Brand new ET5410+ emits high pitch noise

[hp3310a]

Hi everybody,

  I got a brand new East Tester ET5410+ electronic load today and immediately ran into a problem that I didn't *hear* about in any video review: It emits a constant, loud and mind numbing high pitch noise at around 8kHz (give or take). I'm attaching an audio recording where you can hear it in the background (you may need to crank up the volume above normal). In reality this noise is much louder and the fan noise is not as loud as it sounds in the recording.

So first question: Did anyone else experience this with their new ET5410+? If so, does it go away? Is there a fix or do you have to tolerate that?

I also noticed that the fan runs on slow revs when idling after boot up and never shuts down. It's in a 10°C environment so heat can't be an issue.

Any pointers really appreciated!

[hp3310a]

I uploaded a video for more context:

[thm_w]

I can't really hear anything, you can use a FFT app on your phone to see what the frequency is.
Then maybe try adding a load and see if it shifts/increases/decreases.

Its not unusual to see on electronics, and some units may have it worse than others.

[gamerpaddy]

fft shows a peak at 6660hz
older ET used a transformer.. this one uses a switching psu. 12V 1A 
one of the lowest quality ones available, not a single filter and junk caps on it

[DaneLaw]

fft shows a peak at 6660hz
older ET used a transformer.. this one uses a switching psu. 12V 1A 
one of the lowest quality ones available, not a single filter and junk caps on it

That also, explains a good part of the weight reduction, which went from around 3.7kg on the older ET5410 down to 2.9kg on the new "enhanced" ET5410+.. also seems they made quite a few other downgrades... according to users should the metal framing around the front, be gone on the +version.
I don't know if anything else has been changed.
- but in East Tester corporate wording-limbo, they just enhanced it, but "enhanced" for who.. the users, or their financial bottom line, likely the ladder.

I haven't noticed any high pitch on my unit, but it's the older variant, so not that comparable.
A brief video from each side of the internals of my unit' if you wanna analyze parts that differ between ET5410 and ET5410+.

 

[hp3310a]

Here's the FFT, mic next to the air intake at the front of the idling unit. One peak (the one that I can hear clearly) is at 7kHz, another almost equally powerful one is at 13kHz (that one I probably can't here). On the low end of the spectrum is the fan.




Not sure yet were the noise comes from, I need to wait what the vendor has to say about it before I open the unit. Could very well be the power supply at the rear end but equally well the LCD. I don't see what would make such noise on the two main boards (but what do I know).

The noise stays the same under load.

[hp3310a]

So, the vendor responded and suggested I repair the unit. Not sure if I'm going to do that, but I was curious whether what he said was right.

First of all, this is my power supply, doesn't look as shabby as it was suggested earlier.  I disconnected it and let it  run without connection to the board. It's completely silent, also when it is connected.


I powered the board from a lab supply to rule the noise is coming from the power supply, maybe indirectly due to ripple. Nope, still there, loud and clear.


It is this chip that makes the noise, next to the power supply socket.


The vendor suggested as much also said what it is: ME7660, a charge pump DC-DC voltage converter (https://www.ic72.com/pdf_file/m/421650.pdf). I put a finger on it and the sound changed. Tried the same with the capacitors nearby but that didn't change the noise.

So I guess the culprit is identified, but will a replacement be any better? With my luck it will be exactly the same or worse.

BTW: I keep failing to insert  images the way other people do with a small expandable image inline, expanding inline to a big version when clicked. My images always open a new page when clicked instead of just changing size. What is the correct board markup to do that inline expansion thing?

[Grandchuck]

C78 or one of the other ceramics?  They can "talk" due to the piezoelectric effect.

[beatman]

Maybe try to resolder all the SMD componeds there and the chip to see if something change.

[ledtester]

BTW: I keep failing to insert  images the way other people do with a small expandable image inline, expanding inline to a big version when clicked. My images always open a new page when clicked instead of just changing size. What is the correct board markup to do that inline expansion thing?

