EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: Leuams on February 20, 2016, 05:33:15 pm
-
I was digging through the electronics trash at work a few weeks back and found this beauty, a Keithley 2700 Multimeter/Data Acquisition System. Since the last piece of test gear I pulled from the trash bin only needed a new power switch I came away with a shit faced grin thinking I had scored an easy to fix new piece of test gear.
At home on the bench I opened it up and had a look around paying close attention the electrolytic caps or anything that looks like it had released some magic smoke. I did not find anything obvious so I moved on to applying power. The unit powered on and didn't throw any errors. Where it fails is in its main task of measuring. Applying a voltage gives readings significantly less the expected, in the mV range. Measuring resistance yields the same results with readings that are significantly lower than the expected resistance.
I removed the cover again with the intent of checking the voltage rails when I found that the magic smoke had been released from U209. A quick google search gave me the service and user's manual.
The service manual I found is here: https://www.element14.com/community/docs/DOC-52772/l/keithley-service-manual-for-model-2700-multimeter-data-acquisition-system (https://www.element14.com/community/docs/DOC-52772/l/keithley-service-manual-for-model-2700-multimeter-data-acquisition-system)
The service manual initially looked promising at helping me identify U209 but the replaceable parts list stops at U192.
Does anyone have a service manual that includes these parts? If it helps the main board says Keithley 2700-102H.
I have attached a couple more pictures of the board after removing U209. I also had a go at the footprint with my DMM with the intent of tracing out the connections to U209 (maybe I will upload a simple schematic if I complete this) and found that pads 5 and 6 are connected. I do not think they were originally connected. Looking at the via directly connected to pad 5 it appears to have burnt through the board (the photo showing the underside shows this a little better) maybe causing pads 5 and 6 to short. It also looks to me like the PCB is delaminated in that photo. This seems to be a 4 or more layer board and that via could have shorted to other unknown traces.
Is this multimeter even worth repairing if it is in this condition? I still have to find the original source of failure. Any advice or help would be appreciated, thanks.
-
Ouch. Can you post overall board photo? It does not seem like an opamp, perhaps we can guess based on references to other parts on the board.
-
Yep, that part is dead...
The board will likely be very crispy under it but it's certainly still worth repairing. Someone else here has got to have a 2700 to open up and grab a part number for you.
-
I am afraid my photography skills are lacking to be able to take clear close up photos.
I did find this forum post with a picture of the board but mine does not look like that: https://www.eevblog.com/forum/buysellwanted/fs-ebay-auction-keithley-2700-6-5-digit-dmm-and-data-acquisition-system-(usa)/msg839126/#msg839126 (https://www.eevblog.com/forum/buysellwanted/fs-ebay-auction-keithley-2700-6-5-digit-dmm-and-data-acquisition-system-(usa)/msg839126/#msg839126)
I ended up looking on ebay after my initial post and bought a working 2700 for a reasonable price. I can use that to help me fix this one or at the worst have a spare unit for parts. I am still trying to reverse engineer the connections and am drawing up a quick schematic in KiCAD. I have a suspicion about what the part is based on the power connections and surrounding ICs but without an understanding of how the circuit functions it may be more of a shot in the dark.
-
I have an 2700, but unfortunately my pictures of its guts are bad, and I'm unable to open it right now (wife and kids, you know).
It's definitly worth it to repair the meter. It's basicly a Keithley 2000 with support for 2 optional switching modules. Specs wise it doesn't differ much from the 2000 too. We have some running for almost 13 years 24/7! The VFD is something like 60euro's and quite easy to replace.
Next time I take it apart I will make some better photo's and save the FW for Tin's FTP.
-
U209 is an LM393 dual comparator. As another respondent has said the 2700 is a Keithley 2000 class DMM so is certainly worth getting fixed but it really comes into its own when you use the datalogging modules and the excellent Excelinx software for it. This enables you to set up and control a complex datalogging operation using Excel, no Labview or other esoteric programming needed. And it's free!
JoeD
-
Thanks JoeD!
Before reading your response I was trying to reverse engineer the connections, I had my suspicions it was a LM393 based on the IC appearing elsewhere on the board and the voltage supplies lining up. I posted the schematic I had come up with for U209 and a few surrounding connections. It was not a complete waste of time since I found out why pads 5 and 6 were shorted, Q203 seems to have failed. It is not in the parts list but luckily there are a handful SOT-23 packages with the exact same markings, 7E ARJ, that are in the parts list (Q127, Q131, Q132, Q133) that gave me the part number VN0605T N Channel MOSFET.
-
I have a 2700 with a 20ch card and love it. I would be willing to put some real effort into fixing it if I were in your shoes.
-
I didn't need much convincing to attempt to fix it anyways, haha.
I have ordered a few LM393 and replacement N MOSFETS. VN0605T is obsolete so I went with the suggested replacement on Vishay's website and ordered a handful of 2N7002K.
-
Thought I would update this post on my progress for anyone interested.
I replaced the LM393 and the nearby failed NMOSFET. My handy work is attached. :P The unit turned on with no magic smoke. I recently purchased the DMMcheck Plus so I was using that to determine if the meter was operating correctly, It was not. DC voltage measurements were off by about 2 orders of magnitude and resistance/ACV voltages were not responding. Surprisingly the DC current was measuring correctly. I got smart and looked up how the analog circuitry operates in the service manual, figure 4-3.
(https://dl.dropboxusercontent.com/u/24004644/KeithleyAnalog.png)
Following that diagram I could focus on the parts that deal with the voltage/current/resistance/frequency measurements. I also got even smarter and pulled out my Seek thermal camera and had a look around. I found a few more damaged components. Below is a list of additional components I have replaced, most of them were truly damaged.
- U105 and nearby U103: DG211DY CMOS analog switch
- U115: LM393 quad comparator
- U102 and U118: TLE2081CD J-FET Op-Amp
- Various SNJ132199 JFETS that are shown in figure 4-3
- I also replaced all the electrolytic caps after observing a few of them getting uncomfortably warm
To make a long story short the unit now is able to measure DC voltage correctly along with resistances below about 1Kohm. DC current works too.
I still have a few more repairs to try to fix AC voltage measurements and enable the full range of resistances. Looking at figure 4-3 I'm planning on replacing U123, Q124 and Q125 to repair the ohms current source. A few rough measurements using another multimeter of the resistance current suggested that this may be the issue.
For the AC voltage I have U112 on order but it was not in stock through Digikey so hopefully eBay will pull through and get me functioning products.
Guess this can be considered a partial repair for now. That shit faced grin has returned to my face after restoring the ability to measure DC voltage. ;D
-
Wouldn't the burnt area of PCB have issues with leakage?
-
I would scratch off as much black carbonized substrate as possible, carefully (there might be traces on inner layers!) before putting replacement chip back in.
It's not like this is very sensitive analog part or signals there, but better be safe than sorry..
-
I scraped as much off as possible but I was concerned with damaging the pads and traces (I lifted one of the pads and had to do a bodge job). I understand that this meter will never be 100% the same and may have calibration issues but for my purposes and the low low price of (almost) free it is better than anything else I have.
-
I'm also struggling with K2000, a question to replace JFET SNJ132199? with IF1320.
-
AFAIK the JFETs in the K2000 are used for switching. So the FETs need to have rather high voltage and low gate leakage. To switch a signal with +-15 V signal and some 5 V of so of gate drive, something like a 35 V rating should be minimum.
With only 20 V rating the SNJ132199 is no good.
The more likely candidate would be a MMBF4393.
The strange part number could be for selected ones for lower leakage.
-
AFAIK the JFETs in the K2000 are used for switching. So the FETs need to have rather high voltage and low gate leakage. To switch a signal with +-15 V signal and some 5 V of so of gate drive, something like a 35 V rating should be minimum.
With only 20 V rating the SNJ132199 is no good.
The more likely candidate would be a MMBF4393.
The strange part number could be for selected ones for lower leakage.
Thanks Kleinstein.
MMBF4393 for JFET input, or for all snj123199 JFET?
-
I scraped as much off as possible but I was concerned with damaging the pads and traces (I lifted one of the pads and had to do a bodge job). I understand that this meter will never be 100% the same and may have calibration issues but for my purposes and the low low price of (almost) free it is better than anything else I have.
A fiberglass brush can be great for removing this kind of burnt carbon mess. It will take off soldermask as well so leakage may even be improved.