Author Topic: Please recommend an electronic load.  (Read 1324 times)

0 Members and 1 Guest are viewing this topic.

Offline Saneoc

  • Contributor
  • Posts: 8
  • Country: ro
Please recommend an electronic load.
« on: September 22, 2018, 08:17:48 am »
Hello.

Does anybody have experience with electronic loads loading PWM outputs?

I am searching for an electronic load with extremely fast saturation recovery time.

There is a high side switch with overcurrent protection. It is driven by PWM with the frequency of 200-400Hz (automotive light bulbs controll ). I need to use the electronic load to check its current sense behaviour.

Untill now I used an electronic load from BK Precision (model 8514) but there is a problem:
  Imagine there is a setting of 1A on electronic load and it is on. When the high side switch is fully on - there is a current of 1A pulled from the electronic switch. During  the off period the eloads output goes into saturation because it tries to pull 1A but the switch is off.
Right at the next activation of the switch there is a huge spike in pulled current ( during the period of saturation recoverry time). This spike causes the overcurrent protection to deactivate the output.

I did some measurements with a workarround of using two series high power diodes parallel to the switch. When the switch is active - current goes through it and when it's not active -the curret goes through the diodes. In this case eload never goes into saturation but also the output voltage is always high. That also affects the behaviour of the system.

Can anybody please recommend me a fast eload about 40A and 40V?

What is your experience with electronic loads and pwm signals?

Thank you for your time.
 

Offline JohnPi

  • Regular Contributor
  • *
  • Posts: 73
  • Country: us
Re: Please recommend an electronic load.
« Reply #1 on: September 23, 2018, 08:04:14 am »
40 V * 40 A is a lot of power, but if you keep the duty cycle low, it is manageable.

Rather than use a commercial e-load, I would use a power FET or BJT (e.g. 2N3055) as a current source. If you don't need great saturation voltage, you could just put ~ 2V/40A = 50 mohm (low inductance  resistor) in the FET source to GND, and hold the gate at about 5 V. It will regulate a reasonably constant current of (VGATE-VTHRESHOLD)/50mohm -- you can calibrate as needed.  With 40 A, you'll have to be careful about ground offsets, but I assume you understand this if working at these power levels.

Alternatively, if you have another 40 A supply available and a 40 A diode (or schottky) -- GND the supply to the e-load; connect the diode anode to the supply and cathode to the e-load. This will keep the e-load always operating. Choose supply voltage (few V) to match the diode drop + e-load VSAT requirements. Depending on your switch turn-on and -off rates, you may still run into problems because of parasitic inductances and need to keep the wiring compact and short.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 15430
  • Country: nl
    • NCT Developments
Re: Please recommend an electronic load.
« Reply #2 on: September 23, 2018, 10:12:54 am »
For these kinds of tests I have build my own DC load which uses fixed resistors which are switched by MOSFETs. I can set it to a fixed resistance which is the same regardless of the load without the lag of a control loop.

Another option would be to use a DC load with a combined constant voltage / maximum current mode (CC+CV). In the constant voltage mode the DC load will switch itself off when the voltage falls below the set voltage. With an additional current limit the DC load will allow the voltage to rise. This way you setting limits to the region where the DC load works.

This will probably get you into the range with the more expensive ones. GW Instek sells DC loads which may do the job: https://www.gwinstek.com/en-global/products/detail/PEL-3000E Maybe an Agilent/Keysight N3000 series (mainframe + modules) could do the job as well in constant voltage mode.
« Last Edit: September 23, 2018, 10:25:10 am by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Saneoc

  • Contributor
  • Posts: 8
  • Country: ro
Re: Please recommend an electronic load.
« Reply #3 on: September 23, 2018, 10:01:00 pm »
Thank you for the suggestions.

The equipment is intended to be used for testing automotive ECUs therefore we have to use propper test equipment.

I'll take a look on eloads CC+CV function.
 

