Rigol DS1102E Failed - again!

[BobbyK]

I posted about my failed Rigol before - it failed to boot with a black screen and all lights stayed on. This happened last summer as soon as the weather got warm here. Last time I sent it in for repair - took two weeks, but it got back working. Today (weather's warm again) the same thing happened.

Not wanting to send it back for repairs every summer, I opened the thing up and measured some voltages coming from the switch mode supply, and they were all much lower than they should be (4.2V, 9.2V, -3.1 V...) ...poor regulation? Couldn't find any obvious problems on the switch mode supply, and since I don't have a second scope, it's hard to check. Has anyone had similar problems (Black screen, lights come on and stay on, no booting)? And have you been able to fix it without sending the thing in for repairs...?

This is getting really frustrating,...I need my test equipment to work and my circuits to fail(!) not the other way round 


Based on the supply diagram from "A Helene" in another post, I measured the psu output voltages while connected to the main board. Here are the results....




As you can see, all the voltages are too low. Any ideas what could cause this? Any help is greatly appreciated.

[kunc]

http://rigol.codenaschen.de/index.php/Schematics

Capacitive or resistive check tl431 and nearby

[Hagis2k]

Does this only happen when it is warm? try cooling down some components and see what happens you should contact rigol maybe they are kind enough to send you a replacement power supply?

[GiskardReventlov]

After reading your other post

https://www.eevblog.com/forum/testgear/rigol-ds1102-black-screen-problem-fails-to-boot/

if I was in the same situation, I'd send it back for repair. I'd make sure they knew this was the second time and you don't want a third time.

If they didn't do the proper repair on the FPGA - BGA, maybe they just reheated it, then it's likely to show up again and again.  The PS3 suffered a very similar failure (epic) and the repairs involved sony giving new PS3 instead of repairing and returning your PS3.  Most likely because they had to be removed and reballed.

Best of luck to you on getting it worked out we all know the feeling.

My only concern is that now Dave's going to suggest that we really all should have 2 oscilloscopes.

[markce]

Dave's going to suggest that we really all should have 2 oscilloscopes

That's not a second Rigol though ( ).

[BobbyK]

Here is a quick update for anyone who might read this in the future:

The low output voltages of the Rigol 1102E SMPS were the result of leaky/faulty output capacitors - NO NOT THE ELECTROLYTICS - BUT THE MLCCs! 

This is the first time I am seeing almost every single MLCC in a power supply fail, with nothing else exploding!! Rigol must have bought a batch of doozies, I cannot explain why they all had really low capacitance (e.g. 42nF~65nF for 100 nF caps, and 3nF fr 10nF ones!). I exchanged ALL MLCCs and the scope came back to life. Voltages match those of the reverse engineered schematic, and all seems good again!

Can someone explain this phenomenon? Can Heat destroy MLCCs faster than small Electrolytics? Or was it just a bad batch of caps they stuck in their products? 

[Kleinstein]

MLCC are somewhat sensitive to mechanical stress. However this usually results in a short - though reduced capacitance might very well go unnoticed in many circuits.  Heat alone usually is not a problem unless it causes mechnical stress.

[tom66]

I have seen MLCCs fail, but never that many at once. Usually they fail leaky (often times short) but will pass a capacitance test, if not shorted, within 20% or so.

[wblock]

Here is a quick update for anyone who might read this in the future:

The low output voltages of the Rigol 1102E SMPS were the result of leaky/faulty output capacitors - NO NOT THE ELECTROLYTICS - BUT THE MLCCs! 

This is the first time I am seeing almost every single MLCC in a power supply fail, with nothing else exploding!! Rigol must have bought a batch of doozies, I cannot explain why they all had really low capacitance (e.g. 42nF~65nF for 100 nF caps, and 3nF fr 10nF ones!). I exchanged ALL MLCCs and the scope came back to life. Voltages match those of the reverse engineered schematic, and all seems good again!

Can someone explain this phenomenon? Can Heat destroy MLCCs faster than small Electrolytics? Or was it just a bad batch of caps they stuck in their products? 

The smallest I see on that schematic are "u1" caps.  Are these not shown there?  Are they the orange ones on the left of the first photo, or SMD on the bottom of the board?

If you have a list of part labels and values, please post it.  With so many failing, it will probably be happening to more people.

[BobbyK]

Here is a quick update for anyone who might read this in the future:

The low output voltages of the Rigol 1102E SMPS were the result of leaky/faulty output capacitors - NO NOT THE ELECTROLYTICS - BUT THE MLCCs! 

This is the first time I am seeing almost every single MLCC in a power supply fail, with nothing else exploding!! Rigol must have bought a batch of doozies, I cannot explain why they all had really low capacitance (e.g. 42nF~65nF for 100 nF caps, and 3nF fr 10nF ones!). I exchanged ALL MLCCs and the scope came back to life. Voltages match those of the reverse engineered schematic, and all seems good again!

Can someone explain this phenomenon? Can Heat destroy MLCCs faster than small Electrolytics? Or was it just a bad batch of caps they stuck in their products? 

The smallest I see on that schematic are "u1" caps.  Are these not shown there?  Are they the orange ones on the left of the first photo, or SMD on the bottom of the board?

If you have a list of part labels and values, please post it.  With so many failing, it will probably be happening to more people.

u1 is another way of writing .1u which is 100nF. There are 10 and 100nF caps near the outputconnector, and spread throughout the board. The small green MLCCs were also bad! I have marked the bad ones, as far as they can be seen in this picture with triangles:

[BobbyK]

MLCC are somewhat sensitive to mechanical stress. However this usually results in a short - though reduced capacitance might very well go unnoticed in many circuits.  Heat alone usually is not a problem unless it causes mechnical stress.

I did some more research, and you are right. Thermal stresses, resulting in mechanical stresses, seem to affect MLCCs very negatively. The soldering process can be a major cause of failures similar to what I had, and so can shock and/or a layout that promotes bending or stress in areas where MLCCs are installed. I have a new respect for MLCCs now - I always assumed they are very safe to use and that their placement is not as crucial as e.g. Electrolytes or References, zeners, etc.

[wblock]

u1 is another way of writing .1u which is 100nF.

Well, yes, but there aren't any 10nF (u01) shown on that schematic.

There are 10 and 100nF caps near the outputconnector, and spread throughout the board. The small green MLCCs were also bad! I have marked the bad ones, as far as they can be seen in this picture with triangles:

And they all failed the same way, open rather than shorted?  That's just weird.

[BobbyK]

u1 is another way of writing .1u which is 100nF.

Well, yes, but there aren't any 10nF (u01) shown on that schematic.

There are 10 and 100nF caps near the outputconnector, and spread throughout the board. The small green MLCCs were also bad! I have marked the bad ones, as far as they can be seen in this picture with triangles:

And they all failed the same way, open rather than shorted?  That's just weird.

I'll have to correct myself - the 10n is C7 I believe (in the picture, not schematic) and is not next to the output connector.

And the failing open thing is completely new to me too. I have seen caps fail open before, but never all caps on a board - it sounds crazy, and I checked every single other component on that board, out-of-circuit, just to be sure, before I even looked at the MLCC caps. It was that "weird"ness that made me write that last post - otherwise, most people know that with SMPS's it's always the electrolytes and the diodes/Mosfets, right?

BTW, the schematic I believe is only partial, and I don't know if it is from the early 1052e's or from the later boards (with some unpopulated areas, and changed components). I used it as a guide for troubleshooting, but my board certainly had deviations from that schematic.

[wraper]

(e.g. 42nF~65nF for 100 nF caps, and 3nF fr 10nF ones!). I exchanged ALL MLCCs and the scope came back to life. Voltages match those of the reverse engineered schematic, and all seems good again!

Can someone explain this phenomenon? Can Heat destroy MLCCs faster than small Electrolytics? Or was it just a bad batch of caps they stuck in their products? 

Did you measure them right after desoldering while they were still hot or at least warm? If they are from Y5V, Z5U (what they most likely are) or similar ceramic this would be what they should normally measure at elevated temperatures as they are very unstable and insanely drop their capacitance when heated. You can just take them with the warm fingers and look how the capacitance drops. And IMO fault were caused only by one cap? in the feedback loop, somewhere around TL431, and failure wasn't because of the capacitance drop but probably leak/short (which can disappear when you heat the MLCC).