Datron 1041 vintage multimeter

[Gyro]

An excellent learning opportunity for the kids  . A nice diversion from the single chip world of Arduinos etc. Having all the ICs socketed is a real bonus.

Hopefully the arrival of the new HC193 will bring a smile to their faces and yours! 

The end result will be worth the effort, a nice high impedance meter with lovely 4 1/2 digit display, excellent accuracy, stable 1uV resolution and responsive 10 rdgs/sec (see the timing diagrams at the end of the user manual). "Vintage" in all the right ways!

[Zeitkind]

An excellent learning opportunity for the kids  . A nice diversion from the single chip world of Arduinos etc.

Well, doing "old school" electronics is harder than when I was young. We used to grab old broken stuff from the garbage, took them apart and build other stuff out of the parts we found. My first stereo amp was totally made of parts I salvaged from old TVs and radios. Of course it was horrible. But the speakers I had were even worse..
Nowadays - when you open a device you prob. find a switching something for power and a second PCB with a single chip solution.

Hopefully the arrival of the new HC193 will bring a smile to their faces and yours! 

Kids need a feeling of success when doing such stuff, so I really hope we get it fixed...

[Zeitkind]

Step by step..

The replacement 74HC193 arrived and we changed the chip today. Improvements: Range switches work - kind of.
The output of 74HC193 is now as expected and the range switches cover 001 to 111 (ABC). But still the meter doesn't really work:

10mV range: 1,23mV is shown as .1230 V  ;wrong unit and wrong decimal point
100mV range: 12,34mV is shown as .0123V, sometimes as mV ; wrong unit and wrong decimal point
1V range: seems fine, 1.234V is shown as 1,2340V
10V range: 1,2V is shown as .0120V ;wrong decimal point
100V range: same as 10V ;wrong decimal point and prob. wrong range
1kV range: same as 1V-range! ; weird
In Auto mode, the ranges are shown as in manual mode, so Auto increases and decreases the ranges somehow, but in 10V it's going nuts turning the relay on and off.
http://www.mein-zeugs.de/pub/VID_Datron_1_288p.mp4
mp4/H.264 file, Firefox doesn't like them much, use Opera, Chrome, IE or Safari.

So I now have a working 1V-range and the other ranges show correct values but with wrong decimal point. Waiting for new NAND gates (M4 & M7), to lazy to test them somehow on a breadboard. We tested the old 74C193 on a breadboard (with some LED's and switches) and all 3 outputs are indeed dead, so prob. another chip down the output line is bad too.

[Gyro]

Doh!

Still, it's progress.

The Decimal point decoding is done by M6 (74C42), followed by the M5 inverter package and switching transistors. I suspect M6, but worth adding both to your shopping list. M6's outputs feed back into the autorange gating which could also explain the 'hunting' behavior. There are so few ICs in that section of the board that I'd probably look at getting replacements for all of them at this stage.

It's encouraging that the actual measurement function seems to be working though and apparently in spec.


P.S. Curious that it seems to be multiple ICs on that board - they are the only 74C series ICs in the unit, everything else is TTL. It sounds like there may be some dodgy aging mechanism with 74C.   I've re-used lots of older 4000 series CMOS ICs and never had a failure with those.

[eblc1388]

The integrity of IC M6(74C42) and M5(14069) can be easily verified on a breadboard.

74C42 is a BCD to Decimal decoder and M5 is an HEX inverter. Should be a fun and good exercise for your boys in understanding logic function.

[Zeitkind]

Was thinking about that. A little fun project over the next months is a LED lamp (or better a LED-wall..) with a bunch of 74HC595 and a sound input to control the "LED snake" crawling forward. No idea yet, but I prob. add some more IC's like 4011 and others to that lamp to get some "effects" or several ways to control the "LED snake". We will try to build one lamp for each, but that is not decided yet.
Chips like the 74595 are great for such projects and are easy to understand. Logic stuff like NAND and XOR.. well.. we'll see.. ^^

[Zeitkind]

Finally. We replaced the two 74C00, the 74C10 and the 74C42 today with new (NOS I guess..) HC series. Everything is working now , DC is in spec (9.999V and 2.501V on 10,000V and 2.500V) and AC is working too. Can't say if AC is in spec, I don't have any reference for AC here, but it's fine with my other meters, so I think it's also in spec. 
Not sure if both NAND gates are bad, I just distrust all of them and replaced all IC's on the DC board with new HC-versions. I'm a little bit disappointed that there is no Ohms-boards in this meter, I will look for one, but they are for sure hard to find.. Perhaps I will find a broken meter with a working Ohms-boards..
Many thanks to all!!

Repair history: Replaced AC mains filter, replaced all tantal caps with new tantals (with higher voltage rating) or electrolytics, replaced all 74Cxx on the DC board.

[Gyro]

Hey, congratulations to you and the kids. 

I think you're right, those 74C parts are only good for the bin! Useful to know that it does all work properly with HC.

You have a nice meter there, easy to maintain, accurate, and no worries about EEPROMS, backup batteries etc. Yes, it's a shame you don't have the Ohms board, especially as it does 4 wire guarded measurement.  There's little chance of finding one on its own, but you might want to look for a broken parts donor. You can also use one from a 1055 1051, it is electrically identical, just using closer tolerance parts. You could maybe try to build one from the schematic I suppose, but your time would probably be better spent making an accurate external current source instead.

The 10mV range is very nice for showing the kids thermocouple effects in the uV region and for use with low drop current shunts. Note that different meters can take different times to settle to 0uV after power up with the input shorted (the spec requires 2 hours warm-up to meet full performance).

There are a couple of quick mods that can make it much nicer to use.

1. The horrible blank display on overload:
This is easy to change. On the A/D board, connect a diode (ought to be schottky, but 1n4148 works fine) between M15 pin 8 (cathode) and M9 pin 6 (anode), no track cuts needed. The output of inverter M9 is open-collector so the mod doesn't cause contention. This will make the display show '+/- leading 1' on overload (the same behaviour as most DMMs).  An easy mod to do on the bottom side of the board.


2. 10 Reading per second display (this also makes autoranging performance very fast):
As shown on the timing diagram at the end of the instruction manual (Fig 5.4, Closed loop timing) it is possible for the meter to run closed-loop at its full speed rather than being internally triggered at 2 1/2 reading/sec. To do this:
/HOLD needs to be held low (to disable internal triggering)
/DELAY needs to be held low (this sets the conversion rate to match the Filter setting)
/PRINT COMMAND needs to be connected to SIGNAL INTEGRATE (to cause immediate start of a new conversion as soon as the previous one is finished).

All of these signals are available on the bottom side of the unused edge connector J4, where digital output option boards would fit. You can implement this on an offcut of 0.1" stripboard and carefully plugged in. Please double check these, but /HOLD is on 6L (Lower), /DELAY on 2L, /COMMAND on 5L, and SIG on 3L. GND is on 7L.

This mod can be a little temperamental on the 1041 (works every time on the 1045), it can sometimes freeze on startup (needing power cycling) or if the Hold button gets pushed. I suspect that there is a marginal timing in the logic somewhere, but if it doesn't work for you then you can simply unplug the stripboard (or investigate further, the presence and value of C49 seems partly responsible).

J4 is a very useful connector. It gives you access to all output data in parallel BCD form, together with range, polarity, status etc. It would be very easy to put a proto board in there (needs double sided access to the connector fingers) and with some logic and an Arduino or PIC, convert the data to serial or USB serial for data logging experements. If you also connect a header to J4 on the DC Isolator board (coincidence) you could also remote control the ranges, AC/DC, Filter etc. I keep meaning to do this myself.

I hope this helps. 

[eblc1388]

J4 is a very useful connector. It gives you access to all output data in parallel BCD form, together with range, polarity, status etc.

Yes, but I don't seem to be able to locate a pinout mapping for the connector J4 in the service manual.  Therefore I decided to roll my own, as in image attached.

Hope someone will find it useful.

[Gyro]

You're right, the only ways to decode the pinout of J4 (A/D board one) are to either pull out the sometimes cryptic signal connections from the schematic or work back from the 50 pin Amphenol description. I have a scrappy pencil marked-up o/p board schematic, but I've grabbed a copy of your nice neat one, Thanks.

BTW. I was playing around with the 10 rdgs/sec mod again yesterday, I suddenly realised that tying the /FP override signal (J4 L8) low disables the buttons in the 'Control' block and prevents accidental pressing of the Manual button (manual trigger not manual ranging) from locking it up. I still can't get it 100% reliable from power-up though (which it is on my 1045). Something fails to trigger the first conversion I think.

If an output board option is fitted, the 10 rdgs/sec option can all be implemented on a 50pin Amphenol plug of course and easily fitted/removed as required.


One other mod that might possibly be of interest... 'Autorange-up' happens at 100% overrange (19999 rdg.) but 'autorange-down' doesn't occur until 10% or range (1000 rdg.). For a 4 1/2 digit meter this is quite a loss of useful resolution. This behaviour can be changed by cutting the track between M31 pin 11 (down latch clk) and M20 pins 1+12, and then adding a wire to connect M31 pin 11 to M20 pin 9 instead. The track is accessible top-side next to M31, or alternatively pin 11 can be lifted from the socket.

With this mod there is no longer any hysteresis between ranges, so it may hunt if the input is hovering between 19.999 and 20.00 for instance but I haven't found this to be a problem in practice. On the rare occasion that it is, it can be overcome by going to manual ranging anyway.

[eblc1388]

'Autorange-up' happens at 100% overrange (19999 rdg.) but 'autorange-down' doesn't occur until 10% or range (1000 rdg.). For a 4 1/2 digit meter this is quite a loss of useful resolution.

You've lost me here about the 'loss of resolution'. When the meter autorange-down at 10%, isn't the signal now displays at 100% of the new range?

Does your proposed mod triggers the autorange-down operation at 20% instead? 

[Gyro]

Yes, it triggers down at 20% (rather than the original 10%). It makes more difference than you might think - in practice if you are measuring in the 10-20V range (eg. a 10V reference) you tend to get an extra digit on the display compared to the original autorange behavior (I think because the autorange normally counts down through the ranges). It makes a difference for me anyway. 

[eblc1388]

With this mod there is no longer any hysteresis between ranges, so it may hunt if the input is hovering between 19.999 and 20.00 for instance but I haven't found this to be a problem in practice.

Would you consider the following mod which would only trigger the autorange-down operation if the meter reading is below 1800 (18% of current range), instead of your 20%?

M31-11 pin is kept low, which allow auto range down to occur. If the count reaches 1800 or more, then the auto range down operation will be disabled. After the range changes down, the meter will display 17999 or less. This would add 2% hysteresis to the range change operation.

 

[Gyro]

Perfect! I searched around for a spare gate to try to implement something of the sort but didn't think of a discrete solution. Yes that looks like the optimum solution for maximising the number of digits displayed whilst allowing some hysteresis between (auto)ranges. Thanks. 

[Gyro]

@Zeitkind:

Looking back at your pictures, I noticed that your DC isolator board uses a temperature compensated and selected LM312 (under the black thermal shield) as used on the 1045, rather that the LM394/LF355 combination used on later 1041s.

There is little or no difference in performance between the two implementations but the LM312 variant was much harder to manufacture. It required multiple temperature cycles to determine the 'select-on-test' (FSV) resistor values. The zero-drift compensation procedure is over a page long and includes a couple of recursive loops! Very time consuming. 

Hopefully you will never have to touch this section but I have attached a photo of the equivalent 1045 schematic page for reference.

[Zeitkind]

1. The horrible blank display on overload:
 [...]

2. 10 Reading per second display (this also makes autoranging performance very fast):
[...]

These hacks sound interesting. Don't have much time atm, but will likely try it.
Accessing the data in binary form or controling the meter - hmm, not sure. I can attach eg. my Sanwa DMM to a PC and it's quite accurate, so well.. The meter will have its place in further electronics tinkering with kids, though we normally do not need 4,5 digits.. ^^ But who knows. One brought his dad's dead Metrawatt DMM, that was killed by an electric fence, another kid has a dead Sony amplifier, but we will first repair an old Sega Gamegear etc. etc.. Enough work.. 

[Zeitkind]

Hopefully you will never have to touch this section but I have attached a photo of the equivalent 1045 schematic page for reference.

When checking at the beginning, I already noticed that some parts may be hard to find if broken. So yes, I'm quite happy, that the meter is still in spec and the problem was "only" in the ranging TTL section. A complete recalibration ... *yikes*
But well, tbh, it's nice to have a working vintage 4,5 digit meter, but new modern DMM are fine for me too. I like to repair old stuff built of descrete parts, simply because you can indeed repair it and not just change boards. But if something is broken beyond reasonable efforts? I have eg. a working Tek 453A (but rarely use it now). If it will break some day - it will be dead. To complicated for me, to complicated to take apart, high voltage (8kV) I don't really like to work on etc. And you can get a used CRO for much less the spare parts will cost..

[Gyro]

1. The horrible blank display on overload:
 [...]

2. 10 Reading per second display (this also makes autoranging performance very fast):
[...]

These hacks sound interesting. Don't have much time atm, but will likely try it.
Accessing the data in binary form or controling the meter - hmm, not sure. I can attach eg. my Sanwa DMM to a PC and it's quite accurate, so well.. The meter will have its place in further electronics tinkering with kids, though we normally do not need 4,5 digits.. ^^ But who knows. One brought his dad's dead Metrawatt DMM, that was killed by an electric fence, another kid has a dead Sony amplifier, but we will first repair an old Sega Gamegear etc. etc.. Enough work.. 

In usage terms, the strengths of the Datron (apart from its beautiful display  ) are its 1uV resolution and >10Gohm input resistance. These can be used to advantage for educational experements, high impedance circuits, tracing shorts by voltage drop method etc. Most of the time a DMM is fine but there are times when you'll be really glad that you have it.