Help with Sinclair DM235 Multimeter repair

[overvoltage]

Hello everyone,

I have recently bought a Sinclair DM235 Multimeter for very little Money.
It seems to be a pretty rare multimeter since not much info is available.

Sadly it has two Problems:
1.The case was completely destroyed in shipping.
2.The Display shows negative overload no matter what setting is selected, or what is connected.

I was able to track down the service manual:
http://dasarodesigns.com/~medasaro/DM235_SM.pdf

I have already checked the PSU voltages and they seem to be fine.
No traces look corroded or otherwise damaged.

In Fault Finding, step 4 the Service Manual says: "Check TR9 and associated circuitry"
I have replaced TR9 with a BC547 and nothing changed, but I still don't get what associated circuitry means.

The Service Manual also advises to connect 1V to the Input and check if pin3 on IC4 reads 0.8V but in my case, it reads close to -0.9V
I have also checked R25 and R21 and they are fully in spec.

I will now try to purchase a new CA3130E to replace IC4.
But in the meanwhile is there anything else I can try, anything I'm missing?
Thanks for the help!

[Armadillo]

Test C7 or try replace it.

[amspire]

Welcome to the madness of Clive Sinclair!

The National Semiconductors LM3900 was an interesting device that no-one really wanted to use. So of course Clive had to immediately jump on it to form the heart of a precision multimeter. Then there was the question of how do you find a matched pair of P channel mosfets? In a logic IC of course. The result is the feared word "selected" beside the 4007 IC. Some poor sod had to sit down and test every single 4007 device. One batch might have a 90% success rate. Another batch might be 100% fail.

As long as the main logic IC is working, this is not a bad repair project and learning how a Clive Sinclair design works  does show how you can think outside the box.

The topology of comparing the input to a single slope ramp is not a great one. It needs the extra rc filters on the input that will make it a sluggish meter when compared to the modern multimeters. It will be interesting to see how well it actually works.

A gate of the 4007 is permanently connected to the input and it might have blown. Look for where you are getting voltage drops. Across R20? Across R21? R25?. If it is across R20 and yet not across the other two, the 4007 is probably the culprit.

[bd139]

Being Sinclair, just as you get it working, it will probably catch fire or something!

As for the selected ones, there is a rumor that Sinclair was so cheap, he actually filled his driveway to his house with discarded semiconductors that failed his selection process. These were all discarded already but were good enough for Sinclair.

Really it's not a terrible design apart from the selected CMOS IC. It mostly works. The later Thurlby branded ones, derived from the same manufacturing base, instantly substituted this for the ICL7106 IC which was probably a good idea. Instead, however they designed them so that when the batteries inevitably leaked, they'd destroy the main board in two minutes flat.

[Simon]

Please don't post the same topic multiple times, it really annoys people and won't get you help any faster.

[bd139]

IC2 is a MOSFET array. It's basically being used as two matched MOSFETs in a differential amplifier role as an impedance converter. This is a lower cost substitute for matched JFET ICs which can get bloody expensive.  High impedance at the input in so the internals don't mess around with the signals you are measuring or load the circuit under test. Output is low impedance so it can drive more demanding circuits, like IC1. With IC1, this has a couple of different roles including an AC converter and adding a DC offset. Quite clever really.

You can probably match your own easily enough. Any random one will allow you to see if the circuit is broken or not. Also the legendary Bob Pease wrote about the analogue properties of the CD4007: http://www.electronicdesign.com/analog/whats-all-cd4007-stuff-anyhow

Next time I see one of these DMMs on eBay I'll buy it and write up an article on the design I think as it's quite interesting.

edit: found some CD4007's in a tube in my cupboard. I might try and clone the front end of this and see where it gets.

The LM3900 was chosen by the looks because it allows current driven operation which is easier to handle when you have limited rail voltages at the time typical of battery based applications or wall wart powered devices. Rail to rail amps I'm not sure even existed when this was designed. Terrible noise performance and linearity though.

[bd139]

Discrete MOSFETs vary wildly in parameters. You'd have to replace with a CD4007 in this case as they are made on the same substrate at the same time so probability of a reasonable match is high. Personally I'd buy 5 or so and a socket and just swap them around until the least crap result occurs.

If you wanted to match discrete MOSFETs you would have to use a curve tracer to plot and match Vgs against Ids

[glarsson]

So the CD4007 is made up of 6 N-Channel MOSFETs and they are used as a differential amplifier.
In theory, I could replace it with 6 discrete MOSFETs to test if it is at fault.

You don't need six discrete MOSFETs. Only two of the 4007 MOSFETs are used. You need to match them and probably also clamp them together to try to get them to track thermally.

[bd139]

Yes you can test it easy enough. Read the following for how it works (read the first chunk as it's an odd little chip): https://wiki.analog.com/university/courses/electronics/electronics-lab-28

For each pair of MOSFETs in it, you can set up an inverter and see if it inverts as a test case. This is described in the above article.

[bd139]

Possible that the main display IC is knackered now.

To test an op amp, you need to set it up as a unity gain follower (google this). I can’t get to the schematic at the moment as I’m on a cruddy 3G data phone. If it’s an LM301 which it might be, can’t remember, then you will need to set it up properly with a compensation cap and negative and positive rails (two 9v batteries will do fine for this). Use a variable resistor as the input across +/-9v and the input and make sure the output matches the input. It’s not a perfectly conclusive test but it’ll pick up an obviously broken one.