Glitchy Enertec 7150 ( Solartron / Schlumberger )

[jtu]

Hi,

I've got my hands on one a few days ago. It is not working stable. It can be working for hours one day, but the other day it has one of the 3 behaviours:
1. Upon turning it on it continuously clicks relays very fast, buzzing actually. "Fail 1" on the display.
2. Upon turning it on it shows "Fail 1" on the display. No relay sounds.
3. But most often it seem to be resetting itself repeatedly in a matter of seconds to minutes. Sequence of action on the display is: OL, Fail 1, reset, normal. Relays clicking slowly meanwhile.

I have checked power rails, those are within specs and seem stable over time. I have not logged voltages, but I have measured on multiple occasions either cold or warm, good or failed state.

IC103 clock is stable.

I've got impression that it generates some signal on the HI input which causes overload or autorange fail. It can be vice versa of course. Autorange works normally when it is in a "good mood".

I have seen the following. Usually voltage fluctuates a bit (nothing connected). OL happens when the fluctuations go from last 2 digits to next one(s). I tried to catch it on the scope, it goes up to 0.2-0.7V when that happens.

I have read posts here about this model. I have not found any with similar symptoms. Elsewere out there too.

I would like to continue to debug it, but have not enough knowledge.  Any advice, suggestions, ideas?

Schematics http://www.ko4bb.com/getsimple/index.php?id=download&file=06_Misc_Test_Equipment/Solartron/Solartron_7150/Solartron_7150_Maintenance_Manual.pdf. I have newer model from the two described, first power schematic from two.

Veiksmi,
J?nis

[jtu]

Today I put some attention to data transfer between floating analogue and earthy logic parts. This was because I read here that optocouplers might be the culprit.

Logic -> analogue seem almost normal TTL levels, but the opposite direction seems weird to me. See pics. It's almost like something is pulling the line down by more than 2V. Is this normal behaviour?

[Armadillo]

Check the reference voltage at;

Measure;
R14 with respect to the anode of D20, can you get 13v or 14v [either Leg of R14].

[jtu]

Thanks for the clue!

R14 measures 12.9V and 13.9V (I'm dropping last digit of my multimeter, it is no precision device).

[Armadillo]

Does the meter measures input correctly?

Example apply a known voltage to the input terminal, what voltage is read out?
Apply a known resistance to the input, what voltage is read out?

[jtu]

Yes, it measures voltages and resistance correctly when it is in "a good mood". Accuracy might be questionable, but it compares with other meters.

[Armadillo]

Yes, it measures voltages and resistance correctly when it is in "a good mood". Accurecy might be questionable, but it compares with other meters.

Resolder all dull solder joints, check connectors, WD40 contact cleaner on all mechanical switches and contact surfaces.
If the meter is expensive, I would begin to change all the coto relays or parts therewith.

 

[Mickle T.]

1. Upon turning it on it continuously clicks relays very fast, buzzing actually. "Fail 1" on the display.

This is absolutely normal start up feature - "relays contacts refreashing".
"Fail 1" has nothing to do with the voltage reference, COTO relays or anything in the analog signal path. The most common cause of Fail 1 is optocouplers LED degradation.

[Armadillo]

1. Upon turning it on it continuously clicks relays very fast, buzzing actually. "Fail 1" on the display.

This is absolutely normal start up feature - "relays contacts refreashing".
"Fail 1" has nothing to do with the voltage reference, COTO relays or anything in the analog signal path. The most common cause of Fail 1 is optocouplers LED degradation.

Hey Dude, did you not see the strong optocoupler pulses in the attached oscilloscope traces attached above?
Since you are so sure about yourself, then why don't you tell us how do you propose to test the optocoupler degradation to justify your assertion.   

[jtu]

Took it apart today. Looked for solder problems.

Noticed that somebody has already done a repair on this. IC15, IC5511, C4 has been replaced. Soldering job did not look convincing (some pins were not wetted on the other side). Resoldered those. No other visible soldering problems.

Cleaned it a bit.

Same behaviour. Thinking about the next step now...

Meanwhile, take a look at the pictures

[jtu]

Noticed that Schaffner filter makes noise sometimes. I should replace it.

But it would be good to find the problem first.

[jtu]

"Fail 1" has nothing to do with the voltage reference, COTO relays or anything in the analog signal path. The most common cause of Fail 1 is optocouplers LED degradation.

Not a big problem to replace optocouplers. It would be good to confirm it first somehow...

[Mickle T.]

1. Upon turning it on it continuously clicks relays very fast, buzzing actually. "Fail 1" on the display.

This is absolutely normal start up feature - "relays contacts refreashing".
"Fail 1" has nothing to do with the voltage reference, COTO relays or anything in the analog signal path. The most common cause of Fail 1 is optocouplers LED degradation.

Hey Dude, did you not see the strong optocoupler pulses in the attached oscilloscope traces attached above?
Since you are so sure about yourself, then why don't you tell us how do you propose to test the optocoupler degradation to justify your assertion.   

