Author Topic: EEVblog #398 - Lecroy 9384C Oscilloscope Repair  (Read 33336 times)

0 Members and 1 Guest are viewing this topic.

Offline cidcorp

  • Supporter
  • ****
  • Posts: 105
  • Country: ca
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #25 on: December 17, 2012, 03:42:31 am »

I have to say - I do really enjoy sitting back and reading the threads on these troubleshooting missions.  I really don't have much experience in troubleshooting devices like this
and I learn alot each time one of these threads pops up.  Wish I could offer some advice.

BTW - Those thermal imaging cameras can't be cheap...are there ones out there that don't require you to mortgage your home?  Just curious...cause it makes sense they *could*
save you alot of time searching for a fault.

Anyway, just wanted to say I'm enjoying this thread.  :-+ :-+
 

Offline lewis

  • Frequent Contributor
  • **
  • Posts: 704
  • Country: gb
  • Nullius in verba
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #26 on: December 17, 2012, 03:51:15 am »
It's extremely unlikely the 3.3V rail is fine, it should not read that low on an ohmmeter even if it is designed to take 11A. I don't know what the output voltage of that Agilent multimeter is on the ohms range, but it's unlikely to be enough to want to power up everything connected to the 3.3V rail, and it's certainly not magical enough to turn on any semiconductor junctions when the probes are reversed. Semiconductors are not linear and will draw no current if the 'supply voltage' (from the ohmmeter) is low enough or negative.

Also, as someone mentioned, the low resistance reading suggests the rail should draw 20A or so at 3.3V. So if Dave's getting 11A, something's heating up fast and increasing it's resistance, reducing the current. That suggests to me it's not semiconductory, probably tantalumy, ceramicy, bonding-wirey, or circuit-boardy. Or the resistance reading is erroneous. Either way, as Dave says, there is definitely a rat there somewhere.
I will not be pushed, filed, stamped, indexed, briefed, debriefed or numbered.
 

Offline ecat

  • Frequent Contributor
  • **
  • Posts: 296
  • Country: gb
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #27 on: December 17, 2012, 03:53:26 am »

...

BTW - Those thermal imaging cameras can't be cheap...are there ones out there that don't require you to mortgage your home?  Just curious...cause it makes sense they *could*
save you alot of time searching for a fault.

You could take a look at ThermoPhone (heh, this is what I would call it, or maybe iPhone uHot :) )
http://www.kickstarter.com/projects/andyrawson/ir-blue-thermal-imaging-smartphone-accessory?ref=card
 

Offline lewis

  • Frequent Contributor
  • **
  • Posts: 704
  • Country: gb
  • Nullius in verba
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #28 on: December 17, 2012, 03:58:57 am »
BTW - Those thermal imaging cameras can't be cheap...are there ones out there that don't require you to mortgage your home?  Just curious...cause it makes sense they *could*
save you alot of time searching for a fault.

They're not cheap - http://www.rapidonline.com/Test-Measurement/Fluke-Ti10-Thermal-imager-82499

But they're coming down! http://www.alpha-electronics.com/Items/10930/flir-i7-educational-package.html
I will not be pushed, filed, stamped, indexed, briefed, debriefed or numbered.
 

Offline excess_heat

  • Newbie
  • Posts: 1
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #29 on: December 17, 2012, 04:08:04 am »
Why not try your AIM-TTi I-Prober 520 in field mode to track down the short?
 

Offline jammit

  • Newbie
  • Posts: 1
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #30 on: December 17, 2012, 04:21:23 am »
Here's my two cents. I posted to the Youtube video, but finally decided to get an account here.
3.2V at 11A is 0.29 ohms. The resistance you were measuring was about 0.110 ohms at the DC power plug (I subtracted the 0.060 leads resistance). The resistance almost tripled when when you went to 11A. This is telling me the traces are heating up and increasing resistance. You might want to desolder the DC plug and check for shorts under it, and you might want to measure the voltage at the tantalum bypass capacitors when powered with this 11A. An ohm meter uses a small current to measure the voltage across resistors, and the changes in resistance is small. Imagine if the ohm meter uses 11A instead.
 

Offline htassell

  • Supporter
  • ****
  • Posts: 29
  • Country: au
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #31 on: December 17, 2012, 06:01:40 am »
Hi,

I just posted on Youtube, but I thought I'd mention it here. As others indicated on the video, you can see damage to one of the brown packages in the analog front end of the left most channel (when the BNC's are facing towards the bottom of the frame at 12:36-12:54.

I guess it could just be physical damage due to handling the board and getting it in and out of the chasis etc. rather than an electrical fault. It may well be unrelated, but it might explain some of the other weired behaviour if it isn't the cause of the short.

I thought I'd just mention it in case it was worth looking at...
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29891
  • Country: au
    • EEVblog
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #32 on: December 17, 2012, 07:08:23 am »
I just posted on Youtube, but I thought I'd mention it here. As others indicated on the video, you can see damage to one of the brown packages in the analog front end of the left most channel (when the BNC's are facing towards the bottom of the frame at 12:36-12:54.

Yep, aware of that, done by accident during the teardown.
Not related in any way.

Dave.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29891
  • Country: au
    • EEVblog
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #33 on: December 17, 2012, 07:12:25 am »
BTW - Those thermal imaging cameras can't be cheap...are there ones out there that don't require you to mortgage your home?  Just curious...cause it makes sense they *could*
save you alot of time searching for a fault.

You won't get any change from $1500 for a bottom of the range one.
That's why I don't have one, and even if I did, I'd be a bit loath to use it in a troubleshooting video knowing it's not a tool Joe Average is likely to have.

Dave.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29891
  • Country: au
    • EEVblog
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #34 on: December 17, 2012, 07:25:54 am »
But they're coming down! http://www.alpha-electronics.com/Items/10930/flir-i7-educational-package.html

AU$990 "educational" pricing.
http://www.flir.com/cs/apac/en/view/?id=56018

I'm an "educational" resource, surely?  ;D
But still, even at that price it's a tool that's hard to justify.

Dave.
 

pir

  • Guest
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #35 on: December 17, 2012, 08:22:45 am »
Here's my two cents. I posted to the Youtube video, but finally decided to get an account here.
3.2V at 11A is 0.29 ohms. The resistance you were measuring was about 0.110 ohms at the DC power plug (I subtracted the 0.060 leads resistance). The resistance almost tripled when when you went to 11A. This is telling me the traces are heating up and increasing resistance. You might want to desolder the DC plug and check for shorts under it, and you might want to measure the voltage at the tantalum bypass capacitors when powered with this 11A. An ohm meter uses a small current to measure the voltage across resistors, and the changes in resistance is small. Imagine if the ohm meter uses 11A instead.

 :-+
well i'll take it a little further....
the scope is 1G so FOR SURE there is a "symmetrical tracing lengths or symmetrical impedance" all over the place (as possible..from the designers point of view..especially over the memory storage circuitry)
i bet that the "paranormal activity" over the 4Th tantalum cap measuring 0.100 Ohms @ 23:00 was also the same to the other 3 caps...board bending....
so it is a 3.3V-GND short from a trace...
and where is the bigest possibility for this to happen? ...THE SCREW HOLE NEXT TO THE PLUG(pin no3  and no2 3.3V...pin no1 is not shorted)
and why this was to happen?....because THE PLASTIC CASE of the scope is broken...it fell off !!!  :-//
and why is this also possible to be the teardown video problems with that glitches?....the 2 memory boards hung-over from their plugs and their weight put stress on the plugs and the board...
and what is the "mechanical lever" to absorb that behavior?.... the screw next to "MAIN POWER PLUG"  |O
DESIGNER'S FAIL !!!!

i suggest Mike's method with the voltage drop and interrogate that screw hole

p.s.
damn broken keyboard...that's why the editing
« Last Edit: December 17, 2012, 10:31:45 am by pir »
 

Online IanJ

  • Supporter
  • ****
  • Posts: 1007
  • Country: scotland
  • Pro EE guy many years ago, now a hobby/home biz.
    • IanJohnston.com
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #36 on: December 17, 2012, 09:47:01 am »
Hi all,

Just because it's easy to do and just to rule it out I'd hook up the Lecroy PSU but substitute your bench PSU on the 3.3v rail and check the current consumption. I suspect the current will drop down a bit. If it drops a lot and to within spec of the Lecroy PSU then there's some ideas for you.

Yes, we've all been there........and frustrating it certainly can be.

Ian.
Ian Johnston
www.ianjohnston.com
Manufacturer of the PDVS2 & PDVS2mini
 

Offline T4P

  • Super Contributor
  • ***
  • Posts: 3706
  • Country: sg
    • T4P
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #37 on: December 17, 2012, 09:47:45 am »
The resistance you were measuring was about 0.110 ohms at the DC power plug (I subtracted the 0.060 leads resistance). The resistance almost tripled when when you went to 11A.
He had his leads rel'd already.
 

Offline JackOfVA

  • Supporter
  • ****
  • Posts: 350
  • Country: us
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #38 on: December 17, 2012, 04:37:39 pm »
As a rough estimate of the 3.3V rail consumption, we can work backwards from the heat sink Rth and observed temperature rise. I saw 8 ICs with heat sinks in the area identified as 3.3V rail powered and I believe Dave said the heat sinks measured around 60 deg C after a few minutes with the high current 3.3V supply.

For a typical room temperature of say 28 C, the temperature rise is 32 deg. As a really rough estimate of Rth, take  6 deg/watt -- I imagine there is fan cooling and 6 deg/watt is a guestimate of Rth without the fan running. Probably closer to 3 deg/watt with the fan operating.

(The 8 heat sinks are 4 each of two types, but for this purpose we'll assume all are 6 deg/watt.

Each heat sink is thus responsible for 32 deg rise / 6 deg/watt = 5.3 watts per device, or 43 watts for all 8 devices.

At 3.3V supply, current required is 43/3.3 = 13 amperes.

This is not wildly off the 11A observed draw.

This estimate could be refined with data on the heat sink dimensions and measured temperature rise, of course.

Jack


 

Offline M. András

  • Super Contributor
  • ***
  • Posts: 1020
  • Country: hu
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #39 on: December 17, 2012, 04:47:37 pm »
btw at the beginning of the video there is tons of dust and god knows what under the little soic chips near the power connector maybe a good dusting with compressed air would help
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 7234
  • Country: us
    • SiliconValleyGarage
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #40 on: December 17, 2012, 05:17:39 pm »
video at 11:14 : you can clearly see the deformation of the pcb around the screw hole. Take an xacto knife and scrape that metal off the top and bottom layer. then carefully scratch the fr4 away.. if you hit another plane : peel the plane away as well...

i have fixed an agilent scope once. they used long smetal standoffs from the board to tthe case. the case got thumped exactly where there was a standoff. this compresses the board enough that it actually permanently dented the board around the hole. there was a short between planes. by peeling off the pad of that hole i managed to get rid of the short. i also rpelaced the metal standoff with a nylon one ( as the internal 5 volts plane was exposed after peeling off )

scope still works five years later.
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline lewis

  • Frequent Contributor
  • **
  • Posts: 704
  • Country: gb
  • Nullius in verba
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #41 on: December 17, 2012, 05:34:33 pm »
But they're coming down! http://www.alpha-electronics.com/Items/10930/flir-i7-educational-package.html

AU$990 "educational" pricing.
http://www.flir.com/cs/apac/en/view/?id=56018

I'm an "educational" resource, surely?  ;D
But still, even at that price it's a tool that's hard to justify.

Dave.

Never take any list price at face value, there's always some haggling to be done even if you're not strictly an 'educational' resource (university/school)  ;) Failing that, if you know someone at a uni, get them to buy it for you. In fact, I'd bet Flir or Fluke would give you one!

It's an extremely good tool for the price, especially when compared to a $12000 scope, not just for troubleshooting but for design too. I use mine for thermal modelling and heatsink evaluation during the design phase of pretty much everything and it's very very useful.
I will not be pushed, filed, stamped, indexed, briefed, debriefed or numbered.
 

pir

  • Guest
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #42 on: December 17, 2012, 06:15:43 pm »
video at 11:14 : you can clearly see the deformation of the pcb around the screw hole. Take an xacto knife and scrape that metal off the top and bottom layer. then carefully scratch the fr4 away.. if you hit another plane : peel the plane away as well...

i have fixed an agilent scope once. they used long smetal standoffs from the board to tthe case. the case got thumped exactly where there was a standoff. this compresses the board enough that it actually permanently dented the board around the hole. there was a short between planes. by peeling off the pad of that hole i managed to get rid of the short. i also rpelaced the metal standoff with a nylon one ( as the internal 5 volts plane was exposed after peeling off )

scope still works five years later.

 :-+
 

Offline jancumps

  • Supporter
  • ****
  • Posts: 1205
  • Country: be
  • New Low
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #43 on: December 17, 2012, 07:03:02 pm »
video at 11:14 : you can clearly see the deformation of the pcb around the screw hole. Take an xacto knife and scrape that metal off the top and bottom layer. then carefully scratch the fr4 away.. if you hit another plane : peel the plane away as well...

i have fixed an agilent scope once. they used long smetal standoffs from the board to tthe case. the case got thumped exactly where there was a standoff. this compresses the board enough that it actually permanently dented the board around the hole. there was a short between planes. by peeling off the pad of that hole i managed to get rid of the short. i also rpelaced the metal standoff with a nylon one ( as the internal 5 volts plane was exposed after peeling off )

scope still works five years later.

I feel (yes, not based on anything better than feelings) that we have a winner here.
 

Offline PuterGeek

  • Regular Contributor
  • *
  • Posts: 88
  • Country: us
    • SolutionsPLUS engineering
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #44 on: December 17, 2012, 07:18:47 pm »
Dave,

Don't you just love all the 'free' advice you are getting (me included)?  ;D

-Joe
 

Offline Electr0nicus

  • Regular Contributor
  • *
  • Posts: 121
  • Country: at
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #45 on: December 17, 2012, 07:53:24 pm »
Dave,

Don't you just love all the 'free' advice you are getting (me included)?  ;D

-Joe

Means we all want to see a part 2, where Dave actually fixes that broken LeCroy.

I also suspect that hole as the culprit, as many people here also do. Maybe the 3.3V power supply layer on one of the inner layers has a very small clearance to the plating of that hole. And as you can see in the video the screw has somehow fractured the plating and then maybe pressed the splitters against this 3,3V power supply layer, shorting it out to ground. Because when you look at that screw hole closely, you can see, that the PCB is there somehow de- laminated.

That wouldn't surprise me, because if you watch Dave's first video, you can see that this lecroy has some really big damage to the side of the case. This indicates that it may have been dropped. So that would put very high stresses on the PCB, particularly at the points where it is screwed to the case.

I would recommend to drill out the plating of that hole first, and then see if something changes.

The high current experiment Dave performed also tells me, that if that screw hole has somewhat to do with that problem, then it can't be a dead short. Because then the 8 Chips wouldn't have gotten warm, which means that they were actually powered up.

Also a easy method check, if some sort of high current action is going on around that screw hole, is to test the temperature with a finger. If there are significant amounts of current flowing trough a resistive fault (because the screw hole can't be a complete short because of the above mentioned reason) then it has to get warm. A easy calculation confirms that: Let's assume the circuit consumes 6A in normal mode, and also assume that all the chips are OK. So if there are 11A going in, then trough the fault, 5A have to flow. Even at a resistance of 0.15Ohms at 5A this equals 3.75W of power- dissipation. This screw hole has to get warm, even if there are big power Planes on that PCB.
 

Offline Pat Pending

  • Regular Contributor
  • *
  • Posts: 160
  • Country: us
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #46 on: December 17, 2012, 09:20:05 pm »
Anyone else with access to the same LeCroy and can measure the impedance of the 3v3 rail?

Just a suggestion but maybe some of that 11A is due to internal protection diodes or substrate diodes where the 3.3V and 5V systems interface to each other
within the custom chips, maybe 2.7V can be found on the 5V side?

On a side note, this particular LeCroy PSU OEM resembles the design used in the TDS420. Except in this case the LeCroy service manual
has the PSU schematic whereas the Tektronix SM does not.
« Last Edit: December 17, 2012, 09:22:42 pm by Pat Pending »
 

Offline Dread

  • Regular Contributor
  • *
  • Posts: 109
  • Country: jm
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #47 on: December 17, 2012, 10:07:50 pm »
Lots of good suggestions, I also loved the video, keep these coming  ;D

Mike beat me to it in that I knew you were wasting time with using that meter and trying to find milliohms of difference, that’s futile and compounded by the fact that that you say  the meter has a high resolution but as you stated it's accuracy is in question! So at those resolutions it begs the question what is it's repeatability of reading at the same point twice or more, it seemed to be really poor. Which goes to the point that a lack of accuracy does not mean it will always be off by plus X or minus X but can drift all over the place.

My second piece of advice is don't be afraid to use that exacto Knife and cut a surface trace or two.  I use to do that when the voltage ran all over the board.  I could then isolate which side had the problem.  A second or third cut can isolate it down to a very small exact area
.




The Optimist says the glass is half full, the Pessimist says its half empty, an engineer only see's a glass that’s twice as big as it needs to be!
 

Offline merser

  • Contributor
  • Posts: 11
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #48 on: December 17, 2012, 10:36:45 pm »
Quote
My second piece of advice is don't be afraid to use that exacto Knife and cut a surface trace or two.  I use to do that when the voltage ran all over the board.  I could then isolate which side had the problem.  A second or third cut can isolate it down to a very small exact area

That's ok, but I'd check there aren't jumpers to desolder first that serve this purpose of isolating sections.

I don't see how you can't follow this through and fix the damn thing in part 2 Dave. It is like I am about to climax and she says take it out.  I'm all in favour of happy endings.  ; - )
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29891
  • Country: au
    • EEVblog
Re: EEVblog #398 - Lecroy 9384C Oscilloscope Repair
« Reply #49 on: December 17, 2012, 10:37:58 pm »
Mike beat me to it in that I knew you were wasting time with using that meter and trying to find milliohms of difference, that’s futile

No, it's not. This is a very common and successful technique to quickly locate shorts. It is also safer than powering the board up, so is a common first option.
It just happened to have not worked in this instance.

Quote
and compounded by the fact that that you say  the meter has a high resolution but as you stated it's accuracy is in question! So at those resolutions it begs the question what is it's repeatability of reading at the same point twice or more, it seemed to be really poor.

Watch the video, the repeatability of the meter and probes is excellent.

Quote
Which goes to the point that a lack of accuracy does not mean it will always be off by plus X or minus X but can drift all over the place.

The accuracy of the meter has nothing to do with repeatability, that's not how it works.

Quote
My second piece of advice is don't be afraid to use that exacto Knife and cut a surface trace or two.  I use to do that when the voltage ran all over the board.  I could then isolate which side had the problem.  A second or third cut can isolate it down to a very small exact area

Once again, a last resort method. And something that is not possible with multi-layer boards.

Dave.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf