Farnell DM131 Multimeter Thermal Probe

[WaveyDipole]

I have here Farnell DM131 DMM, which appears to have been manufactured circa 1978. It is a mains powered bench type DMM, using LED numerical displays. It measures both AC and DC volts, current as well as resistance up to 20M. It also has a temperature sensing function.

Unfortunately, it came without the temperature sensing probe. Usually, this would be a thermocouple, composed of two differing metals which generates a potential when heat is applied to the junction. In this instance, the temperature probe is shown on the circuit diagram as a transistor with the BE junction connected and C n/c. It does not mention the type of transistor though. The probe is described in the manual as having a semiconductor tip (microtransistor) encapsulated in an epoxy resin ('araldite') bead, 3mm dia x 6mm length. Thermal capacity 0.25 joules per deg C approx. Temperature range is not surprisingly quite limited, but still manages a potentially useful -55deg C to 125deg C.

It seems it would be a simple matter to DIY the probe by connecting the BE of the transistor via a length of wire to a 3.5mm jack (the small matter of calibration notwithstanding). But what transistor would I use? Clearly must be Si, because a Ge junction would be destroyed before reaching 125deg C. Would any Si jellybean work or would I need a specific type? Also, is the collector used as the tip or is it simply NC and buried in the araldite, so that ambient temperature through the araldite is required to reach the BE junction?

[MotorBikeLes]

The way the circuit is drawn, it would appear that pressing the (linked I assume) switches will simply put a S/C across the transistor. I guess they must be wrongly drawn.
Upper contact to be NC, lower NO makes more sense surely.
I agree you need a silicon transistor, but different doping levels would probably give different temp curves. Just try a small general purpose NPN and check what sort of accuracy you get. In fridge, at room temp, and in boiling water should give you three easy reference temps. You will need a red alcohol thermometer for the fridge and room temp. Does it require a constant temp for instrument environment? Any thermocouple based unit does, though a few degrees of "cold junction" not to important if measuring say 500 degrees with a thermocouple type.
Les.

[Bazz4CQJ]

Way back in the 70's I constructed a digital thermometer (hot stuff then!) which just used a run-of-the-mill silicon diode as the sensor - and quite sensitive it was too. I'll see if I can find the article, as it also included a calibration procedure. I think there will be one pre-set to adjust the offset and one the slope, using an ice bath and steam alternatively.

B

[Radio Wrangler]

It's just an odd drawing of a jack socket wit switch contacts. The contacts close whenthe plug isn't inserted.

There is a JFET current source, and the DVM presumably measures the transistor Vbe at that current. Si gives about -2mV/C.

Something jellybean would seem fine. Try an ordinary 2N3904 or BC108. if it works, find a more suitably packaged variant.

David

[WaveyDipole]

Sorry, I cropped off the connection with the IC and perhaps should have mentioned that the switches appear to be on the jack socket and the diagram seems to show two separate single pole switches. The 3.5mm jack socket in situ is 3-way (i.e. a stereo type), it did not properly fit in the opening in the chassis and has only a single switch on the middle contact, which was left n/c. hence my suspicion that it was replaced at some point and not wired correctly. The line going straight up in front of the FET goes to a pin on an IC. When the jack is withdrawn, this pin is connected to 0v which is essential for the correct operation of the DMM on all of other functions - and hence the reason it malfunctioned. When a probe is connected, the IC is disconnected from 0v and thermal circuit comes into play. I re-wired the socket so that the switch now does its job and the meter works on all functions.

Bazz, since only one junction of the transistor is used, I did wonder whether a diode would do the job, although David suggests that the meter measures the transistor Vbe so maybe not. Either way, your article would be of interest. If you can dig out then it would be appreciated. The manual does, in fact, suggest placing the probe in ice water to set the calibration to 0deg, although it doesn't mention steam!


David, I will give a jellybean and maybe a diode a go when I get a chance and report back, although I wonder whether a BC108 with its metal case would be a more sensitive candidate.

[Herald1360]

Try something like BC859 (surface mount version of BC559 jellybean).

Transistors used as sensors like this usually have B-C short rather than open but the meter may be set up to allow for this.

[Bazz4CQJ]

Hi, I scanned the article I referred to, but had to do it in hi-res, so the pdf file size is too big to load on the forum; I could send it to you by email if you want.

It really deals with setting up a prototype, and it may well be that with something that has gone in to production like yours, the calibration will have been simplified. It may give you some background; let me know if you'd like it.

B

[WaveyDipole]

Bazz, thanks. I have PM'ed you my e-mail address.

Chris, I have no surface mount transistors to hand at the moment although I had wondered whether to try one as they are much smaller. I will experiment with the ones I have for now and maybe include the one you suggest in my next order. I will also try with BC short and open. Thanks for that bit of information.

[Herald1360]

PM your address if you'd like a few to play with.

(Might be MMBT 3904 or 3906)

[WaveyDipole]

That's a very kind offer. I will experiment with what I have for now, but I might take you up on it later.

[WaveyDipole]

I had a bit of time today so I tried making up a temperature probe and experimenting with different devices. The transistor shown on the circuit diagram is an NPN device so the first device I tried was a 2N2222A that I had to hand. I wired it as per circuit diagram and used the thermocouple on a Fluke DMM for comparison of the ambient temperature in the room, which on the Fluke read 19.7deg, which seemed reasonable.

The 2N2222A on the DM131 gave me a reading of around 43deg. When the transistor was held between a finger and thumb, or placed near the tip of the soldering iron the temperature almost immediately started to increase which confirmed that the DM131 temperature function seems to operate in principle. I then tried shorting the connector to base and the temperature shot up to 88deg! The DMM is clearly not happy with that arrangement. Disconnecting the collector retuned the temp to 43deg. I also tried wiring up the BC junction instead of the BE junction and found that this gave the same result as the BE junction, and shorting to E approximately doubled the reading in the same way.

Next up was a 1N4148 diode. This gave a reading of 88deg when connected on way, and -90deg when connected the other but there was no sensitivity. Placing it near the tip of the soldering iron did absolutely nothing so I guess the transistor with its additional junction is needed, even althought it is n/c. I will however build the circuit for a diode sensor that Bazz sent me. Thanks for that.

Finally I tried a 2N2907. This is a PNP transistor so I wired it with opposite polarity to the NPN. This gave me a reading of 22.8 deg, which is only 3 deg more than on the Fluke. There is a calibration setting on the DMM which could probably be adjusted to compensate for that. I haven't adjusted it yet as I want to try some other devices.

Obviously different transistors do give different readings which presumably depends on the characteristics of the device. I will experiment with further devices to see whether i can get even closer to the correct reading. I found that it takes time to experiment as it is necessary to wait for the transistor to cool down after heating it up or touching it, but placing it on a pair of pliers and placing another one on top speeds this up quite a bit - I guess due to it acting like a heatsink. I also need to do some more tests to determine whether the temperature increase is linear - i.e whether it matches the Fluke thermocouple for a given heat source, but since the soldering iron it too hot (can only go up to 125deg) I'm going to need a constant heat source. Maybe an incandescent lightbulb that has been on a while?

[karesz*]

Hi WD,
Here is a good AppNote over "Accurate Temperature Sensing with an External P-N Junction". the ca 2mV/C has a variance between types and exemplares too and is more at 2,2mV as less, but you need to calibrate it.
Karl

[MrBungle]

Thanks for the appnote - really useful. I've just acquired a Philips PM2524 DMM without a temperature probe and they appear to be unobtainable so the next step is build something

[WaveyDipole]

Likewise, thanks for the appnote and associate information. Very helpful. I have downloaded for offline reading at some point. The DMM131 manual does describe the procedure for calibrating at zero deg by using ice in water, but also states that the calibration temp is between 20-23deg which is a bit confusing. The project note for the diode based meter sent to me by Bazz suggests using boiling distilled water for the hot reference point.

It looks like the Philips uses a 5-pin DIN socket for the thermal probe so maybe a bit more sophisticated than the DM131?

Incidentally, does anyone know how to stop the resistance reading from fluctuating? It seems that the problem is the probe wire. If I strap the resistor across the terminals then the reading is stable.

[MrBungle]

I'd probably throw the probes away or try with some different ones to start with. Probes are technically classifiable as consumable items. I personally don't use probes put prefer Pomona minigrabber leads which seem to last 2-3 years a go before I knacker them: http://uk.farnell.com/pomona/3782-48...-5a/dp/1231212 (cost about £14 for a pair of black/red ones with VAT added)

I looked at the Philips unit and while it looks like it might be interesting with a DIN socket, it just uses a thermistor apparently



P-N junction with a DC amplifier looks like a generic option that might work for all meters.

[WaveyDipole]

Well my probes did come at a fairly budget price from Maplins. The useful bit was the various attachments. That being the case I did try the AVO probes leads as well with the same result. They work fine on the Precision Gold DMM. It would seem that EMI being picked up by the probe leads is causing the ADC to fluctuate. I do pick up a background 50hz signal when the leads are plugged in, but not on the terminals when they are not connected.

So your Philips uses only two of the five pins! Perhaps the design is a result of the Philips European heritage. It will be interesting to see whether a P-N junction probe will work and what adjustments might be required to get a reasonable calibration (I have yet to read that AppNote). The advantage of the thermocouple is that it will measure much higher temperatures so I guess the P-N junction will be something of a limitation in this regard but perhaps ought to be enough to measure the temperature of heatsinks or components.

[MrBungle]

That's interesting. Might be corroded or dirty sockets then. I have the same problems with my Heath power supply. I had to go at them with a fiberglass pencil for half an hour to clean them up when I got it. Cheap banana leads from Rapid fall out all the time. The Pomona ones don't.

I doubt the P-N junction will work without some kind of amplifier but worth a try. I'd prefer a thermocouple personally. You can use a standard one with any meter, you just need a calibration table you can print out from here: http://us.flukecal.com/Thermocouple-...ure-Calculator

[Bazz4CQJ]

Originally Posted by MrBungle

I'd prefer a thermocouple personally. You can use a standard one with any meter, you just need a calibration table you can print out from here: http://us.flukecal.com/Thermocouple-...ure-Calculator

My recollection re thermocouples was that you needed a reference junction; looking at Wikipedia https://en.wikipedia.org/wiki/Thermocouple it shows a Fluke thermocouple meter with internal reference junctions which have thermistors in them to measure their temperatures independently.

If you simply measure the emf of a thermocouple with "any meter" surely there are errors in converting that to temperature? Thermocouples' big advantage (only advantage?) is that they will work at very high temperatures (~1200'C) and very low temperatures; I have an old Comark 1602 thermocouple thermometer which were very expensive when they first appeared. For temperatures from -10 to +100'C, my impression is that almost all modern meters use either thermistors or semiconductor junctions, for which the supporting circuits are so much simpler/cheaper.

[Radio Wrangler]

Actually thermocouples are quite common on multimeters. I needed a meter in a hurry and didn't have one with me, so I wound up with a seventeen-quid job from B&Q and it even came with a length of thermocouple cable with a fused end.

David

[Bazz4CQJ]

Ok, but assuming it's accurate, it has the right reference circuitry in it and it's not just "any meter". I've bought quite a few digital thermometers of various kinds in recent years (usually with internal and external sensors and a max/min memory), usually costing £3-14 and every one has been thermistor-based, but limited to the range ~-10-100'C. I guess that your multimeter covers a wider range? Surely it's the case that a thermocouple circuit must be more complex than a thermistor circuit/junction sensor?
Of course, these days multimeters of astonishing specs do come cheaply, e.g. Maplin's analogue item at £9.99; 20kohm/V, all the usual stuff, plus transistor tester, capacitance meter,loaded battery tester and audible continuity tester... but no thermometer function . For a given amount of money, there has to be a line drawn somewhere, even in China.

B