Hey Dude, It's not my assertion. Did you not see the service manual?

[Armadillo]

1. Upon turning it on it continuously clicks relays very fast, buzzing actually. "Fail 1" on the display.

This is absolutely normal start up feature - "relays contacts refreashing".
"Fail 1" has nothing to do with the voltage reference, COTO relays or anything in the analog signal path. The most common cause of Fail 1 is optocouplers LED degradation.

Hey Dude, did you not see the strong optocoupler pulses in the attached oscilloscope traces attached above?
Since you are so sure about yourself, then why don't you tell us how do you propose to test the optocoupler degradation to justify your assertion.   

Hey Dude, It's not my assertion. Did you not see the service manual?

The optocoupler was "CHECKED" by OP and if it failed, it won't be doing any measurement reading and the "fail 1" will be shown upon power on and remain shown, won't you say, however
the texts also said to "Check the connecting wires and plugs", which you conveniently left out.
So, would you propose how to test your assertion that is it "optocoupler DEGRADATION despite the oscilloscope traces shown to you?

From the OP description, fail 1 appears somewhat cyclic and furthermore coupled with relay cycling, so I reckon the "possibility" of relay stiction or intermittent failures of contacts which is a common failure mode of such reed relays causing the microcontroller to time-out and failed on communication intermittently. It could be a degradation of the optocoupler but the strong pulses have been shown to written it off. What do you think about the oscilloscope traces shown to you.

You are absolute with what you "read". So would  you propose the tests to complete this repair other than what you "read". And do comment on the traces in your reply.     

[Kleinstein]

I have seen LEDs fail in a kind of intermittent way, likely due to a loose bond wire. It is possible that optocoupler can show a similar type of failure - working sometimes and not working if at a different temperature / board stress. So even if the scope measures Ok there might be still a failure in the Opto-couplers. The DMM seems to work sometimes. The failure is of an intermittent type, like a bad solder joint, connector or similar and according to the manual in the path of the optocouplers.

Compared to the relays optocouplers are cheap and easy to replace.
The relays used are difficult to get and soldering it somewhat critical as it is in the low leakage area - so the relays are among the last parts to replace, unless one knows for sure they are bad.

[Mickle T.]

Among many of the repaired DMMs, I has a Solartron 7150+, 7081 and Datron 1071 with the same (or similar) problem. 7081 was very old (revision F) and has a two failed optocouplers: 1/4 of ILQ74 in the analog section and one 6N136 in the floating link (similar to 7150+). The quick and dirty o/c checking is very simple: decreasing a current limiting resistor, e.g. by 50%.

[Armadillo]

I have seen LEDs fail in a kind of intermittent way, likely due to a loose bond wire. It is possible that optocoupler can show a similar type of failure - working sometimes and not working if at a different temperature / board stress. So even if the scope measures Ok there might be still a failure in the Opto-couplers. The DMM seems to work sometimes. The failure is of an intermittent type, like a bad solder joint, connector or similar and according to the manual in the path of the optocouplers.

Compared to the relays optocouplers are cheap and easy to replace.
The relays used are difficult to get and soldering it somewhat critical as it is in the low leakage area - so the relays are among the last parts to replace, unless one knows for sure they are bad.

Based on your cheaper economical theory hypothesized with element of stress and temperature variants might be a convenient way out, but a way out, nevertheless. No arguing about it.
Replace and change the cheaper components is the way of repair, so to speak.

But since you agreed the scope measures OK, but still chance the opto-couplers failure. I would add that, technically speaking, one should also understand how the pulses are generated in the "integrator" in the first place that could also mask as a fail 1.

With that said and judging from the age of the instrument and the retrospective failure and repair of the optocoupler, I should say, experience do count and so be it.
 

[Armadillo]

The quick and dirty o/c checking is very simple: decreasing a current limiting resistor, e.g. by 50%.

Its quite an interesting dirty method actually, except I don't understand if you want to increase If or decease it. Can you elaborate further.
But on the hindsight, this method definitely warrant a lot of failed optocoupler, nevertheless. It that a standard practice?

[2N3055]

The quick and dirty o/c checking is very simple: decreasing a current limiting resistor, e.g. by 50%.

Its quite an interesting dirty method actually, except I don't understand if you want to increase If or decease it. Can you elaborate further.
But on the hindsight, this method definitely warrant a lot of failed optocoupler, nevertheless. It that a standard practice?

With optocouplers, with age CTR of coupler starts to degrade.. Mostly LED degrade with their light efficiency.
So you increase LED current by some margin and check if that will cleanup signal on other side...
You can do that by temporarily connecting a resistor in parallel with existing LED current limit resistor and see if it helps..
If it does, CTR od optocoupler is getting out of specs marginally, just enough for circuit to intermittently  stop functioning reliably, but not fail completely so it is obvious...

But it is ready for replacement..

[SeanB]

The dropping resistor trick was very common on HP printers where the opto switches lose coupling ratio with time, so increasing current will gain a good extension of life.

[Armadillo]

At least we have 3 different versions of the poor-man's trick, I mean the poor man days without an oscilloscope.

Well, facing with the OP measurement of strong oscilloscope traces with a pk to pk swing exceeding 5volts,
do you think that the Pk to Pk swing looks like a problem of CTR Opto degradation to you?

What's the threshold of standard TTL logic or CMOS logic?

We heard 3 version of it.

One say O/C for open circuit, [don't understand what's he talking about?]
One say intermittent, on the basis that parallel the resistor will alleviate the problem, if it doesn't then its marginal opto degradation. [not positive, may miss other causes of failure].
One say life extension [I agree, depends on how hard the Opto LED was driven, depend on the opto collector load resistor value  also].

But I am not against the proposal to doubt the opto and change it on the basis of the age of the instrument.   

[2N3055]

Those scope traces doesn't look like anything to me.. First, they look like they were done with AC coupling, and you can see baseline of signal moving as capacitor is charging...
Average DC component of square wave that is not 50% duty cycle is not 0V...So it looks like it has negative component and goes below 0 volts...

Second, timebase is too long, can't see edges at all... If CTR of optocouplers is marginal, you might get distorted edges, and also unstable jitter on timing of edges and shortened pulses... Timebase is such on those snapshots that you can't really see any detail..
So while voltage swing seem ok, it is far from saying that everything is OK...

Like Mickle said, he should check both Rx and Tx,  on optocouplers and after TTL buffers.
Communication could very well be TTL level RS232, it might even be possible to decode traffic...

And since somebody was already inside, check whether jumper SP1 is shorted as it should be.....

[Armadillo]

Those scope traces doesn't look like anything to me.. First, they look like they were done with AC coupling, and you can see baseline of signal moving as capacitor is charging...
Average DC component of square wave that is not 50% duty cycle is not 0V...So it looks like it has negative component and goes below 0 volts...

That is because there is a floating Board and a Earthly Board, and the reason why optocouplers are used for intercomm. They don't call it floating for no reason. On the floating board, the common is not better defined, so irregardless of how he hooked the probe, as long as there is a delta V. forward voltage onto the LED of the optocouplers, it will transmit those pulses..

Second, timebase is too long, can't see edges at all... If CTR of optocouplers is marginal, you might get distorted edges, and also unstable jitter on timing of edges and shortened pulses... Timebase is such on those snapshots that you can't really see any detail..
So while voltage swing seem ok, it is far from saying that everything is OK...

Frankly, its appalling of what you are saying about those edges but I think you are trying to say the rise/fall time of the pulses that is deemed acceptable. For the fun of it, what's the value?

he should check both Rx and Tx,  on optocouplers and after TTL buffers.

He should, might as well.   

[jtu]

The quick and dirty o/c checking is very simple: decreasing a current limiting resistor, e.g. by 50%.

Thanks for the advice!

I tried it today.

First, I made a test rig, see pic

Second, I tried to play with resitances.

But I must say that the meter had "a very good day" today. It worked without problem for couple hours before I started on it. I got Fail 1 only three times today in total around 5h time. First, before I attached resitances, I poked RX and TX lines on both sides of the optocouplers with an oscilloscope proble. Fail 1 happened when I touched pin 3 IC519, diode part of the optocoupler responsible for the direction earthy logic -> floating analogue. Later, out of the blue, no action from me at that moment. That was in between and after my play with test resistances.

No action to watch, no conlusions for today.

Just a feeling that I should replace optocouplers anyway, it's cheap and simple. I'm going to replace mains filter because I don't like noisy parts which should be silent anyway, that's much more expensive.

I'm leaving it overnight measuring cheap 5V USB supply.

PS Btw, have anybody stumbled on 7150 service manual with Fail 1 explanation? I can see the explanation in the 7150+ manual, but not in 7150's. I still think it is the same, because that part of the schematic seem identical.

[jtu]

Those scope traces doesn't look like anything to me.. First, they look like they were done with AC coupling, and you can see baseline of signal moving as capacitor is charging...

Average DC component of square wave that is not 50% duty cycle is not 0V...So it looks like it has negative component and goes below 0 volts...

Second, timebase is too long, can't see edges at all... If CTR of optocouplers is marginal, you might get distorted edges, and also unstable jitter on timing of edges and shortened pulses... Timebase is such on those snapshots that you can't really see any detail..
So while voltage swing seem ok, it is far from saying that everything is OK...

Like Mickle said, he should check both Rx and Tx,  on optocouplers and after TTL buffers.
Communication could very well be TTL level RS232, it might even be possible to decode traffic...

And since somebody was already inside, check whether jumper SP1 is shorted as it should be.....

Sorry, my bad, it really was AC coupling I just got this scope, this is actually the first project with it. I'm still used to my old analogue... Much to learn.

Attaching a few more pics, DC coupling this time. Measured on pin 6 (collector) of the optocouplers. 1st channel (yellow) is floating analogue -> earthy logic direction.

But it's getting too late for me to continue today.