Counter repair : Enertec / Schlumberger 2618

[Vince]

Hi folks,

Next in line : the repair of an Enertec / Schlumberger counter, model 2618.

Old French brand(s) so I guess not so well known as the usual US big names.

8 LED digits, 120MHz, with a 1.2GHz RF input, and a big TCXO, lucky me.

It dates back from the late 70's, so all through-hole and discrete logic chips. No proprietary chips, so there is a good chance that it can actually be fixed, crossing fingers ! 

There is one "large scale" (for the day ie..)  integration circuit though, which does all (most of) the counting, and latching and multiplexing of the 7 segments LED displays, but it's an off the shelf component and still available today, so even if this is faulty, I should be able to fix the counter.
Chip is an LS7031.

What's more...   the front panel looks digitall-y with LEDs all over the place, and switches to cycle through the various ranges and functions/modes, but it's all show : there is actually no micro in this thing, too old for that. The illusion comes from a couple state machines/counters that cycle in a round-robin fashion, that's all.


Symptoms ? 

- When powered up, all digits are off, doesn't even display a trailing '0' at the end/far right. Feeding a signal doesn't change a thing.
- The Mode and Range switches do work, if you play with them they do activate the appropriate LED indicators on the front panel.
- When I power the unit OFF, for a split second, you can see the display come to life and display all zero, as well as all the decimal points.

So, from that I think that the digital power rail is fine, and that the problem probably doesn't lie in the analog/front-end of the counter, for if the counter failed to sense a signal, AT LEAST it ought to display a trailing zero, one would think...
The LED digits themselves seem to be in working order.

Soooo, it looks like the problem lies in the core of the instrument : the counting sub-system, ie... that dedicated LS7031 chip... or maybe it's fine but some related discrete external circuitry is keeping it from operating properly. That will be my starting point...

I did check the TCXO using another counter : it's alive and spits out a stable 10MHz as expected, so at least this important bit, works, that's a good start ! 

I attached the schematics below. It's all there except for the RF daughter board sadly. But it's simple enough so if need be it could be reverse engineered.
The core of it is the usual multi-stage pre-amp made up of a few (5 here)  cascaded RF discrete transistors (BFR 90 here) followed by an ECL frequency divider. Here it's a two part job : first a ceramic package, a SAF1534 , 1.5Ghz divide by 4 chip, followed by another but slower divide by 4, an MC1697 1GHz chip, in a plastic package this time.

Google failed to find the datasheet for these two chips, too old I guess...   A gentlemen managed to find the datasheet in his collection of old printed databooks for the MC1697, and scanned it for me, but no luck with the SAF1534 sadly.
Anyway I am not too worried as I doubt the RF board has anything to do with the problem at hand...
I do understand however, that RF frequency dividers do "age" and lose performance over the years, so when come time to test the counter (if I can fix it that is...), if it fails to achieve its specified 1.2GHz limit, I will know where to look... I hope it wouldn't be too hard to find an adequate substitute for these dividers.

That's it for now. Some pics of the thing and then as usual I will post as I go...

[Vince]

Schematics here :

[voltz]

The very first thing to do, if you haven't already, is check all the power rails for correct voltages. Are they within spec? You may also want to take a look at them with a scope and see how much ripple they have. A small amount (a few milivolts) is acceptable.

If thats all good, and the TCXO is working, then follow the 10mhz from that into whatever dividers, counters there are. High frequency chips do tend to fail first generally. At what point does the signal fail? See how far that gets you.

[Vince]

Hi Voltz,

Yep, supplies first.... I am just finished going through this first step, so before I move on, here are the results.

1) Visual inspection. There are massive amounts of some residue in the vicinity of one of main filter caps, but the stuff looks weird to me, yellowish and bubbly. Nothing like your typical electrolyte damage, no sign of corrosion anywhere... and there is none of the stuff on the cap itself ! Weird. Then I looked all over the board and found a couple spot, nowhere near a big cap, where this stuff aggregated somehow on IC's pins and even managed to "climb" atop an IC socket !  Actually you can find the stuff, in smaller/thinner traces, pretty much everywhere...  So I don't know what it is but it's not leaked electrolyte I think ?!  I used liberal amounts of IPA and soft ESD safe brush, and the stuff dissolved instantly/very easily.   So if someone has ever seen this kind of stuff and know what it is...
My best assumption being : the whole board looks like it has been assemble hastily by a not so competent worker, almost as if the unit was somehow supplied a in kit form and assembled by a DIY user ! But this counter was a professional piece of gear not a toy, so I find it weird. Anyway, maybe this stuff was lots of flux residue that got never cleaned, and degenerated over the years, into this horrible substance I see today...

2) I checked all the big filter caps : 5 of them. Only one of them measures correctly at 5,100uF for 4700uF.  The other 4 measure about TWICE their nominal value ! Even for an electrolytic it's quite absurd... I have thousands of old electrolytics in my parts bin and 95% of them are within 10% of their marked value...
So clearly these 4 caps must be dead. I don't have an ESR meter just yet (it's on the cards though, of course) but I guess the ESR must be pretty bad. 
So they would need to be replaced I think.

3) Measuring DC levels : they all seem good to me.  As you can see from the schematic, we have 5 regulated (hence the 5 caps.... ) supplies. +/- 5V, +/- 12, which all use a 3 terminal linear regulator, and then their is this 4V supply which feeds only the LED displays, nothing else, and uses it's own dedicated transformer winding, and is regulated only very crudely with a Zener diode and a pass transistor.

Numbers :

+12V : 11,92V
-12V : -12,06V
+5V :  4,98V
-5,2V : -5,13V
+4V : 4,34V

So looks good to me. The +4V is 0,35V higher then nominal value but that was to be expected when you rely on the Vbe of the transistor and the accuracy of a Zener... as it turns out the Vbe is about 0,6 here, and the Zener drops 4,95 instead of 4,7.  Not great but I think still within spec for a Zener !  0,25V over 4,7V that's about 5%, I guess you can't complain. Anyway they did it on purpose obviously. This supply only drives the ELD displays so it's hardly critical.

3) Ripple:  despite the state of the filer caps, the ripple on all 5 rails is good. See example waveforms below (looked similar on all 5 rails). It's about 10mVpp at worse, using the full 100MHz BW of the scope. When turning BW limit on, 20MHz IIRC, it drops down to around 3 to 5mVpp.
So I am quite happy with that.  I would still replace all 5 filter caps I think, but before I spend money on this I will first to make sure I can actually fix the instrument !    The ripple being good, I can keep trouble-shooting the thing I think !


OK so now that the basics are covered, I can start probing around the digital stuff.
The display fails to even show a trailing zero, which it should show regardless of mode of operation, even in event counting mode where the 10MHz is irrelevant/unused. So either counter chip is bad, or it works but the display board is faulty and fails to show us the counting. so will probe around in this direction, as a starting point.

Stay tuned... 

[Vince]

Alright, worked a bit on it last night.

I concentrated first on the display board, to see if I could rule it out for good, or not.

- Checked (continuity test) all 30+ solder joints that actually, physically connect the display board to the main board, as there is no cable or connector linking the two boards, as you can see from the pictures, grrr..   All checked fine.

- Then checked 5V to the BCD decoder chip U11 and 4V to all the LED displays.   All is fine.

- Checked the decoder proper operation : 0000 at the input, and at the 7 segment outputs, it gives the appropriate pattern to be able to display a zero, so all is well there too.

- Then checked all the transistors (Q2 to Q10) and displays : grounded the base (via a 270 ohm resistor as used on the board) of all of them to see if the displays could actually show me that zero. Again all is well : each and every transistor/display combo, is able to show me a nice and bright zero digit.

So from there I now know that if the display is blank, it's entirely due to the main board/counter chip not driving the base of the transistors, and that the display board itself is fine.

The solder job on the display job is god awful just like the main board, maybe worse... see pics as an example.  Of of the pins of the decoder chip was bent and did not actually make it through the pin hole... but there is solder and it makes good contact anyway.  Then there these two resistors with horrible joints as well....   


So then I could start working on the main board (refer to schematic). Traced the 10MHz TCXO as Voltz suggested. All is well, it goes through various divider chips and multiplexers and gates, all behaves as expected, based on what I understand from the schematics anyway : first Q10 (intended as a buffer I guess ?) then to  U42 MUX chip, then to U25 divider (by 5 then by 2), then fed back to the same MUX chip, then goes to U24, a divide by N, and the scope display s a 1kHz signal at its output so it divides by 1,000 no less. Then this is fed to U12, an ECL discrete logic gate. It's differential output is not being take advantage of :inly the logic true half of the output is being used. Then the signal makes it safely to the door of U15, the counter gate. 
So is well for the TCXO reference, it snakes its way all th way to the counter, no worries at first glance.


Then I had a quick look at the counter chip (LS7031) : all supply voltages are there : +5V and  -12V.  Had a quick look at the pins of the oscillator that's used for the multiplexing, there is activity there too, though I registered only 3kHz or so which seem slow to me ?!  Well I checked only with my DMM not the scope so maybe the DMM didn't do a good job..; will check later with the scope of course.  But well, there is activity, so clearly the counter chip is not dead, not completely at least ! But it's definitely not driving the common anode of the LED digits.

So in order to start troubleshooting that counter chip, I now have to feed it with something ! So I started to look at the analog front-end, for both the A and B channels. I used a x10 probe and applied a low frequency signal so that I can probe it everywhere hopefully without disturbing the circuitry enough to keep it from operating properly. So I fed a strong (2Vpp) square wave, a few kHz, 5 or so IIRC, just enough that it is comfortable to look at on the screen when the scope is set to analog mode.  Refer to the appropriate schematic now...   This is how I understand it.  Channel A and B are similar so let's concentrate on Channel A, the top half of the sheet. From let to right :

1) A first relay K1 to select DC or AC coupling (C1 cap)

2) A second relay K4 to enable or disable the attenuator (R1 R2 voltage divider)

3) A third relay K6 to enable or disable the HF filter (33pF C7)

4) Q1 ... I am puzzled here, not in my comfort zone, it uses these old UJT "transistors"... it is "shaped"/laid out like a differential pair, but it looks strange to me...
I am thinking maybe it's some sort of mirror rather. The second UJT of the pair, on the righ hand side, then feeds the section responsible for the trigger out at the back panel. First goes through some trimmers to adjust offsets, then goes to U5, an op-map providing some amplification before it finally goes to the BNC connector at the back.

5) Q2/Q3.  Another pair... "inside" the first one ?! Looks wird but why not..  Now this looks more like your traditional diff pair, using bipolar trannies and loads in the collectors, and taping the collectors to provide the output.   So this must provide the amplification when signals are really low.

6) CR3 and CR3 must be clamping the output to avoid damaging the inputs of the logic chips that follow.

7) Then the signal, still analog at this point, goes through U1, a cascade of 3 ECL buffers that shape the signal and give it sharp transitions times so it can then be processed by the digital part of the counter circuitry.

Digital processing starts from here. first we have a couple NOR gates that let you select between positive and negative edge

9) Then another couple NOR gates, still from the same U2 chip, that let you select whether you want to feed the counter with the analog input you have just amplified and shaped... or if you would rather switch to the RF input.



OK so once I got that figured, I could then feed a signal to the inputs and probe along the signal path, step by step, to make sure it would find its way to the counter chip LS7031.

Result ? Well, the signal didn't get very far, on either channel !  So looks like this machine doesn't have only a digital problem on the counter side... it also has problems on the front end, pfff.... how sad...though OTOH it will make for a longer / more interesting repair !!! 

- Channel B : the poor signal goes nowhere : one of the relays can't remember which one today, is unresponsive... signal gets lost very early.

- Channel A : all three relays work fine and the signal can therefore get to the input of Q1.  However... at this point, my square signal is quite distorted, kinda like a probe compensation signal on a scope, would look like when the probe is way under compensated.  So I don't know what's happening, is Q1 defective and "loading" the input signal ?  If I turn on the input filter, it make things worse, signal now then looks like pretty much a triangle. Well you know what I mean, still RC exponential in nature, but overall looking close to a triangle...  I was not expecting that much at 5kHz and a filter using a 33pF cap ?!    So again not too sure what's going on here. Either way the end result is that there is no signal anymore when I probe at the input of the first ECL buffer/ diff pair output ! 

