HP 1740A power supply burn up

[jklasdf]

Some pictures. 50MHz sine with flood gun going too. I'd forgotten how nice the displays on these are!

The displays on HP's old analog scopes really are quite nice. It's apparently because the CRTs don't use a scan expansion mesh, like a lot of Tek scopes of the same/slightly later era.

[bd139]

Agreed. I’ve had a fair few teks and the HP CRTs are much better for sure.

[bd139]

Looking good now. Last new LM723 arrived so I chucked it in. Fortunately this one is socketed so it didn't mean removing the power supply again and was really easy.

So firstly the new LM723 regulators installed. Fortunately one of them was still good:



The repaired and re-connectored power supply. 6-way connectors were cheaper so I just got them. Also made for easier debugging which was handy:



Fired it up and voltages were slightly too high so I trimmed them down again to 15.000V and all the rails are now spot on.

Anyway first time this is running under its own power supply:



Seems to work ok. The timebase needs a bit of a squirt of contact cleaner however. Sort of expected with the ganky switches in these scopes.

Next steps are:

1. I'm going to add a small heatsink made out of a bit of bent alu sheet to the 15V rectifier as it runs very hot. This should stop it taking the rest of the damn thing out.
2. Clean timebase out and calibrate it.
3. Feeeet. beanflying's 1740A feet will be printed by a friend of mine at some point in the next couple of weeks. He's got a relatively decent printer and is doing it for free if I set up his WiFi extender

[bd139]

Heatsink done. As mentioned elsewhere I have virtually no metalwork tools so this is not pretty 

Bingo - looking alive again:

[bitseeker]

Woot! Looks great, bd. Congrats.

[jklasdf]

Looking good! The timebase switch is probably the worst part about these scopes, as great as the display is.

[bd139]

Yeah the timebase switch on this isn’t totally healthy yet. Need to do some work on it. The delay sweep doesn’t turn off properly and the rotation is gappy.

[tggzzz]

On my 1740A the PCB tracks under the timebase switch were abraded by the plastic part of the switch. The parts contacted by the switch fingers were fine.

I soldered a thin wire across the gap, ensuring no solder flowed onto the part contacted by the switch's fingers.

[bd139]

I shall bear that in mind. I suspect something similar is afoot. Thanks for the heads up

[bd139]

Ok so I managed to get inside the failed LM723 and found that one of the bond wires had melted. This happens to be the internal pass transistor's collector so it looks like it went over-current at some point, probably when the connectors melted and the bridge went short.

Photo inside LM723 showing the melted bond wire. You can see the blob on it.



Blown up bit on datasheet:



On a positive note, the reference part of it still works so I'm going to attempt to put its hat back on and play around with it and my 34401A 

Also a hack. This could be repaired by changing the pass transistor to outboard. Snip off the collector and emitter pins and use a 2n2222 or something with base connected to the frequency compensation pin.