Hewlett Packard by Salvador Dali (HP 3562A Dynamic Signal Analyzer)

[TerraHertz]

Today's parcel arrival - a HP 3562A Dynamic Signal Analyzer. Bought relatively cheaply because of a couple of dings, and 'no power up.'

Here's one of the diecast side frame pieces. 


Hmm... I'd been hoping all the mechanical damage was in the outside covers (that can be easily reshaped) but no...
It's been my experience that diecast metal can bend a little, ONCE. Trying to bend it back always results in fracture.
This piece is HP PN 5021-5838, but unfortunately ebay has nothing to offer.
It's 530mm long, if anyone has some HP instrument of the same size that they are scrapping.
If no replacement turns up, has anyone had any success with heating diecast to anneal it, or straightening it while not far off the melting temp?

The probable cause of 'no power up' was easy to spot. Can you see it?



So some of the wires to the mounting pins on this transformer had snapped as the pins pulled out of the plastic during the impact that dinged the frame. It's a bit of a design weakness I think, to have a quite heavy laminated iron transformer mounted only by a few pins press-fitted into the transformer plastic.

Anyway, fingers crossed it looks like I can fudge repair the transformer.

[TerraHertz]

Phooey. The transformer is fixed and soldered back in. I could try the machine out except... there is one tiny RIFA cap on the power supply board and yes it can potentially get mains voltage across it. (For once I actually have a service manual with power supply schematics.) So I pulled it out, it's only 2200pf, and... I don't have that value in stock.

I have some ceramic 2n2 1000V caps, but due to this cap being in a critical part of the mains switcher (primary current feedback loop), I don't quite feel confident to substitute type. And it's something of a pain to install/remove the PS board, so I don't want to try it with the old RIFA, then replace it later.

Not going to be able to power up the machine tonight.

About the transformer fix. Kind of microsurgery. Two of the fine wires had broken off flush with insulation structures. Used a scalpel to excavate just enough to expose a tiny bit of the wire. Used fine wire strands to extend those and the easier to access ones. Then secured those delicate fudges in place using UV-curing hard resin. It sets firmly in a few seconds, then is fully hard in about 20 seconds more.
Then replaced the pulled-out legs, soldered on the new wires, and finished with another layer of resin to secure the mounting pins fairly well. Multimeter tested and the windings are all there and sensible resistance ratios.

I'd have drilled a couple of holes in the PCB and used a big nylon wiretie to secure the transformer solidly, except it's a 3 or 4 layer PCB and there aren't any clear areas of PCB around the transformer that I could drill holes through. Well I suppose I'll just have to avoid dropping the machine onto concrete, as the previous owner appears to have done.

[Haenk]

Haha, that RIFA already has some nice visible cracking marks on it. 100% perfect decision to not power it up (who knows how long it has been sitting in storage and attracting humidity).

[tpowell1830]

If the metal from the bracket is iron or steel based, which you can test with a magnet, heat it up to stress relief temp, about 485 C and then try to straighten while hot.

Otherwise, if the metal is pewter or other amalgum, YRMV. Heating to 70 C may soften enough to straighten but possibly not work, depending on the material mineral makeup. It may already be cracked too much.

If you power up, use the 60 Watt light bulb in series method first to see if there are issues.

Hope this helps...

[TerraHertz]

Haha, that RIFA already has some nice visible cracking marks on it. 100% perfect decision to not power it up (who knows how long it has been sitting in storage and attracting humidity).

It sure does. I'm surprised you can see the cracks in that photo though.
Painfully, the only nearby 'go buy it tomorrow' place is Jaycar, and they don't stock 2200pf / 2n2 in these.
A few months ago I bought quite a big kit box of mains-rate metalized poly caps, specifically to cover this problem (replacing RIFA caps.) But_that_ doesn't go to low enough values either.  This part is a critical circuit component, not just some mains noise filtering thing.

If the metal from the bracket is iron or steel based, which you can test with a magnet, heat it up to stress relief temp, about 485 C and then try to straighten while hot.

Otherwise, if the metal is pewter or other amalgum, YRMV. Heating to 70 C may soften enough to straighten but possibly not work, depending on the material mineral makeup. It may already be cracked too much.

It's zinc-based diecast, as used everywhere by everyone where cheapness is #1 objective. Everyone who repairs HP gear, with the kinds of mechanical damage these frames suffer, probably hates the stuff as much as I do. It's really brittle. I'm quite surprised that frame bent as much as it did without fracturing. Perhaps because the force was mainly compression along the rail.

If you power up, use the 60 Watt light bulb in series method first to see if there are issues.

Variac with current meter.
These days I even replace the equipment fuse with the correct one for 240V operation. I used to often not bother.

[Kleinstein]

The 2.2 nF capacitor should likely be a class Y type. So rated for improved safety against a short. 2.2 nF is also in the usual order of magnitude for this types, also in smaller supplies. Modern class Y caps are often ceramic (often blue case).

[TerraHertz]

It looks like I'll be able to obtain a new side strut. To my amazement Keysight seem to have it available as a spare part!
   https://www.keysight.com/my/fapHomePage   Find A Part
   https://www.keysight.com/my/partDetail/5021-5838

I'm first going to try buying one from them, to see how it goes. If they can actually supply (some) standard structural parts for old HP gear that would be great.

In the thread over at groups.io [HP-Agilent-Keysight-equipment] WTB: HP diecast side strut PN 5021-5838
the issue of the screws came up too. Here's a pic:



The rounded-lip ones are the HP originals, the longer screw is a commercial standard M4 countersunk head
(but too long.)

The strut securing screws on mine are stainless M4, with a posidrive countersunk head.
Only slightly unusual in that the CSK head has rounded edges rather than the usual CSK
head which is wider. But normal CSK screws fit perfectly well. I don't know why the HP
ones have that rounded lip. Maybe so if something brute-force pulls the head through the
diecast, then there will still be enough of the frame metal around the hole so that a
standard CSK screw will hold? Or is it something to do with absorbing impact energy via
partial destruction, like vehicle frame crumple zones?

One thing to beware is that HP changed from imperial to metric at some point, and some
machines have a mix of the two. It's a good idea to have sample screws of both types,
to carefully feel out holes to determine the correct thread. Screwing the wrong one into
diecast threaded holes will wreck the thread.

If your frame used imperial threads, those strut screws will be UNC #8.

Some potential suppliers (Sydney, Oz and online):

Industrial Fasteners  67 Barry Ave Mortdale. (right at end of street)   ph 9534 6844
http://www.hitechfasteners.com.au/distributor/industrial-fasteners-pty-ltd-indust

The Stainless Shop. 170 Taren Point Road Taren Point. NSW 2229 ph: (02) 9525 1777
https://www.thestainlessshop.com.au

http://www.smallparts.com.au/
PO Box 2161, MANSFIELD, Queensland 4122
3239 Old Cleveland Road, Capalaba West (Brisbane Southside), Queensland 4157

https://www.aliexpress.com   "stainless countersunk M4 screws   EG:
https://www.aliexpress.com/item/1005002205917013.html?spm=a2g0s.9042311.0.0.323c4c4drgWRX5
  M4 8mm   (the HP ones are 7mm long, but 8mm will do.)   
  100 for AU$15.62 incl delivery. Bought! But will take 4 weeks.

[amyk]

That part looks like something anyone with a CNC could easily make out of a "real" metal, although far more expensive than what Keysight is willing to sell you one for.

[TerraHertz]

That part looks like something anyone with a CNC could easily make out of a "real" metal, although far more expensive than what Keysight is willing to sell you one for.

True but... thankfully I won't have to ever consider that extreme resort. If Keysight doesn't work out, I've received not one but TWO kind offers of salvaged frame struts.

Irony - I'd expected that to be difficult, and this one other tiny thing to be easy. But it's the reverse.
The rear fan filter is held on by 4 small knurled nuts. One is missing. I'd assumed they would be M3 or UNC 6 or something simple, and I'd probably never find an identical knurled nut so would just use one (or all 4) plain nuts.
No such luck. Firstly the threaded shafts are pressed-in studs. Quite difficult to replace. And the thread... I have NOTHING that fits it. Not metric, not BA, not 1/8" W, not UNC... A UNC #6 is about the right diameter but too coarse a thread pitch. So it's UNF?

[TerraHertz]

As expected, I can't find any source of those chromed brass thumb nuts with UNF 6-40 threads.
So improvise. I found something similar on Aliexpress - https://www.aliexpress.com/item/32892002445.html
Slightly larger, too thick, and all metric thread sizes.
Bought some with 2.5mm threads, and a 6-40 thread tap. Once they arrive I'll cut them down in thickness and tap the new thread in. Too bad they are not chromed.



The replacement 2200pf main-rated caps arrived, to replace the doomed (cracked casing) RIFA cap on the power supply board.
With that and the repaired transformer all put back together, first power up. Runs self tests OK.



A replacement side strut is on the way, but will take a while to get here.

Trying to run the machine, the Source output works, level setable and runs a frequency sweep correctly.
For a while there seemed to be something wrong with the two inputs. They appear to be seeing random noise. Then I discovered they have 'floating' and 'grounded' modes. In grounded mode they work as I'd expect.  More investigation needed there. Maybe even reading the manual!



Today is a good day. Also because... one of my cats (Charlie, in pic) has been missing for several weeks. I'd assumed he'd been run over or something. This evening he turned up again.

So far I have the the programming guide and two volume service manual, but have not been able to find an original of the operating manual, PN 03562-90001. Only a PDF which I find horrible to read, hence avoiding it.

Did HP go out of their way to be sure you can't use this thing as an oscilloscope? You can view the inputs in X=time mode, but it autoscales everything and I can't (yet) find any way to set the time/div.



Top: Spectrum of a 5KHz square wave (generated by the sig-gen in the scope.)
Bottom: Spectrum of the 3456A's own Source gen outputting white noise. The sine sweep output works too.

I'm very glad the heavy impact this machine took (that bent the frame strut) didn't damage the CRT internals. The focus and positioning are still fine.

Next: Capture all the EPROM contents.