Solartron7061

[Dek]

Hi Kleinstein,

I was looking at the voltage ip at the 15v and -15v regulators and was wondering why they weren't balanced.
Yes the ip to the +15v reg is about 34v 
A quick check of the resistance between the ip of the 15v and ip of the 36v regulators and they are dead short......getting somewhere :-)
I'll have PCB5 out and check C902. 

Dek

[Dek]

If yes, this could indicate a problem (near short) with C902 (the one before the 36 V regulator).

Spot on Kleinstein.  :clap

I swapped out the cap and now the 36v rail is a rock solid 35.5v
Quick check of the old cap and on my £15 ebay tester reports the cap as a 0.31 resistor.
Using my 2aa battery box (fluke 187 reads as 3.2243v, the Solartron measures 3.2241v (floating between 3.22404 and 3.22410) not sure if that is indicative of a problem of just my un-shielded hacked up test lead picking up noise
Checking Ohms ranges with the Dekabox, all looking quite good.

The display still flashed up Glug Death, but I'll take that as a win for tonight :-)

Dek

[grizewald]

Nice result and great diagnosis from Klienstein.

Considering how R901 has been fried, you probably want to pull that and check its resistance. To save yourself future grief, a wholesale replacement of all the electrolytic capacitors would probably be a good idea. If one has failed short like that, then there's a very good chance that the others won't be far behind it.

[Dek]

I'm an idiot

Just testing the 32v unrequlated supply, slipped and touched  R510 and R511 causing a short between the + and - rails 

Display now blank, no keypad leds and no key press beep.

The led D509 near T501 is off.

All psu rails appear to be OK including 32v unreg and the 40v for the VFD.

Looks bad .... was getting so close.
I'm off for a beer and start again in the morning....

[grizewald]

Really sorry to hear that!

Shorting those resistors would do nothing on one end, but short +32V with -32V at the other end, quite possibly taking out the bridge rectifier. I'm having trouble understanding why this would have taken out the 24KHz filament drive (as indicated by the LED not coming on), but that would certainly explain the blank display.

I'm always paranoid about probing live circuits and prefer to turn the machine off, make the connections with test clips and then power on again. At the very least, I'll use probe covers on my multimeter probes so that slipping off a pin or probe point can't cause a short. It's a lesson that is very painful to learn the hard way.

[Dek]

Hi Grize,

Painful lesson indeed, blood and guts paid for by the instrument
I was doing one last check on all the rails before investigating further the Glug death on start up.
I'm kicking myself because the 32v unreg supplies (I think) are only needed for the expansion ports (I have the memory removed at the moment) so it was an unnecessary test....

Not getting the CLR on IC315.
No CS on the PIA - looks like the processor / digital side is toast. Can't see any activity on the data or address bus.
No output from the intergrator.
No response from the keyboard
Glugs and Glug drive appear to be OK which is ironic because the reason for tracing Glug Death in the first place!

I fear that multiple components have been damaged and the meter is scrap, BUT there might be some hope - processor reset is floating about 2v
...now off to try to see if it is the power fail circuit...... or am I 

Dek

[grizewald]

I'll keep my fingers crossed that it's just something simple. Are you sure you only shorted the 32V unregulated supplies? I really can't see how that could knock out the CPU and filament drive.

I guess you just need to go methodically through the various power rails and see if any are missing or being overloaded.

[Dek]

I'll keep my fingers crossed that it's just something simple. Are you sure you only shorted the 32V unregulated supplies? I really can't see how that could knock out the CPU and filament drive.
I guess you just need to go methodically through the various power rails and see if any are missing or being overloaded.

Yes definitely R502 R503.
Looking at the photo, I wonder if I inadvertently shorted the trace between the two resistors to R502? (not sure what this trace is).

Also spotted a bent leg, but as this is A5 address line, I don't think this is related to the original problem.
I'll carefully check all the rails again, then go through the power fail and watchdog circuit to see if that is holding the processor reset.

[Kleinstein]

The bend pin could still cause some problems  - possibly only after bending the board a  lot.

AFAIK the computer chips (6809 and 6821) are not so rare and should still be available in some form.

Bend back the A5 pin and checking the supply and reset part would be the next points.  If the short was from the + or - 32 V to one of the computer inputs this could definitely cause some damage. So it may be worth to follow the trace in question.

[grizewald]

If I had to hazard a guess, I'd say that the track which runs between those two resistors goes to the GPIB chip.

What bothers me is that you say the LED on the board doesn't light up. That implies that the +5VD rail at TP509 isn't coming up for some reason. That rail also powers the clock circuit on PCB 8, and if you don't have a clock and you don't have the filament drive which is derived from +5VD and the 24KHz signal from the clock divider circuit, then that would fit well with +5VD not coming up.

What do you see at TP509?

[Dek]

Hi Grize,

TP509 = 5.06v
49mhz and 24mhz signals are alive.
I see the 24mhz on the filiment drive fets Tr501 and 502.

I pulled the GPIB interface adaptor to see if I could see the trace - the socket base  is almost solid, so checked the continuity of the trace between the resistors to the IC socket - it goes to pin 18 which is the clock signal.
This  is generated directly from the processor (pin 34 E) and is also to nand gates in IC118 which generated /OE and /WE
This line appears to be 3 to ground, so it looks like I've most likely fried everything connected to this signal line.

I think I will take out any IC that has direct connection to the E clock line which I think is the processor, the GPIB and the 74LS00 IC118 and see if that clears the short.
Then it's a case of a new processor and 7400 - I'll leave the GPIB out for now and see if it comes back to life.
So there might be still some hope left.
How does that sound as a plan?

What a muppet

Dek.

 

[Kleinstein]

The analysis sound logical. Removing chips from a socket should be relatively easy, and LS00 are still available. For a first test one could just cut the leg in question and add it back if the 74LS00 was not at fault.

[grizewald]

That really is exceptionally bad luck to short 32V to the E line.

You have almost certainly fried both the GPIB processor and the main processor. For the soldered stuff, Klienstein's suggestion is the right way to go. Just snip the pin on the NAND gate mid way and you can just solder it back together with a blob of solder if the chip is OK.

A little hint on replacing soldered logic ICs from the days when I used to do it for a living: just cut all the ICs pins close to the package body to remove the IC. Then with a pair of fine tweezers and your soldering iron, melt the solder on the pin and swiftly pull the remains of the pin out. Then resolder each hole to get fresh solder in the hole before you then clear each hole with a vacuum desoldering gun. This way, you limit the amount of time that the hole and via is heated and eliminate any mechanical stress during the process.

Other methods, like trying to desolder the IC complete risks damaging the pads, through hole plating or tracks. You really don't want to make things any worse than they already are.

[Dek]

Thanks Grize and Klein,

Your help (and encouragement) is much appreciated

I've found suppliers of all the ICs and not too expensive, should have these in about 5 days.
I also noticed that IC105 (138 decoder) is also connected to E so I have orders spares for that too.
Hopefully the damage is limited to those ICs and not propagated further.
As you have suggested, I'll snip the IC leg connected to E to check which IC is taking this line to ground.

Dek

[grizewald]

Well spotted that E also goes to IC105, I completely missed that one when I was looking for other parts connected to that node.

Still, the big plus with the Solartron meters is that all of the various ICs are still available. If you'd fried the ASIC in an HP meter, finding replacement parts could be a major headache. Hopefully the damage is limited to just those components connected to the E node. I'll keep my fingers crossed for you!

[Dek]

Now, just waiting for the 138 decoder to arrive and then time to see if I can resurrect this puppy.
One thing that is bugging me is the filament drive led is still off. As far as I can tell, with the 5v and 24khz /24khz lines live, the LED should be on....

It was the processor that was holding the E line to ground, wonder if it's possible to test if the 7400 and the 74138 survived without taking them out and without socketing  the new processor?

Dek

[grizewald]

Now, just waiting for the 138 decoder to arrive and then time to see if I can resurrect this puppy.
One thing that is bugging me is the filament drive led is still off. As far as I can tell, with the 5v and 24khz /24khz lines live, the LED should be on....

It was the processor that was holding the E line to ground, wonder if it's possible to test if the 7400 and the 74138 survived without taking them out and without socketing  the new processor?

Dek

??

Your processor looks like it is already socketed.

As you don't have low impedance to ground with the dead processor removed, then the other two chips are at least not shorted and shouldn't damage the new processor. To test them in circuit other than looking at the impedance on the pins which are connected to the E node sounds kind of tricky.

I still don't understand why the LED in the filament drive went down. Maybe it's just a coincidence and the LED itself is actually dead? If you have the 24KHz signal coming into the transformer T501, do you have the expected waveforms appearing on the output?

[Dek]

??
Your processor looks like it is already socketed.

As you don't have low impedance to ground with the dead processor removed, then the other two chips are at least not shorted and shouldn't damage the new processor. To test them in circuit other than looking at the impedance on the pins which are connected to the E node sounds kind of tricky.

I still don't understand why the LED in the filament drive went down. Maybe it's just a coincidence and the LED itself is actually dead? If you have the 24KHz signal coming into the transformer T501, do you have the expected waveforms appearing on the output?

I should have worded that better... is it possible check the other ics in circuit to prevent damage before installing the new processor in its socket.
I guess other than a simple "check the inputs are not  shorted" there is no easy way to tell.

I'll check the F1 and F2 output  waveforms....
F1 appears to be accessible, but F2 only looks line I can access it from the underside.
Guessing F1 and F2 should be approx 5v square-ish with a slight DC offset bias and 180 degrees out of phase?

Dek

[Kleinstein]

A first test would be checking the supply. However the xx138 may already be needed for such a test. These of often used to generate the chip select signals - so a broken / missing xx138 could cause the RAM and ROM to work against each other.

Checking for signals that are shorted could be possible - so bring the data-bus to a defined state, e.g. where the CPU can write, or at least no other chip gets control over the data bus.

[grizewald]

I'll check the F1 and F2 output  waveforms....
F1 appears to be accessible, but F2 only looks line I can access it from the underside.
Guessing F1 and F2 should be approx 5v square-ish with a slight DC offset bias and 180 degrees out of phase?

Something like that. I'll see if I can get the 'scope on mine later and capture the waveforms for you.

[Dek]

I took the LED out of circuit and tested on the bench PSU, it works OK !
TP509 4.95v

Hooked up to F1 and F2 look OK.

I'm guessing that there needs to be a filament current for the LED to be on? as at the moment the display is all off.


Dek.

[grizewald]

That looks good, but I'll still try to probe mine in a bit.

Have you replaced the 74F138 yet? I notice that there's a signal "-DSPDAT" generated by that chip which goes to the display. I also noticed (rather ominously) that the E node also goes to IC208, the 63B21 PIA. Did you replace that chip as well as the processor?

[Dek]

That looks good, but I'll still try to probe mine in a bit.

Have you replaced the 74F138 yet? I notice that there's a signal "-DSPDAT" generated by that chip which goes to the display. I also noticed (rather ominously) that the E node also goes to IC208, the 63B21 PIA. Did you replace that chip as well as the processor?

The 74F138 should be with me tomorrow or Saturday, I have a new processor, PIA and the 74LS00 ready to install.
Nothing changed yet, the fact the filament LED was off was bugging me, so needed to check that first before moving on.
I was thinking of just trying the new processor and PIA before pulling out the two logic ICs but worried that if one or both of the logic chips are fried it could damage the new proc or PIA.
So the plan now is to replace the 2 logic ICs as well as the proc and PIA all together and see if the meter comes back to life.

Dek.

[grizewald]

Here's what I'm seeing on the F1 line:



About 3V DC offset and 10V P-P. I couldn't see anywhere I could safely put the channel 2 probe to measure F2, but I'd say yours is probably OK. The voltage levels could well be different if the display is actually on, so you're probably right about the LED not coming on unless the display is actually on.

[grizewald]

I was thinking of just trying the new processor and PIA before pulling out the two logic ICs but worried that if one or both of the logic chips are fried it could damage the new proc or PIA.
So the plan now is to replace the 2 logic ICs as well as the proc and PIA all together and see if the meter comes back to life.

I think that's a sensible plan. There's no need to take additional risks.