HP 6236B Repair/Restoration

[garrettm]

So I picked up an HP 6236B a couple of months ago for around 60 USD from an online auction. It looked pretty grimy, but I figured it would clean up okay.

Here are the original listing photos.









After half a magic eraser, some elbow grease and a final scrub with a microfiber cloth we have a respectable little power supply. Surprisingly, the writing in marker on the side came off without any fuss. I also couldn't bear to look at the thumb tac jammed into the top right knob, so I dug through my spare HP parts and thankfully found a replacement. (I hope who ever pulled off the cap and stuck that thumb tack in there slips on a banana peel...)





After powering it up I noticed that the displays didn't work right. When the selector switch was set to 6v it read back the voltage and current decently enough. But when set to either +20v or -20v things were sketchy at best. If you held the knob just right, the display might work for a second or two.

Turns out there was some significant oxidation on the switch contacts. Holy mackerel Batman, look at that patina!





Grudgingly, I pulled the finicky bastard apart and used some baking soda paste and a tooth brush to clean the contacts, rinsed with distilled water, then applied some Deoxit to the contacts to hopefully keep them clean for the foreseeable future.



I apparently forgot to photograph the whole cleaning process... So here's the finished result after I'd put everything back together. Of note is the shinier silver contacts.



Now that the displays were working as the good folk at HP intended, I noticed the +20v meter flutters rapidly when I adjusted to about 8v... Oh great, right after I just reassembled everything, I need to open it up again.

What you are seeing is my lazy attempt to observe the rapid changes in the output voltage while slowly turning the 20v adjustment pot. Holy moly Batman, that's an oscillation!



Here's what adjusting the tracking knob for the -20v rail looks like. Notice the lack of oscillations while turning at the same leisurely pace. Clearly, we have a sick pot. I'm blaming that damn Coronavirus.



So here we go again.



Wait a minute... Is that--yup. It's a RIFA time bomb. I apparently missed it the first time I opened up the unit to clean the switch contacts.



Surprisingly, it actually looks okay. But I replaced it anyways--some random polypropylene Chinese job. Can't possibly be worse than the time bomb it replaced.

Well its time to bust out the goot wick and Kester 186 and get those pots outa there.



But wait, it’s time to break some stuff first. The rubber stand-offs near the pots had turned rock hard. As they say in Texas, there will be blood.



Meanwhile, I wicked off the excess solder, bent the pins vertical and then did some gymnastics getting the pots out.





Here they are. Hulking Clarostat wire-wound, single-turn units.



This guy looks like someone had opened it up already...



Well, let's see what the DMM shows. Yup, bad pot. And this is the one that had been tampered with. I don't know if it was worse than this and someone tried their hand at "fixing" it or what. All I know is that this little guy is going into the garbage can.



And here's what the other one looks like. Silky smooth. I'll put it in my parts bin in case I ever find a use for it.



A major gripe I have with the 6236B is the use of single turn pots. You get terrible resolution when adjusting with only a single turn, and its finicky too. So I decided to drop some 10 turn pots in. Its nice to gracefully dial in the correct voltage without overshooting by half a volt from twichy fingers. Saves delicate ICs that don't like to be overvolted.

Anywhoo, I figure some Bourns 3590S will do the job nicely: 10-turns, 50ppm/C, IP65 rated and cheap too. It's kind of nice to see something that isn't made in China. Gracias me amigos from across the border.





With those dropped in, I needed to replace the rubber spacers I broke earlier in my rampage. Luckily I had two rubber feet that were the same height. So I slapped them on and put everything back together.





After turning the PSU on, I quickly noticed I hooked up the pots wrong. Turns out I wired them for counter clockwise adjustment... oops. I had originally connected them such that pins 3 and 2 were shorted, but HP wants pins 1 and 3 shorted. So I pulled the blasted thing apart and rewired the pots. I didn't think to take photos of the fix while still under the berserk status ailment.

Putting everything back together for the umpteenth time, I was rewarded with clockwise adjusting knobs. Finally, the meter scales move in the same direction as the pots! However, the adjustment range for the 6v rail is out of tolerance: full scale is 7.3v--that’s off scale Batman! The +-20v rails at least stayed within the scale markings, clocking in at 24v.

I should point out that this problem existed before I put the new pots in.

That's all for today folks.

Tune in next time for some digging through the schematic and swapping out resistors to fix the full-scale output. Should be an easy fix. But keep your fingers crossed.

[Zucca]

well done!

[WattsThat]

Nice work. Thank you for the baking soda/toothbrush idea 

Good call on replacing the Rifa. I bought a 6237B a few months ago. I repaired one of the plastic meter retainers, cleaned it up and put it on the bench. I put it under a 500mA load on the 18v supply and left it on for awhile. About an hour later, it went boom, well more of a pop and an extended hissing noise and given the amount of smoke that poured out of it, my thought was holy crap, the transformer blew? When I got in front of the supply to turn it off, the output was still at 500mA, 18 volts. What the heck?

Yup, it was the Rifa cap. It stunk up the lab for about a week. Sure takes a lot of work to get to it...

[garrettm]

well done!

Thanks Zucca! Despite my grumbling, I try to do good work.

Nice work. Thank you for the baking soda/toothbrush idea

It's very effective, but also messy. I'd suggest covering everything that’s not being worked on in that stretchy cling wrap used for packaging or the thinner stuff for food. Also you'd want a syringe and two cups nearby, one with distilled water and an empty waste cup. Then gently irrigate the contaminated areas with warm, or better yet hot, distilled water into the waste cup. The hotter the water, the easier it is to dissolve the dried and remaining clumpy baking soda. I threw my cup in the microwave for a minute or two before irrigating and that seemed to work great. Also, be very careful with bending the wires, the wire at the solder joints is prone to mechanical fatigue and can snap off or have a few strands break. I had a wire on the board that I had to cut and resolder as it had a couple of broken wire strands. Easy fix, but annoying all the same. 

Yup, it was the Rifa cap. It stunk up the lab for about a week. Sure takes a lot of work to get to it...

Dear lord! That's got to be frustrating. Thankfully the transformer blocked the aluminum foil from raining down and shorting out the exposed traces and components. Though I bet cleanup was still a mess. People say these RIFAs have good specs, but if they spontaneously explode, why ever put them in a product when there is a comparable part? I mean, I get that it takes 30 years for them to crack enough that moisture can ingress into them for failure to occur. But it still seems dumb when a new product uses them.

I have another repair project on the bench (Racal Dana 5002 wideband level meter) where I'm replacing two cracked and fully exposed RIFAs that were used for mains switch debouncing. Thankfully I live in a desert where ambient humidity is essentially nonexistent. Otherwise these little suicide bombers would have blown up already.

[bitseeker]

Hi garrettm! Nice restoration job. Regarding the "cheap" Bourns pots, where did you get them? There are counterfeits that are inexpensive and work, but feel quite different (and probably won't be as durable as a real one). Depending on your use, it may not be a concern, but it's good info for those who may be fixing a similar supply and not realize that there are genuine and look alike versions.

[garrettm]

Well... I did it again! Caught another HP 6236B while fishing on eBay, this time for 55 USD. This little fella looks to be in better shape than the last one. Apparently someone felt like cutting off the power cord... While dumb, at least this madman didn't throw the business end in the trash after liberating the power cord. It's small miracles like these that restore my faith in humanity.



Meanwhile, in my secret underground laboratory, I was going full CSI Miami.

While pulling off the top cover I noticed a faint white residue hiding on the underside nearest the radial capacitors. If that's cocaine, someone must have had one hell of a party. Pulling the fuse showed no obvious signs of electrical waterboarding. But maybe it's a replacement put in after the PSU was mercilessly tortured? A small chip in the glass looks to be an innocuous manufacturing defect. The markings show LF 313 2A: Little Fuse, model 313 (slow blow) 2A rating. Seems sensible. However, HP says to use part number 2110-0002, which is actually a Little Fuse 312 (fast blow). Checking my other 6236B shows a 312 fuse. Interesting. Looks like it might have been replaced by the previous owner. I'll swap it out later, but it's good enough for the moment.

Aside from a slow blow being used, no clear signs of foul play were noticed so far.



Glancing at the components on the PCB, I couldn't see anything that exploded and let out the magic electron juice. However, the dust on the PCB was pretty unusual: it glistened as if wet in certain areas. Taking a finger swab, I could detect a slight oil like consistency. Not good. But maybe not bad either. I'll figure this little mystery out after I replace the power cord.



But before all that, here's some more circuit porn for the eyes. HP really had some of the best looking PCBs of the late 80s and early 90s. Apparently HP stopped using those hermetically sealed metal can solid tantalums in favor of these less costly plastic encapsulated types in yellowish orange. Can't say I blame them. Those caps were/are incredibly expensive.



Looking at the date codes, it would appear my newly adopted baby is 30 years old. He's all grown up now.



And now the fixing begins! After looking through all my spare power cords, I found what was apparently either an original HP cord or a cord made by the same manufacturer that HP purchased their cords from! I'm big on making repairs look stock whenever possible, so this was a big win for me. The exception to that rule is if I can improve performance while tactfully keeping things looking professional or at least "pretty looking".



Here's a side-by-side comparison of the proposed replacement with the cord attached to my other 6236B. Look like long lost twins finally being reunited at age 30. A real Oprah Winfrey moment in the field of test equipment.



Anyhoo, let's get that old cable out of there! The trick to getting these things off without destroying everything is to pull out all the wires from the jacket, one-by-one of course, and then grab the empty jacket with a pair of needle nose pliers and gently tug till it pops out. You can also compress the bushing if you need to, but a little tug-o-war worked for me.



Markings on the original jacket match the new one, nice!



Now we just need to remove those pesky old wires, cut the jacket on the new cable, strip the ends... And done.



And now the hard part: putting that strain relief / bushing on the new cord and getting it to fit back in the hole. In the immortal words of Happy Gilmore: "That's your home! Are you too good for your home? Answer me!" But in all seriousness, you will want to use a vice to compress the jacket. There is no way you will do this with just pliers without breaking and/or marring something. The tolerances are too tight with the way the heatsink fins surround it. Even after compressing with the clamp, the elasticity of the cable caused it to re-expand and become slightly too large while trying to insert the cord back into the hole. After the third time in the vice I managed to get the bushing installed but then needed to use some pliers to orient the cable such that the flat part of the hole lined up with the compressible part of the bushing. I wrapped the pliers with masking tape to avoid scuffing the bushing and heatsink. Probably wouldn't see the scuffs anyway, but that's how I roll: Like a pro. Yo. Alright, alright I'll stop.



Now it's time to bust out the soldering iron and some good quality flux and solder. First you will want to pull the hot and neutral wires through the clear plastic tubes, then wrap the exposed copper around the solder lugs. And finally give 'em some heat! Seriously though, you will want a high wattage soldering iron when doing the ground lead, it's heatsinked pretty good. Probably could have unscrewed it to avoid that problem, but I didn't. I live life like I drive my car: Dangerously.





And here's the final result. Looking factory fresh, if I may say.



The time of truth was now upon me: I donned my protective eyewear, put the top cover back on, plugged it in, flipped the power switch...and heard a light hum from the transformer. No exploding caps or blown fuses today it seems. To be honest, I wasn't entirely sure it would turn on. The white residue on the top cover, light oil on parts of the PCB and slow blow fuse alluded to a possible problem. Connecting the PSU up to my DMM showed it to be A-okay for the most part. Like my other 6236B, it goes slightly off-scale when turning the adjustment pots to full-scale. So they both need to be recalibrated.

But before calibrating the two power supplies I need to do some more work on this guy: replace the RIFA mains cap, clean off the oil on the PCB as well as test the electrolytic capacitors. I hope the oil isn't actually leaked electrolyte and somehow made its way into the unit from the vent holes on the top cover. I also probably need to clean the switch contacts and put some Deoxit on them for good measure.

So much to do, and so little time to do it.

[tkamiya]

Wow...  excellent job!

That's just a result of cleaning?  It looks like front was repainted and refinished!

[garrettm]

Thanks tkamiya!

A little cleaning can go a long way. I generally prefer to use citrus based cleaners (De-Solv-it, GooGone, etc.) and soapy water as opposed to IPA. But sometimes IPA is the only thing that can remove certain types of sticker residue, tar and other gunk. Though it can be pretty harsh on paint. Thankfully, a tiny amount of silicone grease (sometimes called plumbers grease or dielectric grease) mixed with some citrus based cleaning oil can restore the original sheen of the paint if dull. I let that sit for 24 hours and then remove any excess that wasn't absorbed by the paint with a microfiber cloth. The result is pretty impressive.

Silicone grease also works wonders at lubricating old plastic gears and switches. I recently restored my HP 6110A 3000V power supply and used liberal amounts of this stuff on all the mechanical parts. The voltage dial moves so smoothly now I can't believe it's the same power supply. The thing was a squeaking mess before the restoration.

[mag_therm]

That is a good job.

I have two of the HP6253A (dual 24V 3.5 Amp) rack mount units here.
The instruction manual revision is 1967.
By your photos, your unit appears to have metal can op-amps.

The error amps in the 6253A  are un-named in the parts list,  long tail pair dual Si npn.
The main rectifier diodes are MR1032b,  but I could not find a datasheet.

On purchase about 10 years ago, I replaced some of the Sprague electro caps with Nichicons  and they have worked ok since.

I use one over the lathe to power a toolpost spindle with DC motor field and armature, and for battery charging.
The drift of voltage is only  a few millivolt over a day charging a battery.