Help conducting a proper ripple test for DC power supply

[Syllith]

Our company recently purchased thousands of counterfeit power supplies by accident. The client we supply them to is scared that if the DC signal it's providing isn't "clean", it will damage the batteries for their backup system. Because of this, they're wanting me to conduct a ripple test to see how much AC voltage is on the line of the counterfeit compared to a legitimate one.

To set this up, I took my oscilloscope and set it to AC coupling and set the probe to x1. I have the scope set to 200mv per division at 10uS. I also enabled bandwidth limited, although it doesn't let me choose the range and there's no indicator. Hopefully it's 20mhz? The power supply we're testing outputs 54VDC and can handle 2000w. We testing with a max load of 33 amps. We have a large load bank that can has a decent range of amperage we can pull. It has an 54 volt input, negative input, and a ground. I have attached my probe to the +54 terminal and the ground of the probe to the ground of the load bank.

After testing a few known good units, I was getting about 30mv RMS of AC. With all the counterfeit units, I was getting 130mv RMS AC.

Does it sound like I'm performing this test correctly? Does 30mv RMS of ripple sound right? How bad is 130mv in comparison? Would 130mv even be capable of destroying a huge 48 volt battery? I'm a bit new to performing this kind of test, so I'm not exactly sure what I'm looking for and google hasn't been helping much as far as specific values I should be expecting. Any help would be greatly appreciated.

[tunk]

Are these any help:
https://www.youtu.be/OiAmER1OJh4
https://www.youtu.be/Edel3eduRj4
Could also be interesting to look at the peak values.

[jwet]

Fundamentally you're just looking for the AC content on the output so your test is fine.  A 1x or 10x scope probe is fine, you might double check voltage ratings but good probes are usually fine to over 100v.  Ripple is usually specified with a 20 MHz bandwidth, one of the reasons scopes have a BW limit at 20 MHz.  You probably want to load the supply at different DC loads, ripple is usually worst at full load but not always- switchers can do funny things at light loads.  Sweep it over the range. Also, if you can, look at the frequency content with a spectrum analyzer or FFT mode on a scope.  100 mV vs. 30 mV on 56v DC doesn't sound like much or sound bad but it all has to do with the next guy in the chain and what his needs are- someone specified the allowable ripple input for a reason.  The acronym EMC is electromagnetic "compatibility".  The system is designed to let systems operate with one another.  I used to do military shipboard stuff- I could really appreciate the problem of putting all this electronics right next to each other on a big dirty ship- Mil Std 461 seems like a PIA but it what let's systems work.

[Syllith]

Ok, so I tested 5 good units and 3 bad units. The internal components of each unit is nearly identical. Only thing I really notice is the genuine one is using some higher quality jianghai caps and the ingenuine one is using these cheaper looking ones from "RUC"? Not sure, googled for 20 minutes and couldn't find a single mention of them. Could this be the cause for the higher ripple voltage?

[ejeffrey]

Definitely.  An output capacitor that is too small or too high ESL/ESR could easily increase ripple voltage.

[tunk]

No expert on this, but if you go to the following page, the two tested apple chargers have
100-400 mVpp ripple. So I'd guess your PSUs are acceptable (at least concerning ripple).
https://lygte-info.dk/info/indexUSB%20UK.html

Edit:
Maybe you could post PCB photos of both? Maybe someone could comment on build quality?

[Syllith]

Good idea, anybody see anything noteworthy between these 2?

[Syllith]

Also, one thing to mention, we're in a lab environment. All my test were conducted in a room with tons of other electronics going at once. Everything is properly grounded, but I'm wondering if this would be detrimental to my test accuracy.

[bdunham7]

I don't think that the differences in ripple are going to matter to your battery, so that would be the least of my concerns.  If that is the only concern, I see nothing to worry about.  However, if they truly are counterfeit, you may have EMC issues from a noisier unit, as well as safety and documentation of compliance.  I can't imagine anyone would allow delivery of these if they are known to be fake.

Your sample size is pretty small and you are only doing one test, so you might want to expand on both of those.  b/t/w, how did you determine that they were counterfeit and not just a production variation?  These things typically sell for not much more than the BOM and I don't see a lot of missing or substandard parts in the fake one.  As far as the capacitors go, you'd have to remove a few and test them in detail.  You also might want to use an IR imager and see how hot things get in each unit.

[rteodor]

anybody see anything noteworthy between these 2?