I think it has to do with the resolution of the images. In a lot of cases you can reduce them down significantly without sacrificing clarity.

[thm_w]

I've never had the IC itself generate significant noise, its always the capacitor or inductor.

That said there are tons of 7660 clones, likely with varying frequencies/rise times/etc. So you could swap that out, or swap the caps out as mentioned.

[hp3310a]

Thanks for the comments. The seller and I came to an agreement and I will try to fix the unit by replacing the charge pump chip and its surrounding capacitors. I will probably also replace the power supply because it only produces 11.7V and while using the lab power supply at 12V to see if that makes a difference didn't change anything wrt the noise, it will probably not hurt either. Plus a 12V 1A supply isn't very expensive. I'll report back once this is done.

[hp3310a]

So the replacement chips arrived and I went to work.




After desoldering the old one - which was actually a TP7660 not an ME7660 (has the same function and is pin compatible) - and soldering a new ME7660 in its place I discovered...

I've never had the IC itself generate significant noise, its always the capacitor or inductor.

That said there are tons of 7660 clones, likely with varying frequencies/rise times/etc. So you could swap that out, or swap the caps out as mentioned.

that you were right, the noise was still there and not a bit less loud. I also found that...

Maybe try to resolder all the SMD componeds there and the chip to see if something change.

this was a good suggestion, too. The old chip had a cold solder joint (on the output pin) that was clearly visible when I lifted it.

I don't have the necessary 0805 10µF capacitors here but I have some electrolytics, so I tried those.



Looks terrible but it works, sort of. The noise got a lot less loud but I can still hear it. So one of the ceramic capacitors was involved (both were about the correct capacity), but there's at least one still producing unwanted sound.

Since I don't know which I would have to desolder each one, measure it and order a replacement. That's rather error prone, does someone know the right values?

Also which ceramlcs are less likely to produce such noise, is there a brand were this doesn't happen as a matter of principle (or quality)? The two through-hole caps will be replaced too, once I found the other culprit(s).

Before all this happened I also discovered (under the microscope) that the TP7660 apparently was hand soldered and so was a second capacitor on top of the 1206 to the right of the chip. I didn't notice that in the earlier pictures I had taken. So this was not a new unit but a returned one with a failed repair attempt, great.


Not sure how to determine the intended capacity of the stacked cap, the value that my LCR tweezers spit out likely are not what it is supposed to be.

[wraper]

Before all this happened I also discovered (under the microscope) that the TP7660 apparently was hand soldered and so was a second capacitor on top of the 1206 to the right of the chip. I didn't notice that in the earlier pictures I had taken. So this was not a new unit but a returned one with a failed repair attempt, great.

Could be a factory thing. Something being hand soldered (unless you bought used) usually does not mean it's a returned unit. Use tantalum capacitors instead of MLCC for 7660 if those are what's singing. Although it should be ~10kHz so it's likely LDO problem. HT75A0-1 LDO although not explicitly stated, from datasheet application circuit looks like it is not compatible with using MLCC output capacitor (C78), so it may oscillate and MLCC may sing in such case. My suggestion is to replace C78 with 10-22 uF tantalum capacitor. Also if C79 is more than 0.1uF, it's better to remove it. Using MLCC with incompatible LDO is a common engineering mistake.

[hp3310a]

Could be a factory thing. Something being hand soldered (unless you bought used) usually does not mean it's a returned unit. Use tantalum capacitors instead of MLCC for 7660 if those are what's singing. Although it should be ~10kHz so it's likely LDO problem. HT75A0-1 LDO although not explicitly stated, from datasheet application circuit looks like it is not compatible with using MLCC output capacitor (C78), so it may oscillate and MLCC may sing in such case. My suggestion is to replace C78 with 10-22 uF tantalum capacitor. Also if C79 is more than 0.1uF, it's better to remove it.

That helps, thanks a lot! I'm beginning to understand how this works: 12V in > LDO > 10V > Charge Pump > -10V.


According to the data sheet of the HT75A0-1, it wants 10µF polarized caps too, so if I replace all 4 of them - C53 & C54 for the ME7660 and C49 & C78 for the LDO, thats what you are suggesting? Certainly according to the data sheets for both of them. I wonder what the smaller 0604 caps (C50 & C79) are there for, decoupling maybe?

[wraper]

I don't think replacing 7660 capacitors is really worth it unless C78 does not help. But it won't hurt either. The thing with MLCC is that they lose capacitance under voltage bias, so tantalum may be better there as you'll get more real capacitance with the same capacitance rating.

[hp3310a]

OK, thanks again. Looking at the datasheet of tantalums right now, apparently they are a different form factor and bigger than 0805. The smallest one I can get quickly is only rated for 16V, is that enough head room at +12V supply? 20V ones are a little bigger and those may not fit.

[wraper]

For usual tantallum at least 50% voltage rating margin is a must. But it's fine to use tantallum polymer near max rating.

[hp3310a]

Then I will have to use wired versions because a C-size case which appears standard for 25V is 6mm long and doesn't fit on a 0805 footprint.

[wraper]

If you use wired part, then you can use usual aluminum electrolytic. 20V tantalum will be fine, especially as you need only one. The thing with SMD tantalum is that they get wounded during soldering. They have a self healing mechanism, but when used at high voltage, they will have high failure rate as quite a bit may fail short instead of healing properly. There is no such problem with tantalum and aluminum polymer, aluminum electrolytic.

[hp3310a]

At least for a test I'll use electrolytics, for a test that should work. Not sure if that's a good permanent solution because they hang upside down (the board is facing downwards normally). Maybe I get some of this while glue that is often found in power supplies to keep the components from moving.

[ptluis]

C79 has a bad solder on the IC side, you should check carefully all soldering on the dc-dc circuit.
Soldering on that dc-dc circuit looks really shitty

[hp3310a]

Replacing the caps for the LDO, including the stacked cap, brought almost complete success. There remains a faint noise at the same frequency that I can't really pinpoint the source of, maybe the small 603 caps C49 and C79 - cant really tell.


I also fixed the reversed polarity of the leftmost one I soldered yesterday. The voltage on the -10V test point is only -8V, so not sure what this means in terms of functionality of the board. The -10V rail is used for the FET-input opamps, so they not really secondary I guess but I have no idea how these things work and how critical the -10V really are.

The removed stacked cap had a capacitance of about 25µF-40µF combined, so the replacement is a lot less at 10µF, not sure if that would bring down the voltage further towards -10V.

The reversed ET5411 schematics from morris6 (https://www.eevblog.com/forum/testgear/east-tester-et5411a-(and-et5410a)-electronic-load-schematic-reversed/msg4882754/#msg4882754) put the ME7660 caps at 47µF and the output cap on the LDO at > 22µF. All except the stacked one were 10µF on my board. All those values are way beyond what the data sheets for ME7660 and the LDO say is "common".

I'll turn in on and load it a little to see how it behaves.

[wraper]

There remains a faint noise at the same frequency that I can't really pinpoint the source of

I see another 3x of AMS1117 LDO incorrectly used with MLCC on the output as well. They may oscillate too. You can check it with oscilloscope.

[hp3310a]

With the unit completely closed I can't hear the offending high-pitch noise any longer - in part thanks to the fan that's always on. So yeah, problem solved! I guess I'll leave it at that.

When the tantalums I ordered arrive I will replace the electrolytics, maybe for higher capacitance versions. I managed to find case-B types, those might fit the 805 footprint if I'm lucky. I can choose from 10µF, 22µF and 47µF.

I did run a short test with up to 10A at 20V and the device didn't seem to mind the weak -10V rail, at least I didn't notice a malfunction.