Well these were my very quick observation from last night.  I will do it all again and took my time and better document it, to make sure I am not chasing red herring as always do.  I spent most of my time on the display board, the front end I did kinda in a hurry at the very end, so I might have screwed up here and there.

However, if things are confirmed, then I am screwed because I have no way of feeding in a signal to the counter chip, with both input channels out of service ! And I don't have an RF generator so I can't use that input either.
I am really not at ease with this UJT things, so I need a way bypass this so I can feed the counter and fix the problem there first. Then once it's counting/working, I can scroll back and sort out the analog stuff in the front end.

So my plan/ideas to get round this wold be, probably :

- Channel B : bypass the faulty relay so the in put signal can get to the amplifier, hoping that this one does work...

- Channel A : bypass the amplifier and inject a signal directly into the digital chain. can't inject it in the buffer cascade since these ECL chips have symmetrical inputs/output and my generator obviously doesn't have that. So need to inject it to some single ended input. So that would be somewhere inside that group of 4 NOR gates, U2, that we discussed earlier. Test point  TP1 looks good ? It goes to a single ended input, ECL of course. not familiar with ECL so I am bit nervous about it ....   I probed these gates to figure out voltage levels. The two logic states for these ECL chips appear to be about -0,.8V and -1.6V.  That's easy enough to generate. I guess I would rather not use my function generator for this ? TI's an old 1MHz analog, very basic unit. The rise times on this thing are not that great, I fear it might upset the ECL inputs.  I think I would rather use my pulse generator. It uses ECL chips itself, and rise/fall times are pretty fast, 5ns minimum, should do it ? Actually I might reduce that a little bit (it can adjust rise times precisely) to avoid ringing ? I mean, these ECL apparently need proper input and output termination... I would use an ultra signal as I did previously, a few kHz, so I hope such a low frequency will keep me out of trouble. However the fast rise times might still cause problem no ? Will see... will adjust accordingly, based on the ringing I will observe on the scope...

So the question is, what value of a resistor might I use to inject the signal ? I mean, the injected signal will collide with the ECL ouput of the preceding chip !  I don't know how the output stage of an ECL chip is made, don't know how one might go about protecting such an ouput when you apply a signal to it ?

From what we see on the schematic, all ECL input and output pins of these NOR gates are terminated with a 1200ohms resistor, and half that when two outputs are tied together, which make sense...

Any help welcome on how this sensitive matter, before I blow this ECL NOR chip !   .........

- And last option : disregard the two input completely and instead set the counter to "timer" mode : on the rear panel there is a BNC giving access to the gate, either to probe it or to drive it externally. I could just drive it externally to provide a time interval (just a super low frequency square wave from my old sig gen), then the internal 10MHz, which we know does work, would just increment the counter !   Yeah might try that.... might be the safest way to go about it, as it does not involve messing with the guts of the front-end, no need to inject a signal into the ECL stuff.. and maybe damage an output if I do it wrong


OK so I know what to do now..... but no this evening, spent too much time typing this message, not much time left to play with the counter ! 

Anyway, as always, any question/comments/suggestions/help welcome 

[Vince]

Worked on it last night. I focused on the analog front end, this time I took my time and probe time and time again to make sure I was not chasing red herrings. Glad I did. Turns out the amplifier on channel A is just fine... problem turned out to be my coax cable that's developed an intermittent contact... 40 year old or so, I excuse him...   Was my only cable, need to get some decent brand new cables and accompanying accessories (Tee connectors, 50 ohms loads and through terminators, the usual...).
I traced the signal all around the front end schematic, all is well. See the annotated schematics and waveforms below.  I fed a 2Vpp square signal, make its way to successfully to test point 'A', left side of the 1Mohm resistor, at which point the signal is still intact. Then at 'B', on the other side of the resistor, right at the input of the amp, signal is kinda rounded. Switching the filter on doesn't make it significantly worse, doesn't make any change at this 1kHz frequency.
I was a bit worried about this but I get strictly identical waveforms on the other channel. so I kept probing. At 'C', the output of the amp/input of the first ECL buffer/shaping circuit, I do get my signal, back to a nice square, with voltage levels around ECL levels but not quite. Only 300mVpp on the non-inverting ouput, and a bit more, 500mV on the inverting input (represented below). But at the ouput oif the this bfirst ECL driver we do have nice ECL levels, hence about a 1Vpp signal. Then it traverses the otrher two ECL drivers just fine. Then goes to the 2 NOR gates that select the edge type, that works fine. Then goes to the 2 remaining NOR gates of that chip, to let you select between the LF and RF input, that works fine too. And that's it, it then goes to the main board, on the other schematic sheet.

About channel B, I was right though : there is indeed a problem with one of the 3 relays involved. It's the one that selects the attenuation.  The relay won't click anymore, however no big deal since the signal actually goes through the relay just fine, and unhurt so to speak. The relay is stuck in the  x1 position, so I don't mind for now. I will investigate this later, maybe the relay itself is just fine and it's just the switch or wiring, who knows... if I am lucky.

So that's one good thing done : power supply checked OK, display board checked OK, and now the front-end checked OK as well.
So now only digital board is left. I have already check the 10MHz reference from the TCXO, works just fine and finds its way through the various frequency dividers and muxes.  So what I did was apply a signal to channel and try to trace it. As you can see the schematic is a see of discrete logic chips, all over the place, not so easy to follow, I have seen better looking schematics for sure...
Anyway, turns out the input signal actually is represented twice on this sheet, see the green starting point. The green lines represent what part of the path I have managed to trace the signal to.
The main counter gate is circled in purple. both green paths converge to it, each snaking its own way. I guess the two paths are necessary depending on whether you want to measure frequency, or period/time.

