Help with Keithley 224 Repair (Mysterious 1.5uA Leakage)

[garrettm]

I picked up a Keithley 224 current source a while back and finally got around to testing it. The 224 is the 220's little brother, with only 100mA to 10uA ranges installed on the analog board. The two are otherwise identical aside from the change in lower leakage protection diode in the 220 and the firmware.

First off, the unit turns on and responds to the front panel controls. The voltage compliance, while not terribly accurate, seems to work fine. It will also output current on each of the 5 current ranges. However, there is some significant leakage on the mA and uA ranges of about 12-15uA. This cannot be removed by adjusting the Zero Span trimmer (R392) or through a complete calibration.



1) Leaky Triax / Loose Bulkhead Assembly -   Not Fixed Yet

After some poking around, I noticed the triaxial cable is very "loose" near the bulkhead connector and most importantly that the leakage will drop to around 3uA if held just right. Some nickel 2+ corrosion (green film) seems to be present on the connector as well. So some work needs to be done here.

A newer zip tie holding the output cable to the fan and some heat shrink joining the thin fan power wires to some thicker wires at the base of the fan connector suggest the cooling fan had been replaced. And seeing how quiet the fan is for a unit made in 1985, I'd say that was done fairly recently.

My guess is that during the fan installation the bulkhead assembly was damaged.



2) Mysterious uA Leakage - Unsolved

After removing the shield and connecting some good triax up to the output with guard connected to the inner shield the leakage dropped to about 1.5uA. Success! But still not good enough, that leakage basically kills the 10uA range calibration and programming accuracy.

Upon further poking around, I noticed some more problems:



3) No Negative Output Current - Solved

Setting the output negative does not work correctly: output drops back down to leakage levels.

Probing about I found that the unity gain buffer U312 (Intersil ITS6435, likely a better spec 741) was not being set to inverting configuration since U306 (a quad SPST CMOS switch) had failed open on switch D which is used to invert the logic to the input of switch C that sets U312 as an inverting buffer. Thankfully switch B is left disconnected. Paralleling it with switch D corrects the issue and negative output is now working.





U306 is an Intersil IH5052 and is apparently pin-for-pin compatible with the DG211, which is still in production from a number of manufacturers.



4) Output/Guard Relays Always On - Unsolved

Another issue is that the reed relays K306 (output high) and K307 (guard) do not turn off when the "operate" button is set to off: They are always connected when the unit is powered up and disconnected when the unit is powered off.



Looking at U320 (TI ULN2003AN, 7ch darlington array) shows that pin 7 (base of ch7 NPN) never goes low. Going through the trouble shooting tests listed in the manual show that U320 works fine.



It also shows that the manual has some pretty big typos: The 224 does not have a 1uA range. So pin 15 will never go to +5V (ch2 NPN is off) as it would in the 220. It is on (+0.7V) for all ranges (10mA - 10uA) except 100mA (in which it is +5V and pin 11 is +0.7V). Also, the schematic shows nothing connected to pin 15, but it is in fact connected to something. What that something is I don't know.

The base of ch7 is connected to pin 4 of U305B which is also used as an Enable line to the shift registers U301-303 driving the DACs. I'm not sure if this should always be HIGH or if it is supposed to be LOW when the output is off. The manual is pretty terrible in regards to this.

[garrettm]

I should point out that all of the power supply checks and analog board checks were okay (outside of the typos in the manual as pointed out earlier).

Reference: Output Common.
Q319 Collector: +143.8V, good
Q318 Collector: -144.3V, good

Reference: Analog Common (Guard).
+15V: +15.35V, good
-15V: -14.68V, good
+5V: +5.046V, good

Configuration: 100V compliance, 100ohm load, output on.

Range: 100mA.
U320 P11: +0.68V, good
U315 P2: +9.80V, good

Range: 10mA - 10uA.
U320 P15: +0.68V, good

Range: 10uA (manual lists 1uA, but this isn't a selectable range on the 224).
U320 P11-15: P12-14,16 0V (floating as in schematic); P11 5V (as expected); P15 +0.68V (as expected) **
U309 P1: -4.99V, good
U311 P16,24: +6.3V, good
U310 P6: -6.31V, good

Those checks don't really help with K306 and K307 always being on or finding the source of the uA leakage.

** U320 ch1 is not shown in the schematic or talked about in the 224 manual. Ch1 collector (P16) is left floating as is the case with channels 3-5. In the 220 channels 3-5 drive relays not present in the 224. Again, ch1 is not shown in the 220 schematic.

[garrettm]

A quick update on the repair.

Playing around with the Guard Zero trimmer (R343) I can get the leakage to mostly go away. However the 10uA range drifts +-300nA with some 10-20nA noise on top.

So clearly there is a leakage path somewhere between guard and output high. The question is where.

I'm curious if anyone has suggestions or any tests that might help find it. Is this a leaky semiconductor junction, board contamination or something else?

For reference, the above test is without the faulty triaxial cable connected. I've used a new Keithley 3-lug to alligator cable backwards to connect my DMM to the output, fully guarded. I can bust out the Keithley 617 if anyone thinks that might help. Just need to find my 2-lug alligator cable...

[MathWizard]

I am also trying to track leakage currents on an old Keithley DMM. I can see myself buying broken equipment for fun, but I better tell my wallet and clear a space for boxes of stuff.

[garrettm]

Finally found my 2-lug triax to alligator cable. After hooking up the Keithley 617 electrometer I am starting to get some different numbers than with the Tektronix DMM4050.

With the output off I am measuring about -16uA leakage on the output of the 224. This leakage is persistent when the output is on and set to 00.000uA or 000.00uA. When the current source is powered off, the electrometer jumps down to sub picoamps. So the test setup seems fine.

Interestingly, when changing ranges from 1mA to 100uA on the electrometer with the 224 set to 0.0000mA I get different measurements, this was also seen on the Tektronix DMM and made me question if the DMM needed to be sent in for cal. I also observed this same anomaly with my Leeds Northrup 4385 current shunt. I was measuring around -550uA with the 617 set to the 1mA range and about -30uA when set to the 100uA range. My guess is that small sense resistors (<=100ohms) cause the 224 to act differently than when using larger values (>=1kohm) on the mA ranges. I don't think that this is from the output hitting the compliance limit as I usually set the limit to 10-30V while testing.

After some fiddling with the Span Zero and +/- Span Zero adjustments I got the electrometer to read 0.0000mA and -22.00uA when jumping between ranges. Setting the output to off always returns to -16uA regardless of the value of any of the calibration potentiometers. Which is strange since previously the Guard Zero trimmer had affected the measured leakage on my DMM4050. So that must have been an artifact of my test setup at the time.

The most confusing thing is that the measured zero offset on the 1mA range will change slightly when going to the uA range and back. Something like -40 to -60uA instead of -22uA. Setting the output to +1mA and then back to 0mA returns the offset back to its initial measured value (-22uA). So this might have something to do with the j-fets used for range switching.

But before I open that can of worms I'm going to properly replace U306 and U312 as a pair since whatever damaged U306 may have affected the op amp (as the switch configures the op amp for inverting or noninverting unity gain). So it will probably be a few days before I can do some more rigorous testing.

[garrettm]

Turns out I had a DG211 in my junk box, so I figured I'd swap it out today.



While pulling the rear shield off the analog board I noticed one of the rubber spacers had liquified and contaminated the board near the 2mA Positive Span and  +/- Span Zero trimmers.

 

It was pretty easy to clean off with a cotton swab and some isopropanol. I ended up swapping out both spacers incase the other also decides to turn into goo. Unfortunately this fix had zero affect on the leakage issues.

After swapping out U306 I put the shield back on and ran my 2-lug to alligator triax cable inside for testing.

Good news is that the positive and negative current output is working again. Finally!

Bad news is that the leakage has not gone away--if anything it has magically increased to -25uA after the unit has warmed up. I wore gloves while working and only touched the edges of the board. I also cleaned off all residual flux residue from the repair with 99% IPA.  Maybe I'll order a can of compressed IPA and a can of compressed air to give the whole board a thorough cleaning.

With the shield back on I decided to revisit my previous tests and see if I could reproduce the aforemetioned measurement anomalies.

Keeping the Keithley 617 on the 20-2mA ranges, the output of the current source seems to work fine. The test points were:

019.00mA (20mA range 617, 100mA range 224)
19.000mA (20mA range 617, 20mA range 224)
1.9000mA (2mA range 617, 2mA range 224)



The uA ranges also appear to work fine when the electrometer is on the mA range.

190.00uA (2mA range 617, 200uA range 224)
19.000uA (2mA range 617, 20uA range 224)



But! Things get weird when I drop down to the 200uA range on the Keithley 617.

The only thing I can think of is that the output voltage must rise to source the programmed current for the larger sense resistors. When there is a smaller sense resistor the output voltage is lower and doesn't cause the leakage to occur. It's possible the protection diodes are being biased on from the guard circuit not tracking correctly. I'll post another update after I investigate this further.

190.00uA (200uA range 617, 200uA range 224, )
19.000uA (200uA range 617, 20uA range 224)
00.000uA (200uA range 617, 20uA range 224)

[leighcorrigall]

Hi Garrett,

I have a Keithley 224 and just got around to calibrating it today. This current source is an entirely different beast than the Keithley 23X Series, despite being of similar vintage, and I have noticed a severely delayed settling time at the 100 mA range. When I calibrated it, I left it operating for 30 minutes, with the cover on, before adjusting the potentiometers housed under the shield. Despite the aforementioned drawback, I am impressed by the precision of the instrument. Check out the calibration procedure I setup for MS Excel  2021 01 11 - Calibration Procedure.xlsx (14.11 kB - downloaded 43 times.).

Have you noticed the delayed settling time with your unit?

12 minutes to reach a setpoint of + 100 mA

17 minutes to reach a setpoint of - 100 mA