Author Topic: EEVblog #398 - Lecroy 9384C Oscilloscope Repair  (Read 33192 times)

0 Members and 1 Guest are viewing this topic.

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29665
  • Country: au
    • EEVblog
EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« on: December 16, 2012, 08:23:56 pm »


Dave.
 

Online mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 11993
  • Country: gb
    • Mike's Electric Stuff
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #1 on: December 16, 2012, 09:23:49 pm »
Here are my tips for finding supply shorts.
Using resistance measurement, even on a high-precision meter is generally fairly futile due to contact resistance variation
Instead, apply about an amp or two across the supply and measure mv levels across the rails at various places. The voltage will reduce as you get nearer the fault. This effectively gives you a 4-terminal measurement, avoiding the issue of contact resistance

I've also use the 'blow the arse out' method. Now I have thermal imager this approach should be a lot easier in future!
An IR thermometer can also be used by scanning it carefully across the board.
 
The other useful tool is the i-prober. You need to switch the supply with an external MOSFET so you're looking for an AC signal to avoid the position sensitivity.
 
This could be a game-changing tool for faultfinding : a $175 thermal imager
http://www.kickstarter.com/projects/andyrawson/ir-blue-thermal-imaging-smartphone-accessory?ref=card
« Last Edit: December 16, 2012, 09:26:03 pm by mikeselectricstuff »
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline ceecrb1

  • Contributor
  • Posts: 27
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #2 on: December 16, 2012, 09:46:07 pm »
A great video and very well timed for me..
At work, I am currently being beaten by a robotic moving light (martin mac 2000) which refuses to move its pan stepper motor but will "brake" it..
So your video just was great for a "back to basics" remind me what I am missing, remind me of the stages I am missing out and the corners I have cut.

I did find you complaining about spending an hour or so on it really funny.. As I am no where near as good an engineer, it can take me a lot longer.. If I had gotten that far with so little head scratching etc in an hour or so I would be going home very happy with myself!

Isnt it funny how self crittical we can really be!

Really hope you get to the bottom of it, and that you upload the video of showing how you got there!

 

Offline adrianblack

  • Contributor
  • Posts: 27
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #3 on: December 16, 2012, 09:56:35 pm »
Silly question, but what if it was the power supply the whole time? Maybe the model you have actually users 11 amps on the rail normally? Nothing was actually wrong? The low impedance was just a red herring  .... And a dying rail might have been causing the original fault.

I was curious if you can power the board with both the original power supply and also your bunch supply for the 3.3v. You would just have to pit the caps back of course.

I'm probably way off here but thought it would toss it out there.

Sent from my Nexus 7 using Tapatalk 2

 

pir

  • Guest
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #4 on: December 16, 2012, 09:58:32 pm »
well ....there is no way a comp not blown up with 11Amps...
also there is a big probability that a trace would blown up or heated with 11Amps...

also...
taking in mind that you connected and disconnected the power plug into the board many times ...and you put 11Amps...and this is an old horse...
IT IS THE F..... PLUG of the board !! unsolder it and inspect the boar under it
a sort under the plug ..you can't see it...you can't imagine that

it happened to me... O0

Murphy is hidden behind the simplest things
 

Offline M. András

  • Super Contributor
  • ***
  • Posts: 1020
  • Country: hu
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #5 on: December 16, 2012, 10:09:35 pm »
worth take a look at the ps itself too 11 amps would warm up the area around the short even if its the connector
 

Offline JoannaK

  • Frequent Contributor
  • **
  • Posts: 337
  • Country: fi
    • Diytao making blog
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #6 on: December 16, 2012, 10:23:32 pm »
I like that people dare to show the less successfull attempts too (be it design, repair or whatever). Debugging (even Hw) can be PITA, and sometimes you'll just spend hours with dead-ends, wrong leads and none progress what so ever.

Well, at least you'll know that those tantals are OK..   8)

Mike: that IR-imager looks really nice idea.. Would really like to have one of those.
 

Offline david77

  • Frequent Contributor
  • **
  • Posts: 793
  • Country: de
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #7 on: December 16, 2012, 10:24:31 pm »
Can you run the PS without it being connected to the board? My guess is also a flaky 3,3V line on the supply.

Isn't that usually the first step: Check if it's actually plugged in?  ;)
 

Offline hans

  • Super Contributor
  • ***
  • Posts: 1038
  • Country: nl
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #8 on: December 16, 2012, 10:28:56 pm »
I don't see why it would be the power supply, as it's a ATX-like power connector which has shrouds between each contact. If a short is going to exist between the two, it would be noticed at the plastic is severely damaged.
And ofcourse the power supply of the unit isn't very suspicious as there is still a short without the cable plugged in.
But I guess an inspection can never hurt.

I'm very interested whether the unit 'works' if Dave would power the 3.3V from it's bench supply and the rest from the original unit. But I guess that's one hell of a setup to achieve, and still wouldn't narrow down the search (except maybe if channel 2 or 3 is completely dead and the others are working, you know one of those ASICs there is very well done inside).

On college we used to have a short locater. It gives an acoustic beep where the tone depends on the resistance. The ranges were quite small (200 ohms to 200m ohms), so it's very audible to hear differences. It had some very very sharp probes.
Looks like this is the exact unit: http://doc.es.aau.dk/fileadmin/doc.kom.aau.dk/labs_facillities/control/manuals/polar850-sheet.pdf

I think it's probably even nicer or atleast a good alternative to a very expensive  multimeter. You don't have to look at a display all the time, so you can entirely focus on the board and how you grab the probes.

But.. if the short is ~0.16 ohms, and you get ~11A@3.3V, it means you're getting something non-linear. Although faults can be all over the place, wouldn't this suggest its something semiconductorish because would expect to get well over 20A-30A with 0.16ohms short?
« Last Edit: December 16, 2012, 10:31:54 pm by hans »
 

Offline maca_404

  • Contributor
  • Posts: 10
  • Country: au
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #9 on: December 16, 2012, 10:40:36 pm »
Whilst I doubt it is related to the current problem,  When doing the visual inspection on the input you did not comment that there is a section which seems to be broken out of what I assume is one of the brown input relays might be worth a quick look.  Good idea to  run the supply and look for hot spots though if you had a thermal imager I think you may have had more luck with this.
« Last Edit: December 16, 2012, 10:42:58 pm by maca_404 »
 

pir

  • Guest
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #10 on: December 16, 2012, 10:54:06 pm »
I don't see why it would be the power supply, as it's a ATX-like power connector which has shrouds between each contact. If a short is going to exist between the two, it would be noticed at the plastic is severely damaged.

well... ask the grounded screwhole next to the 3rd 3.3V pin
also take in mind the "paranormal activity" while measuring the last cap and bending the board...@23:00
« Last Edit: December 16, 2012, 11:11:56 pm by pir »
 

Offline PuterGeek

  • Regular Contributor
  • *
  • Posts: 88
  • Country: us
    • SolutionsPLUS engineering
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #11 on: December 16, 2012, 11:02:50 pm »
Silly question, but what if it was the power supply the whole time?

I was wondering the same thing. If the 3.3V supply in the power supply is flaky it could do weird things. Also if the resistance is really 0.16 ohms, the power supply should be drawing 3.3V / 0.16ohms = 20.6A instead of 11.5A.

I think I would start by opening the power supply for a visual inspection. Then plug the power supply connector into the board but disconnect the 3.3V wires from the connector and measure the supply power using the dummy load.

If the 3.3V supply can't generate enough amps, I would use the bench supply for 3.3V and scope power for everything else. I might do a partial reassembly to 'running' configuration and see if it works.

Of course, I would probably should solder the caps back down!  >:(

I would hate to change from chasing a power supply issue to chasing a noise issue!
 

Offline peter.mitchell

  • Super Contributor
  • ***
  • Posts: 1568
  • Country: au
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #12 on: December 16, 2012, 11:07:06 pm »
one thing I have found that helps is using canned air and freezing suspect components, powering it up and checking.
 

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2681
  • Country: gb
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #13 on: December 16, 2012, 11:08:55 pm »
I haven't seen the video yet but the 9354 I have pulls 220-240W from the wall and a lot of that is dissipated from the motherboard so I'd expect pretty heavy current flow on the main supply rails.
 

Offline EEgalitarian512

  • Contributor
  • Posts: 11
  • Country: us
  • It is better to be loved/hated for who you are.
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #14 on: December 16, 2012, 11:40:32 pm »
Hi there Dave. Great video, as always. Not sure if this suggestion is doable or helpful - but what about the possibility putting things back together and substituting your 40 Amp power supply just for the 3.3V and or 5V supply lines to see if the unit operates normally. If it does, then that could possibly move the troubleshooting path in the direction of the Power Supply - with the overall assumption being that the 11 Amps is normal for the 3.3V rail. Good luck, whatever approach you choose. Peace. John Martin
 

Offline PuterGeek

  • Regular Contributor
  • *
  • Posts: 88
  • Country: us
    • SolutionsPLUS engineering
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #15 on: December 16, 2012, 11:43:58 pm »
Blowing the short out reminds me of a time I tried that.

