Keitley 238 restoration

[Ice-Tea]

I acquired one of these (amongst a lot of other stuff), it was labelled as dead.

First attemp: well, yeah. It's dead. So what would be the second step? Yep. Check the fuse. Can't be that easy, right? But the fuse is broken. So, toss in a new one. It boots! Huzah! Display lights up, everything normal. Well, maybe not everything. The fan is making a ruckus.

Everything? What's that smell? Burning electronics, that's what it is! So, turning it off, popping it open. Couldn't see it right away, so I turned it on again for a fraction. A small string of smoke escaping from a large resistor told me all I needed to know. That's one large resistor that puked its internals out. Checked the service guide, 15K resistor in the -15V supply rail. Farnell to the rescue, FFWD a few days, popped in the new resistor.

Still boots (yay!) No longer pukes anything out (yay!) So now I can check the output. Something comes out of it (yay!). It's off by a LOT.

Oh noes. What now? Well, what I should have done from the start. Check that zener. Which acts like a 3-ohm resistor. That might be in its job description somewhere, but it seems to miss a few other points. Farnell again!

And voila. Functional now. Checks out against my Keysight 1252B. Haven't checked the entire range quite yet (need to install a new fan first) but there it is: restored Keithley 238. Ready to serve once more, probably in an eBay outlet near you shortly!

I'll post a few picks over the entire range later.

[cncjerry]

I've wanted one of these for a while.  I have a 4 quadrant sink/source that's good up to 100w.  It's great for cycling batteries as it can be a load and then programmable charger, within limits.  It's a keithley 228.

You got lucky with yours I think.  Mine has had a few problems since I got it but it is working great now for a few years.

Jerry

[alm]

I have seen the same issue on a 237: shorted zener and dead 15k power resistor. Haven't looked into why. The resistor looked like it got hot and damaged the insulation of the nearby electrolytic.

[Ice-Tea]

No damaged caps here.

My guess is that the fan gave way first, leading to the overheating -15V stage.

[alm]

I recall the resistors in the 237 having about 180 V across them. 150 V across 15k is about 2 W. They looked like more than 2 W resistors. So I would not expect the resistors to overheat without a fan. Not sure about the 238.

[Ice-Tea]

Eh, sch2  wat supposed to be the BOM, is listed as 2W. But you're right. Gonna look forward something beefier.

[Ice-Tea]

It's actually a 150V rail (gonna doublecheck, but that's what's on the schematics).

So 150-15 = 135/15k=9mA for 1,215W. The resistor I put in was rated for 3W (the original BOM has a 2W part even though it looks bigger than the 3W part).

The "new" resistor runs at 170°C. The old at 120°C. The new one is rated up to 235°C, so that actually makes sense but it's still way too high for my taste. Within operating limits or not, I won't have it. Ordered two of these:

http://be.farnell.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=10001&langId=32&urlRequestType=Base&partNumber=1174440&storeId=10154

I'll be replacing the non-broken original resistor for the +15V rail as well.

[Ice-Tea]

From left to right: the old one, the rejected replacement and Mr. Right (yes, size does matter).



Old temperature:



New (both resistors in view):



Accuracy is pretty good...




... but as can be seen in the picture, open loop current is off by about 10uA. Which seems like a lot. As calibration is supposed to be over the 488 interface and I have no gear for it, I'm going to close her up "as is". If anyone is interested in her, let me know!

[alm]

I agree that a larger resistor that gets less hot is better. 10 uA on the 100mA range is within the +/- 20 uA sourcing and +/- 6 uA measuring spec if my quick glance at the data sheet was correct.

[Ice-Tea]

You are probably right, but it doesn't feel great if it measures 10uA (which is two digits) when there's nothing going on. Then again, I'm trying to sell this thing so: yeah, totally normal, nothing to see here folks! 

[HalFET]

So I actually bought this one from Ice-Tea. Luckily I do have access to the equipment necessary to calibrate it.

Current game plan is quite simple:

• Replace the wirewound resistors with a cermet with characteristics similar to the replaced carbon composite resistors. (Suggestion from a friendly Keithley/Tek service engineer)
• Clean up a few dirty spots on the board with a good IPA rinse. (Think there are some leakage currents that could use IPA + blast of nitrogen.)
• Calibrate

I'll also try to post the calibration part, since I figure that might actually interest people slightly more. 

[Ice-Tea]

 

So, you'll replace the replacement? Why? Just a series resistor in a library PSU?

Anyway, good to know Shell be Well taken care off!

[HalFET]

So, you'll replace the replacement? Why? Just a series resistor in a library PSU?

Anyway, good to know Shell be Well taken care off!

Two reasons actually, carbon comp resistors have very low self inductance, given that similar wirewounds were already available at that point I can't help but figure keithley had a reason to go for this type. Especially if you consider how noisy they are, and Keithley's complete disregard for the concept of compromise.The second reason is safety. The 2W resistors blow because something is drawing very high currents. If I'd leave the 10W in there it'd be something down the line that  self-widlarizes instead next time. So there is method to this madness! (And a third reason is that I'm worried they're flapping around a bit, so I'd have to silicone them down a bit.)

But still got to inspect the board completely, I think I saw signs of a previous repair attempt in the measurement section, but didn't yet have time to lift the cover off that one.

[HalFET]

So, you'll replace the replacement? Why? Just a series resistor in a library PSU?

Anyway, good to know Shell be Well taken care off!

It's not going to be as quick of a fix as I had hoped, it's actually a small miracle that it worked after having inspected the board thoroughly. Found a few bodges and unclean repairs, took a gentle IPA scrub and touches of the iron left and right. And apparently the 15V zener there was simply pressed against the pads, the solder cracked and connection was intermittent. Someone seems to have snipped it off with side cutters and then tried to place it back on the board in a bodge fashion. I repaired the bodge as good as possible with some desoldering braid for now, going to order new zener diodes on Monday.

And just out of curiosity, where did you get the schematic if I may ask?

[Ice-Tea]

So, you'll replace the replacement? Why? Just a series resistor in a library PSU?

Anyway, good to know Shell be Well taken care off!

Two reasons actually, carbon comp resistors have very low self inductance, given that similar wirewounds were already available at that point I can't help but figure keithley had a reason to go for this type. Especially if you consider how noisy they are, and Keithley's complete disregard for the concept of compromise.The second reason is safety. The 2W resistors blow because something is drawing very high currents. If I'd leave the 10W in there it'd be something down the line that  self-widlarizes instead next time. So there is method to this madness! (And a third reason is that I'm worried they're flapping around a bit, so I'd have to silicone them down a bit.)

But still got to inspect the board completely, I think I saw signs of a previous repair attempt in the measurement section, but didn't yet have time to lift the cover off that one.

Well, not sure if you'd have to worry too much about noise on a carbon vs wirewound resistor in a linear zener PSU but then again: I don't get to design nA PSUs, so there...

As for the reason why it blew: my guess is these run hot anyways and rely on airflow to stay alive. When the fan gave way, so did the resistor. That, or the zener blew first, essentially overloading the resistor further.

As for the bodge on the zener: that may have been me. I may have done a quick repair to see if it works at all and forgot to clean it up after. My appologies in that case. Honnestly can't remember, it was quite some time ago (and late too, at the time).

As for the flapping: you have a point. At some point, I figured I'd strap it to the elco next door but I don't think it would have liked such a hot neighbor 

As for the schematics: a few posts down this thread: https://www.eevblog.com/forum/testgear/keitley-236-teardown-and-review/

Also: I posted from my phone before. Library PSU should be linear. Obviously.

[HalFET]

Well, not sure if you'd have to worry too much about noise on a carbon vs wirewound resistor in a linear zener PSU but then again: I don't get to design nA PSUs, so there...

As for the reason why it blew: my guess is these run hot anyways and rely on airflow to stay alive. When the fan gave way, so did the resistor. That, or the zener blew first, essentially overloading the resistor further.

As for the bodge on the zener: that may have been me. I may have done a quick repair to see if it works at all and forgot to clean it up after. My appologies in that case. Honnestly can't remember, it was quite some time ago (and late too, at the time).

As for the flapping: you have a point. At some point, I figured I'd strap it to the elco next door but I don't think it would have liked such a hot neighbor 

As for the schematics: a few posts down this thread: https://www.eevblog.com/forum/testgear/keitley-236-teardown-and-review/

Also: I posted from my phone before. Library PSU should be linear. Obviously.

Well, depending on where it's located it could influence it, I'm more worried about the power rating though

Nah, I don't think you did the zener, the lead that was snipped was actually the one at the side away from the resistor

Thanks!

[alm]

Two reasons actually, carbon comp resistors have very low self inductance, given that similar wirewounds were already available at that point I can't help but figure keithley had a reason to go for this type. Especially if you consider how noisy they are, and Keithley's complete disregard for the concept of compromise.

Not sure how a little inductance would affect this circuit since this circuit is pretty much all DC. There is a component in the power supply that is often used as noise generator, and it is not the carbon comp resistor . So I can't imagine noise being much of a consideration either way.

The second reason is safety. The 2W resistors blow because something is drawing very high currents. If I'd leave the 10W in there it'd be something down the line that  self-widlarizes instead next time.

Make sure to get a fusible resistor, which is guaranteed not to catch fire when overloaded. The one thing I'd be worried about is cooking the nearby electrolytics. Which is where a lower thermal resistance (larger physical size) might help.

[HalFET]

Haven't seen the complete schematic, so no idea! But these should fail open, hence why I replaced it with these. Might add some kapton foil in between the resistors and nearby circuitry as an extra fail safe now that I think about it.

But overall there were a few bodges so I'll just fix up the board completely.

[HalFET]

Current status: Fixed the solder bodges, cleaned the digital board with IPA, gave the analog boards a blast of nitrogen, installed the new zener and resistors.

The leakage current is up to 400 µA now, but I'll calibrate it later. The fan Ice-Tea installed is only rated 115V, and while these Sunon fans in theory can run of 230V I'm not quite willing to trust it. (This one will be running unattended for prolonged periods at home.) Sadly you have to take almost the entire thing apart to get to the fan. Lets just say I'm not a fan of this unit's construction. Additionally I'd like a fan that pushes a lot more air through there considering how hot this thing is running, compared to the noise of the R&S CMU200 and SMH it won't even be audible    Also have to laser cut some polyimide foil to slip underneath the resistors to shield the components and the board from them in case they do decide to go boom. So much work, so little time to do it all.

And the new resistors are running rather hot (130°C), as expected, but they're rated up to 220°C so still got 90°C margin there:


More worried about what it'll do to the PCB underneath since I did mount them down on the board, might go in again and lift them slightly off the board if I do open it up to install another fan. I'll also give the triax connectors a rubbing with methanol then.

[Ice-Tea]

Sure about the fan? My guess is you'll measure 120v on the leads...

[HalFET]

Sorry, my bad, read the schematic incorrectly.

Still think I'll have to go for something with slightly higher air flow though. I'll actually run a test with a few thermocouples and a polycarbonate sheet on top to see how hot it runs.


Note to self: don't attempt to read schematics with transformers on 5 hours of sleep.

[tagchen]

Hi,
i have three defect 238. 1. transformer going hot and smokes 2. no action 3. start, but not check.
new transformer cost 390euro  , so i will repair the other first.
in 1. the resistors between transformer was very hot, can by other defect more.

[Ice-Tea]

1) If the trafo goes hot, chances are something is drawing a *lot* of power somewhere? If you can get hold of thermal camera, see if anything else heats up real fast..
2) Stating the obvious: check the fuse

[tagchen]

1) If the trafo goes hot, chances are something is drawing a *lot* of power somewhere? If you can get hold of thermal camera, see if anything else heats up real fast..

thanks,

i think its not a problem from drawing power, more the transformer has eh short circuit. i will little open it and  with acrylic lacquer and vacuum sealer press better inside. The 2. think can by the same problem.

[MiDi]

1) If the trafo goes hot, chances are something is drawing a *lot* of power somewhere? If you can get hold of thermal camera, see if anything else heats up real fast..

thanks,

i think its not a problem from drawing power, more the transformer has eh short circuit. i will little open it and  with acrylic lacquer and vacuum sealer press better inside. The 2. think can by the same problem.

I had hot transformer in my 237, as it turned out it was the Japan version (different transformer marking).

Seems there are no hi-res pictures of 238, attached a collection of my 90s 238.
In idle it takes ~40W.

There is Firmware rev A06 installed, is this the latest release?
Would be nice if you could tell what revs there are (it tells fw rev on power up).