As you can see I have not yet managed to figure out how to get the upper left path to get to the counter gate, but I get the feeling it probably works just fine its just me not understanding in enough detail how this whole mess/soup of logic and modes of operation, work exactly.

So I am not overly worried about it. so I then switch to the last part of the board, the actual counter chip, that LS7031. Probed all 40 pins, helped with the datasheet (attached) to help me understand what each of them was supposed to do.  Seems to confirm what I thought from the start : it's faulty I think ! 
All eight digit strobe lines, and all four BCD outputs are stuck low no matter what.  From what I understand, the only thing that could potentially make this normal, would be if the counter was blanking leading zeros... then again I think it ought to display at least the LSD ! Which it does not.  Regardless, even if it did blank all 8 digits, then in this case it is supposed to signal this fact by asserting one of its outputs, which is devoted to this purpose. Said line is active low, and it is stuck high in my case, ie the chip is telling us its NOT blanking anything...
Even if the chip was not receiving any pulses, was not counting anything, it should still display a trailing zero.

So looks to me like this chip is indeed not in great shape...
I probe its internal  oscillator, used for scanning/multiplexing the display. I DO get something (see screen capture below), a 3+ kHz clean square wave. So my DMM was right then, it was measuring 3kHz too.  I was wrong ti be suspicious about it. Not sure why the datasheet states that this frequency can go up to 500kHz !!!  If 3kHz does it, why would one want to go half a MHz ?!.. would only generate more power consumption and more switching noise/EMi problem I guess !   Still, the oscillator working doesn't mean much I think. The thing would probably still work even if the chip was completely dead, as long as power supplies were present !

So, I shall be ordering one of those chips. There long obsolete of course, but found one on Ebay, one on Aliexpress and one on banggood IIRC. Prices vary wildly from 15 to 50 bucks...
Yeah, no choice but getting them from dubious sources then, but what else can I do... I will just cross fingers ! 

While its in the mail, which could take a month or more from China as always, I will fix the relay problem on Channel B, and try to probe the digital board some more to try and learn more about it.  Will also give me time to order caps to replace the 5 big filter caps from the power supply section. will also use that time to order some BNC cables and accessories.

That's it for now...

[Vince]

OK worked some more on it, probing the digital board again to get a better understanding and try and check as many things as possible.

Previously I followed the signal path for the input signal, from the channel A to the input of the counter gate. That was quite straightforward. That's highlighted in blue in the block diagram and digital board schematic below.

I was having much more trouble understanding and following the signal path for the other input of the counter gate (highlighted in green) : the gate signal per se.   Spent some time on it, block diagram helped a lot, lots of probing too...  I know where my confusion came from in the end, and in the process it cleared another problem I had :understand what the hell this "sampling" circuitry/feature was for. Don'"t laugh, never had a counter before... I am discovering this thing...

My problem was that I had assumed, not knowing any better, that if the gate opening time was say 10ms, that signal would be applied straight to the counter gate, and the counter would therefore be able to measure the input signal every 10ms, hence 100 readings per second.

But after a while probing around with the scope and the playing with the sampling pot at the back of the unit at the same time, I eventually realized that, well you know of course... that sampling rate is called like this for a reason ! LOL  It sets the speed at which the counter takes a measurement. No matter how narrow the gate opening is, the counter will never take benefit of it to take more measurements...  so when using the printer/recorder port at the back of the unit to log data... you will be stuck to the ultra slow speed dictated by the sampling circuitry, which is about 0.5s at the fastest,  up to 7 seconds or so at the slowest. I checked the user manual for my other counters, seems  they all revolve around that figure.  roughly from 0.1s at best, up to 10s or so for the slowest speed.

Now why would one want to slow the update rate of the display I am not sure, ideas ? 0.1s is only 10 readings per second, can't see that being a problem.

There is an external gate trigger input, maybe this one allows to bypass the sampling circuitry (highlighted in yellow) to get measurements as fast as the gate opening time allows, would have to look at the schematics in more detail.

So at the counter gate I had my input signal on one input, and  "bursts" of the sampling signal applied to the other input, rather than the sampling signal itself.
Then I checked the output of the gate (highlighted in red),  pulses were coming out, great. Folowed the path through U16 then U21, pulses still there. Then we are U30 and U33, the two external LSD counters. The LSI counter chip / LS7031 is not fast enough to go up to 120MHz, so it has only 6 digits/counters in it. The two least significant/faster counters must be supplied externally. Pulses are coming out of U30/U33.... great.

So the main counter chip does have pulses coming in, power supplies are there, the sampling circuitry is providing the reset pulses, and the built-in oscillator to scan the LED displays does run. So this counter chip has everything it needs... I am getting now quite confident that it's the culprit....
Obviously there is still a slim chance that the chip is fine and some less obvious problem in its vicinity keeps it from operating properly, who knows, I might need to do further probing/investigating for less obvious things. But after all the testing done so far, I can safely say that 99.5% of the counter is working, all the essential bits. If the counter chip is no the problem then it can only be something very minor, some glue circuitry somewhere around the chip. TO start with, the chip is not directly socketed to the main board 40pin DIP socket as you might expect. Instead, as you can see from the pictures at the beginning of the page, the socket holds a little daughter board which carries the counter chip of course, but also 4 logic chips which do god knows what. I don't have the schematic for this little board sadly. So it's still a possibility that something is going wrong on this little board, I guess.

Anyway, I have already ordered a replacement LS7031, seller shipped it a couple days ago, now waiting... but as usual it's coming from China/Taiwan, so I will have to be quite patient... a couple weeks at best, 4 or 5 if I am not so lucky, so stay tuned.

Quite impatient to receive it and plug it in !!! 

[Vince]

Spent some more time on the thing the other day, again probing arouind to get a better understanding of the thing while waiting for the replacement counter chip, units way from china.

Started with the sampling rate circuitry, after scratching the surface previously.

I highlighted that section of the circuit, in blue.

