Smart Tweezer DMM Hack

Add voltage-measurement mode to your Resistor/Capacitor/Diode with one submini DPDT switch.

Alternate title:   NEVER USE THE TESTE ON LIVE CIRCUITS

Fig 1 - The MS8910 Smart Tweezers

During our circuit hacking here at Openschemes, we often find ourselves poking around at the internals of some random electronic gadget.  We've seen consumer electronics evolve from good old through-hole components to all SMD devices.  Although cool, SMD's have the drawback of tiny, illegible marking codes which can really slow down the tracing of an interesting circuit.

Enter the Smart Tweezers!

As you may know, "Smart Tweezers" are a relatively recent development in the DMM field.  They essentially consist of a simple DMM circuit powered by a 3v lithium cell, and built into the form factor of a pair of oversized tweezers.  The metallic tweezer probes act as the DMM terminals.  Picking up, or contacting an SMD device allows the smart tweezers to measure resistance, capacitance, diode Vf, or inductance (in higher-end models).  We recently received a free set of the MS8910 smart tweezers with the purchase of a Protek D620 sampling multimeter.  They are extremely handy for measuring SMT components on PCB's, and although the accuracy was initially pretty terrible it is very simple to adjust yourself as we will see later.

The Drawback..

During a recent hack (of the aforementioned D620 DMM, actually) we still had to grab our trusty Fluke and wield it's oversized probes in order to measure a few power supply voltages.  Since these were measured across their respective input and output capacitors, it occurred to us - Hey, we should hack our $30 smart tweezers to allow THEM to measure these voltages for us!  And off we went.  The total hack time of this project was roughly 4-6 hours.  3-4h of research and circuit tracing, and 1-2h of modification and testing.  Our fastest yet!

A Quick Aside

The back of the MS8910 smart tweezers has one of the most hilarious pieces of ch.. err, broken english we've ever seen - and that's saying a lot based on how much business we do in China.  Enjoy...

Fig 2 - Back of the MS8910 Smart Tweezers showing our mode-selecting DPDT switch and the most awesome warning ever:

CAUTION - NEVER USE THE TESTE ON LIVE CIRCUITS

A truly chilling thought, but very good piece of advice for budding engineers.  Don't do it, you'll be sorry.

Disassembly

Disassembly of the MS8910 is trivial - remove the battery screw to expose two hidden screws.  Remove these, and one more near the probes and the case can be easily slid open.  The PCB resides in the LCD side, held by a few plastic clips.  The battery resides in the battery compartment, obviously, and is connected to the PCB by red/back wires.  Be careful not to break them during disassembly.  One interesting point is that the red wire (battery positive) is considered the GND terminal of the device and is directly connected to the top probe (near HOLD button).  The black wire (battery negative) is considered the V- negative (-3v) supply for the DMM chip.  The positive (+3v) supply for the chip is generated  by the DMM IC itself with a capacitor-based charge pump.  Makes sense, when you consider that a typical meter needs to measure both positive and negative voltages so a bipolar supply will be required.

The DMM IC

Finally!  A company that does not attempt to obscure their product, but rather engraves the web address right onto the IC package - kudos!  The DMM IC is made by Cyrustek, a Taiwanese IC company specializing in DMM chips whose company motto is presently:

 " Cyrustek - of the global leader in Digital Multi-Meter IC and beyond "   Outstanding.

Fig 3 - The Cyrustek ES51925, 3000 count AutoRange Smart DMM IC

Download ES51925 datasheet from Cyrustek

Fig 4 - ES51925 Function Table - MS8910 uses RCD mode only, listed on line #8

From inspecting Line #8, we see that RCD mode measures on the standard input VR1, also used for Voltage measurements.  Since one smart tweezer probe is used for GND, this must mean that the other probe connects to VR1.  So with a little luck and a little skill, we SHOULD be able to add voltage measurement capability to the MS8910 without a lot of trouble.

Delving Deeper

Let's look at exactly how voltage and RCD components are measured.  First, voltage is easy - apply a voltage to the VR1 terminal.  This terminal has a 10MEG series resistor used as the top resistor in a range-selectable resistor divider.  You can see the other resistors connected to the "VRx" terminals.  As the input voltage rises past a volt or so, the IC switches in the next lower resistor to divide the input down to a more reasonable level.  DC voltage measurements only need this divider ladder.  AC voltage measurements need the rectification circuit shown on the ADI ADO pins.  ADI/ADO are terminals of an op amp, connected around the diodes to make a precision full-wave rectifier - look it up in opamp application notes like these.  An example schematic for the voltage measurement function is shown in the datasheet, page 19.

Fig 5 - Voltage Test example diagram from ES51925 datasheet

If I recall correctly, the AC to DC rectifier/filter is NOT on the MS8910 so it cannot measure AC voltage... yet.  Maybe we will add it later.

Next, we compare the typical voltage measurement to a component measurement - here we use the diode mode as an example.  In diode mode, the forward voltage of the diode is still measured using VR1 (thru 10MEG, of course), but the stimulus current must be provided by another pin.  We can see this below in the diode test example from the datasheet.  Stimulus is provided from the OVH pin (here just a small current is needed) and the meter basically does the same old voltage measurement to display Vf.

Fig 6 - Diode Test example diagram from ES51925 datasheet

Resistor measurement is fairly straightforward, and capacitor measurement probably measures the frequency of an RC oscillator made from an internal resistor and your external cap, OR..  Applies a current stimulus to the cap and measures AC ripple.  The latter is not very likely since the AC rectification circuit is missing, but that's how a lot of meters do it - store that in your trivia bank.  Either way, stimulus comes from the same old OVH network and is fed to the component under test at the probe.  The response is measured by VR1 and the result is displayed on the LCD.

Adding The Voltage Function

So how do we get this beast to measure voltage?  Two things are needed:

1) A logical signal to tell the DMM IC to use Voltage measurement mode

2) Removal of the OVH stimulus circuit so it will not screw up our measurement.

Sounds Logical

First things first - the logic.  Another inspection of the Function selection table shows us the pin settings for each mode.  We HAVE the pin settings in line 8 (RCD mode), and we WANT the pin settings in line 5 (Voltage measurement mode).  It's time to take a peek at the live circuit.

Fig 7 - Highlighted Function Selection Table from ES51925 datasheet

As was mentioned earlier, the "low" level is -3v - essentially the black battery wire.  A 'high" level seems to be 0v, the red battery wire.  In no case did I see the charge-pumped +3v used as a logic high.  Checking the state of the interesting pins with a meter gives us the following table.  Any pins that are "don't care" (X) in both our modes are left out of this table.

Since FC2 becomes a don't care when FC1 goes high, and VON and CLAMP are already logic high, we luck out and only need to pull FC1 high (0v, battery red) in order to change modes.  This is going to be easier than we thought!

Inspecting the FC1 pin, we see it is tied low with a diode.  Why a diode and not a resistor, We're not sure.  It may be due to the voltage sequencing requirements introduced when the charge pump starts running.  But for our needs, we require a resistor so you will have to break the trace between the diode and the IC pin, and insert your own resistor.  We used 100k but anything from 1k to 1MEG should be fine.  In order to enter Voltage Measurement Mode, we will have to pull the FC1 pin high.  To pull the FC1 pin high means to short it to the RED battery terminal.  We do this with a switch, so we can toggle back and forth between the new Voltage Measurement Mode (FC1 shorted to RED) and the normal, RCD mode (not shorted). To use RCD mode, we open the switch and the 100k plus diode will pull the FC1 pin low.  To use Voltage mode, we close the switch and pull the FC1 pin to 0v.  For our modification, we soldered a fine wire from the IC side of the new 100k resistor to one pole of the DPDT switch and another fine wire from the switch to the battery RED.  It doesn't matter which position you use on the DPDT, only that the FC1 is shorted to RED when the switch is in position A and FC1 is floating when the switch is in position B.  Position A means Voltage mode and Position B means RCD mode.

Next, we flipped our switch to position A and reassembled the device.  Power on, and SUCCESS!  We are no longer in RCD mode as the LCD never displays the ohm or farad symbols.  We next applied a 0.5v DC signal to the probes with a 1MEG resistor for safety and bingo! - the LCD read 0.530.  Sweeping the voltage up and down proved that it was really measuring voltage, although somewhat inaccurately.  :)  We will address that later.  This particular LCD was never meant to use Voltage measurement mode so you will not have a Volts or AC icon, but it's no big deal:  Upon first entering V-mode you are in auto-detect.  The DMM will choose DC or AC based on the input signal.  Since our AC mode is incomplete, we can push FUNC one time to force the DMM into DCV mode, or simply avoid applying AC signals and let it figure it out itself.

Testing the Logic Hack

Although we know that the stimulus circuit SHOULD be disconnected, why not test the device right away?  In our opinion, you should be able to glean SOME useful information even for a partial or failed test.  So damn the torpedoes - full steam ahead!  We set up the smart tweezers with a DC power supply and a known good Fluke 87 DMM to check the quality and range of the V-mode measurements when only the FC1 bit is changed.  Here's the data.

Fig 9 - Graph of the Voltage Measurement Comparison showing nonlinearity above 3v - Logic change only

 

Fluke	Smart Tweezers in Voltage Mode	Error
0.043 V	0.043 V	0 %
0.800 V	0.814 V	1.75 %
1.089 V	1.109 V	1.84%
2.007 V	2.045 V	1.89 %
3.093 V	3.14 V	1.52%
4.114 V	6.79 V	65.05%
5.114 V	10.69 V	109.20 %

Fig 10 - The raw measurement data.  Tweezers are running about 2% accuracy up to 3v

The verdicts?

A) This piece of crap is way out of calibration.  At some point during the initial tracing, we did note that the reference voltage was -395mV instead of -400mV so we will go back and re-calibrate later.

B) When the input voltage exceeds the positive supply voltage, the measurement is "clamped" by the OVH pin.  V-mode measurement is severely affected by the OVH network, and is basically no good for inputs greater than about 3v.  We see a lot of 3.3v, 5v, 9v, and 12v rails in systems we look at, so this will have to be corrected.

A little more surgery

By inspecting the traces on the PCB, we see that the VR1 input and it's associated 10MEG resistor are located just to the right of the IC, when the probes are pointed towards you.  For the OVH network modification, we will need to cut 2 traces and add 3 jumper wires.  These edits are shown in the lower image below and will be discussed in detail momentarily.  The edits from step 1 are also shown for reference.  Also, here is the IC pinout with the FC1 and VR1 pins highlighted.

Fig 11 - The pins we are interested in.

Fig 12 - Edits made to the tweezer DMM circuit.  Edits: Red X = CUT.  Orange and Purple = NEW WIRE

First, we must disconnect the 10MEG from the PTC resistor (part of the OVH network) and reconnect it directly to the probe.  The trace we will cut is on the BOTTOM of the PCB and is denoted in blue.  It runs from a via near the 10MEG all the way to one of the PTC resistor terminals (The orange ceramic-cap looking guy, marked PTC).  Cut the trace with an exacto, or scratching with tweezers.  Make sure it is broken well by checking with a meter - it's a good idea to remove about 1mm of trace to be sure.  Now you need to add a jumper wire from the 10MEG resistor directly to the probe.  This jumper is shown in orange.  We soldered directly to the via-side (NOT the IC side) of the 10MEG resistor pads, and soldered the other end of the wire right to the probe anchor, a big blob of solder where the probe meets the PCB.  You can see this fine trace of magnet wire in the pic, although it's connection to the probe anchor is obscured.

Next, you must disconnect the OVH network from the probe by cutting on the TOP of the PCB right between the probe and the potentiometer VR2.  This trace is shown in GREEN.  We used tweezers to scratch a deep channel between the probe pad and the potentiometer pad.  Last, the potentiometer and the probe need to be jumpered to the other side of the DPDT switch.  This connection will be closed when the switch is in the opposite position from the Voltage mode we discussed earlier.  I connected the probe to the second pole of the DPDT and the pot/PTC node (yes, they are shorted together) to the B-side throw of the DPDT.  Remember, when the DPDT switch is in the A position, the pot/PTC is DISCONNECTED from the probe for V-mode.  And when the DPDT switch is in the B position, the pot/PTC is CONNECTED but the logic line on the other side is floating.  In the picture, you can see that we took our POT/PTC connection from the leg of the PTC - that's fine, it's an easy point to solder to.  Our second connection (to the probe) is seen in the picture as a second connection to the 10MEG.  Either one is fine - as long as the probe is jumpered to the 10MEG, you can either take the DPDT lead from the probe or from the resistor pad.  It's up to you.

 Fig 13 - DPDT Wiring as seen from bottom. 

We define Position A (V-mode) when FC1 is shorted to BATT, and Position B (Normal, RCD-mode)when Probe is shorted to PTC

Testing and Calibration

We personally placed the DPDT switch so it did not obsure the hilarious safety warning, but you are free to place it wherever you like.  Now that all the switching is taken care of, it's time to test the device again and make sure we didn't screw anything up.  First, we adjusted the reference voltage from 395mV to 400mV by twiddling the reference voltage pot as shown below.  We reassembled the tweezers, gently folding our new wires so they were not pinched by the closing case, and began testing.

Fig 14 - Location of calibration potentiometer

Next, we check the accuracy of resistor mode. 

Last, we check the accuracy of Capacitor mode.  Fairly bad at the low capacitor range, but the datasheet specifies the part as 3nF - 300uF.  This might be improved by adjusting of the OVH potentiometer, but we will leave that for another time - it's better to do something only AFTER you understand why you would do it, rather than just turning knobs and not knowing what they do.

Conclusion

So there you go - for the cost of a 25 cent DPDT DIP switch, we added the $10 dollar voltage measurement function to a cheap pair of smart tweezers.  We're already using ours to check all sorts of voltages - logic levels across pullup resistors, input, output, and error voltages on DC-DC converters and LDO's, and offset voltages of a few opamps.  Not bad for a day's work. 

Happy Hacking!

Update!

We've just seen the same MASTECH MS8910 on ebay for $16 - we highly recommend it.  Buy it from ebay - we DO NOT recommend circuit specialists, webtronics, or any of the other names used by the garage operation run out of Mesa, AZ due to their poor customer service.  

Back to Openschemes Projects

(c) 2008, Openschemes