DC electronic load BK8500 oscillations in CR mode, not broken but bad design.

[Zucca]

First if all sorry for the bad video quality and the wire which is crossing the BK8500 display. It is my first video so please have mercy. If you like it I will buy a new camera.
Now to the problem: only when the BK8500 is going to be equal or less than the equivalent resistance of the circuit in R and P mode, it starts to oscillate. I have no idea if it is broken or not.
Here the video, the scope is at 10.0V/ and 500ms/:



I tried also to change the  value of the series resistor but same results, only if R(BK8500)<=R equuivalent of the circuit in P and R mode.

Hope you can point me in the right direction: broken or bad design?

Thanks!

[jadew]

I'm a noob, so don't take my word too seriously, but I think it's bad design.

Most likely because the resulting voltage divider is generating too much of a change in voltage while both constant power and constant resistance mode are basically a function of voltage.

They consume X current based on the input voltage, in order to keep the W or R constant, which changes the input voltage in return.

[Zucca]

Thanks jadew,

I agree with you. I think I will not use the R and P mode in those extreme condition, probably the ADC have not enough resolution to manage that situation.

[saturation]

All programmable DC loads oscillate under the right conditions.

http://www.hoecherl-hackl.com/index.php?main=RV9GQVEuaHRtbA==#Schwingen

[Kevin.D]

 I am not going to say thats a design problem  unless/until  some one else with the same model tests there's and gets the same response . If  someone else does post similar results then   it could be the C/R and CP control loop is to slow and is causing the oscillation your seeing (it could be these two control loops are done digitaly and the CC is  analogue feedback ,but without a seeing a schematic we dont know).
 See if you can get someone else with the same to test theres ,then post back .

quote :"All programmable DC loads oscillate under the right conditions" .
 
So do all power supplies but those conditions should be the extremes and rarely met .Setting a constant resistance  value to a lower value than the source impedance isn't  really a condition that should cause a problem for a well designed eload .   

(p.s I just tested my own diy eload design which also has a CR mode (analogue feedback loop) it  also oscillated when it has a large load impedance and small  set value constant resistance ,but it was not a bad as yours (when my set value was 0.1 smaller than source impedance it oscillated) ,I fixed it easily enough with some extra compensation.

[MarkL]

On constant P, you were giving the load an impossible situation.  With 20V and a series resistance of 16 ohms, it's not possible for the load to draw more than 6.25W.  The observed behavior was it searching for a solution.

As for the constant R test, I think it should have worked.

I've written before about BK8500 problems with constant R and P modes:

  https://www.eevblog.com/forum/testgear/bk-8500-dc-load-erratic-output

It seem to be a half-baked feature.  If you need CR and CP in your work, I would suggest returning the unit and finding something else more stable.

FWIW, I tried your test on an Array 3723A and it works fine (CP > 6.25W notwithstanding).

[Zucca]

Kevin.D

very clean way of thinking.

Now I have to wait until another guy with a BK8500 who can reproduce the problem. Is your diy eload a programmable one? Does it have an ADC? I'm still thinking that one bit more or one bit less in the ADC can cause this oscillation in some way in my case. If your eload is 100% analog it should be more robust in my theory.
When I have time I will try to connect a C in parallel with the eload and see if the oscillation goes away.

I also looked in the inside circuits but everything looked just fine. May be it´s a calibration problem... I can´t fine any information regarding how to calibrate this device, it should not be too much complicated for a eload.

Take care,

[Zucca]

Thanks MarkL for cleaning up the air in the CP mode case, yeah you are 100% right. Stupid me  .

Well I can live without CP and CR, not big deal. May be in the future I will mod this eload to improve the CR and CP...
Now I will read your post.

[pomonabill221]

Why don't you try your tests without the 16 ohm resistor, and connect the electronic load directly across the power supply?
 Set the current limit on the power supply to say, 2 amps so you don't go into CC mode on the power supply.
Then run your tests again.

I also agree with Markl that the load was trying to draw more than was available in your circuit because of the 16 ohm resistors.

Without seeing the detail on the load, and the camera angle, it is hard to see what the load is doing, also focus is off.
I know this is your fist video, but try and work on camera angles so we can see the load a little better.

[Zucca]

Thanks pomonabill221,

I did what you suggested and I think my unit need some repairs. Basically the CR is not working when the current is constant and the eDC load has to regulate the voltage.
Here a follow up video:



(yes I changed the 1991 camera, in the video I forgot the parasitic R in the cables so the BK8500 has probably not a big offset)

It seem to be a half-baked feature.

This is not half baked in my planet, but really broken. If this is expected/design then shame on BK forever.

All programmable DC loads oscillate under the right conditions.

After the video, I tried to put a capacitor in parallel to the eDC load but no mercy.

BTW before I will start to attempt a repair I need to know if it is reproducible in another unit or not. I keep you posted.

Any schematics somewhere?

[Kevin.D]

The test you showed in your second video doesnt really prove anything or show that something is wrong ,  that  cycling is just the current limit of your power supply kicking in at the level you set ,  and the slower counter reaction of your eload to the sudden reduced  output voltage (the eload then overcompensates and so the current  drops below the set current limit of the power supply) .

Just a thought about your previous video which does appear to show a real problem, could you have had a current limit set on your electronic load when your also in CR mode .?
Some eloads should have some automatic upper current limit set even though your in CR mode , maybe your model has a combined Variable current limit + constant resistance when in Cr mode .  Just check your user manual to see if it's has so .

[Zucca]

*Title Updated*: Surely a bad design!

I just finished to test a second BK8500 unti I have here at work. SAME RESULT, the problem is reproducible  !

This tells me the following two golden rules:

1)  In CR mode the eDC load can adjust ONLY the current and not the voltage to achieve the desired V/I ratio.

If the unit is attached to a constant current Is and the CR<VeDC/Is (where VeDC is the voltage sensed by the eDC load) then
1) constant current limit of the PSU kicks in
2) the eDC load reacts by decreasing the current in order to get the set CR value
3) the current limit of the PSU switch off
4) the eDC load reacts by increase the current in order to get the set CR value
5) back to 1)

This make sense if I imagine an immense MosFET in the eDC load which can only change the Ids betweent the terminals. The eDC load can not generate a voltage in order to get the desired CR with a constant current set.

2) In CR mode the eDC load can ONLY work in situation where an increasing current will cause MORE power to dissipate

Now back to the case with a eDC load in series with a R and a constant voltage PSU Vs. As long CR>R the eDC load will dissipate more energy by increasing the current in the eDC load.
If CR<R the eDC load should dissipate less energy by increasing the current at the terminals. For some stupid design reasons this can not apparently happen.

Why in a over $1000 device BK did not implement a big RED LED (or event a buzzer) in the front to signalize a error condition like this? It could also be used in the CP mode where the outside network can not deliver the required CP to the eDC load....

Moreover it is surely worth ot mention that there is a reverse pn junction between + terminal and - terminal of the eDC load (I read this in this forum somewhere, I can find it anymore where sorry). So even with a power off unit if you inver the voltage on the terminals you will get a diode conducting current from the - to the +.

And of course not a word in the manual  , on the other side I did not have to repair the unit  .

So thank you guys to guide me here, I hope it was useful for somebody out there.

[MarkL]

It seem to be a half-baked feature.

This is not half baked in my planet, but really broken. If this is expected/design then shame on BK forever.

Who knows for sure, but it certainly didn't work as per the spec or manual for us.

I can only say that if it doesn't work the way you need, and you're still in your return window, I would strongly urge you to return the unit and look for alternatives.  Don't expect B&K to fix any bugs.

We found the CR/CP problems with our B&K 8500 within a day of receiving it, and B&K technical support strung us out for months.  Eventually they admitted they weren't going to fix it within any given time frame.  And because of the time elapsed, B&K corporate wouldn't give permission to the distributor to accept a return.

From the time we reported the bug to getting our money back it was almost a year (and after many letters and phone calls).  It was one of the worst customer support experiences I've ever had.

Ultimately it's your decision if the unit is usable for your application.  Just a fair warning if it isn't and you're going to try to get them to fix it.

[engineer_in_shorts]

I think you are trolling here.
 

 Both examples you give run into the current limit of your psu. You have given us a demonstration of the interaction of the psu current limit and the eload dropping the load.  TTi eloads in this case disable the input and the user has to re-enable the input..... as another  perspective to this situation.

[Zucca]

@MarkL
Sad story to tell. In my case I will use 99% CI and CV so I will keep the unit. Now at least I know what are the limitations of the BK8500, what it can do and what it can not do.
However shame on BK to put on the market such a CR function with no proper documentation in the manual. That function was so strange to my eyes that I thought the unit was defective! Just one page more in the manual and I would have been quiet and cool.

@engineer_in_shorts
I am sorry if you feel trolled by me, it was not my intention.

TTi eloads in this case disable the input and the user has to re-enable the input

... and may be in the manual TTi is telling you about that situation, and how to react. Thank you to confirm that it is possible to implement a better design taking care of situations where the eload can not deliver what the user expect.

[Kevin.D]

I think you are trolling here.
 

 Both examples you give run into the current limit of your psu. You have given us a demonstration of the interaction of the psu current limit and the eload dropping the load.  TTi eloads in this case disable the input and the user has to re-enable the input..... as another  perspective to this situation.

I dont think thats so ,in the first video (when he is testing CR mode ) It doesnt look like the psu current limit was being hit ,(in the second vid yes it clearly was),.
 Zucca can you just confirm  that you tested the CR mode with your psu current limit set much higher so that it didn't interfere with the test .

[Zucca]

Zucca can you just confirm  that you tested the CR mode with your psu current limit set much higher so that it didn't interfere with the test .

No, and I can prove it!

In the first video at about 3:28 we have the following situation:
1) eload in CV at 0.1V
2) PSU is showing 20.13V and 1.230A

According to the math, the resistance attached to the PSU was about (20.13-0.1)/1.23=16.2Ohm (witch matchs the nominal value of the two resistor 8+8 Ohm) in series between the PSU and the eload.

So no matter how I modify and change the setting of the eload the current max that the PSU can deliver in that circuit is 1.23A (basically eload in short circuit, see also time 5:33) and in this case the current protection is NOT on (no red led on under the current PSU display).

I never touched the current knob on the PSU for the rest of the video.

In the second video the current limitation was on, that´s for sure.

[pomonabill221]

Your second video was much better, thank you for doing that.

I would have thought that direct connection from PS to Eload should have worked also, since when you hit the ps CC limit, the voltage should just decrease while the PS maintains the max current.
Similar to placing a low value resistor across the ps terminals and forcing the ps into CC mode.  The supply shouldn't oscillate, but just start reducing the output voltage to maintain the current limit value.

The Eload has a problem in this situation, and has been verified by several others here.
Sad that this problem exists with this load but glad that it isn't "broken" and as long as you realize this when you use it, you still have a very usable instrument!

[Kevin.D]

I posted a few messages ago that some eloads also have a upper current limit setting even though you are in CR ,CP modes ,and asked you to check your manual.You didn't reply  .
So I checked it and  it's pretty clearly in there  . it says  :- page 43 in manual
 "Over Current protection
When operating in CR, CC or CW mode, the load current will be limited by a current limit value set by the user. The maximum current limit value is equal to the maximum rated current for each model.
Once the maximum current limit is reached, the DC Load will enter the over current protection state and the current will be limited to the set value. (The input will not turn off). If the DC load previously operated CR or CW mode, the DC load will automatically revert to CC mode and the VFD display will indicate CC."

And watching your video you can see this is exactly whats happening ,(every time you reach 16 ohms the display reverts to CC mode and displays Amps .

So you must have set this upper current  limit yourself  to coincide exactly with 16 ohms and 20V (you said you reproduced it on another unit at work ?what with exactly the same values ? )  . 

Conclusion :- theres nothing wrong with these at all and you must have gone into settings menu to set these  up to limit at that current in CR/CP mode .     
R u aTroll ? 
 

[Zucca]

Hi Kevin.D,

Thanks for your investigation and your effort.
I did not reply because I checked the manual and I skipped to read the "Over Current protection" since I did NOT set one, so when I read the title I thought: "this is for limiting the current, I didn´t use it, I don´t need it, I don´t need to read this section". Same thing with the "remote sensing" section and others. I read the CR and the CP ones plus other parts. This is the truth and yes I should read the entire manual I agree with all of you. On the other hand does the section "Saving and Recalling settings" pag. 45 help me for a CR function issue?

On the second BK8500 tested by me at work I saw the same oscillation problem (10V PSU, and 5 Ohm R), and again I did NOT set the over current protection.
After making Pomonabill221 happy it´s your turn now. I will do another video just the BK8500, one R and the PSU, I will show the over current setting of the eload and change the R between 15Ohm and 8Ohm see what happen (why I didn´t test it in the video second? ).
I also plan to do it live, so I will not do something offline before and than record the video with the results after. It should be much better if it is live and a real life cut.

PS: I don´t have time to fool people over the EEVBlog, it´s boring and stupid. On the other side trolls here must have a very short life. That´s good to know.

[Zucca]

Here we go Kevin.D,
Real life, real mistakes. This is the uncutted video:



This time I made the mistake to assume R/2 and not R as the starting point for the oscillation (please don´t think I was messing around on the BK8500 during this time, I did not touch the unit). It looks like the unit had an overcurrent protection set to 15A, so it was not kicking in none of the past tests. I also checked the manual further for SYSTEM SET:VOLTAGE ON SET and  SYSTEM SET:VOLTAGE OFF SET (pag. 42):

The DC Load can be set to only turn on if the voltage is at or above a set value. Additionally, the
load will turn off if the voltage drops below a second set value. This feature works for any mode.

So it was correct to keep them to 0V in the video .

And watching your video you can see this is exactly whats happening ,(every time you reach 16 ohms the display reverts to CC mode and displays Amps .

Unfortunately I can not confirm that in that video and in this one.

Moreover according to the manual across page 6 and 7:

The DC Load can operate under the following modes:
....
- Present a constant resistance to the DC source (this behavior simulates a perfect resistor whose resistance doesn't change as a function of current or voltage).

it seems not true to me.

[pomonabill221]

Hi Kevin.D,

After making Pomonabill221 happy it´s your turn now.   I am sorry if you felt that way.  I was just trying to help you find out if the load was bad,, and to try another method...

[Jay_Diddy_B]

Hi,

You need to be careful when testing the constant power mode, especially when the source has a resistor in series with it.
In this model I have a source with 10 Ohms of series resistance. I am sweeping a resistive load from 1 Ohm to 29 Ohms.



Here is a plot of the power in the load:



Note: that there are two solutions for each power. The maximum power is dissipated when the load resistance is equal to the source impedance. This is called the Maximum Power Theorem.

If you consider that you want 2W dissipation in the load there are two solutions. This is met if the load is 3.82 Ohms or 26.3 Ohms.

The load does not know which solution to use.

Jay_Diddy_B

[Kevin.D]

Well that was clear enough in that last video you posted,I can  see that you hadn't set a max current  limits on in the eload  . Zucca , sorry for thinking  you might be a troll .
 I Agree with your conclusion , from those two value resistor test you showed it very much looks like theres a Bug in the BK eload CR/CP modes  .

[Zucca]

Hi Kevin.D,

After making Pomonabill221 happy it´s your turn now.   I am sorry if you felt that way.  I was just trying to help you find out if the load was bad,, and to try another method...

I felt happy that someone was trying to help me. There was 0% irritation to make a follow up video for you as for Kevin.D. I mean people are helping me for absolutely nothing, zero, nada, 0$ back and I should be not cool with that?

If you consider that you want 2W dissipation in the load there are two solutions. This is met if the load is 3.82 Ohms or 26.3 Ohms.
The load does not know which solution to use.

Wow, the legendary Jay_Diddy_B! I was impressed by your ESR meter design and with your LTspice ninja skills. Thank you so much for yours efforts!
Your are right in everything. I gave up with the CP investigation since the BK8500 was not able to handle situations where the outside circuit could not provide the required power to the eload.
I also got in my mind the possibility of two solutions, but I think the circuit will drift in one or in the other depending by the initial conditions. So I imagine if I connect the eload to open the PSU-R network (the R current is 0 at the beginning) the voltage Veload will descrease and the Ieload will increase until the solution with the bigger resistance is reached (in your example  26.3 Ohms). How to get the circuit working in the 3.82 Ohm solution? If I initially short circuit the PSU and the R, the initial current in the resistance will be VPSU/R. I connect then a eload one terminal to the R and the other one to the PSU without removing any cable, so the eload has a shor circuit between the terminals. Then I remove the cable which is holding the eload in short and let the VPSU/R current flowing into the eload. Now in the initial condition is different and the Ieload current will decrease and the Veload will increase unitl the solution with the smaller resistance is reached (in your example 3.82 Ohm). For the rest it depends in which direction the circuit wants to drift I think. This is how I expect the system should work, but with the BK8500 who knows?

FYI in the previous videos I was investivating mainly the CR function (eload simulate a perferct R, regardless the power dissipate by the eload) and not the CP where the eload is trying to dissipate a Constant Power.

Zucca , sorry for thinking  you might be a troll .

No problem at all, please continue to be smart like this. I need people like you in this forum.