're not generating 

forrestc - Thanks for posting the video as I think it answers my question. It appears that the DC load cannot adjust quickly enough to handle a switching source. Looks like I might have to go for some of the more advanced and thus more expensive models out there.... or I can look at creating a resistive load bank with some smarts like what nctnico has done.

I may also play around with an DIY electronic DC load circuit and see what kind of response I can get with it with some high speed component. I wonder what sort of speeds I could get?

Mike

[MarkL]

My experience is that a DC load doesn't really like a dynamic power source.  For instance, to illustrate I hooked my Array load up to my Rigol Power Supply, and set the load to constant resistance mode, and set the rigol to deliver at most 1A at 24V.   See the video below to see what happens.....  in short the load can't adjust correctly with this type of power source - it keeps trying but it doesn't succeed.

It depends on the design of the control loops in both the power supply and load.

I tried the same experiment on an ITT MX100TP set for 24V @ 1A, and an Array 3723A load set for 20 ohms.  It worked with no instability.

While I agree it's hard to beat the simplicity of a real resistor, I wouldn't rule out an electronic load just yet.


dolomike: What range of duty cycle for 5kHz-20kHz is needed?

In a quick test the above Array settles in about 40us to a 10A target setting.  It settles in about 10us to a 1 ohm target setting (@ approx 16A).  These numbers are from a simple switch closure, and they're too long for 20kHz, but it might do better with a MOSFET-driven steady state PWM.

Edit: Fixed model number.

[MarkL]

Here's a little more about applying PWM to an Array 3723A load.

A 10kHz square wave was generated by a beefy LIPO 4S battery pack (approx. 16V) put through a MOSFET switch.

Here is the load in 10A mode.  Blue is current, red is voltage across the load.  Until it settles at 10A, it's unreasonably high at almost 65A.



Here is the load in 1R mode.  For whatever reason, it settles much faster and with much less of a current spike.  There's an odd spike in voltage above the idle battery voltage indicating some inductance inside the load.



Just for comparison, here is the same plot but with the load disabled.  The PWM generator was manually triggered.  There's some current being drawn in the beginning due to a capacitor inside the load on the input and likewise, the voltage does not return to 0V because the capacitor is holding it up.



So, I think in resistance mode, this load could be usable at 20kHz as long as you know what to expect out of it.  Because it overshoots, it's certainly going to stress your circuit for testing.

You'll have to be the judge.  Maybe it's worth a try.  No idea on the parallel requirement, though.  I only have one.

Edit: Fixed model number.

[SharpEars]

Here's a little more about applying PWM to an Array 3732A load.

As far as I can tell, there is no such thing as an Array 3732A load. What is the correct part number?

[rx8pilot]

I have a 3710A - looks like the one in the video a few posts back.

incidentally, it's for sale if anyone needs one.

[Circuitous]

Here's a little more about applying PWM to an Array 3732A load.

As far as I can tell, there is no such thing as an Array 3732A load. What is the correct part number?

Well, I have an Array 3732A, it works quite well.  I got it from here: http://www.circuitspecialists.com/array-3723a-electronic-load.html

[MarkL]

Here's a little more about applying PWM to an Array 3732A load.

As far as I can tell, there is no such thing as an Array 3732A load. What is the correct part number?

Oops - sorry.  It's a 3723A.

[Makerman69]

We use several ARRAY products including three different models of their DC loads at our test facility. Can't say anything bad about them and there is significant cost saving compared to so called legacy brands. My understanding is this is a Taiwanese company manufacturing in China. Maybe the engineering is a bit on the upside compared to some of the mainland China makers.

[Zbig]

I've just got an ARRAY 3721A. I'm quite impressed with it so far but I'm a bit surprised by one design decision. Peeking under the heatsink assembly, I could see eight pass transistors in what looks like a TO220 packages to me. Granted, the heatsink is beefy, there are two fans and the airflow is tunneled within a metal sheet shield but isn't asking the poor TO220s to dissipate 50W each maximum (the unit is rated 400W) stretching it a bit? I'm pretty sure they knew well what they were doing but still, why not TO247? I could imagine the thermal compound going bad on one of them and the heatsink-mounted temperature sensor not even noticing as the localized hot spot from a single TO220 wouldn't affect the average heatsink temperature much. Or am I overthinking it? I realized the last sentence didn't make much sense but anyway, seems quite strange to use TO220 in such device.