Noticed a few minor things but one that I find relevant is the thickness of the PCB.

Much like the thickness of the car doors a thin PCB usually indicates a cheap and underperforming product.

[blackdog]

Hi,

Just a tip....
Use a coax connection to the power supply for your scoop, RG58 cable is low cost and OK, keep the coax away from other wires.
Use a floating scoop, i use a battery scoop to be shure that i have no commonmode problems, look for the scoop @ MICSIG

Maybe test is outside...
A bunch of wire of a certain length in a bucket of water can also serve as a load, but is somewhat inductive.
Or a number of car lamps in series/parallel is also a cheap solution for a low interference measuring system.

I hope this helps you a bit.

Kind regard,
Bram

[trobbins]

If you want to assuage your client about any impact on their batteries per se, then ask for the battery ripple 'current' capability spec and compare that to a measurement (although that will require you to test with the same or similar battery and at different SoC).  The client may have other unforeseen issues due to ripple/noise voltage, but that would likely be more about their load equipment or battery management circuitry.

[Faringdon]

The client we supply them to is scared that if the DC signal it's providing isn't "clean", it will damage the batteries for their backup system. Because of this, they're wanting me to conduct a ripple test to see how much AC voltage is on the line of the counterfeit compared to a legitimate one.

Sorry , but others have mentioned it.....your customer is totally and utterly wrong.
A power supply with a bit of voltage ripple on its output wouldnt do any harm to a battery.
A charger  controls the current...of interest would be the current waveform into the battery...but batteries are usually not bothered by high levels of AC in the charge current...as long as the average charge current is below about 1C or so.

Ask for more detail from customer...it sounds like they    are without a clue.

You can measure output ripple if you want...of the units into a dummy resistive load say...but 130mV ripple on a 54v output for something thats going to end up being used as a batetry charger is no problem.......whats the charge current being delivered?...is it too much for the battery...thats your main concern.

[tunk]

If you know who made the old PSUs, contact them and ask.

[Faringdon]

Vout ripple testing...



https://www.edaboard.com/threads/smps-output-ripple-voltage-measurement-with-10-1-scope-probe-is-best.358562/

https://www.edaboard.com/threads/home-made-coaxial-probe-for-outo-measurement.254629/

+ similar ones on  EEVBLOG

[Syllith]

Thanks for all the responses. Sorry it took me a while to get back to them. I'll try to elaborate more.

These power supplies are for telecom powering cell tower equipment and we have a few thousand of them that were ordered from a 3rd party with the expectation that they are legitimate. Well upon receiving them, we had a technician plug a few of them in to test functionality, but got an alarm.

Upon further investigation, it seems they were never assigned a serial number from the "factory". The controller showed a blank spot where the serial should have been. Because of this, the cabinet cannot differentiate between the 2 power supplies, thus it's unable to communicate properly. So that means if we have 2 or more of these bad units installed into 1 cabinet, it throws an alarm. They work completely fine on their own, or even mixed with other legitimate OEM units. As long as there aren't 2 of these counterfeit units in 1 cabinet.

Now, we sent them back to the real OEM to have them take a look and they confirmed they are indeed counterfeit and even sent us documentation to identify fake ones. But here's the weird thing. The counterfeit units are VERY close to the real thing. So much so, we're thinking a possibility is that one of the 3 manufacturers their company uses have been producing and selling these behind the OEMs back. Perhaps the OEM writes the serial number once it gets to their warehouse, but since the counterfeits never made it there, that's why they're missing it? They're also missing a barcode on the bottom of the PCB.

I'm not sure if the OEM will help us since we didn't actually purchase it from them. We do work closely together though, so there might be a chance they can help fix the serial number issue. Anything else I should be looking into to help shed some light on this situation?

[Faringdon]

Thanks, i dont see any counterfeit problem.....from what you say, they are just PSU's that never got a serial number....maybe they can use a "RF wand" to quickly input the serial number to the unit?

[trobbins]

It seems to me like you have shot yourself in the foot, presumably for a lower cost supply option.  If you have no recourse to your 'supplier' of that counterfeit equipment, then presumably you have no warranty pathway except for what you have warranted to the client.  You also likely don't have any factory test documentation or any traceable confidence that the counterfeit equipment complies with any standards or specs, and you may even be in breach of your supply agreement with the client.

If I were your telco client then I would not have any confidence either.  Imho, there are just too many performance concerns that can't be quickly assuaged by simply waving a test 'wand' over the equipment.