It's not very complicated... First there is U23, a 555 timer wired as an oscillator. Its RC constant is adjsuted by the sampling rate pot at the back of the unit, as you would imagine...
It can produce a square of a period ranging from about 0.5s to 7s.
This periodic signal then feeds U31a , half of a 4528, a dual monostable chip. I guess the purpose of this is to produce calibrated pulses of a constant and well defined width, regardless of what the sampling rate is set to, or if you set the pot all the way to "manual" trigger, which you can then trigger single shot either by pressing the RESET push button on the front panel, or by sending some pulse to the external trigger input BNC at the back of the unit.
No matter how you trigger the measurement, it will always produce a well calibrated pulse. Looked at it on the scope of course. Can't remember exactly but it was something like 10 or 20us. That is, plenty enough to reliably reset the counter(s) and latch the data, yet way less than the fastest gate setting (10ms on this unit), so that there is no overlap between two reset sequences. Well that's my understanding anyway.
All this circuitry works just fine, no worries.


That being done, I could turn my attention as planned, on the daughter board that hold the counter chip, to look at it into a bit more detail. I now know that pulses are coming to it, as well as the reset signal, and that the supply voltages, +5V and -12V, reach it to.  So I wanted to check a few more little things about it, to be even more sure that the surrounding circuitry was fine, that it was indeed the counter chip being faulty.

First I wondered about the 3,2kHz signal I saw no it oscillator pin (used for scanning the display). It was the only thing that indicated that the chip was not totally dead..... or is it ? The datasheet of that LS7031 chip says that the scan signal can be set either with a timing capacitor... or be fed externally. So I wanted to see if the last breath of life I saw in this chip was real or not.... and it's not ! That chip is dead as a dodo ! I unplugged the chip from it's socket, and the 3.2kHz signal was STILL there to be measured on the socket pin ! So... this signal was hardly saying anything about the chip's condition. I guess I should have guessed this sooner : if there were a timing cap, it would have to be right next to the counter chip, on the daughter board, not on the main board, I would assume. And there is not a single cap on that daughter board so....
Anyway, I then probed around the circuitry on the main board, that's responsible for generating this 3kHz signal for the counter chip. I highlighted it in red on the schematic.  Again it uses this 4528 monostable chip, another one that is, U41. Both halves of it are used this time, no waste.
As you can see, it all starts with U41a, which is wired in an interesting/unconventional way.. nothing that's suggested in the app notes of the data sheet at any rate !  I mean, this monostable is actually wired to behave... like an astable not a monostable ! LOL  Dear. It oscillates and produces the base 3.2kHz signal. The way they did it was to disable the clock input, its tied permanently to Vcc. Then they trigger the chip using one of the enable pins rather than the clock pin. That enable pin is driven by the negated output, which feeds and RC circuit (Waveform #1).  The positive output (Waveform #2) then feed the other monostable of the chip, U41b. This one is used as a monostable, what its supposed to be, so they again disable the clock input and prefer to used the an enable pin to trigger it, whatever works...  It produces 70us wide pulses (Waveform #3) which then go to transistor Q12 which does voltage level translation : the signal now swings from -12 to +5V rather than 0/++5V. That's Waveform #4.

These 70us pulses also feed another input of the counter chip, the INHibit input, via U40 a dual D flip-flop, then a NAND gate, U37. I checked that gate, works fine. checked the flip-flop too, its output is quiet of course for now, because it's data input is fed by on of the outputs of the counter chip... and since its dead, well not much is happening. But I am not too worried about this flip-flop. There a re two of them in that package, and the other one, used in a totally different part of the main board, works just fine. So I would have to be very unlucky that only one the two flip-flops in it, is faulty !

So all controls and data inputs signals are present on the socket in which the daughter board is plugged. Also probed directly on the counter chip pins, all signals find their way where they should. Confidence is hight then... that my new chip when it arrives, will fix this counter.

Then only thing left to do was to have a closer look at the daughter board. As you can see it comprises of the counter chip of course, plus 4 more chips (datasheets attached), they are :

- one  N8T97N : a hex buffer
- one MC14049 : another hex buffer
- two MC14016 :  quad bidirectional 2 to 1 multiplexers.

Looked briefly at the traces on the PCB. Looks like these buffers are tied to the 8 strobe lines and 4 BCD outputs.
8+2 = 12... 2 hex buffers.... all adds up. Only 8 switches though, so I guess they handle only the 8 strobe lines and that BCD outputs don't need to.

So my guess it that the multiplexed outputs of the counter, are being switched between to either feed the display on the front panel, or... or the printer port at the back of the unit, for the daughter board (in red on the picture) is sitting right along the connectors in which the optional printer interface board plugs into (in green). The printer port has to provides individual BCD output for all 8 digits. On my older nixie counters, this is easy, the printer port just taps directly the BCD outputs of the counters... but on this more modern counter with a fancy "do it all" LSI chip that multiplexes the display.... one then needs to undo this multiplexing before it can be sent to the printer port !  So I guess that's the main purpose of this optional board, which I don't have and for which I don't have the schematic either.

Oh, and there was one last thing that fooled me into thinking there was any life left in that counter chip : the fact that when you power the unit off, for a split second all 8 digits come to life and display all zeros.  So, I unplugged the daughter board.. no more counter chip in the unit. Powered it off.... and I still see the digits briefly light up ! So, it had nothing to do with the counter chip and this thing is most likely as dead as it can be.

So that's it... spent a full week probing around all around the unit : power supply, analog front end, front panel/display board, and each and every section of the digital/main board. Was nice, I now have a much better understanding of the unit, and am quite confident that a new counter chip is all it needs to come back to life. Can't wait to receive it !   

Enough rambling for today 

[Vince]

Hi there,

Worked a bit more on the thing last night. Still waiting for the LS7031 chip so I though I would do something useful in the meantime... recapped the power supply section.  First time I do this. I feared it would b next to impossible to find replacement caps of the appropriate dimensions and taht it would cost a fortune, them being big and numerous but... not at all.  I usually order from Farnell, the only big sellers with operations in manland europe I knew of so far, which bothers doing business with private individuals. Yeah I sometimes wish I lived inthe US and could cheaply order anything from Mouser or Digikey.
Someone recently told me about " TME " another big retailer in Europe I had never heard of. Good prices and a plentiful offering. I could easily find exactly what was needed. 