[forrestc]

I've just got an ARRAY 3721A. I'm quite impressed with it so far but I'm a bit surprised by one design decision. Peeking under the heatsink assembly, I could see eight pass transistors in what looks like a TO220 packages to me. Granted, the heatsink is beefy, there are two fans and the airflow is tunneled within a metal sheet shield but isn't asking the poor TO220s to dissipate 50W each maximum (the unit is rated 400W) stretching it a bit? I'm pretty sure they knew well what they were doing but still, why not TO247? I could imagine the thermal compound going bad on one of them and the heatsink-mounted temperature sensor not even noticing as the localized hot spot from a single TO220 wouldn't affect the average heatsink temperature much. Or am I overthinking it? I realized the last sentence didn't make much sense but anyway, seems quite strange to use TO220 in such device.

I just picked a random FET datasheet with TO-220 and TO-247 options.  This one has 0.45*C/W junction-case thermal resistance listed for TO-220, TO-247 and D2PAK.   So, for heatsinked applications, it seems like it should be good (50W *0.45C/W = 22.5*C temperature rise, with a 175*C junction temperature max means the case needs to stay under 152.5*C)...   So, this doesn't seem like a lot of a stretch.  One would have to look at the thermals for the exact transistors and also the thermal resistance of the heatsink (case to ambient).

[Zbig]

Thanks for checking that. Actually, it's case to heatsink resistance I'm more worried about. Do they specify that as well for the one you've checked?

[nctnico]

IIRC you have to count on 0.5C/W for direct heatsink mounting (non isolated) when it comes to a TO220. After that you have to add the thermal resistance of the heatsink which can be another 0.5C/W even with air cooling. All in all that 50W easely adds up to a 75 degree C difference between the air and the die of the transistor.

[Jester]

Hello all,

I came across this thread while googling for a manual for a Kikusui PLZ300W electronic load.  I need an electronic load for a project I took on, and am on a tight budget.  I looked at the Array units and they seem nice.  I ended up buying the PLZ300W, on ebay, for $360US, including expedited shipping.

My main reason for making this choice was budget, and, in a previous life, I shared an identical unit with a couple of engineers and a couple of techs.  I was in that lab for ten years, and that load was no spring chicken when I got there.  It performed flawlessly for all the time I was there, and it was in high demand (lab was engineering department for a battery and charger manufacturer).

I didn't over-think it, and went with what I know.  I doubt it has the features many of you seek, but, as far as dependability and accuracy per dollar, it's a worthy consideration (if buying used).

I think we all love new gear, but I've been so burned by Chinese made goods, I generally do what I can to avoid them.  It may seem racist but, really, I've thrown away A LOT of Chinese made tools; frankly, I think there is some cultural differences, and they do not embrace the meaning of the word "quality" in the same way.  Regardless, I did just buy that 1052E.

I don't do much engineering these days, but have this one job, and need to set up a mini lab at home.  I'm trying to do it for under $1000US, so as not to suck all the profit out of the job.  I bought a Rigol 1052E direct from the manufacturer, two HP bench supplies (6289A, 6291A), and the PLZ300W, and, so far, with shipping, I've spent $833US.  I still need a simple, low-speed function generator, and an array of test leads, but I think I'll make my budget.  Considering that I've not been watching ebay for good prices, let alone hit up the HAM radio folks, but rather selected and purchased these items today, at the "buy it now" premium, I think one could do even better.

Anyway, my point is, 30 year old equipment doesn't tend to inspire confidence, and usually lacks great, modern features, but the old thru-hole gear is very dependable and very repairable.  Best of all, you can choose quality levels you couldn't otherwise get into the budget.  The one caveat is the dark years of electronics: the previous decade.  When Europe introduced the RoHS and WEEE legislation, it all really went bad.  The lead free solders, tin whiskering, delaminated PCBs (from the higher heat for new solders) made for some of the worst lived electronics in the history of, well, electronics!  I avoid gear from that era, whether it be an electronic load, or a car stereo.  You can keep it

So, there are some opinions from some random guy on the internet.

Best of luck in your endeavors.

Regards,
Mike O'

FWIW, I have a couple of old school Kikusui loads (PLZ72W and PLZ152WA), still working fine.

The one thing I really like about this old school equipment is the ease of use. Much like my go to Agilent E3615A power supply, you turn it on and adjust to the desired value. I had and sold an Array power supply nearly identical to the Array load being discussed, it worked very well functionally however the user interface sucked. The loud beep on every key touch was also annoying until I put a piece of tape over the transducer.

The one thing I did not like about the PL72W was the single line display, so I replaced it with a cheapo but accurate  ebay dual V/I display that fits in the original cutout.

[Peter_O]

I've a 3710 electronic load and just got the 3312 USB adaptor 2nd hand.

Simple and maybe stupid question: Do I need the 9male-9male cable null modem or 1:1?

[OiD]

Guess I'm going to do some necromancy here in this old topic.

I was doing some maintenance on my Array 3710A and did a complete tear-down. I solved the issue of bad voltage measurements by adding dome shaped washers.
I'm very happy with it, got it back around 2014 for 80€ when a company I worked at shutdown.

I've been looking at tear-downs of the 3711A and... they are identical. I have often run mine at 200W continuously and had no issues. Any higher than that and it disconnects the load.
So if the schematics are identical... and so is the actual device...

Is it only software that is limiting the power?

I think I saw an EEPROM next to the MCU, I doubt the actual code is stored there, even the 150W or 300W keyword... but maybe we could do EEPROM dumps of the devices and see if there is anything?

If anyone has the schematics for the 3710A and 3711A and they are correct/identical... 

Doubt I'll have time to dump my EEPROM in the coming weeks as it's in use.

 

[jacky1975]

Array 3710 can use Array 3711 software to adjust to 300W load.

[neverendingstudent]

I have a number of Array / Circuit Specialists 3710A electronic loads I got from a pallet of T&M gear I won at auction - I've been cleaning / testing / selling them one at a time on eBay.  They are pretty nice and, at least according to the specs, designed to be pretty hard to accidentally damage.  Over current / over voltage / over power protection, reverse polarity protection.  The 3710A is rated for 150W and I've tested them up to the 190W range when stress testing before eBay sale and they've held up fine.

With all that said, I also have a bunch of rather nice Japanese Kikusui PLZ334W loads... feel free to guess which ones I keep to use on my personal bench.

[MathWizard]

So I've been modelling and modding the Array3700 (maybe it's 3710), for now to use 8x IRFZ44N, so I'll only go up to say 40V max (they are from ebay too so I don't exactly trust their ratings).

For now, at least with 40V, I think I can make a version that goes up to about 150W continuous, and I figured I'd double it to 300W for really short tests or pulsed loads, I better calculate if the junction temp is still ok.

One thing I see is in this Array3700, is a lot of capacitance around the control loop. So like on the main Vref input , and on the voltage control op-amp for all the mosfet driver op-amps, and again on the output itself, they have chosen large caps.

So doesn't that really limit the testing it can do ? With all the caps removed or greatly reduced, the gain of the main voltage driver drops to 0dBV at 1MHz. With those caps in the pdf, the main voltage driver is 0dBV by 100Hz. In both cases the DC gain is up around 70dBV with the OP07's and a mod or 2 I've made.

I have tons of little switches, so it wouldn't be hard to low and high bandwidth setting. IDK how fast high current DUT's can respond tho, so maybe you don't want to try changing the load too fast.

But don't people hook up their signal gen's to E-load and do all kinds of HF stuff too ?