Why does Electronic Load have capacitance?

[treez]

Hello,
We cannot properly do sudden full-load-to-no-load testing of our 300W Buck converter because the electronic load has capacitance. Why is this? I mean, why does it have capacitance when we programmed it to be a 34 ohm resistor?

The electronic load is part number EA-ELR 9750-66 (0-750V,0-66A).
EA-ELR 9750-66 Electronic load:
http://www.powerfactor.co.uk/power-supplies/electronic-loads/ea-elr-9000-series/ea-elr-9750-66

The Buck converter has vin=400v, vout=100v, Pout=300w
We need to do the full-load-to-no-load test to check that the Vout doesn’t spike up and overvoltage the output capacitors when suddenly no-loaded, but we can’t do it because the Electronics Load has capacitance, which prevents the vout from spiking up as it would do in real life with the customer..

 

[srizbf]

AFAIK electronic loads are emulated using power MOSFETs and high dissipation resistors.
(can't say what is the type in your environment)

may be that is the reason why the capacitance is showing Up in the circumstance of sudden change in load.

 

[FvM]

Are you asking us for help to read the specification of your instrument?

It's specified with < 1.5 ms "response time" (10/90% rise time). Technically, it's a switched mode power supply with recuperation, so it surely involves output inductors and filter capacitors. If you want considerable faster rise/fall time, use a purely analog electronic load respectively a real resistor with a power switch.

 

[treez]

Thankyou, I will place the order for the fet switch and power resistors.....to think in 2015 and you cannot even get a electronic load that can correctly imitate a 34 ohm, 300w resistor.
Indeed one can see that it has a capacitance because there is about 1A pkpk of ripple current flowing to it at the switching frequency, and yet the vout is 100v+/-1%, so there is indeed an output capacitance in the electronic load.....which to me is cheating of them because we programmed it to be a "34 ohm resistor"

 

[FvM]

to think in 2015 and you cannot even get a electronic load that can correctly imitate a 34 ohm, 300w resistor.

Yes, to simulate a 300W resistor, you don't need a 10 kW switched mode electronic load. It can be easily modeled with analog means. But a variable 10 kW load is a different thing. There are good reasons why it's designed as is.

 

[treez]

Thanks,
From this thread , do you agree that we can formally declare that Electronics Loads MUST NOT be used in transient testing of SMPS?...(due to the fact that the electronics load has a capacitance which will change the transient response of the smps, because the output capacitance of an SMPS is a feedback loop parameter)

 

[FvM]

Would agree that this specific electronic load isn't well suited for fast transient tests. There are purely analog electronic loads as well, and also many test cases that can work with 1.5 ms rise time.

 

[treez]

There are purely analog electronic loads as well

You mean such as making one with some resistors and a fet and a switch?.....surely these kind of things are not off-the-shelf though?....they have to be hand made in the lab?

 

[dick_freebird]

There's e-loaads and there's e-loads. The rack mount ones
are complex and DC ones often use feedback to get the
arbitrary and fine resolution offered by the front panel.
This makes them slow to respond and to apply load, which
might appear capacitive or might appear inductive. I have
had some issues with e-loads applied to DC-DCs when
transients are applied to the output. Enough that I built
my own coarse-grained e-load using small discrete FETs,
binary weighted drain resistors and a gate driver apiece.
This worked great for step-load testing, as abrupt as can
be (I did not try to degenerate the gates to get more sane
risetimes on the load current, I was interested in worst-
casing the step response and tuning compensation to get
it to best transient overshoot / undershoot).

 

[treez]

Thankyou , I think we can say that anybody using an off-the-shelf e load for smps transient testing is walking on thin ice....and if they cant prove that it can be used for smps transient testing, then they shouldn't use it, but use bare resistors and fet switches instead.

 

[treez]

There are purely analog electronic loads as well

Being asked to do transient test for stability on a 150W SMPS using power modules....i told them we cannot use electronic loads due to the capacitance of the e-load.....may i ask what you meant by the "analog electronics loads".

 

[FvM]

I would translate "purely analog load" to switched resistors or current source with sufficient fast current controller.

It should be noted that real loads also involve reactant impedance components or even negative impedance, in so far it's not clear a priori what's the suitable load impedance for stability tests.

 

[treez]

Page 4 and 5 of the following states how an Electronics Load (B&K precision 8500 series Electronic load) can be used to do a transient response test of a B&K 9130 Power Supply….
https://www.mouser.com/pdfdocs/DC_Electronic_Load_Application_Note.pdf

However, we have discussed this on this forum and we came to the conclusion that because these Electronic Loads have an Effective capacitance associated with them, they therefore are not suitable for transient response testing because the output capacitance of the power supply is a feedback loop parameter, and therefore we must have it defined, and not be adding extra capacitance via the Electronics Load. Also, the Electronic Loads can never provide such a sudden no_load_to_full_load transient like a switching FET and a resistor can.

The datasheet for the 8500 series Electronics Loads says that during a transient test the current can only slew at a maximum of 2.5A/us. -Nowhere near as fast as a FET and a resistor bank.
Datasheet for the 8500 series Electronics Loads:
https://bkpmedia.s3.amazonaws.com/downloads/datasheets/en-us/MDL_datasheet.pdf

Is it to be taken that the particular power supply concerned in this test (B&K precision model no = 9130) has such a slow feedback loop, and large output capacitance, that in this case the slow slew rate, and extra added capacitance of the Electronic Load is insignificant?...and therefore transient response testing is OK with this electronics load?

- - - Updated - - -

Also, the following Agilent document on page 6 shows an Electronic Load being used for Power Supply Transient response testing….
https://www.testworld.com/wp-content/uploads/power-supply-testing.pdf

 

[Easy peasy]

you get what you pay for with electronic loads, the good ones have near zero input capacitance, the others, well they need some bus capacitance for stability...

 

[treez]

https://www.richtek.com/en/Design Support/Technical Document/AN038
fig 12 of this (above) shows how slew rate is too slow in most E Loads.

- - - Updated - - -

We see from near fig 12 that the response time of say a current mode smps with 10khz crossover frequency, is going to be about 22us……also that the step load rise time should be about 5 times faster than this at least…so that’s around 4us…….E loads usually go up to about 2.5A/us at the affordable end….so if the load step was much more than 10A then an E load wouldn’t be cutting it.

- - - Updated - - -

you get what you pay for with electronic loads, the good ones have near zero input capacitance,

Thanks, yes, ..the E load in the top post is a very good quality expensive one, but it was startling how much capacitance it had….when you compared the AC content going through the E load set up as a resistive load, with the AC content that went through a plain resistor of the same value.