- Decent brand (Nichicon)
- 105°C rating
- same capacitance and voltage rating
- Low ESR

Actually having 105 Low ESR Nichicon was already christmas to me, but I was even given the choice between 2 Low ESR series of products form Nichicon : 'PS' and 'PW' series.
https://www.tme.eu/fr/katalog/condensat-electrol-tht-a-faible-imp_100264/#id_category=100264&s_field=niski_prog&s_order=ASC&visible_params=2%2C2%2C10%2C31%2C32%2C39%2C42%2C115%2C118%2C118%2C120%2C120%2C130&used_params=2%3A106660%3B118%3A25369%3B120%3A24868%3B

Similar price. Compared their respective  datasheets, the PW is even lower ESR than the PS series, so went for the PW series. And even within the PW series, I had the choice between two form factor.  So, for all the fellow living in Europe, if like me you didn't know about TME, well worth adding to your address book.
No minimum order, just a flat 7,90 Euros shipping fee. Very fast too : ordered sunday night, was delivered to my door Tuesday 2PM ! Thei warehouse is located in eastern Europe in Poland. Helps keep the prices low I guess.

Anyway, replaced all that : 5 caps in all, all 25Volts rated.   two 2,200 uF and three 4,700uF caps.

As far as footprint, these modern caps are somehow literally twice smaller than their older counterparts, despite being exactly the same value and voltage rating. so I had no trouble fitting them to the board : even though they don't have a center pin like the old ones do, hence were off centered, they still fitted fine because their diameter is way smaller. So all in all they only marginally exceed the original foot print, as can be seen in the pictures below.
another lucky thing is the terminal pitch : they are the same !  So no bending required, the new caps sit perfectly flat on the PCB, nice and secure.

I measured them all before fitting them, they all read real close to the nominal value. From memory they were all slightly under the nominal value, by no more than 50uF worse case. So something like for example 4675uF for 4700. Can't expect better than this for something that's supposed to have extremely bad tolerances...

Had a brief scary moment though : after replacing all the caps, I checked DC levels for all 5 rails. They were all spot on but for the -12V rail which somehow read positive 1V ! Eh ?!   Had to be something easy though, related to the cap replacement. Turned out to be a wiring problem : the old caps had 3 ground terminals tapped from the can. It just so happened that for in the particular case of this -12V rail, the designers actually relied on the presence of all 3 terminals, when they routed the PCB : two of these terminals were actually used as a "jumper". So me not having soldered these terminal, made an open circuit. Basically it cut the link between two ground points : ground at the input and output of that -12V regulator. So I just soldered a piece of wire on the underside of the PCB to restore the link that used to be there, and they presto, back to normal.

Then I checked the ripple on all 5 rails with the scope. Not much change there, same kind of waveform and amplitude, only a tad lower. 2mVpp for the best of them, 5mVpp for the worst. So I am happy man I guess.

Then to finish it off, never did it so could't resist the temptation : I autopsied one of the caps. I expected to have nasty electrolyte dripping on to the bench, but no. No such thing. Was not dry either. I unrolled the foil.  Looks like it's a paper insulating foil impregnated with whatever electrolyte stuff. Paper is hardly soaking wet, hardly dry either. It was just "damp" to the touch. It's made of two individuals metal+ paper strips, superposed. Each about 4 feet long.
I keeps the other 4 caps intact, I want to measure their ESR for educational purposes.. once I manage to get an ESR meter, fairly soon I think, it's on my short list of affordable but useful stuff since I like to repair/work on all gear like this.

At least I now have the datasheet for the Nichicon caps (attached), so that gives me a huge table of ESR values for all possible combinations of capacitance and /voltage rating. A good reference point I guess. Obviously old caps will have high ESR, but still, it's valuable reference material.

Now I just need to replace the power cord, as the plug is badly designed : once plugged into the socket, it gives you zero grip to pull it out !

Anyway, that's it for today... just waiting tor the LS7031 now...

[Zucca]

Vince, fantastic!

Keep going I am enjoying your repair a lot, don't be sad if nobody replies we are watching your work with pop corn and coke..

 

[Vince]

Hi Zucca,

Nothing wrong with a bit of pop-corn ! LOL

I am not sad at all ! ;-)

The beginners that might be interested in seeing the repair are like most of the time, lurking silently in the background (but page and picture views attest that they are there!) .. and as for the more experienced that might post to help me with my repair... do not post because well, I have not yet asked for any help ! LOL
Since this particular repair happens to go smoothly, so far I have managed to figure it out by myself... however when I was in dire need of help with my two Tek scope repairs, the 2232 and TDS 544A, the experts were there to help !   

[Vince]

Well, success, the thing is fixed ! 

Received the replacement LS7031 counter chip today ! Boy that was quick, barely 2 weeks, usually takes a full month or more, whenever I order stuff from China !
I got lucky then 

I first replaced the power cord as mentioned earlier. Makes a world of a difference, this thing is now user friendly.  The plastic grommet securing the cable was a real PITA to get out and even more so getting it back into place ! Boy... I wonder how they manage to fit them at the factory.. maybe some special little tool to make their life easier...

Then put the new chip into place. Old one was dated early 1980, "new" one/old stock is dated mid 1995, 15 years younger, and brand new, so hopefully I will die before it does.

Did a quick sanity check using the "test" mode of the instrument, which basically sets the instrument as a Frequency counter and loops the TCXO reference to the input, measuring itself. It's therefore supposed to display an accurate and stable 10,000,000 on the screen, which it does. yeahhhhhh.... looking good !  So I put the cover back on and did some more testing. Used my old analog function gen to test low frequencies, from 1Hz up to 1MHz, the fastest it will go. Worked fine.
Then to test higher frequencies, I switched to using my pulse generator, as it goes up to 50MHz. Set it to the max, 50MHz, counter still there no worries. Displayed frequency as you can see, is hardly accurate, and it is also very unstable, but this I expected : nothing to do with the counter, and all to do with the pulse generator. I had already noticed this problem when I used it to test my old Nixie 160MHz counter. First guess is the PLL which has trouble getting locked, hence the frequency is wandering all over the place. That will make for another interesting future repair ! LOL  Even more "interesting", challenging I should rather say, because there is no service manual/ schematics for this thing !   