SLA batteries have a ripple current limit - that is the cause of concern wrt the battery per se.  However your installation appears to be in a telco environment, so your equipment runs the risk of disturbing other site equipment in so many unforeseen ways.

[Syllith]

Well it wasn't us who ordered them. It was actually our client that made the order. We're trying to salvage them due to the sheer amount that was ordered. I'd say they're going to have to contact the OEM and at least get some technical documentation. I doubt they'll fix the problem themselves, so that's why we're here. We're willing to perform whatever tests, we just need to know what do look for, which isn't really possible without OEM specs. Best we can do is compare it with a certified PSU. Also the software problem will need to be fixed as well before anything else can happen. If they can release a sort of software fix for it, we'll be in charge of installing it onto every unit.

[trobbins]

Hmmm.  Obviously don't make yourselves the meat in any technical or performance sandwich.  I can appreciate that you are being requested to take on a technical expert role here - just beware that that may have unforeseen consequences if you provide advise or expert opinion as compared to just test results.  Just my 2c.

[Faringdon]

Sorry to harp on, but this just sounds like a normal situation to me....you are buying PSU's from a company, you never wrote a spec for the PSU, you have given "BOM freedom" to the maker of the PSU.......and they have just  gotten them built by someone who can do it cheaper at the time......this is the very nature of that market philosophy...which is the philosophy of the western world now.
Yes i think its bad...and yes mistakes happen which can be very costly........but its unavoidable with our "sub it out" mentality........it just sounds like they forgot to put the serial number in it to me......any power supply you get from anybody may be counterfeit, if you mean by counterfeit that the person you buy it off got someone cheaper to make it......There is a middle man somewhere  here making big bucks from this "system".

I mean, most meaningful woudl be to do a comparative thermal test on a genuine, and a "counterfit" PSU.....see if the counterfit lasts less time.
As its a telecom power supply...youll need to be sure that  the "counterfit" isnt producing electrical noise which ruins the telco data transmision/reception.
It woudl be interesting to do an inrush test, so you can see if they skipped the inrush limit on the "counterfit"
You coudl always send counterfit and genuine for "destruction testing".....this is where they both go into  their own chamber  on max load...and the temp and humidity is raised incrementally on both till they fail...then you can see which one is more rugged.

A quick test woudl be to get them side by side on the lab bench......into max load......take the cover off (already) , and see with a thermal cam if one is getting much hotter than the other.
You coudl send counterfit and genuine for a mains transient test..........and incrementally increase the transient each time...see which one cracks first.

You coudl send counter and gen for a conducted and radiated emc test...and compare their performance.

I woudlnt worry about the normal switchign frequency ripple..........btu if you see "ripple" which is much lower than f(sw), (at max load) then that is feedback loop oscillation and more of a problem...

Also, compare the waveform shape of the current that charges the battery from both conter and gen.......but im sure it will be flat DC current in each case.
Compare the magnitude of the charge current in each case...is it what was set?
You coudll always pull out the main  electrolytics...do an ESR check on each...but even then, if the couterfit was higher esr it still doesnt mean it fails the sepc, because what was the spec?.....you coudl even measure the ripple current in each cap, but from what you say, the only bom differences were the electrolytic cap, so it shoudl be the same cap  ripple in each.
The caps shoudl have hours and temperature ratings on them (eg 3000hrs at 105degc)...so check this is the same for each cap.
You can also do a lifetime calc for  the caps.
You can also do a thermal test, where you put a thermocouple on the top of each cap, and compare. (max load, max ambient temp in a chamber)

But it just sounds like a normal situation to me......when the western world subs out its entire  power supply business to the Far  East, so that some western  middle man somewhere can make more bucks...this is what happens....and it happens all the time.

Dont you worry about this...the one who should worry is the (western) middle man....thats if its not just a case  of somebody forgetting to put the serial number in....which i  suspect is the real problem here.
______________________----------------------------------___________________________
Sorry to harp on...counterfeit PSUs is a interesting subject to me.

Could i summarise?

Your co buys PSU's for a telco.
The PSU is designed by an OEM
But you buy it from  "known manufacturer_A"
....But you think some  unknown "Manufacturer_B" made it instead.  ?

But it all sounds normal to me.....swapping out big el caps is a standard build to build way of saving money.
And it just looks like someone forgot to put in the serial number.