Siglent Oscilloscope SDS2104X repair help

[Gaktyt]

Hi.

I unfortunately blew one channel on my scope with a bit of overvoltage, and now the trace doesn't appear at all. So where do I start?

Does anyone have any experience with Siglent scope repair? Does anyone have any schematic? Does anyone know if the front-end is repairable or is it all just one big custom chip?

Siglent quoted me €700 for the "repair" (which turns out is just a main-board replacement), and I have about €100 in shipping cost - which lands me just shy of €200 for a completely new scope including shipping. Unfortunately I can't afford the repair nor a new scope at the moment, so I really need to figure out a way of repairing it.

Your help is much appreciated.
Gaktyt

[tautech]

Welcome to the forum.

Dave's teardown shows the front end clearly so if you have other working channels it shouldn't be too hard to follow the signal path until you find the fault.
https://www.eevblog.com/2016/03/31/eevblog-864-siglent-sds2000x-series-oscilloscope-teardown/

[Gaktyt]

Great, thank you for the link.

Yes I have three working channels left :-) , so let me have a try.

[tautech]

You might also find this video Dave did on the front end of the later 4c X-E Siglent scope.
https://www.eevblog.com/2019/07/10/eevblog-1228-do-digital-scopes-have-real-verniers/

It may not give you much more than a better understanding the function of some of the IC's in the front end signal path.
I'm sure if you write up your findings and methods used it can also help others too that blow a channel.
Good luck with your fix. 

[Gaktyt]

Thank you. Great idea, I will try to write something up.

Would this be a sensible approach? I crack the scope open, power it up, put a signal on ch 4 (the blown one) and e.g. ch 2, then follow the signal on both channels with my spare scope, and when ch 4 does not match ch 2 I would have found the first defect component?

[tautech]

Thank you. Great idea, I will try to write something up.

Would this be a sensible approach? I crack the scope open, power it up, put a signal on ch 4 (the blown one) and e.g. ch 2, then follow the signal on both channels with my spare scope, and when ch 4 does not match ch 2 I would have found the first defect component?

Essentially yes, until you get to the differential pairs where they won't be ground referenced so some thought and care need be used for where to place the probe reference lead and how to do this. You may need to use a differential probe if you have one.
Hopefully you will find the issue before then and just connect the probe reference to one of the BNC shells.
The probe compensation square wave output should be fine to use for the test signal.

But before we get ahead of ourselves. the scope must come apart and yet parts need still be connected on the bench so it can be powered ON and be safe to work on. For extended ON time you need to provide cooling for some of the parts so you will need a bench fan.
Depending how it all lays out on the bench and if you can see the display you might need a mirror too.
There's every likelihood you can use one of the channels of the SDS2kX to trace the signal if you can see the display. A short custom link between the probe compensation output to the ch4 input could be handy to reduce the clutter on your bench.
And pictures, everyone loves pictures ! 
Take an analytical approach and work out where you need to probe, make drawings, study datasheets and mark test points with a fine Sharpie.

Where you in 50 Ohm or 1M input mode when it was damaged ?

[Gaktyt]

Hi tautech

Sorry for late reply, you know - holidays, other issues and general procrastination delayed me. But anyway, I cracked it open and there is not much to add to Dave's teardown, it's basically the same piece of equipment with some minor hardware revisions. Also I did not find any hand soldered diodes like in the teardown.

Where you in 50 Ohm or 1M input mode when it was damaged ?

I was in the 50 Ohm mode when it happened. I noted today that ch4 (the blown one) in 1M mode will only go up to 10V/DIV when turning the knob, whereas the other three channels go up to 100V/DIV. Also ch4 trace is not displayed at all on the screen.

I have attached a picture of ch2/ch4 I was not able to identify the AUD 6683 chip. The KAQY214S is a ss-relay.

I connected ch2 and ch4 to the reference signal. On ch4 most notably the signal is missing on AD8370 pin 16, AUD6683 pin 6 and the transistor I marked with a circle. But it is impossible for me to actually "follow" the signal, it's a 4 layer pcb and my hands are shaking from old age. I asked Siglent for a schematic, but oh noooo - it's a big secret.

I don't know what to do but just replace the LMH6552 and AD8370. I can't identify the AUD 6683 and the four transistors in between, not even from the marking codes.

[tautech]

The resistors are what I would check first and that's a bit of work to understand the coding/markings as they appear to be in A11 format.
Have a look here:
http://kiloohm.info/eia96-resistor/10X

[Gaktyt]

Thanks, but why the resistors? (asking out of ignorance)

When it happened I was measuring the output of a current transformer which yields only a few micro amps, the voltage however was a few hundred volts. I would have thought semiconductors would be the prime suspects? And I would need to de-solder the resistors to check them, unless there is a another novel way of doing it that I'm not aware of.

[tautech]

Ah, OK then you are on the right path.
In 50 Ohm input mode find the signal path and study the datasheets for each of the semiconductors in that path for their max input voltages and that may offer clues to the most fragile parts.

I've never had a 2000X apart for repairs but for sure there will be someone here that has.
Keep bumping this thread with your fault finding steps and discoveries until hopefully someone jumps in to help.

[DaJMasta]

If there's actually no trace for the channel displayed at all (not just clipped to one rail or something), I'd actually start with checking the ADC.  If you see a channel trace and it is just unresponsive, then it's likely something in the front end (which is much more typical of overvoltage damage).  Now there's a chance that if one of the chips in the frontend doesn't respond to commands it just disables the trace - in which case looking at them would be a reasonable choice - but if you're really not seeing anything, it may be a digital domain issue, which would mean pretty extensive damage.  If it were just blown passives/dumb programmable gain amps/relays/etc., I would still expect to see some kind of trace displayed, it just wouldn't show any signal, or it would show an entirely wrong signal.

[Gaktyt]

Hi DaJMasta!

If there's actually no trace for the channel displayed at all (not just clipped to one rail or something), I'd actually start with checking the ADC.

There is NO trace whatsoever, nada, nista, niente, ничто, nichts. And the only way I can think of to check the ADC would be to swap the inputs of ch3 and ch4 (which goes to the same ADC), but it's a BGA component on a minimum 4 layer pcb, so chances for success are slim.

Now there's a chance that if one of the chips in the frontend doesn't respond to commands it just disables the trace - in which case looking at them would be a reasonable choice

I guess that is the chance I'm hoping for. The most tangible and workable fact so far is:

On INLO (low input of diff amp, pin 16) on the AD8370 there is no square wave signal, only something that resembles a floating pin. Which obviously means there is no square wave on the output of the AD8370.
 

- but if you're really not seeing anything, it may be a digital domain issue, which would mean pretty extensive damage.

Or it could be the worst of worst case scenarios, both digital & analog problems.

[DaJMasta]

Yes.... if an overpower event happened and you managed to cook your ADC, the whole input section would likely be charcoal...


Hoping for the PGA or something in the path just not returning usable info to the monitoring systems!

[Gaktyt]

After some more measurement I decided to try replace the following three components, but I need some help to translate the smd marking codes into part numbers so that I can order them.

K52G - SOT-23 I believe it's a transistor of some sort NPN/FET
AUE 6683 - SOIC-8 I believe it's a op-amp
T4 -  I believe it's a diode

Can you please assist?

[goaty]

K52G could be BAV199
T4 could be 1N4148

[thinkfat]

AUE (6683) might be a TI OPA727 in VSSOP-8 package.

[Gaktyt]

Hi goaty

K52G could be BAV199
T4 could be 1N4148

Thank you for your suggestions, the BAV199 was also suggested to me from others, so I will order that. And the T4 yes I was also thinking some small signal diode like the 1n4148.

[Gaktyt]

Hi thinkfat

AUE (6683) might be a TI OPA727 in VSSOP-8 package.

Yes I think so too, but still a bit suspicious because the TI datasheet says smd marking is only "AUE" no mentioning of "6682".

[tautech]

Hi thinkfat

AUE (6683) might be a TI OPA727 in VSSOP-8 package.

Yes I think so too, but still a bit suspicious because the TI datasheet says smd marking is only "AUE" no mentioning of "6682".

6682 will be a production/date code.
Datasheets normally show what it means.

[DaJMasta]

Hi thinkfat

AUE (6683) might be a TI OPA727 in VSSOP-8 package.

Yes I think so too, but still a bit suspicious because the TI datasheet says smd marking is only "AUE" no mentioning of "6682".

There is some chance that the other marking is on the bottom side - if you end up pulling it to replace it, it's worth a check.

[marq12345]

Sorry to revive an old thread, but I recently damaged channel 2 on my 1104X-E and a google search brought me to this thread while I was trying to identify components. The trace for this channel was off-screen no matter what I did with gain, vertical offset, or AC coupling. The front-end also has a SOT-23 package marked K52G, which was shorted. Turns out it is actually the input JFET to the discrete amplifier, 2SK508G, but I couldn't easily find it from Digikey etc... so I ordered it from LCSC (Which I think the is parent company of JLCPCB):
https://www.lcsc.com/product-detail/Bipolar-Transistors-BJT_UTC-Unisonic-Tech-2SK508G-K52-AE3-R_C258268.html

After replacing it, the scope seems to work normally again, so this JFET is at least a good enough match. Also, while it was still broken, I attempted a self-cal (which didn't help), and after replacing the JFET, I had to self-cal again for the trace to appear back where it should. Surprisingly, in my case, the two protection steering diodes to 5V and 0V at the input of the JFET are still ok. Hopefully this helps someone in the future.

[FringIdeas]

@marq12345, and everyone else with the pics, links, etc, thank you!

I had the same scope, 1104X-E, with the same problem, the trace would not stay on the screen. I found nothing while internet searching the first few times. Then accidentally found this thread. I went and ordered a few 2SK508G-K52-AE3-R from the link. Took about a month to get to me, but I got them. I opened the scope and the K52G did seem to check out bad. So I replaced it with the new part, fired up the scope, did a self calibration, and I got my trace back!! So happy.. so happy..

Thank you for sharing the information!