Here's a teardown of the C.E.M. DT-21 multimeter. It's a cheapie I got for the price of a takeaway lunch at an auction, mainly because it reads 0.1°C. The stated accuracy for temperature is actualy fairly good too, at ±(3% reading +1°C). We shall see how it faires. Since the internal construction is a bit surprising, I figured I'd post the photos here. Plus I don't think anyone has done this one yet. Then again, it's probably sold under different names, so I might just have missed it.
As you can see, it's a multisensor meter. It's a nice gimmick I suppose, I don't have one of those yet.
It's rather small which I like, smaller even than the notorious DT-830, although a bit chunkier. The detachable holster features the shock protection system copied from Fluke, and a very flimsy tilting bail. It works though, since it only has to support a very small meter.
The meter is held together by a single self-tapper, which at least has a good length. The back of the case has some shielding, a bit is folded back on itself, to be contacted by the spring on the PCB. Not exactly beautiful, but I suppose it kinda works. What's not great though is that the case has no molded groove, there's just a single lip on each part with about 1cm of overlap. Then again, if you ever use a device like this one on anything vaguely energetic, you deserve the Darwin award coming your way.
On to the main PCB! Whoah, I didn't expect that. That's a lot of passive components! All those SOIC-8 chips are TL062C dual opamps. And there's a lot of trimpots, about a dozen or so in total.
And is that? Oh indeed, dual board construction! How curious. Here's the daugtherboard:
And with a detail that makes me go "hrmmm". Notice the solder bridges on two pins of the opamp. It might be a piece of wire placed there intentionally, doesn't look like an accidental solder bridge. Those pins are opamp output and inverting input, both times. The marking's scrubbed off, but that's probably not intentional, the remaining part of the marking is the same as the other opamps. That cap in the corner doesn't look great either.
Now on the main board. Under the daughter board, you'll find a couple more opamps and trimpots, and the single PTC which let's them claim they have input protection. As expected, there's a single glass fuse for the mA range, while the A range is unfused. It's what you get in this price range.
The on-end resistors at the top looking like a band of drunks ought to be the measurement resistors. The markings claim them to be 0.5% accurate, which would make sense, given that the stated DC accuracy of the meter is also 0.5%.
That bit of foam on the left is where the battery rests. Yup, it just lies loosely in the case, sandwiched between the main processor die and the outer case. At least the foam thickness seems to have been chose correctly, there is no rattling.
The input connectors look unusual. Instead of the usual stamped metal, we get tubes riveted to a piece of metal, which is screwed (self-tappers) to the PCB. In theory it's a rather durable construction, in practice corrosion and fatigue in the plastic as the PCB moves slightly against it will cause contact issues. Adding a bit of solder before putting the screw back in might work well though.
If we flip over the PCB, we find gold plating for the switch, and the environment sensors. The cabling is hand-soldered of course, but rather shoddily. Here's a tip if you don't know yet: don't use metal tweezers to hold cables when soldering, you'll very easily melt the insulation. Wooden ones, or tweezers with metal tips work, and in a pinch you can wrap the tweezer tips in heatshrink or even pieces from paper towels.
So, after putting it back together, Let's test it! The meter does come with leads, which are quite pointy and very scratchy on the continuity test (which funnily enough is a shared function with the 200Ohm range... If you ever have to measure many low value resistors with this meter, you'll drive your coworkers insane.) The probe rating of 1000V Cat III is a lie given the lack of tip insulation, but the cable insulation feels rather nice actually, quite a bit less stiff than PVC usually feels. It's probably chock full of highly poisonous plasticizers
The labelling on the cables reads 80°C 18AWG 2000V. Yeah, better not test that voltage rating.
And there's the temperature probe, which is the immersible type and one of the reasons I bought the meter, I don't have one of those yet. The way I've been measuring water temperature is a bit of a hack, with a standard unshielded bimetal temp probe in a plastic bag:
That old Mastech is probably accurate to a degree or two, at least it gives me the same temperature as my IR meter gives me when measuring a piece of tape on the outside of that pot. The C.E.M... yeah not so much. Something's obviously wrong with the full degree temp range. The smaller range might be salvageable though given the many trimpots. We'll see.
Now, I took some more measurements for those interested. I'm using my Uni-T UT61E as a reference (first column), it should be accurate enough for testing the cheapie. All voltages are DC:
3.003V: 2.97V (1.1% low, pass - stated accuracy is ±(0.5% +2 counts))
12.044V: 11.89V (1.3% low, fail)
19.088V: 18.84V (1.3% low, fail)
17.879mA: 17.7mA (1% low, pass - stated accuracy is ±(1.2% +2 counts))
118.42mA: 117.5mA (0.8% low, pass)
98.96uA: 98.3uA (0.7% low, pass)
148.45uA: 147.5uA (0.6% low, pass)
1.003A: 0.99A (1.3% low, pass - stated accuracy is ±(1.2% +5 counts))
5.024A: 4.96A (1.3% low, pass)
9.9909kOhm: 9.9kOhm (1% low, pass - stated accuracy is ±(0.8% +2 counts))
0.9739kOhm: 975Ohm (0.1% high, pass)
The lux meter seems to kinda agree with the lux meter sensor/app on my phone (Nexus 4), which isn't saying much. I suppose I can conclude it works. I don't have instruments for testings its accuracy though. The humidity and decibel functions seem to show a different random value whenever I switch to them. They're broken.
The continuity tester is the scratchy type of course, but it's instant. The diode tester will only show overload for this green LED here, but it'll light it.
Power consumption at the 9V battery is 2.3mA, which means it'll only last around 200 hours. The low battery indicator comes on at 6.9V, the reading starts going off at 2.8V and the meter turns off at 2.75V. I suppose that's acceptable.
Screen update rate is 2Hz, and while there's no overshoot, on dropping voltages it sometimes takes 3 or 4 readings to settle.
Well to summarize, this meter unfortunately is not fit for the purpose I bought it for, which is measuring temperature somewhat accurately, but I might be able to fix that using the trim pots.
As for electronic work, I suppose it's ok if you tinker with Arduinos and predominatly digital stuff, it'll tell you whether you have it running on 5V or 3.3V
And it's very small, which is nice.
Home
Help
Search
Login
Register