Anyway, as far as the counter goes, it looks like like it's working fine. I then fed that 50MHz to the RF / 'C' input, because well... the RF input is actually the main reason I bought this counter in the first place !  I don't have any RF generator, but I had a go with my 50MHz pulse gen. Usually, from what I can read, the RF input on counter typically requires at the very least a 100MHz signal to "prime the pump" so to speak.  but well, 50MHz is not that far off the scale, so maybe I stood a chance. Glad I tried : it DOES work ! Counter displays the same result as it does on the low freq / 'A' input, yeah ! 

Of course I would still have to test it to up to its 1,2GHz limit, because from what I can see on TheSignalPath blog, the RF frequency divider chips don't age well and lose performance over the years, they don't quite go as fast as they used to when new. Still, at least I know the RF board is alive and working, I am relieved ! 

So that's it, end of this repair ! 

Was worth it, gained some valuable experience with this new kind of device, was my first encounter with a  counter (pun intended, forgive me), a universal one at that, where trying to follow a signal path is a recipe for headaches. Was also a good introduction to the practical implementation of these fancy ECL super fast logic chips, and how to interface them to TTL level chips and also to analog signals (output of the differential pre-amp).
Was also my first power supply recap. Also my first device with an RF board in it, a whole new world of electronics in its own right, yet it somehow looked instantly familiar to me, thanks to the many great videos I watched on TheSignalPath blog, most notably the one where he describes the RF board  kit he fitted to upgrade his modern HP counter.  The construction of the RF board in my old counter is extremely similar... looks like there is not really a million ways to tackle this particular problem...

So, I am happy. I bought this counter, for repair, for  only 50 Euros  (+ 18 Euros shipping), and fixed it  for the price of that LS7031 (around 15 Euros shipped).  So for 80 Euros or so, I have an 8 digit counter with a 1.2GHz RF input, which I can now use as a valuable stepping stone/lever to help me some time later, when money permits, to fix some old broken spectrum analyzer. I will be able to check the many LO and IF signals that populate these complex instruments, verify that I get the correct frequencies at all points, and with 8 digits I will also have enough resolution to be able to tell whether a given PLL is properly locked or not. So, a useful tool I think.

Still need to check the counter at high frequencies obviously, to see how far it can actually go / if I need to replace the divider chips. To do that, I am thinking of getting one these cheap programmable PLL boards I came across recently. There might be others that are better I don't know, but this one has may fancy at the moment :

http://www.ebay.fr/itm/35M-4-4GHz-PLL-RF-Synthetiseur-de-Frequence-de-Signal-ADF4351-Development-Board/182662863864?hash=item2a878e1bf8:g:QHMAAOSwXY5ZQjOB

... because it costs only 20 bucks or so, delivered, is ready to go, goes up to 4,5GHz so plenty enough for my purpose and future needs, yet it can go down to very low frequencies as well : 35MHz. Can even be programmed to do frequency sweeping as well. So quite a versatile little gadget.
Only needs a 12V supply and a way to communicate with it via a SPI bus... which means I first need to put together a computer in the lab so I can stick my old Atmel AVR programmer to it and write some code to get that PLL board going. Not sure I can wait to get that done... I shall investigate a way to put together a quick and dirty VCO using a few discrete components and a simple pot to adjust frequency... so that it's standalone and doesn't require a computer to get going.  That would also be a good excuse to do my first RF little project/board, it's as good an excuse as any, I guess...

Well, that's it for this repair, a success. I won my gamble, I got my cheap RF enabled counter to help me later with fixing a spectrum analyzer, I am happy.

I can now switch to the next repair.... I have planned to start work on restoring and repairing that ancient 1959 Vintage vacuum tube Tektronix 317 scope I recently bought.  Stay tuned to the repair topic of the forum, for those who might be interested in that.

The other day I ordered a few bits and bobs that I know for sure I will need for this project, but again some of it is coming from china, so might take a while to get to me. Well who knows, maybe I will get lucky again and receive the stuff in way less than a month.  Said stuff is not vital though, I can start working on the scope right away actually...

Anyway, hope this thread was of some interest to some newbie lurking in the background, and if you like old tube stuff then see you soon for the Tek 317 restoration ! 

[TheUnnamedNewbie]

I'm potentially getting one of these myself. It has a broken channel C though. I had a few questions:
Do you have a service manual you are willing to share?
If not, it could be that I missed it but I can't seem to see a channel C schematic on this page...
Is the OCXO standard or an option? All pictures I've seen from the unit online seem to have an OCXO but I don't find any mention of it being standard or an option.

[Vince]

I'm potentially getting one of these myself. It has a broken channel C though. I had a few questions:
Do you have a service manual you are willing to share?
If not, it could be that I missed it but I can't seem to see a channel C schematic on this page...
Is the OCXO standard or an option? All pictures I've seen from the unit online seem to have an OCXO but I don't find any mention of it being standard or an option.

Hallo the unamedBNewbie,


- Schematics : I posted all I have, and it was hard enough finding them. So no, indeed I unfortunately don't have the schematics for the C input / RF board. However as you can see from my pics, that board is fairly simple, shouldn't take that long to reverse engineer... assuming you actually need to in your particular case. Would be interesting to see you trouble shoot your counter... maybe you can post on my thread if you want, if really it's the same model it would be relevant and interesting... 

- Time base : I have failed to find any pic of this particular counter on Google, from what I remember, so you managed better than I did ! 
Mine has a TCXO not an OCXO, so I would therefore say that the OCXO was an option !   ...  could you post a pic of this counter with an OCXO  ? Just out of curiosity...
My TCXO "looks" like an OCXO because of its large metallic case which is somewhat misleading, but it's clearly labeled as being a TCXO...
Also, the counter is only 8 digits so I don't even see the need for the high accuracy of an OCXO, that of the TCXO is adequate I think... maybe a hair on the verge, so OK, an OCXO would give you the littel bit of extra accuracy you need to make the 8th digit really reliable... but 95"of the extra accuracy of the OCXO would be wasted/unused, a bit sad.  So I think the TCXO is just fine and if one need really high accuracy, I guess it's best to use a different counter with more digits and an OCXO.


Anyway, I am very interested in seeing you trouble shoot your RF board, please don't hesitate to post photos, waveforms, diagnosing process... lets fix your puppy ! 

I assume you have a decent multimeter and scope to help you in this repair.

[TheUnnamedNewbie]

Hey, Thanks for the reply! I'm sorry for not reading better... In the middle of my exams at uni, I'm gonna just blame it on that, eyes do enough reading when I'm not wasting time taking a break browsing the forum.

I payed the unit earlier today, so I should have it in a week or so according to ebay. Item was listed as having a functional channel B, but A and C not ok. I hope I can fix C as that would be the main addition to the lab, I already have two counters (though those aren't that great) in the lab that can go to about 150 MHz. However, seeing as B works, I suspect it's gonna be a problem in the amplifiers or prescales so if need be I'm sure I'll find some kind of compatible parts that can do the same function.

In my haste earlier I thought I read OCXO, and then seeing the big case on the picture... Kinda got a little excited, thinking I managed to find a device with a OCXO for 50 bucks... Oh well, TCXO is still a nice addition. If I find time I'll post a log here!

I think I have everything I need to try and fix. Only thing would be something to measure at 1.2 GHz, but I can just injecct lower frequency signals and measure those with my scopes.

[Vince]

Ok, good luck with your exams ! Yeah if you find the time to post about your repair I would be interested 

From what you say though, I think you are wrong : it looks like the RF board/prescaler is fine. The fact that BOTH the A and C don't work, is the clue !
I mean, the RF board/prescaler sends its signal straight to the A channel (as can been seen in the schematics). So if A doesn't work, C automatically won't work either ! 

So I think the first thing you must do is FIRST fix the A channel, then only once this is done, worry about the RF board... which I bet works just fine 

This also means that the problem must lie in something that's shared among both A and C, which means that the analog front end for A, is not at fault.
Given that B works, it means that the basic functionality does work (the counter unit and time base), so your problem is probably :

1) On the digital board
2) Located in some glue logic that only pertains to A and C : that would be the upper left corner of that schematic.

Check the block diagram to get a clearer picture of the signal path.

Seems like your problem is simple.. a logic chip fried somewhere in the signal path for the A channel, should be easy to pin point.

[TheUnnamedNewbie]

Ah, I haden't checked the schematics that much yet, though I feel stupid for not realizing that it must be something shared (if dr. House has tought me anything it's that you never have two unrelated problems at the same time.... you know, except when you do). I'm happy to hear that it is not (likely) a problem in the prescaler of C because that is the only thing which I can't measure directly, at least not all the way up to the limit of the range. In addition, that also means that the two harderst-to-find-info-on-chips, which are those prescalers, are likely fine... *phew*

All right, I can't wait till I get the unit. Also orderd a big-ass Rohde & Schwarz power supply capable of 10 A... for some reason? Even though I've never needed more than a few 100 mA out of my exisitng stack of supplies I'm sure I'll find some use for it...

But I'm gonna fix this counter first, once I get it. Hopefully get it soon, and I'll keep you updated.

[TheUnnamedNewbie]

Unfortunately I still haven't been able to start the repair... I ordered the device on Monday last week, and seller still hasn't even shipped it. I'm wondering at what point I should go complain?

[Vince]

Well that's 10 days or so then.... not worth "complaining", but definitely worth getting in touch with the seller to get an update... no need to insult him, just ask him politely what the status is ...

[TheUnnamedNewbie]

Finally got it in the mail today. Turns out it is broken differently from how the seller said it was broken. Channel A and B are broken, C works fine. However, B can give me right "interval" timings sometimes.

This is great, because it means that even if I can't fix it, it is a usefull addition, since channel C was the main thing for me. I have a few counters that go up to 200 MHz already.

Will keep updating as I progress through the repair, though now I have a different thign I will repair first (I also got a Rohde & Schwarz powersupply that is broken)

[Vince]

Glad to see that C is working since that's all you really needed it seems.

If C works but not A then, since C feeds the digital portion of the A channel... that means that the counter/instrument as such works just fine, and that only the analog front-end for A is at fault, shouldn't be too difficult to fix. I went through it in my repair so just read my previous posts to get you started (if you decide  to fix it that is...).

For channel B it's less obvious though. You say it's "broken", but works for interval timing though... but only sometimes.  More testing is required to get a better understanding of what it is exactly that works and does not...
The relays in the analog front-end might have dodgy contacts and cause the erratic/unreliable behavior, for example. There are no less than 3 relays in the signal path, before it reaches the amplifier stage.
Also, Channel B is not used to measure frequency to begin with, it's only meant as a "stop" signal, to time intervals. Only channel A(C) is used to measure frequencies.

Obviously as always, first things first : a thorough visual inspection for damage to the PCB or ICs. corrosion, solder joints, signs of overheated chips, whatever might catch your eye... concentrating on the analog front-end since it looks like you problems lie there. Then check all power supply rails for accuracy and ripple. Then only, can you start trouble-shooting anything 

[TheUnnamedNewbie]

I opened up mine today to start the repair. I noticed that where you have a external PCB for the main IC, mine has a ceramic chip in a socket there:

[Vince]

Yeah no worries, it's just fine. I looked at that board and I think it's only purpose is to multiplex the display/counter contents with the "recorder" feature.... which is an optional card. My instrument did not have that recorder board, but strangely had the riser PCB anyway for the counter chip.. go figure.

Anyway, it will only make your life simpler, because the schematics actually do NOT take into account this riser board (so made MY life harder ;-).   So basically what you have in your instrument, corresponds exactly to what's on the schematics... that's a good thing !   

I a bit surprised at the part numbered that your counter chip bears, though : a rather obscure one !: Even Google can't find anything about it !
I am not worried though, it will just be a drop-in replacement for the LS7031, so just use that datasheet. However we already know that this chip works just fine in your instrument, so you don't have to work on it anyhow...