Quick turn prototype PCBs in the mid 80's were very expensive so the large (about 16" x 18"/40 cm x 45 cm) 4 layer boards made with one week turn around cost about $5,000. We hand built a board, powered it up and absolutely nothing!

The external 5V power supply is shutting down so we try to find solder bridges but can't find anything. No parts are getting hot and remove a few part but still and still nothing.  :-//

Resistance between 5V and ground is nearly zero (only had a 3 digit meter). On a whim I check a blank board and it's shorted too! And another and another, all 5 are shorted!

Since the power supply we had couldn't blow it out we decide to use the highest current supply we had - a 12V deep discharge lead acid battery that can supply lots of amps!  >:D

We connect wires to the connector, hook up the battery and the insulation cooks off the wire in seconds! We use bigger wire and the insulation is gets pretty soft but the connector fails before it burns off! We solder heavier wires directly to the board and the solder melts!  |O

At this point I'm more than a little steamed! I take one of the bare boards and remove the laminate to expose about 2 square inches of the power planes and solder 10 AWG (2.5 mm) wire to the power planes and hook it up. In a few seconds the solder melts! So I loop the wire for more area and use a lot of solder with similar results!

Being extremely near sighted at the time, I had a very limited focal distance without my glasses (8" / 20 cm tops). This let me see very small detail which is good since I don't think we even had a loupe. I whip off the glasses, hold the board up to a bright light and get very close. Looking at a via, it looks like the hole is off center and maybe shorting the power planes! And so is that one! And that one! In fact every via I look at is off center so much it looks like they all short 5 volts to ground!

They delivered new boards three days later that worked fine. I don't remember if they volunteered or not but they ate the cost for the whole order.
« Last Edit: December 16, 2012, 11:48:21 pm by PuterGeek »
 

Offline LaurenceW

  • Frequent Contributor
  • **
  • Posts: 258
  • Country: gb
    • It's Time, Jim, but not as we know it
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #16 on: December 17, 2012, 12:41:34 am »
Dave, I was just about to offer a suggestion, but see MikesElectricStuff has beaten me to it! I have used this very approach, successfully  before.

You need something like a 1V 1A Constant current power supply, that is not going to do any damage to your board. Get good quality low impedance power connections (ideally SOLDERED) onto the board, and switch on. Set your meter on millivolts, and go hunting for the lowest value - that will be nearest to your shorted component. this is MUCH easier than trying to differentiate a MilliOhm here or there.

Where this might come unstuck is if there are very low impedance power planes, with no appreciable voltage drop across them.

Hmm, suspecting the power supply is an interesting one, but consider this. If 11A is actually a normal current for this board, that means EACH 3.3V pin on that connector is carrying 6A a piece. That sounds a bit much, don't you think?
If you don't measure, you don't get.
 

Offline beretta

  • Newbie
  • Posts: 1
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #17 on: December 17, 2012, 12:49:59 am »
Johnny come lately...I just reviewed this video and all of the ideas that have been suggested are well in line as what I thought.  My first was... load the 3.3 ps and monitor, second check out the connectors, third freeze components.  Looks like a classic fault but hard to find.  Maybe the circuit is happy with the .158 reading on the multimeter ...  Just food for thought.  I think Dave should carry this to Fix!
 

Online mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 11993
  • Country: gb
    • Mike's Electric Stuff
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #18 on: December 17, 2012, 01:07:30 am »
Quote
Where this might come unstuck is if there are very low impedance power planes, with no appreciable voltage drop across them.
In which case crank up the current. Low voltage drop = low power dissipation, so unlikely to do any damage.
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline Mike Whitenton

  • Newbie
  • Posts: 4
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #19 on: December 17, 2012, 01:27:56 am »
Could it be that the power supply is rated at 6A per pin, times 3 pins =18A? If this is the case 11A would be a reasonable load. In which case the power supply would be suspect. I'd open the supply and do a visual.

I was also thinking that unpinning the 3.3V wires from the plug, powering the unit and measuring the 3.3V wires for voltage. It would tell you if the mainboard is shorting the supply or the supply is dead on the 3.3V wires.
« Last Edit: December 17, 2012, 01:39:02 am by Mike Whitenton »
 

Offline AlfBaz

  • Super Contributor
  • ***
  • Posts: 2007
  • Country: au
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #20 on: December 17, 2012, 02:40:07 am »
Some time ago I was fault finding a board with an FPGA on it with a very low resistance on the power rail. Not being able to find anything obvious I started reading up on VLSI chips and thier power requirements. Since they have millions of FETs, a lot of them in complementry cofiguration, with highly capacitive gates, their static on/off state is unknown. You may end up with many parallel low impedence paths to ground. If the VLSI part doesn't power up correctly or has failed some how large section of the chip may end up remaining as it was before powering up causing large current demands

EDIT: There can also be strict power up ramping timings and sequences that cause the asics not to come out of POR
« Last Edit: December 17, 2012, 02:43:16 am by AlfBaz »
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29665
  • Country: au
    • EEVblog
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #21 on: December 17, 2012, 02:57:08 am »
Hmm, suspecting the power supply is an interesting one, but consider this. If 11A is actually a normal current for this board, that means EACH 3.3V pin on that connector is carrying 6A a piece. That sounds a bit much, don't you think?

Yes, that is why I don't question the manual's mention of it being a maximum of 6A capable 3.3V rail. 6A per pin does not make sense.
So I'm sure the total 3.3V supply is normally under 6A, and that would include the ASIC chips looked at, the memory card (not plugged in), and maybe the CPU card (not plugged in)
That means there is some kind of short on the board as I suspect which is drawing at least an extra 7-8A.
Yes, it could possibly be somehow flowing into another unpowered rail, but IME, never by that much.
Yes, the PSU could have failed, that needs to be checked. But that's another issue.
A 3.3V system rail should not be 0.1ohms on a multimeter, something is up, I smelled a rat and still smell a rat.

Dave.
 

Offline lewis

  • Frequent Contributor
  • **
  • Posts: 704
  • Country: gb
  • Nullius in verba
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #22 on: December 17, 2012, 03:19:24 am »
Mike/Dave beat me to it - this is where a thermal imaging camera is indespensible. We use a Fluke Ti10 for this exact type of thing. Apply 100mA-2A constant current, wait a few seconds, see what gets hot.

We've found all sorts of things you wouldn't expect shorted. Works about 98% of the time, and can find most faults like this in a few minutes.

It's one of those things that once you've used one, you'll never be without it.
I will not be pushed, filed, stamped, indexed, briefed, debriefed or numbered.
 

Online Monkeh

  • Super Contributor
  • ***
  • Posts: 5942
  • Country: gb
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #23 on: December 17, 2012, 03:27:41 am »
Hmm, suspecting the power supply is an interesting one, but consider this. If 11A is actually a normal current for this board, that means EACH 3.3V pin on that connector is carrying 6A a piece. That sounds a bit much, don't you think?

Yes, that is why I don't question the manual's mention of it being a maximum of 6A capable 3.3V rail. 6A per pin does not make sense.

So far as I'm aware, those connectors are rated for up to 13A per pin. Mind you, you'd need about 14AWG for that at 3.3V.

E: Could Molex be more annoying? 9A, 13A, make up your minds.
« Last Edit: December 17, 2012, 03:33:05 am by Monkeh »
 

Offline Geezer

  • Newbie
  • Posts: 1
  • Country: us
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #24 on: December 17, 2012, 03:30:22 am »
  Being a newbie, my advise may be taken with a grain of salt, but here goes.    When you first demonstrated the LeCroy in the first tear down and powered the unit, you blew your circuit breaker.  Upon reset of the breaker, you were able to power the oscilloscope and it appeared to operate albeit with several obvious glitches--errors on the DACs, inaccurate readings, dodgy signal traces, etc.  Also, when you showed us the interior of the power supply, it looked pretty crusty and full of all kinds of potentially faulty capacitors.

  To my knowledge, 3.3V CPUs are notorious current hogs.  Even today's multiprocessor CPU can draw 90 to 125 Watts of power.  For 90W @ 3.3V that's approx 27 1/4 amps.  Although, the processors in the scope don't come close to the component capacity in today's CPUs, you still have 4 processors powered by the 3.3V rail.  So, a 11.2A to 10.2A draw from the 3.3V rail seems a plausible power draw for this circuitry. 

  Therefore, like several of those who posted before me, I think there is strong possibility that the 3.3V rail in the power supply requires closer inspection.  A major portion of the scope's functions is dependent on the components powered by the 3.3V rail.  If this power rail is faulty, it could explain most, if not all, of the errors including the breaker tripping, demonstrated in the first video.
 
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf