Author Topic: Deep Fried LM393 - Keithley 2110  (Read 348 times)

0 Members and 1 Guest are viewing this topic.

Offline beenai2018Topic starter

  • Contributor
  • Posts: 38
  • Country: us
Deep Fried LM393 - Keithley 2110
« on: October 15, 2024, 05:33:52 pm »
Twas supposed to be a simple day  8)  . Check and  Calibrate a BK 1602 Valve power supply (with the keithley 2110 )  to sell.. Got the schematics and the procedure (just twiddling some pots) .   :-DD

I got a whiff off something heating up in the power supply  when setting a meter cal ,and I thought "Hey ,The 2110 has a temperature second display while measuring volts.. So lets turn that on ,and touch this transistor ... " and Zap!!!  :wtf: ?? The temp input has opto isolation and should  be isolated from the Main measurement side ..

Now, The keithley 2110 is going nuts on the display with no input.. and lets look at the BK 1602 .. won't reach full 400V without the current  limit tripping ..(with no load!!) .. Kill two  units with one thermocouple..   |O |O |O


Im wary of  ground short circuits via probes after I watched Dave's "Dont blow the ass out of your scope ...video"  way back ,  and In this case my beliefs were that the thermo inputs were isolated ..


The aftermath  .. I didn't realize this at first but there was probably a mini fire in the scope when it was letting its magic smoke out .... and the smell ..  OMG ,  phenol ,phenols all the way ..

And then I opened the unit (smell to infinity ) ... and prominently left  the space..  Two days of airing out the room and running a mini ozoniser later .. lol


So , it looks like  an LM393 is visibly dead and maybe others bits around it .  Looking at forums it looks like another member (with a 2700)  had one go out as well .



Questions

The pads have delaminated and I have probed the circuit for connections and Annotated them.


I have been slowly scraping away the blackened section  ,but it looks like I may have to clear out the traces and leave the area unpopulated.


I cant cut a hole out in the pcb because there are traces on the bottom . There may be hidden traces as well .


Since the pcb is charred and shows a 6 M resistance on the fiberglass  is it worth reusing  the area  for new parts ?  ,or do a riser / daughterboard away from the burn spot ?

If  I can relocate the circuit ,should I dremel out as much as possible to prevent leakages etc?The top phenolic layer is gone and Im seeing fiberglass mat .

Im thinking this is the start  of problems and more will reveal themselves (whack a mole ,lol)

















 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 15149
  • Country: de
Re: Deep Fried LM393 - Keithley 2110
« Reply #1 on: October 15, 2024, 06:04:57 pm »
The circuit directly at the LM393 is not that sensitive to leakage. These comparators are usually used to control JFET switches, by pulling the gate low to switch them off. The input signals are logic signals and thus not that sensitive.

Chances are that the 2 JFETs controlled from the chip can also be damaged. This are likely some mmbf4392/3.

The signals below the chip look like the supply and digital signals towards the LM393 - so non critical signals when it comes to leakage.
The aera around Q208 may be the most critical part with leakage in the charred area. This part may want a good clean and maybe cuts under the transistor for extra isolation or maybe more this part to a small riser.


It looks like the thermocouple part does not have that much protection and is using the same ground / supply. It would be nice to have that part fully isolated, but is obviously is not. There could be an issue in the TC part of the circuit too.
 
The following users thanked this post: beenai2018

Offline beenai2018Topic starter

  • Contributor
  • Posts: 38
  • Country: us
Re: Deep Fried LM393 - Keithley 2110
« Reply #2 on: October 15, 2024, 07:17:41 pm »
The circuit directly at the LM393 is not that sensitive to leakage. These comparators are usually used to control JFET switches, by pulling the gate low to switch them off. The input signals are logic signals and thus not that sensitive.

Chances are that the 2 JFETs controlled from the chip can also be damaged. This are likely some mmbf4392/3. 

The signals below the chip look like the supply and digital signals towards the LM393 - so non critical signals when it comes to leakage.

The aera around Q208 may be the most critical part with leakage in the charred area. This part may want a good clean and maybe cuts under the transistor for extra isolation or maybe more this part to a small riser.


It looks like the thermocouple part does not have that much protection and is using the same ground / supply. It would be nice to have that part fully isolated, but is obviously is not. There could be an issue in the TC part of the circuit too.

Thanks !!  Gives me a bit more confidence  to tackle  this and trust its  future measurements.

Q208 was replaced previously as  causing an offset on all readings with a 2N7000 .I will look at  further moving it away


The 2 JFET are "M6G" - 4393 . I have discretes to bodge in if needed .


Regarding the thermocouple input .. Whilst the meter allows thermocouple as a second display to voltage ... the  manual warns of dire consequences .. or maybe I'm misreading it.





 

Offline Harry_22

  • Frequent Contributor
  • **
  • Posts: 603
  • Country: 00
Re: Deep Fried LM393 - Keithley 2110
« Reply #3 on: October 15, 2024, 08:15:06 pm »
Hi!

There are no hidden traces visible. Almost the entire topology of this place is made through vias.
You probably need to partially remove/bend the top traces. Remove the coal almost to the bottom layer. Restore thickness with epoxy. Drill vias. And carefully restore the top layer with reference to vias.

Good luck!

PS
Please see in attachment animated Gif.
https://drive.google.com/file/d/1BoWv6qVmAhurE9w8fAfhqcgL8M3zqWl9/view?usp=sharing
« Last Edit: October 15, 2024, 08:35:42 pm by Harry_22 »
 
The following users thanked this post: beenai2018


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf