Author Topic: PSU Constant Current response timing comparison  (Read 3444 times)

0 Members and 1 Guest are viewing this topic.

Offline Peter_OTopic starter

  • Frequent Contributor
  • **
  • Posts: 419
  • Country: de
PSU Constant Current response timing comparison
« on: July 30, 2022, 09:03:13 am »
Having collected some PSUs with CC capability I’d like to know how quick they are responding to an overcurrent condition.

Test setting is like this:

PSU is configured to deliver 25V/1A.
PSU is connected to a 10 Ohms resistor.
This load circuit is switched on/off by a MOSFET-Module, controlled by a function generator.

off: No load, PSU idles at 25V/0A
on: PSU would deliver I=U/R=2,5A but limits current to 1A by lowering the voltage to 10V. The question was: How quick does this happen?



Here are the results:

Agilent 6653A

see picture above.



C1/yellow: Mosfet control
C2/magenta: Voltage over R
C3/cyan: PSU voltage

With the load switched on, the Agilent needs about 20ms to adjust to 1A current.

Siglent 3303X(-E)





The Siglent is slower with about 30ms.
Going down from 25 to 10V it switches the transformator outlet by relay.

HP6114A





For the HP6114A the test runs with it’s voltage limit of 20V.

It needs about 30ms to adjust to 1A too.
Interesting to see, that it needs about 300ms to recover back to 20V when idle.

Voltcraft PS 303 pro







Surprisingly, the low cost Voltcraft is the fastest one: It needs 5ms to adjust to 1A, even with a transformator relay switching, which can been seen in the 2nd shot.


To be continued with the next post about the TDK Lambda Geness 50-30 ...
« Last Edit: July 30, 2022, 09:08:21 am by Peter_O »
 

Offline Peter_OTopic starter

  • Frequent Contributor
  • **
  • Posts: 419
  • Country: de
Re: PSU Constant Current response timing comparison
« Reply #1 on: July 30, 2022, 09:12:11 am »
TDK Lambda Genesys 50-30





The Genesys is by far the slowest one: It needs about 400ms to adjust to 1A.

With the Genesys, a fold back delay can be configured via the serial interface and would be persistent over a power cycle.  I tried to reset any fold back delay, but got an C01 Error (Wrong command) for all fold back related commands. Other commands are fine.

So I assume, this PSU might just be that slow.

Overall, the Voltcraft and the TDK are the surprises of this short shoot out.
 
The following users thanked this post: AVGresponding

Online switchabl

  • Frequent Contributor
  • **
  • Posts: 440
  • Country: de
Re: PSU Constant Current response timing comparison
« Reply #2 on: July 30, 2022, 01:06:26 pm »
Overall, the Voltcraft and the TDK are the surprises of this short shoot out.

To be fair, they do tell you this in the datasheet.
Down-prog response time no-load: 800ms

At a current rating of 50A, 2.5A is basically the same as "no-load". That 800ms spec will be 30V to 0V, so 400ms from 25V to 10V is really spot on.

Current fold-back is really about overcurrent protection. It will turn off the power supply completely when tripped instead of going into CC mode. So it won't help with response time. In fact, it is probably done in software.
« Last Edit: July 30, 2022, 01:08:07 pm by switchabl »
 

Offline maxwell3e10

  • Frequent Contributor
  • **
  • Posts: 869
  • Country: us
Re: PSU Constant Current response timing comparison
« Reply #3 on: July 30, 2022, 04:45:33 pm »
Thanks for posting these plots! Always great to see actual data.
 

Offline alm

  • Super Contributor
  • ***
  • Posts: 2881
  • Country: 00
Re: PSU Constant Current response timing comparison
« Reply #4 on: July 30, 2022, 07:16:10 pm »
Thanks for posting the data, indeed!

If you subtract the time it takes to discharge the output cap down to 10V through the resistor, how much time remains that can be attributed to the regulation? The magenta curve in many of the traces looks quite a lot like exponential discharge to me.

For example the 6114A has a 490 uF output cap according to the schematics. Discharging a 490 uF cap through a 10 Ohm resistor from 25V to 10V takes:
$$10\textrm{V} = 25\textrm{V} e^{\frac{-t}{RC}}$$
$$t = -\ln{\frac{10 \textrm{V}}{25 \textrm{V}}}\cdot 10 \textrm{Ohm}\cdot 490 \textrm{uF} = 2 \textrm{ms}$$

But I imagine the Genesys might have much larger output caps. An active down programmer, which I'm not sure if any of these units have (the 6653A maybe?), would speed up the process of discharging the cap. By how much depends on the current the down programmer draws.

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26906
  • Country: nl
    • NCT Developments
Re: PSU Constant Current response timing comparison
« Reply #5 on: July 30, 2022, 07:25:12 pm »
I agree. This test is very skewed by how much capacitance there is at the output.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline AVGresponding

  • Super Contributor
  • ***
  • Posts: 4666
  • Country: england
  • Exploring Rabbit Holes Since The 1970s
Re: PSU Constant Current response timing comparison
« Reply #6 on: July 30, 2022, 08:05:28 pm »
I agree. This test is very skewed by how much capacitance there is at the output.

Should the testing be adjusted to even out the capacitance (by adding more to the output), or should we accept that the PSU's that need more to ensure a noise free output are inferior and allow them an extra +50% (or whatever arbitrary number suits) to comply?

I would suggest that the answer is "what suits your application best?".

Still, the OP's results are useful; they give an absolute number for response time, which might be important for your use case.
« Last Edit: July 30, 2022, 08:09:40 pm by AVGresponding »
nuqDaq yuch Dapol?
Addiction count: Agilent-AVO-BlackStar-Brymen-Chauvin Arnoux-Fluke-GenRad-Hameg-HP-Keithley-IsoTech-Mastech-Megger-Metrix-Micronta-Racal-RFL-Siglent-Solartron-Tektronix-Thurlby-Time Electronics-TTi-UniT
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26906
  • Country: nl
    • NCT Developments
Re: PSU Constant Current response timing comparison
« Reply #7 on: July 30, 2022, 08:25:18 pm »
If you want to test the constant current response time, then you'd need to change the current limit of the PSU and see how fast it reacts. This would require remote control (either through communication or analog) or a programmable sequence of current limits.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online switchabl

  • Frequent Contributor
  • **
  • Posts: 440
  • Country: de
Re: PSU Constant Current response timing comparison
« Reply #8 on: July 30, 2022, 08:32:22 pm »
But I imagine the Genesys might have much larger output caps. An active down programmer, which I'm not sure if any of these units have (the 6653A maybe?), would speed up the process of discharging the cap. By how much depends on the current the down programmer draws.

Actually... guesstimating from the datasheet (80ms full-load down-prog response time), the downprogrammer in the Genesys must be able to sink something like 5A already. It just has massive output capacitors. Of course it has, it does 50A. Don't connect an expensive low-current board to a huge power supply and expect the current limit to save you.  :-BROKE

I agree. This test is very skewed by how much capacitance there is at the output.

Well, true. But that has very tangible real-world implications because it determines the amount of energy available to cause damage before the OCP kicks in.
« Last Edit: July 30, 2022, 08:34:07 pm by switchabl »
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14199
  • Country: de
Re: PSU Constant Current response timing comparison
« Reply #9 on: July 30, 2022, 08:32:59 pm »
The output capacitor is part of the power supply and should be included. For the user it does not make much difference if the current exceeds the set limit becasue of a slow regulation or because of a large output capacitor. A point may be comparing supplies with different current rating. It is kind of expected that a supply for a higher current would have more capacitance. So the test with a fixed resistor favors the smaller supplies.

The current test the current is limited quite a bit by the resistor and the capacitance is discharged relatively slow. This makes the output capacitor an important part.  The test choosen is not bad, but a bit heavy on the capacitor effect. With a smaller resistor and thus higher peak current the actual regulator response speed may get important. A lower voltage to start with would also make the capacitor less relevant, though the job for the regulator also possibly easier.


In some case with loads that by design do take spikes of high peak current a fast acting current limit can be anoying. So a fast acting limit is not always good.
 

Offline alm

  • Super Contributor
  • ***
  • Posts: 2881
  • Country: 00
Re: PSU Constant Current response timing comparison
« Reply #10 on: July 30, 2022, 08:50:51 pm »
Selecting a power supply for minimum output capacitance to increase the output impedance in constant current mode (faster respond to change in load impedance) is valid for some applications. The 6114A actually has a connection on the barrier strap that allows you to disconnect most of the output capacitance. It might be interesting to see how it performs with the cap disconnected. Based on the calculation I wouldn't expect it to matter much for that particular supply.

I would expect the mobile communication / battery simulator power supplies to perform the best here. Since they are made to measure fast current transients. A dedicated constant current source, like the HP 6177/6181/6186, should be even better.

Actually... guesstimating from the datasheet (80ms full-load down-prog response time), the downprogrammer in the Genesys must be able to sink something like 5A already. It just has massive output capacitors. Of course it has, it does 50A. Don't connect an expensive low-current board to a huge power supply and expect the current limit to save you.  :-BROKE
Very true. If your circuit is sensitive, it's best to match the power supply to the load. Also don't use a high voltage power supply to power the 3.3V rail for your expensive FPGA board.

In addition, switching power supplies (like the Genesys) generally require more output capacitance than linear power supplies, and have a slower transient response. So I'd expect the performance to be significantly worse even if it was designed for the same current.
« Last Edit: July 30, 2022, 09:18:37 pm by alm »
 

Offline Peter_OTopic starter

  • Frequent Contributor
  • **
  • Posts: 419
  • Country: de
Re: PSU Constant Current response timing comparison
« Reply #11 on: July 31, 2022, 08:28:38 am »
Thx to all of you for the input and discussion!

So there is no programmed delay in place, but the down response is perfectly in spec.

Reg. the capacitance here's a picture I found in my archive. Two of the cans of each channel seem to be 330mF.



And I have a measurement with the same load, but with the TDK in fold back mode:
Interesting to see, that it first adjusts the voltage to reach the CC target current, and only then folds back the output.



 

Online tautech

  • Super Contributor
  • ***
  • Posts: 28380
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: PSU Constant Current response timing comparison
« Reply #12 on: July 31, 2022, 08:37:51 am »
Peter_O
 A little tip for better screen shots is to lift graticule brightness to 60+% for the grid to be more visible.  ;)
Avid Rabid Hobbyist
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 

Offline Peter_OTopic starter

  • Frequent Contributor
  • **
  • Posts: 419
  • Country: de
Re: PSU Constant Current response timing comparison
« Reply #13 on: July 31, 2022, 09:39:14 am »
Peter_O
 A little tip for better screen shots is to lift graticule brightness to 60+% for the grid to be more visible.  ;)
Sounds reasonable. thx Rob!
 

Online jjoonathan

  • Frequent Contributor
  • **
  • Posts: 783
  • Country: us
Re: PSU Constant Current response timing comparison
« Reply #14 on: July 31, 2022, 03:37:17 pm »
I (ab)use my SMU when I need fast transient response / low output capacitance. Here are some measurements I took last night.

Milliseconds:
PSU: output capacitor RC, loop responds
SMU: ideal

Microseconds:
PSU: lead inductance L/R, (internal inductance L/R?)
SMU: lead inductance L/R, loop responds

Nanoseconds:
PSU: transmission line sloshing (not pictured), FM radio, wifi packets
SMU: transmission line sloshing (not pictured), FM radio, wifi packets


I went ham with the bandwidth to see transmission line effects in the leads, but those are still a work in progress. At first, I was using a 100 ohm load resistor on the reasoning "hey, ladder lines are 100 ohms, I bet mini-grabber leads are some small multiple of that." This reasoning seems to have been correct and judging from where the plateaus are the actual characteristic impedance seems to be 500-1000ohms -- but bouncing is such an unholy mess on this timescale that just repeating the experiment didn't suffice to generate clean results. Also, loose clipper leads aren't exactly going to have uniform characteristic impedance, so there is probably a map of the cable convolved on top of the bouncing mess for extra fun. I tried making the load resistor 1 ohm on the idea that it would short out transmission line effects (pictured), and I am gratified to see that it worked, but of course it means transmission line effects are no longer visible. The next step is to make a fast-acting load and a driver for it, but that's a project for another day.

Of course, since the leads are unshielded they pick up 2.4G packets too -- but with 1 ohm termination those are pretty minimal. Less minimal is the terrible 2MHz switching noise from my new modem every time it passes a packet. The cable tumors I have on stock don't appear to be up to the task, so that's the next priority. In any case, I've attached the 1 ohm termination results.

Using less bandwidth and ignoring the transmission line timescale is the path of sanity  -- but where's the fun in that?  ;D
 

Offline Frex

  • Regular Contributor
  • *
  • Posts: 118
  • Country: fr
Re: PSU Constant Current response timing comparison
« Reply #15 on: August 02, 2022, 04:30:31 pm »
Hello,
My little contribution to this thread. :)

By curiosity, I have done the same tests with my both main supplies :
The triple outputs lab power supplies ITECH IT6302 (195W 7.1kG) ,
and my home made "DIY" EPSUX3V2 (180W 1.4kG) .

I used the same scope setting as possible for easy comparison (as first post).
The load is 8 Ohms (instead of 10), and the output voltage of the EPSUX3V2
 is 20V instead of 24V.

Both are fast, even if the DIY one is the winner! :-D
The IT6302 use relays for voltage range, with the 10Hz load switch
 is very noisy and I don't think it would love it for a long time..(my hears too).

I added today (06/08) some picts I had made of the setup of both power supplies,
 an the small load switch board.


Frex


1555714-01555720-11555726-21555732-31558984-41558990-51558996-61559002-8
1558990-91558996-10
« Last Edit: August 06, 2022, 09:27:29 am by Frex »
 
The following users thanked this post: Peter_O

Offline jchw4

  • Regular Contributor
  • *
  • Posts: 189
  • Country: 00
Re: PSU Constant Current response timing comparison
« Reply #16 on: June 11, 2023, 01:21:32 pm »
My setup (random parts I had around):

- 10 Ohm 5% ceramic load resistor SBCHE1510RJ
- 0.3 Ohm 5% ceramic sense resistor TWW5JR30
- IRF510 MOSFET

Tools:
- Feeltech FY6600 signal generator
- Digilent Analog Discovery

 

Offline jchw4

  • Regular Contributor
  • *
  • Posts: 189
  • Country: 00
BK Precision 9130 CC response timing
« Reply #17 on: June 11, 2023, 01:28:43 pm »
BK Precision 9130 https://www.bkprecision.com/products/electrical-battery-testers/9130

This one has all brand new capacitors.

"begin" - measurements at the start of a pulse (0.5 Hz)
"end" - at the end.

Every capture is an average of 5 acquisitions.

Note that at the start of a pulse I captured current across the sense resistor, but at the end I captured BK9130 output voltage.

These graphs are from the first channel. I checked the second one and it basically has exactly the same behavior.
 

Offline jchw4

  • Regular Contributor
  • *
  • Posts: 189
  • Country: 00
ITECH IT6322B CC response timing
« Reply #18 on: June 11, 2023, 01:33:20 pm »
ITECH IT6322B https://www.itechate.com/en/info_128.aspx?itemid=548&lcid=90

The same notes above apply:

Every capture is an average of 5 acquisitions.

Note that at the start of a pulse I captured current across the sense resistor, but at the end I captured BK9130 output voltage.

These graphs are from the first channel. I checked the second one and it basically has exactly the same behavior.
 

Offline jchw4

  • Regular Contributor
  • *
  • Posts: 189
  • Country: 00
Agilent 66321D CC response timing
« Reply #19 on: June 11, 2023, 01:57:39 pm »
Agilent 66321D https://www.keysight.com/us/en/product/66321D/mobile-comm-dc-source-battery-emulation-dvm.html

This is a relatively new unit, Agilent-branded.

Surprisingly it has the slowest overcurrent reaction speed in this topic: about 80ms! Exactly as its older brother 66311B (see below).

66321D is capable of 0..15V of output so I measured 15V and 10V (for voltage stability).

This model supports 4 modes of output compensation: HREMOTE, LREMOTE, HLOCAL or LLOCAL.

By default I used HLOCAL. LLOCAL did not make any difference. Remote modes only affected the off-transition (see the photos).

I also had two cables: ~1.5-1.8m AWG16 with local sense only, and a shorter ~1m twisted AWG 14 with twisted sense leads. It made some difference.  :)

Every capture is an average of 5 acquisitions.

« Last Edit: June 11, 2023, 02:32:32 pm by jchw4 »
 

Offline jchw4

  • Regular Contributor
  • *
  • Posts: 189
  • Country: 00
HP 66311B CC response timing
« Reply #20 on: June 11, 2023, 02:09:58 pm »
HP-branded 66311B https://www.keysight.com/us/en/product/66311B/mobile-communications-dc-source-15v-3a.html

It's probably one of the last HP-branded units. Probably made around 1999 - 2000.

The same 80ms over-current delay!  :-//

I did not check the two available compensation modes. Used only the default one, which was probably "Low".

Used only the twisted cable with remote sense as described above.

Every capture is an average of 5 acquisitions.

Please note that some captures measured current and some measured the power supply output voltage.
« Last Edit: June 11, 2023, 02:33:56 pm by jchw4 »
 

Offline jchw4

  • Regular Contributor
  • *
  • Posts: 189
  • Country: 00
RD DPS3005 CC response timing
« Reply #21 on: June 11, 2023, 02:17:21 pm »
RD DPS3005 https://rdtech.en.alibaba.com/productgrouplist-805451526/DPS3005.html

I powered it with the laptop power adapter and thus only 19V maximum voltage.

Here you can see measured both current and voltage. AD has only 2 channels, so these are different acquisitions. But behavior is pretty stable and this is still an average of 5.
 

Offline alm

  • Super Contributor
  • ***
  • Posts: 2881
  • Country: 00
Re: Agilent 66321D CC response timing
« Reply #22 on: June 11, 2023, 09:09:16 pm »
Surprisingly it has the slowest overcurrent reaction speed in this topic: about 80ms! Exactly as its older brother 66311B (see below).
I think the high speed part of these power supplies is about load regulation within the current limit. They were made to simulate a battery and charger. Fast current limiting when shorted would not be a very accurate simulation of a secondary lithium cell ;)
« Last Edit: June 11, 2023, 10:09:12 pm by alm »
 
The following users thanked this post: jchw4

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26906
  • Country: nl
    • NCT Developments
Re: HP 66311B CC response timing
« Reply #23 on: June 11, 2023, 10:03:30 pm »
HP-branded 66311B https://www.keysight.com/us/en/product/66311B/mobile-communications-dc-source-15v-3a.html

It's probably one of the last HP-branded units. Probably made around 1999 - 2000.

The same 80ms over-current delay!  :-//
I have that model as well. I don't know what they where thinking when they designed this unit! The current limit is pretty useless if you want to use this PSU as a constant current source. When you turn the output on, it cranks it up to the set voltage and after a while it starts to limit the current  :palm: FFS!
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Peter_OTopic starter

  • Frequent Contributor
  • **
  • Posts: 419
  • Country: de
Re: PSU Constant Current response timing comparison
« Reply #24 on: December 17, 2023, 04:54:55 pm »
A new PSU to be added ...


Snip:

Having collected some PSUs with CC capability I’d like to know how quick they are responding to an overcurrent condition.

Test setting is like this:

PSU is configured to deliver 25V/1A.
PSU is connected to a 10 Ohms resistor.
This load circuit is switched on/off by a MOSFET-Module, controlled by a function generator.

off: No load, PSU idles at 25V/0A
on: PSU would deliver I=U/R=2,5A but limits current to 1A by lowering the voltage to 10V. The question was: How quick does this happen?


Here's my new acquisition, a HP E3632A:


More pictures in the TEA thread:
https://www.eevblog.com/forum/testgear/test-equipment-anonymous-(tea)-group-therapy-thread/msg5219643/#msg5219643

And the results, showing the PSU voltage (CH2, magenta) at leading and trailing edge of the Mosfet control signal (Ch1, yellow).

 

The E3632 is one of the faster ones with 10ms up and down.
« Last Edit: December 17, 2023, 05:01:25 pm by Peter_O »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf