I'm sure you know this anyway but leaving it off also lets some of the nasty voltages drain away a bit. They will still be dangerously high. It's probably worth making an insulated discharger for the caps from a resistor and the probes from a dead multimeter.
Maybe somewhere aoround 1-10K resistor but possibly someone will have better sugestions.

 

I'm going to try repairing it this weekend.

SgtWookie, I am turning it on, because it has almost fixed itself. The screen is stable, it is not resetting, the trigger logic seems to work, it's just out by about 200mV on all channels, and there is a lot of ripple on the signal (ripple goes away after use for 5+ minutes.) Apart from that, the scope is working ok, so I think that the analog Vref(s) have gone, that is, -5.2V, +3.5V and +5.15V(A), and I'm going to try replacing the caps for those. If not, well, I don't know. Without another scope to debug it, the repair of anything else will be difficult. I have a feeling that the digital logic caps will be ok, but I'll probably replace them as well (good to be sure.)

Just tried the AC cal and it seems to work ok.

I can't say too many bad things about HP for the scope lasting 22 years+. But it is annoying. (Note to self: HP now sucks, Agilent make decent equipment, HP just make crappy printers and desktop PCs.)

Looking at the service manual, the PSU appears so small, and it looks like it will be difficult to get into it.

 

SgtWookie, I am turning it on, because it has almost fixed itself. The screen is stable, it is not resetting, the trigger logic seems to work, it's just out by about 200mV on all channels, and there is a lot of ripple on the signal (ripple goes away after use for 5+ minutes.)

That sounds like a capacitor has pretty much gone dry and stopped conducting a partial short. Look for one that is taller than the label or has the vent opened.

I'm sure you are aware that the input filter caps can hold a lethal charge. They won't discharge except by through internal leakage - use a resistor to get that voltage down.

 

I'm sure you are aware that the input filter caps can hold a lethal charge. They won't discharge except by through internal leakage - use a resistor to get that voltage down.

Of course. I will have to find a 50k 3W resistor.

If I turn the scope on and pull the plug out, will that discharge the cap enough? I think it will to a certain point, but it would probably still have 100V+ on it before the scope cuts out. (Scope will run for about half a second without AC power.) It's a bad idea to use a screwdriver to short out the main input filter cap as I could damage the filter cap, but finding such a resistor in a pinch (two days) could be an issue.

I will have to hope that the PSU is held together using screws, if not it will be difficult to disassemble it without breaking it.

If the repair doesn't work, does anyone have any suggestions for a new, digital oscilloscope? I'm leaning towards the Rigol DS1052E 100 MHz (hacked from 50 MHz) 1GS/a, but I would be ok to spend up to £350 (Rigol scope is £249-299.)

The offset fault appears to be past the attenuators. If I select 1V/div, it shows 1.5V with no input, if I select 100mV/div, it shows 150mV. This is independent of AC or DC coupling. I also shorted the probe to ground - still offset.

 

SgtWookie turned me on to a great way to discharge caps. Use a light bulb.

The resistance changes in the filament as it heats. So at first, there is little resistance, but as is heats up, the resistance also increases.

This will suck a cap dry pretty efficiently.

 

SgtWookie turned me on to a great way to discharge caps. Use a light bulb.

The resistance changes in the filament as it heats. So at first, there is little resistance, but as is heats up, the resistance also increases.

This will suck a cap dry pretty efficiently.

That's a good idea. Would a 230V lightbulb work or would an automotive bulb work well? (Thinking that a 12V bulb will burn out quick with ~325V.)

 

Well, it depends on how large the cap(s) are. You might use a 100 Ohm to 1k Ohm resistor in series with the light bulb, just so you lessen the risk of zapping traces on the board.

 

It's actually fixed itself!

The zero readout is about -2mV. When I first got it, it was about -3mV, so I'm fairly certain that this is normal. There is very little ripple.

But, is this just an imminent failure waiting to happen?

EDIT: SgtWookie, the only dangerous cap is the line input cap, which I expect to be fairly large. The output of the power supply is +15V maximum. Of course, as it is a CRT oscilloscope, I have to be careful of the EHT voltage (green CRT, about 1-2kV), but I'll only be removing the PSU module to do any repair.

 

It's actually fixed itself!

is this just an imminent failure waiting to happen?

This is very typical for a solder joint starting a crack up, or connectors starting to "lose the grip" I am a big fan of the saying. "If it is not broken, do not fix" but the next time this error happened. Turn the scope off, and fix it. Before any real damage to scope occurs. The solder joints most prone to fail is the joints carrying high voltage and/or high current.

 

This is very typical for a solder joint starting a crack up, or connectors starting to "lose the grip" I am a big fan of the saying. "If it is not broken, do not fix" but the next time this error happened. Turn the scope off, and fix it. Before any real damage to scope occurs. The solder joints most prone to fail is the joints carrying high voltage and/or high current.

I will probably take the cover off and have a look. Some of the dust probably should be cleaned out anyway.

 

The solder joints most prone to fail is the joints carrying high voltage and/or high current.

I'd add to that any component that tends to get hot in operation. I can't count how any things I've repaired due to thermal expansion creating a tiny "ring crack" around a component lead or solder pad due to the thermal expansion that occurs due to heat cycling.

 

It didn't last.

It's started power cycling again. Before that, the trace was about 13mV out, and then it maxed out, and then it stopped working all together (with the power cycling.)

I've opened it up (using a Torx/Star T10 - kit of 6 short drivers, very handy for the screws inside - was £4.99 from Maplin.)

I took some pictures of inside.

The first thing that comes to mind is that this unit was definitely build to last. Looking at the circuit board, I see one trace marked 12V, and it is connected to the other end of a 330u/35V electrolytic, which connects to the 'clicker', which produces a sound whenever you press a key or rotate the dial.

The system board covers the entire base of the unit; an additional CRT board runs up the side of the unit and this contains the flyback and a few big caps. I'm fairly certain the CRT board is in good shape. The CRT monitor itself was produced by Hitachi Mizuswa(?).

I was surprised by how large the CPU is! It looks to be a Motorola 68000 P10 processor which is a 16-bit CPU. Must be in an old-style DIP-64 package. I wish you could get newer chips in such packages - ideal for breadboarding, but alas, they must be pretty pricey and probably only hobbyists want them.

The unit has several daunting warnings. Unfortunately, "qualified service professionals" don't exist for this scope any more, and I suspect few will have spare parts.

An overview of the unit, with the PSU installed:

The CRT board is separate. A quick look at it and it doesn't seem to have any problems. The screen is fine, and the caps are ok. The different soldermask colour of the CRT board seems to again indicate it is an OEM part.

I'm going to remove the PSU and take a look inside. From what I hear, it is entirely discrete based(!) There are no PWM or SMPS controllers, it is based around discrete transistors and diodes. Don't see that much any more. It's also apparently some kind of OEM SMPS, and used in several HP instruments. Probably explains the wide variety of voltages.

I was also suprised by the amount of discrete logic ICs. Unlike newer scopes (e.g. Rigol's DS1052E) there are several PALs (at least 3), about a dozen logic ICs, and discrete SRAM, NVSRAM, EEPROM and EPROM memories. The analog section is composed of about 20 or 30 odd transistors, resistors, caps and diodes. This probably explains the $3,300 price tag.

 

I removed the supply and opened it up.

I'm disappointed. There are no bulging capacitors.

The capacitors are quality - Rubycon, and always overspecified. No cheap Chinese generics - it's full of Rubycon. There are loads of them. Nice 680µF/200V x 2 filter capacitors.

That power transistor is massive. Note that the heatsink is apparently live! Is that even legal today?

At the moment, I'm still thinking that the caps are bad. Do I need to get an ESR meter? None are bulging, a few could be leaky and this could be the problem. Unfortunately, it's going to leave me without a scope for a while.

The power supply controller is on a separate board. It contains an LM339 comparator, and as promised, entirely discrete based SMPS control.

For those who are interested, the acquistion side is below.

It is apparent that some low speed, jellybean parts are used (LM358 for example) along with some really high speed parts - NE5539N, which is a 1.2GHz GBW op-amp, as well as Am6687, a high speed comparator (equivalent part AD9687) I can't locate a dedicated trigger IC, so I suspect that trigger logic (using the external comparators - two per chip and one for each channel of the scope) and the analog to digital conversion happens inside the heatsinked IC

There are two oscillators in the acquistion portion of the scope. One 40.0000 MHz and one 40.7232 MHz, which is odd.

 

Bump. Does anyone have any idea?

If I can't find any solutions to this, I'm going to put the unit back together and just not use it when it goes on the fritz. Or maybe I'll sell it on eBay for less than it's worth.

 

Bump. Does anyone have any idea?

If I can't find any solutions to this, I'm going to put the unit back together and just not use it when it goes on the fritz. Or maybe I'll sell it on eBay for less than it's worth.

Do a throughout cleaning using a soft brush and vacuum cleaner.

It is likely that the problem will disappear all together.

 

Before you go selling you ought to look for a service center. While I can repair my own equipment of this type I look for a tech that has already made all the obvious mistakes.

 

Bit of a long shot but you could pull the connectors, give them a spray of contact lubricant and replug them. Could be something a bit corroded and not quite making contact in there.

 

Do a throughout cleaning using a soft brush and vacuum cleaner.

It is likely that the problem will disappear all together.

Good idea - though I've heard that using a vacuum cleaner can cause electrostatic discharge. Any concerns on this? (I don't want to completely break it.)

Before you go selling you ought to look for a service center. While I can repair my own equipment of this type I look for a tech that has already made all the obvious mistakes.

The guy I bought it from might be able to help. It was calibrated by Aeroflex (?) so maybe I can ask them, but that was back in 2005.

Bit of a long shot but you could pull the connectors, give them a spray of contact lubricant and replug them. Could be something a bit corroded and not quite making contact in there.

Any idea on a specific lubricant? Would WD-40 work?

Thanks for the ideas guys, I'm sure I'll get it working. Would it be worth investing in an ESR meter?

 

Any idea on a specific lubricant? Would WD-40 work?

NO do NOT use it There is nothing particularly good/appropriate about using it. It will leave oily film behind that dissolves the grease in moving parts (which should stay in those parts) and will attract dust. Use a real electronic cleaner

 

NO do NOT use it There is nothing particularly good/appropriate about using it. It will leave oily film behind that dissolves the grease in moving parts (which should stay in those parts) and will attract dust. Use a real electronic cleaner

Okay, I'll get some proper cleaner stuff. Thanks.

I figured out some more interesting things about this scope (might as well do that while I have the case off.) First off, the heatsinked IC appears to be another PAL, and isn't an ADC. It forms part of the timing control logic (TCL.) Analog to digital conversion happens inside the MC10319L chips (2 of them, there are two ADCs.) The ADCs are actually rated up to 30 MS/s, apparently HP are only clocking them at 10 MS/s. The ADCs are pretty pricey flash converters - an array of comparators (256) in each chip. The comparators and the LM358s seem to form some kind of trigger logic. Features such as pattern and TV trigger probably happen inside the heatsinked IC. I can only guess this, because the IC is heatsinked, and I cannot determine markings.