No Script, No Fear, All Opinion
RSS icon Home icon
  • EEVblog #398 – Lecroy 9384C Oscilloscope Repair

    Posted on December 17th, 2012 EEVblog 38 comments


    Dave attempts to trace a short “real-time” on the main board of the Lecroy 9384C oscilloscope.
    Teardown video is HERE

    Be Sociable, Share!
    • Tony

      Nice video, sometimes the fun is just in hunting down the problem. Could the short be in one of the big white connectors for the memory expansion boards?

    • http://wattcircuit.com David

      Did you consider trying the TTI current prober? I’ve had some luck with it for this kind of thing, especially when you can put 11 amps through the short, should a piece of piss to trace

    • Ray

      Dave’s comment about a thermal imaging camera being useful caused me to make a mental connection. On the Hack A Day blog there was a post on Dec. 15th titled “A thermal imaging camera for your phone.” It is about a Kickstarter project offering the PCB for US$12 and the assembled device for US$175 (the sensor costs ~US$80).

      The inventor says the original motivation was to track down drafts in an old house. Where I live, we have a similar concern about hot spots and electricity costs for air conditioning.

      It strikes me that a product like this would sell well in big-box hardware stores at a <US$100 price point.

    • Alan

      Could it be a faulty PSU ?
      Not giving enogh current, since nothing blew up, when you connected external PSU ?

    • Alex

      Hello Dave,

      would also think it could be the power supply (given that the information of supplying 6 A is possibly not correct) it is not a real short given that quite some current can be normal on a 3.3 V supply .. hm ..

      • Alan

        totaly agree

      • http://www.eevblog.com EEVblog

        There ma be something wrong with the supply, and it needs checking. But if so, that’s a different issue.
        A normal 3.3V system rail should not measure 0.1ohms both ways on a multimeter. I smell a rat.

    • Miad

      Try giving the 11 amps with the memory cards and the processor board mounted. Also is it possible that the 3.3v plane is shorted somewhere in the board? Maybe it could be visible by placing the board infront of an intense light source and looking for anything suspicious.

      • Alan

        yes, should’ve been tested as a complete unit with external PSU connected

    • Alan

      also, rememember that the initial 11,4 A never increased, but decreased to 10,something in a while ?

    • Rachie

      There appears to be a cracked relay case at 29:19. It seems unlikely a relay could short out like this, but it looks quite suspicious.

      • http://www.gmdsol.com Francisco

        Yes, there is something not right with the relay component, look around minute 12:48…

        I would first check that out! :)

        Thanks

        • http://www.eevblog.com EEVblog

          Nothing to do with it. It was damaged during the teardown.

    • http://twitter.com/ajasmin Alexandre Jasmin

      Would it make sense to power the thing up by supplying the 3.3v rail from your high current source and try to guess what component failed by using the scope?

      • f4eru

        Easier : put the 10 amp
        Probe around the voltage with the multimeter exactly like you did with resistance => you should find quicker wich cap has the lowest voltage

    • http://electrooptical.net/ Phil Hobbs

      3.3V/11A is 0.3 ohms, so if that’s accurate, the leads are also about 0.15 ohms, which seems a bit high.

      However, if you leave the bench supply connected you can trace tens millivolts instead of single-digit milliohms. Also, more importantly, you have a built-in Kelvin connection so you don’t care about the contact resistance. That makes it a much better method than using an ohmmeter.

      Good hunting!

      Phil Hobbs

    • Alex

      Hello Dave,

      yes thinking twice it is still a bit high current for working condition (so yes could be that only your lab supply was able to maintain driving current thru the working chips and the possibly faulty low resistance path) hm ..not easy one..all the best finding the cause. Alex

    • huh

      I see some power 3 terminal devices on that board. Could one of them be a power mosfet? They’re known for often becoming a short when they fail.

      • Zeke

        I’ve had a rash of such shorted FET’s lately. Of course the equipment is 15 years old, and the motors they are driving are freezing up due to lack of lubrication. Lube the motors and add fuse protection. Don’t upgrade, keep the ancient crap running. Billion dollar corporation.. buggers.

    • Kodon

      I noticed also the cracked relay case. Sure it’s not it?

    • swedeham

      Hi Dave Why not try using a tone signal like a 1Khz signal connected at the 3.3 volt socket and then you trace it with a low power amp ie a pair of active PC speakers and use the 3.5mm plug as a tracer. Follow the signal “downstream” and where you loose the signal there is your short. To make it more precise lower the tonegen level or the volume/amplification on the speakers. also touch the circuit when you get close.
      Good luck /Mats

      • f4eru

        an oscilloscope works better for this :)

        • swedeham

          Yes f4eru I agree ,i guess i just answered what i know would have worked and he does have scopes in spades ,but he seem to use moltimeters in a big way.

          Let’s see what he comes up with ;-)

    • Worf

      I think Dave jinxed it at the beginning when he said it was a live debugging and if you’ve come to the end and it’s still not fixed…

      • http://www.eevblog.com EEVblog

        Yeah, it was inevitable!

    • Hans

      Mouser has a sale on Magic smoke-

    • Joe

      Another video Dave!! You can fix it!!! Then sell raffle tickets or auction it off..

    • Sam Reaves

      Dave,
      Glad to see that you are getting back to it.

      I would think about pulling one of the chips on the +3.3V supply. Also you have the TTI Iprobe. You should be able to trace the current and see where it goes. In the meantime I’ll see what I can find out on the 3.3V current requirements.

      What is the serial number of the instrument?

      Sam
      Moderator
      LeCroy_Owners_Group on Yahoo! Groups

    • Ice Skate

      Hi Dave,

      very very good video, showing pratical EE’s skills is awesome. I am awaiting the second part with some genius methods proposed in the forum’s thread to track down this short!

      Keep it on going this greatfull debugging session, revive this old beautifull piece of hardware!

    • KingS Twisted

      Dave…
      Don’t you wish you had some cool device that could trace 11 amps of current through a circuit board. Something like a “AIM-TTi I-Prober 520 Current Probe” (EEVblog #296).

    • Kasper

      Maybe you could spray some cold spray on the board while powering it? That should reveal any hotspots like shorted caps.

      • http://www.siliconchip.com.au Nicholas Vinen

        Or perhaps isopropyl alcohol, it should evaporate much faster when it hits something hot.

    • Alan

      OK, I’m not really into this, but :
      given the fact that an ext. PSU haven’t blown anything, and given the fact that the actual current (pushed by the ext. PSU)only dropped with the time powered, I would assume that nothing is wrong with the resistance on a 3,3V rail.
      If P(Watts) = I2 x R (I squared times the resistance), it should make: 11.5 Amps X 11.5 Amps X 0.16 Ohms = 132.25 Amps x 0.16 Ohms = 21.16 Watts.
      There are at least 4 heat-sinked IC’s dissipating (in a worst case scenario) a total of 21.16 Watts.
      Divided by 4 to approximate each of the heat-sinked IC’s, that gives a 5.29 Watts per IC.
      I don’t find that fact that much unusal.

      Good luck with your repair !

      Best regards,

      Alan

    • schen

      Worst case, nastiest buggiest problem that could have happened is maybe the oscillator or some other common signal going to the four digital ICs has gone bad.

      Could also be a huge red herring. There seem to be many of them when debugging power rails on scopes and complex test equipment. Recently, I fixed (okay, well, debugged, and never got around to fixing) a 4 channel HP ‘scope, and I was following a similar red herring where the power supply drew too much current and kept on going into current foldback. After tons of debugging looking for the short, independently powering each rail, I accidentally shorted the 5V line on the power supply while measuring voltages (and this was while it was plugged in and powered). I got a little spark, and magically, the scope powered on perfectly as if nothing had happened.

      After a lot more investigation, it turned out that this particular scope model was very sensitive to the order in which the supplies were powered up. If they came up in the wrong order, you’d get a dead short in one of the rails, resulting in that fizzling sound. But, if you did this trick where you temporarily shorted (or loaded into current foldback) the 5V line after powering up the other rails, it’d magically turn on again. My guess is that some of the caps in one of the rails started going bad, and that changed how quickly one rail rose with respect to the other one, causing the scope to no longer power up. The “shorting” fix very consistently worked. The difference in rise order / time must have also been marginal, since the scope would continue to function after “fixed” after being left off for several minutes, as the caps kept enough charge to keep it in the correct state.

      Maybe check for ripple from the power supply (while loaded with a dummy load) to see if perhaps it’s bad caps causing different power-up sequences than required? It’s entirely possible that given the correct rail turn-on order, the 3.3V line will actually not short.

      • f4eru

        Good idea.
        The circuits may draw too much current due to non functionnal parts of the circuitry.

        Test to do : power the 3.3 V externally, the rest on the original supply…

    • xDR1TeK

      Won’t powering up the 3.3v circuit, while a component is unloading its net or nodes, blow up something else?

    • xDR1TeK

      I mean if the component were to be removed and hence unloading its nets or nodes.

    • http://Linearlogix.com Chuck

      Dave

      Noticed you had your hands on the PCB while flexing at time 18:30ish and the edge cap read .100 ohm, this may have been you body impedance pressing on the board when you where flexing it..