Measuring SMD Parts With A Home Brew Version Of Smart Tweezers

22 Comments

SMD parts are great; they allow you to pack more parts on a board, do away with drilling dozens of PCBs, and when done correctly can produce a factory-quality board made in a home lab. There’s one problem with SMD parts; troubleshooting and measuring them. The ideal solution would be something akin to the Smart Tweezers we’ve seen before, but this fabulous tool costs three hundred bones. [Kai] came up with a much cheaper solution: home brew smart tweezers that can be built for a tenth of the cost as the professional model.

What [Kai] built is an LCR meter, basically a tool that measures inductance, capacitance, and resistance in a very, very small form factor. The technique of measuring a part’s properties involves feeding a set frequency into the device and measuring the phase, voltage and current coming out. It’s all wonderfully explained by [Dave] over at EEVblog in one of his earlier videos.

The hardware [Kai] is using includes an LCD display from a Nokia phone, an MSP430-based microcontroller, a very tiny opamp near the tip of one of the points of the tweezer, and a programmable gain amplifier used to measure the components. In testing, [Kai] can measure very low-value components with a +/- 2% accuracy, and larger, more realistic components with +/- 0.25% accuracy. An awesome accomplishment, and much better than the common Chinese meters that can’t measure in the nH/pF/mΩ range.

[Kai] hasn’t gotten his pair of smart tweezers working yet – he still needs to get the circuit up and running and write some software. We’ll keep our readers apprised of [Kai]’s progress, though, and gently convince him to work with Seeed Studio or someone similar to get his version of Smart Tweezers onto maker’s workbenches the world over.

22 thoughts on “Measuring SMD Parts With A Home Brew Version Of Smart Tweezers”

    1. My meter doesn’t measure capacitance or inductance. Lacking any sort of write up, the tweezers in the ebay link appear to be no more than convenient test leads. Do you have more info?

      Kai’s probe will do inductors, capacitors, resistors, hence its “LCR” moniker.

      But I do appreciate the link to the tweezer probes. I can do with a set.

      Report comment
      Reply

      1. The tweezers are passive, you need to use them with your own meter (I mentioned that).

        The point of using the tweezers with a good meter is that it is likely going to be more accurate than most things one cobble up at home.

        Report comment
        Reply

    4. 1) The leads of the tweezer probes have their own parasitic LCR, which may be significant if you need to measure small value parts with high accuracy.

      2) My last three “inexpensive” meters with LCR measurement failed like clockwork shortly after the warranty ran out. In every case, I just turned them on one day to find them wholly or partially non-functional; even though they were working fine last time, and suffered no known damaging events. Which made them not so inexpensive. Quality alternatives with a respectable name, reliability, and warranty are *much* more expensive; but accidents can still cause you to lose your investment. So DIY test equipment, with its reasonable price, known quality, and ease of repair/modification is mighty attractive.

      Report comment
      Reply

      1. 1) Parasitic capacitance/inductance is a problem with any test leads, tweezers or not. That is what you have the “Zero” or “Relative mode” button on your LCR meter for.

        2) Don’t use cheap multimeters. Not worth the money. On the other hand, decent LCR meter doesn’t need to cost hundreds, there are some very good ones for around ~100bucks. E.g. this one:
        http://www.eham.net/reviews/detail/4010

        Report comment
        Reply

      1. You are right about that – no inductance. I seldom care about inductors so it’s not important to me.

        One thing to consider is that “in circuit” testing of components is pretty dicey. Caps, for instance, are most often wired across VCC and GND in parallel and measuring one of them is to measure all of them. Resistors are often part of networks that change the expected value dramatically. It’s pretty much all hype the jabber about these things as “in circuit” testers.

        I like the unit I got because it’s VERY handy to check the values of loose smd caps and resistors. It also has fast response continuity testing – the beep has no delay. Even radial lead parts can be tested easily and the LCD autoranges and displays the results lickety split.

        For $30 with free shipping I think you may find that these are a very useful addition to the workbench. I was happy to get mine at $35. I might just get a second pair for my tool bag.

        Report comment
        Reply

  4. Hope it delivers the results he’s going for, this is one cool project.

    Suppose you start from a square wave instead of a sine. The spectrum of this is the summation of sines with an odd harmonic. By using the same sampling technique you’ll have different response spectra seperated from each other, you can look at each one of them by adjusting your fft window.

    Second question: did you look into wavelets? Since the waveform you’re sampling will be non-stationary (you’re sampling in respect to a fixed reference over a window that won’t be perfectly aligned) you could bring down code complexity (it doesn’t get simpler than Haar wavelets) and avoid windowing.

    Report comment
    Reply

      2. yeah, sorry about that. The theory about wavelets is daunting and there are only a few good resources IMO.

        This one is the best tutorial I know of that explains the whole reasoning behind it. It also explains stationary vs non-stationary in a good way:
        http://users.rowan.edu/~polikar/WAVELETS/WTpart1.html

        Re implementation. The best source I found is jWave. It’s a java implementation. Download it and open as a project from source in Eclipse. The actual code for forward and backwards transformation is found in code.google.com/p/jwave/source/browse/#svn%2Ftrunk%2Fsrc%2Fmain%2Fjava%2Fmath%2Ftransform%2Fjwave%2Fhandlers%2Fwavelets
        In that directory is Haar02.java that sets up a few coefficients, the actual forward is in Wavelet.java in the same dir. As you’ll see it’s a few loops with scales and coefficients that can be fixed in a lookup-table.

        There are other wavelet transforms, but Haar is IMO the fastest and best for this goal.

        Report comment
        Reply

    2. @Justin:

      The window will be perfectly aligned as I sample synchronous to the window and to my exitation signal, so I think I can stay with FFTs.

      I already have the most importants parts of the software running: The signal generator and synchronous sampling. Currently the data is output to PC via UART for further analysis.

      Will post an update soon on my site.

      I also thought about using square wave signal or at least multiple sine waves at the same time to measure with more than a single signal frequency. Not sure, if this could be of any benefit, currently I only see disadvantages (less dynamic range for each harmonic), but did not finish thinking about it. But this is step 2, when it is working with the “standard method” and in case I can improve sth. with it.

      Report comment
      Reply

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.