Offline Wolfgang

  • Frequent Contributor
  • **
  • Posts: 816
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: Please recommend an electronic load.
« Reply #4 on: September 24, 2018, 07:12:03 am »
Hi,

I use a RIGOL DL3021A (150V, 40A, 200W). Its maximum switching frequency is 30kHz, however. What you could do is smoothen the load current by an LC filter, and
then go on with your PWM testing. 40A at several 100kHz is a nice longwave or medium wave radio transmitter. RFI and measurement setup is probably a major concern, or not ?
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 1322
  • Country: lv
Re: Please recommend an electronic load.
« Reply #5 on: September 24, 2018, 07:30:24 am »
There is a high side switch with overcurrent protection. It is driven by PWM with the frequency of 200-400Hz (automotive light bulbs controll ). I need to use the electronic load to check its current sense behaviour.

Electronic loads are active devices, with regulation loop bandwidth and so on. They happen to work OK when you need just pure dump of DC power, yet your mileage may vary in dynamic tests. Why don't you measure current sense behavior using real world test bed - automotive light bulb(s)?! It's cheapest load you can get
 

Offline Wolfgang

  • Frequent Contributor
  • **
  • Posts: 816
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: Please recommend an electronic load.
« Reply #6 on: September 24, 2018, 07:42:05 am »
Hi,

Lamps are time dependent, nonlinear loads. Not easy to model (I tried to model them for noise generators).
A "normal" electronic load device will have a hard time to make a realistic substitute for a lamp.

The good thing is that lamps are *slow*. The larger, the slower. You could model those by a pair of resistors (one for hot, one for cold resistance), plus some timing caps to model the inrush current with a time constant. If you compare the model with a real lamp and you like the result, you could then try to teach your electronic load a current over time curve. High speed is no aspect here.

On the net there are SPICE models of lamps of different kinds, modelling nonlinearity, heat capacity and other aspects. Their problem is that obtaining the parameters is quite a measurement chore.

https://electronicprojectsforfun.wordpress.com/making-noise/noise-sources-i-have-built/a-lamp-noise-head/
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 15430
  • Country: nl
    • NCT Developments
Re: Please recommend an electronic load.
« Reply #7 on: September 24, 2018, 08:53:13 am »
IMHO the OP isn't aiming to characterise lamps. The aim is to test a low frequency PWM output for over current protection.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Wolfgang

  • Frequent Contributor
  • **
  • Posts: 816
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: Please recommend an electronic load.
« Reply #8 on: September 24, 2018, 09:46:48 am »
Well, if the purpose of the exercise is just to see when the overcurrent protection trips why not put an LC filter before the load (so its not confused by the fast PWM and runs out of regulation) and ramp up the current ?
 

Offline Saneoc

  • Contributor
  • Posts: 8
  • Country: ro
Re: Please recommend an electronic load.
« Reply #9 on: September 24, 2018, 09:59:56 am »
I see that discussion starts to transform into discussion about light bulbs.  :)

Let me explain the situation better.
I am an engineer at Continental Automotive. We are designing and manufacturing a lot of different kind of ECUs including BCMs ( Body Controll Modules which are responsible fo a lot of functions including lights control).

The light bulb is not a trivial load. It may seem like it's nothing easier than turning on and off a light bulb, but if you consider inrush current, temperature and voltage dependency things quickly start to get tricky.

Automotive light bulbs are usually rated for 12V nominal voltage. In case of overvoltage - their lifetime decreases drastically. When the engine is on, the normal voltage on the battery rail is about 13.8-14V (can go up to 18V and all modules must resist for 2 minutes at 25V).
Due to the fact that we are not allowed to overload the light bulbs, we are continuously monitoring the supply voltage and adjust the pwm drive to get the effective voltage down to 12V. This mean that at normal supply voltage we will always get about 85% duty cycle on the bulb.


The majority of the tests are done using original loads ( real light bulbs ) but there is a set of tests which can't be done with them.
The tricky tests are open load detection threshold and overload threshold. This are two software configurable limits individual for every output.
Open load threshold is the current under which we consider the bulb to be interrupted or missing ( it's not zero because you always have some dirt resistance, offsets, noise and so on. Also often one output can controll 2 light bulbs and we must detect that one of them is burned)
Overload is also a software threshold. Basically it's a short circuit protection but it kicks in much soonet than hardware SC protection from drivers.
The ability of precisely measuring low and high currents as well as being able to configure all thresholds are very important for customers.
This is why we must check that during development phases.

The eload should not reproduce the light bulb current profile but must tollerate PWM.
The solution with LC filter is not practical (especially at high supply voltages). In order to have reasonably low ripple current we would have to use inductors over 1H with saturation currents over 5A. Also this solution may cause the ECU to detect open load at turn on of the output because it expects the certain current within a certaion period of time. Inductor will slowdown the current rise time.

You can imagine that we can't adjust the current of the real load to be right at the limits therefore we use an eload for that.

Until now we did these measurements using a separate power supply and a diode to keep the eload in it's linear region but that may trigger another errors in the ECU which slows down the testing process.

The bottleneck is not in the bandwidth of current eloads but in saturation recovery time.

We can't build our own measurement equipment because later we'll get in trouble with our quality mannagers and we'll have to prove that our test jigs are according to standards.

For sure there are shortcuts to evercome these problems ( this is not the first project which we have done ) but I was hoping that there is an eload which can overcome these problems.

Thank you for your answers. Until now the Eload with CC+CV mode looks the most suitable choice but I'll have to investigate that tomorrow more in depth.
 

Offline Wolfgang

  • Frequent Contributor
  • **
  • Posts: 816
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: Please recommend an electronic load.
« Reply #10 on: September 24, 2018, 10:12:16 am »
Tricky problem. Not sure, but the Kikusui top models could be able to run this.
The point is always that any electronic load needs a minimum voltage to function properly. In a pure PWM environment,
the "bottom" voltage is zero, not what a load likes.
A cheat could be to introduce freewheeling schottky diodes that keep a minimum voltage across the load when the PWM is off.
 

Offline JohnPi

  • Regular Contributor
  • *
  • Posts: 73
  • Country: us
Re: Please recommend an electronic load.
« Reply #11 on: September 25, 2018, 08:04:37 am »
You don't actually need an e-load for this type of test.

Power your ECU from a standard supply (doesn't need to supply the load current). Then connect an additional supply (that can run the load current) across the output driver (i.e. '+' to ECU supply; '-' to ECU's HS Driver output). You may put a resistor in series with this supply. Adjust this supply & resistor so the short-circuit current is at the level you want to test, then run your tests. Use a current clamp to measure the actual current waveforms.
 

Offline TurboTom

  • Frequent Contributor
  • **
  • Posts: 436
  • Country: de
Re: Please recommend an electronic load.
« Reply #12 on: September 25, 2018, 08:51:50 am »
@Saneoc - If oyu like, I'll check a few loads tomorrow. I had the Rigol load for a while for evaluation but due to its poor dynamic behaviour and inaccurate operation at low currents, I decided to return it. I've still got a Maynuo and an Applent. Both consist of virtually the same analog hardware and completely different digital control circuitry. Hence I expect the constant current mode to be comparable between the two while the other operation modes will be quite different. If you need high speed, I'ld recommend to us constant current mode since this is directly "coded" in hardware on most of the entry level electronic loads. Maybe you could use the computer interface to adapt the set current according to the requirements to resemble the behaviour of an incandescent light bulb, this shouldn't be too complex. You could also use the computer program to simulate the failure condition by appropriately programming the load.

Constant voltage, constant resistance and constant power modes are more or less firmware-simulated by modifying the parameters of the analog current sink of the load and are hence usually much slower than CC mode.  At 400Hz I think CC mode should still operate acceptable on a good entry level load. You may want to add a small inductor in series with the load input to reduce the initial current spike when the PWM switch closes against a shorted load passbank (since there was no current flowing before, the load will drive the passbank transistors right to a complete short to establish the preset current) and also a "free wheeling diode" or schottky from the negative load terminal to the switch node to carry the inductor current when the PWM switch opens. But that's basically cosmetics.

I'll let you know my findings tomorrow...

Cheers,
Thomas
 

Offline PedroDaGr8

  • Super Contributor
  • ***
  • Posts: 1252
  • Country: us
  • A sociable geek chemist
Re: Please recommend an electronic load.
« Reply #13 on: September 25, 2018, 09:49:11 am »
I see that discussion starts to transform into discussion about light bulbs.  :)

Let me explain the situation better.
I am an engineer at Continental Automotive. We are designing and manufacturing a lot of different kind of ECUs including BCMs ( Body Controll Modules which are responsible fo a lot of functions including lights control).

The light bulb is not a trivial load. It may seem like it's nothing easier than turning on and off a light bulb, but if you consider inrush current, temperature and voltage dependency things quickly start to get tricky.

Automotive light bulbs are usually rated for 12V nominal voltage. In case of overvoltage - their lifetime decreases drastically. When the engine is on, the normal voltage on the battery rail is about 13.8-14V (can go up to 18V and all modules must resist for 2 minutes at 25V).
Due to the fact that we are not allowed to overload the light bulbs, we are continuously monitoring the supply voltage and adjust the pwm drive to get the effective voltage down to 12V. This mean that at normal supply voltage we will always get about 85% duty cycle on the bulb.


The majority of the tests are done using original loads ( real light bulbs ) but there is a set of tests which can't be done with them.
The tricky tests are open load detection threshold and overload threshold. This are two software configurable limits individual for every output.
Open load threshold is the current under which we consider the bulb to be interrupted or missing ( it's not zero because you always have some dirt resistance, offsets, noise and so on. Also often one output can controll 2 light bulbs and we must detect that one of them is burned)
Overload is also a software threshold. Basically it's a short circuit protection but it kicks in much soonet than hardware SC protection from drivers.
The ability of precisely measuring low and high currents as well as being able to configure all thresholds are very important for customers.
This is why we must check that during development phases.

The eload should not reproduce the light bulb current profile but must tollerate PWM.
The solution with LC filter is not practical (especially at high supply voltages). In order to have reasonably low ripple current we would have to use inductors over 1H with saturation currents over 5A. Also this solution may cause the ECU to detect open load at turn on of the output because it expects the certain current within a certaion period of time. Inductor will slowdown the current rise time.

You can imagine that we can't adjust the current of the real load to be right at the limits therefore we use an eload for that.

Until now we did these measurements using a separate power supply and a diode to keep the eload in it's linear region but that may trigger another errors in the ECU which slows down the testing process.

The bottleneck is not in the bandwidth of current eloads but in saturation recovery time.

We can't build our own measurement equipment because later we'll get in trouble with our quality mannagers and we'll have to prove that our test jigs are according to standards.

For sure there are shortcuts to evercome these problems ( this is not the first project which we have done ) but I was hoping that there is an eload which can overcome these problems.

Thank you for your answers. Until now the Eload with CC+CV mode looks the most suitable choice but I'll have to investigate that tomorrow more in depth.

Being as this is the automotive industry, it sounds to me like you need a much more advanced option than would normally be suggested for hobbyists and lower-end users. You need the background, the certification, the name on the instrument, etc. lest something goes wrong. Unfortunately, since you also need high power, this means a VERY different price point from loads like Rigol and Maynuo. This is obvious based on the price for the BK Precision load you mentioned.


The hardcore "kitchen sink" option would be something like the Chroma 63600 series. It could certainly do almost anything you need it to do.  If you need both 40A and 40V at the same time, their 5 bay 63600-5 mainframe kitted with four or five 63640-150-60 400W modules would fit the bill. Their mainframe loads modules can be run in parallel and have the normal features of Constant Voltage, Constant Current, Constant Power, and Constant Resistance. They also have an added form called Constant Impedance (CZ), which might be useful here. Additionally, it is fully programmable with a variety of profiles and most are able to handle dynamic loads up to 50kHz or more. This format could allow you to simulate multiple independent loads all at once (or gang them together to form a larger load). These loads are used all across industry and are very well known and very high quality, but very pricey.

A second option from them would be the non-mainframe 63201, which can handle up to 2600W and 20kHz dyanmic signals. It doesn't go down to quite as low of a voltage range as the mainframe series, but based on what you describe, you might not need it.

Another option would be the Keysight N3300A series of mainframe loads with three N3306A 600W modules. These have a similar ability to be combined for higher currents (from what I can read) and are also capable of 20kHz dynamic signal.

Tektronix/Keithley loads are rebranded BK Precision loads (which are actually designed by iTech). These share a lot in common with the loads from Array, Maynuo, etc. as the former iTech lead engineer has jumped around a bit. They share a LOT in common with the 8514 and as a result will likely not be what you are looking for.

Other companies that might meet your specifications would be Kikusui, ProDigit, TDI Dynaload. There are likely others that fit this bill as well, but I don't know them.
The very existence of flamethrowers proves that some time, somewhere, someone said to themselves, "You know, I want to set those people over there on fire, but I'm just not close enough to get the job done." -George Carlin
 
The following users thanked this post: ogden

Offline jamodio

  • Newbie
  • Posts: 2
  • Country: us
Re: Please recommend an electronic load.
« Reply #14 on: September 26, 2018, 03:17:18 am »

I'm very happy with the Rigol DL3021.

Not good reason for me to pay the extra $$ for fancy colors and few different specs.
https://www.rigolna.com/products/dc-power-loads/dl3000/

Cheers
Jorge
 

Offline TurboTom

  • Frequent Contributor
  • **
  • Posts: 436
  • Country: de
Re: Please recommend an electronic load.
« Reply #15 on: September 26, 2018, 03:41:53 am »
As announced yesterday, I did a _VERY_ quick and dirty PWM test with my two loads, an Applent AT8612 and a Maynuo M9712. I used a small inductor (nothing specific, just something salvaged from a broken PC PSU) to control the inrush current but as it seems, it got saturated pretty quickly. I still think the inductor will be good to have, also some snubber network may be favorable to reduce ringing. A schottky diode is required to help discharge the inductor. The current ptobe I used is my "prehistoric" TEK P6042 but it does the job...  ;). To simulate your PWM switch, I utilized a substantial MOSFET with an insulated gate driver.

I attached the schematic so you get the idea. I tested the Applent load at 5A, 10A and 15A constant current setting. The voltage supplied was 21V and the power supply was capable of 10A continuous. I also reduced the voltage to the point where the inductor wouldn't saturate to give you an idea what the current would look like if the inductor would have been designed properly. Funny enough, the Maynuo started to oscillate if the inductor didn't saturate. It also produced way more inrush current ringing FYI, I attached the scope screenshots of the current and two photos of the load's displays at 10A, 21V and 80% duty cycle.

I think the Applent load is better suited for the task, especially since internal control loop damping appears to be better suited for that application. In my opinion, even an entry level load like the AT8612 can be used to load a fairly slow (400Hz) PWM source, provided some precautions are taken. Higher range loads liek the suggested Chroma may probably but don't necessarily need to perform better. A sub 400EUR instrument like the Applent probably isn't a waste of money if you get performing as expected.

I hope this information is at least somewhat helpful and gives you some orientation of what to expect from inexpensice DC loads.

Cheers,
Thomas
 

Offline Fgrir

  • Regular Contributor
  • *
  • Posts: 110
  • Country: us
Re: Please recommend an electronic load.
« Reply #16 on: September 26, 2018, 05:18:53 am »
The Aim-TTi LD400 has a dropout setting that disables conduction when the source voltage falls below a settable threshold.  I think this would prevent the turn-on transient that you are worried about with your PWM drive, but unfortunately my lab isn't currently in a state where I can test it for you.

https://www.aimtti.com/product-category/electronic-loads/aim-ld400series
 

Offline Saneoc

  • Contributor
  • Posts: 8
  • Country: ro
Re: Please recommend an electronic load.
« Reply #17 on: October 03, 2018, 12:56:04 am »
Thanks everybody for suggestions.

We ordered PLZ405W from Kikusui.
 

Offline Wolfgang

  • Frequent Contributor
  • **
  • Posts: 816
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: Please recommend an electronic load.
« Reply #18 on: October 03, 2018, 01:07:25 am »
OK, probably the best specs you can get. Much fun with it !
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf