Measure/Projects: Difference between revisions

From OLPC
Jump to navigation Jump to search
Line 235: Line 235:
Thus the probe-circuit must "offsets" the difference between the "+" and "-" inputs with this voltage of approximately +1.18 volts. It derives this from a highly-stable "band-gap reference" (any number of different types are possible -- +1.225 volts +/-1% volt, +1.235 volt +/-1%, 2.50 volts +/-1% etc.) Audio-jack inputs less than this "centering-voltage" will drive the trace lower than center, and inputs greater than this "centering-voltage" will drive the trace higher than center. At the laptop's highest gain the voltages require to do this are quite small -- on the microvolt range.
Thus the probe-circuit must "offsets" the difference between the "+" and "-" inputs with this voltage of approximately +1.18 volts. It derives this from a highly-stable "band-gap reference" (any number of different types are possible -- +1.225 volts +/-1% volt, +1.235 volt +/-1%, 2.50 volts +/-1% etc.) Audio-jack inputs less than this "centering-voltage" will drive the trace lower than center, and inputs greater than this "centering-voltage" will drive the trace higher than center. At the laptop's highest gain the voltages require to do this are quite small -- on the microvolt range.


(ii) Increasing common-mode voltage rejection: Together with the choice of R1* and the gain of the circuit, this +1.18 volt offset dictates the maximum common-mode voltage that the probe can "accept" and still function correctly. For example, with R1* = 1 megohm and an attenuation of 1:50 (i.e. the probe reduces the input to 1/50*( V(+) - V(-) ), a negative "common mode" input will be divided down by a factor of 50. If this common-mode voltage V(+) is negative, it will suck out out of the input approximately Vin/(R1*+R2**) amps, and this will develop a voltage at the input to the amplifier of Vin*R2**/(R1*+R1**)+1.18 volts.
(ii) Increasing common-mode voltage rejection: Together with the choice of R1 and the gain of the circuit, this +1.18 volt offset dictates the maximum common-mode voltage that the probe can "accept" and still function correctly. For example, with R1* = 1 megohm and an attenuation of 1:50 (i.e. the probe reduces the input to 1/50*( V(+) - V(-) ), a negative "common mode" input will be divided down by a factor of 50. If this common-mode voltage V(+) is negative, it will suck out out of the input approximately Vin/(R1+R2) amps, and this will develop a voltage at the input to the amplifier of Vin*R2/(R1+R2)+1.18 volts.
: Example: Common mode voltage is -100 volts, R1* = 1 megohm, R2** = 10K:
: Example: Common mode voltage is -100 volts, R1* = 1 megohm, R2 = 10K:
:: -100*20,000/(1,000,000 + 20,000) + 1.18 = -0.96 + 1.18 or -0.22 volts.
:: -100*20,000/(1,000,000 + 20,000) + 1.18 = -0.96 + 1.18 or -0.22 volts.


Line 243: Line 243:
What about the preferred but expensive LT1413N. In this circuit it can withstand voltages on its (+) and (-) pins equal to the supply voltage (e.g. +5 volts) and -5 volts.
What about the preferred but expensive LT1413N. In this circuit it can withstand voltages on its (+) and (-) pins equal to the supply voltage (e.g. +5 volts) and -5 volts.


'''Trick #3 -- Matched decent-quality (i.e. not carbon-film or carbon-composition) resistors:''' The "common-mode rejection ratio" -- the ability of the circuit to reject noise that is on ''both'' signal-leads simultaneously -- is determined by the precision of matching of the two resistors R1 and R2 to 0.1%, and the tworesistors to R3 and R4 0.1%. The exact R1/R3 ratio and R2/R4 (i.e. 20:1, 50:1, 100:1) is not particularly important, and will be good to +/-2% with typical 1% resistors selected to 0.1%. (All four can be matched for very precise division, but the resistors themselves can drift with temperature and time, and the amplifier's input offset will drift, as will the voltage reference). Ideally The resistors should not "drift" too much with temperature; but since they will, at about 25-50 parts per million per degree C (the CMF type), they all should be of the same metal-film construction. The recommended type is a tempertaure-time-voltage-stable metal-film 1/4 watt resistor such as the Vishay CMF55 or IRC RC55-LF.
'''Trick #3 -- Matched decent-quality (i.e. not carbon-film or carbon-composition) resistors:''' The R1/R3, R2/R4 is called a "bridge" circuit. Its "common-mode rejection ratio" -- the ability of the circuit to reject noise that is on ''both'' signal-leads simultaneously -- is determined by the precision of matching of the two resistor-''pairs'' R1/R3 and R2/R4 to 0.1%. The exact R1/R3 and R2/R4 ratio (i.e. 20:1, 50:1, 100:1) is not particularly important, but the two ''ratios'' should match to 0.1%. (All four resistors can be matched for very precise division, but the resistors themselves can drift with temperature and time, and the amplifier's input offset will drift, as will the voltage reference). Ideally the resistors should not "drift" too much with temperature; but since they will, at about 25-50 parts per million per degree C (the CMF type), they all should be of the same metal-film construction. The recommended type is a tempertaure-time-voltage-stable (50 parts-per-million) metal-film 1/4 watt resistor such as the Vishay CMF55.


One approach is to buy a bunch of decent resistors (e.g. the metal-film industrial Vishay/Dale CMF or IRC RC55LF-D type available from Mouser at www.mouser.com, a quantity of 100 of 10K cost about $7.00 US, a 1000 cost about $25.00 US) and then sort them into 0.1% categories. Unfortunately this requires a digital ohmmeter. But a trial sort of 100 showed extremely tight matching (+/-0.2% around the nominal). Another approach is to just buy precision resistors (at about 8-to-10x the cost):
One approach is to buy a bunch of decent resistors (e.g. the metal-film industrial Vishay/Dale CMF or IRC RC55LF-D type available from Mouser at www.mouser.com, a quantity of 100 of 10K cost about $7.00 US, a 1000 cost about $25.00 US) and then sort them into 0.1% categories. Unfortunately this requires a digital ohmmeter. But a trial sort of 100 showed extremely tight matching (+/-0.2% around the nominal). Another approach is to just buy precision resistors (at about 8-to-10x the cost):
Line 251: Line 251:
* Vishay/Dale PTF5620K000BYBF, 20K 0.1% 10 parts per million per deg C, 1/8 watt at 85C, max 300 volts @ $0.82 ea, or $0.53 each in 100s
* Vishay/Dale PTF5620K000BYBF, 20K 0.1% 10 parts per million per deg C, 1/8 watt at 85C, max 300 volts @ $0.82 ea, or $0.53 each in 100s


With larger-value resistors R1* = R2* = e.g. 1.0 megohm (the recommendation), measurements with a low-cost ohmeter become problematic and suspect due to phenomena such as test lead-noise. The recommended procedure is to buy the precision resistors:
With larger-value resistors R1 = R2 = e.g. 1.0 megohm (the recommendation), measurements with a low-cost ohmeter become problematic and suspect due to phenomena such as test-lead tribo-electric noise. An approach is to sort R3 and R4 to 0.1%, pick an R1 randomly, build a bridge circuit on a plug-in board, and test for R2 matching (the voltmeter goes between R1/R3 and R2/R4. Connect R1 and R2 to a higher voltage supply for more precision -- an acceptable error-reading will be on the order of 0.1 millivolt (100 microvolts).

The recommended procedure, if price is no object, is to buy the precision resistors:
* IRC RC55LF-D-1.0M, 0.1% 25 parts per million per deg C, 1/4 watt @ $1.08 each or $0.72 in 100's
* IRC RC55LF-D-1.0M, 0.1% 25 parts per million per deg C, 1/4 watt @ $1.08 each or $0.72 in 100's
* Vishay/Dale PTF651M0000BXBF, 0.1% 15 parts per million per deg C, 1/4 watt, 500 volt @ $1.68 each, or $1.28 in 100's
* Vishay/Dale PTF651M0000BXBF, 0.1% 15 parts per million per deg C, 1/4 watt, 500 volt @ $1.68 each, or $1.28 in 100's
Line 259: Line 261:
: Example: Differential-mode worst case: the laptop's power is derived from a "floating" source (e.g. the AC-to-DC adapter provided) but the (-) and (+) are plugged into the 230 + 10% VAC receptacle: (230 +10% VAC)^2/(2,000,000) = 0.05 watts, and 180 volts peak/1,000,000 = 180 microamps into and out of the probe's (+) and (-) inputs.
: Example: Differential-mode worst case: the laptop's power is derived from a "floating" source (e.g. the AC-to-DC adapter provided) but the (-) and (+) are plugged into the 230 + 10% VAC receptacle: (230 +10% VAC)^2/(2,000,000) = 0.05 watts, and 180 volts peak/1,000,000 = 180 microamps into and out of the probe's (+) and (-) inputs.


The tradeoff with larger-value resistors will be more circuit error due to leakage across surfaces, e.g. caused by dust and humidity and their combination. An enclosure becomes more important. More about this TBD.
The tradeoff with larger-value resistors will be more circuit error due to leakage across surfaces, e.g. caused by dust and humidity and their combination. An enclosure together with conformal coating becomes more important. More about this TBD.


'''Trick #5 -- Pick a decent-quality "centering" adjustment''':
'''Trick #5 -- Pick a decent-quality "centering" adjustment''':
:''As the laptop warms up the internal reference of the AD1888 "drifts" for about 10 minutes until the voltage stabilizes. This causes the centering to move in one direction or the other.''
:''As the laptop warms up the internal reference of the AD1888 "drifts" for about 10 minutes until the voltage stabilizes. This causes the centering to move in one direction or the other.''


This has proven to be the most difficult and potentially-expensive part of the design. "Trim-pots" can survive only about 200 adjustments (Bourns specification). A 500 full-cycle test of a Vishay/Sfernice trimpot destroyed it. An adjustment potentiometer that can withstand 25 adjustments a day x 200 days a year x 10 years = 50,000 cycles cost as much as the precision resistors and the amplifer. Alternatives:
This has proven to be the most difficult and potentially-expensive part of the design. "Trim-pots" can survive only about 200 adjustments (Bourns specification). A 500 full-cycle test of a Vishay/Sfernice trimpot destroyed it. An adjustment potentiometer that can withstand 25 adjustments a day x 200 days a year x 10 years = 50,000 cycles cost as much as the precision resistors and the amplifer. The favored potentiometers -- a Honeywell pc board-mounted pot and a pin equivalent Bourns pot -- P1 are shown on the Bill of Materials. At least one other manufacturer makes a "pot" that is pin-for-pin compatible.


'''Trick #6 -- Use only parts that can be hand-assembled''': Fewer and fewer parts are designed as "through-hole" with leads on 0.10 inch spacings (2.54 mm), and many very nice parts cannot be hand-assembled because of their tiny sizes.
*(i) A Bourns 3006P trimpot (~$1.50 USD single-unit cost. Works fine, but only 200 adjust life, plus has no "knob").


'''Trick #7 -- Bend the precision resistors' leads correctly (clamp needlenose pliers between resistor body, then bend lead against needlenose0 and do not apply too much solder-heat -- heatsink with needlenose or alligator clips when soldering ''': The IRC and Vishay/Dale resistors will drift +/-0.03% (typical) due to soldering. They also don't like physical shock (+/-0.03% typical). Precision components require precision treatment.
*(ii) A Bourns 3/8" sealed trimpot (~$0.85. but only 200 cycle adjust life, plus has no "knob". Hard to adjust at high gain until the laptop's drift settles down).

*(iii) A Vishay/Sfernice sealed trimpot with a 1/2 diameter knob ($1.92. It failed a 500 cycle hand-test, the specs are ??)

*(iv) A Bourns encoder PEC11-4220F-N0012 plus a Catalyst CAT5113LI-10: 10K, UP/DOWN Digital Potentiometer plus a debounce circuit:($1.20 encoder + $1.50 Digital Pot IC + ~$0.35 74HC132 + 2*~$0.20 + 4*$.07 = $3.88. This does not have a "memory" -- a center-point could be "programmed in" during its build. Is on the edge of too large.)

*(v) Two pushbutton switches + 74HC132 quad-nand schmitt-input plus 4 capacitors, 6 resistors (~$0.50 x 2 + $1.50Digital Pot IC + ~$0.35 quad nand + ~$1.00 capacitors + 6*$0.07 resistors = $4.37).

*(vi) 10K Linear slide-switch. Cannot be easily sealed, don't use carbon comp!!

'''Trick #6 -- Use only parts that can be hand-assembled''': Fewer and fewer parts are designed as "through-hole" with leads on 0.10 inch spacings (2.54 mm), and many very nice parts cannot be hand-assembled because of their tiny sizes.


'''Trick #8 -- Ruggedize the mechanical design''': The pc board itself is rather rugged. But leads into and out of an enclosure will be subject to stress and strain. Potentiometers will suffer blows. Best practices would either "sink" the plane of a control below the surface of the enclosure or as low as possible (a good example is the laptop's controls).
'''Tirck #7 -- Bend the precision resistors' leads correctly (clamp needlenose pliers between resistor body, then bend lead against needlenose0 and do not apply too much solder-heat -- heatsink with needlenose or alligator clips when soldering ''': The IRC and Vishay/Dale resistors will drift +/-0.03% (typical) due to soldering. They also don't like physical shock (+/-0.03% typical). Precision components require precision treatment.


A suggested, alternative design "floats" the pc board inside its enclosure and pushes it (i.e. the potentiometer P1) against the enclosure's lid with a small circle of stiff polyethylene foam. If a child pounds the knob downward, or the enclosure falls on the floor with the knob hitting, there is enough compliance in the potentiometer's shaft to protect it -- the knob hits the lid before the pc board "bottoms out". A sideways blow could be damaging but will probably "flip" the enclosure.
'''Trick #8 -- Ruggedize the mechanical design''': Leads into and out of an enclosure will be subject to stress and strain. Potentiometer adjustments will suffer blows. Best practices would either "sink" the plane of the control below the surface of the enclosure or as low as possible (a good example is the laptop's controls).


== Bill of Materials (BOM) ==
== Bill of Materials (BOM) ==

Revision as of 03:47, 13 February 2008

Temperature monitoring system

How to build

You need-

  1. LM35 or any similar temperature sensor
  2. USB connector
  3. Audio connector
  4. Some wire or a cable. You could use cut a USB cable and an audio cable from the middle
  5. Measure Activity


The USB port has two pins which are of use to us. One is the ground and the other one is the one which provides a constant 5V. When you cut a USB cable in the middle, you could find out by trial-and-error by using a simple Voltmeter which are the two wires corresponding to these two pins. If you don't have a Voltmeter, try using a simple LED and a resistor to do the job.

We provide the +5V that this temperature sensor requires from the USB port. For this we connect the +5V wire to the Vcc of the sensor. We connect the ground to the Ground terminal on the LM35. We connect the Vout of the sensor to the Audio Input jack.


How to operate

Set the Measure Activity to be in the DC Mode and Turn Off the Bias Voltage. Both these can be controlled by using the first two buttons on the 'Measure Toolbar'.

The DC voltage is proportional to the temperature . The voltage that the XO can read varies between 0.9 V - 3V. The output of the sensor would be from 0-5V, so, for for different ranges of temperature one can either use a potential divider for higher ranges or using resistances add a bias to the Vout of the sensor from the +5V of the USB again.


Learn More

The LM35 has an output voltage proportional to the temperature. The scale is 0.01V/C. Read the [Data Sheet] to learn more.

One could log temperature over a period of time - say at an interval of 1 hour for a complete day to know when is it the hottest / coldest.

Video

(put video with flame demonstration here, put pictures here - including ckt diagram)

Intrusion Alarm system

How to build

You need-

  1. A toy laser
  2. An LDR or a Photo-Voltaic cell
  3. Some wire
  4. Measure Activity

Open up the toy laser and remove the batteries. Connect a wire to the spring inside and connect the other wire to the metallic body inside. Connect both of these wires to the USB power supply wires. Click here to learn about how to determine the USB power supply lines.

Connect the LDR to the Audio Jack.

How to operate

Set the Measure Activity into DC Mode and turn ON the bias voltage. Setup the LDR and the laser to be in line and facing each other. When the path of the laser light gets cut, you should observe a marked increase in voltage indicated by the position of the waveform shifting up.

Learn More

The resistance of the LDR - Light Dependent Resistor is proportional to the amount of light falling on it. The potential drop across the LDR is dependent on the resistance of the LDR. When the path of the light gets cut, the resistance increases hence we notice a marked increase in voltage indicated on the screen.

Click here to know more about Bias Voltage and how it works.

Simple isolation and protection amplifier

The following "probe-circuit" allows the student or teacher to "input" both DC voltages greater than +5.0 volts (such as from a 9 volt battery, etc), negative voltages, as well as AC voltage. The probe provides resistive isolation of the laptop from what is being measured (the amount of isolation dictated by the choice of two resistors R1 and R2) The following design may change as it evolves:

  • See revision history at bottom of this section.
Oscilloscope circuit 2.png





















Specifications

These specifications are valid given use of the preferred components as listed in the Bill of Materials:

  • Signal-input plug: mono audio jack (3.5 mm, 1/8", see above)
  • Power-input plug: USB connector
  • Use either a "naked" connector or a USB cable cut and stripped to reveal the red (+5 volt) and black (0 volt -- common) wires. Cut the other two wires flush and tape-over their ends individually.
  • Voltage Gain: R2/R1 (i.e. the circuit divides the input down by this ratio)
  • Accuracy will depend on the ratios of R1:R3 and R2:R4.
  • Common-mode Input impedance: R1 ohms (e.g. 100K or 200K or 1 megohm) each side to circuit common
  • Differential-mode Input impedance: twice the common-mode impedance
  • Laptop settings: Set either (i) MEASURE to DC and OFFSET to OFF, or (ii) MEASURE to AC and OFFSET to OFF
  • No harm will result if these setting are wrong, e.g. the activity starts in MEASURE = AC. The gain will be higher in AC, but the amplfier can be used in AC too).
  • +5 volt power source -- derived from the nearby USB port (red and black wires inside a sliced-open USB cable)
  • +5 volt current draw -- a few milliamperes
  • Minimum power-source voltage -- approximately +4.0 volts [as low as +3.0 volts is possible].
  • Measureable input voltage: See also AC and DC common-mode rejection specifications. The "gain slider" is the adjustment on the right-hand side of the screen. The trace is adjusted to center-screen and then voltage applied to the (+) and (-) probe inputs; the voltage is adjusted until the trace moves to fully up (+ voltage), or down (- voltage):

Gain of 1/100, i.e. R1 = R2 = 1.00 megohms 0.1%, R2 = R3 = 10.0K ohms 0.1%:

Gain slider Maximum +V input, volts Maximum -V input, volts Volts per major division
Full up 0.367 -0.368 approx 0.050
75% up 2.126 -2.036
50% up 13.13 -12.62
25% up 78.5 -73
12.5% up 80 -80 approx 10

Gain of 1/50, i.e. R1 = R2 = 1.00 megohms 0.1%, R2 = R3 = 20.0K ohms 0.1%:

Gain slider Maximum +V input, volts Maximum -V input, volts Volts per major division
Full up 0.185 -0.189 approx 0.025
12.5% up 40 -40 approx 5


  • Frequency specification: DC to 1000 hertz (1 KHz). Beyond this frequency significant aliasing occurs.
  • "Aliasing" appears as wiggles on what should be a pure sine wave (an envelope that modulates the input). This is caused by the sampling of the waveform done by the input-converter (codec).
Trace-speed slider Minimum frequency Maximum frequency
Right - fast TBD TBD
Center - medium TBD TBD
Left - slow TBD TBD
  • DC Common-mode rejection: Voltage injected between (i) the two inputs tied together, and (ii) circuit common (circuit 0 volts), tested to +/-90 volts DC. Measurements TBD.
  • DC Common-mode rejection depends critically upon the matching of resistors R1 to R2 and R3 to R4. Matching of 0.1% is required either by buying 0.1% resistors, or buying e.g. a 100 or a 1000 of inexpensive metal-film resistors and sorting them.
  • AC Common-mode rejection: measured relative to the power-common (circuit 0 volts), 0-1000 Hz, +6 volts AC rms injected: less than 0.001 volt AC rms (observable at the highest gain setting of the laptop: approx 2 scale divisions)
  • AC common mode rejection depends critically not only on the matching of the resistors R1 to R2 and R3 to R4 but also upon stray capacitance in the circuit, the apparent impedance of the reference-voltage source (i.e. the reference-op amp's output impedance), the quality of the op-amps themselves, etc.

Circuit description

This deceptively-simple "probe" circuit is called a "differential amplifer" (differencing amplifier) and although not very fancy it performs the fundamental task of isolating and protecting the laptop from adverse electrical events while allowing the student to measure voltages in excess of +5 volts, minus voltages, etc.

The circuit requires some design tricks and represents some design-challenges, to be described below.

This circuit has the advantage of not presenting an input voltage greater than +5 volts to the microphone input, plus the USB-port's circuit limits its +5 volt output-current to about 1 ampere in case something short-circuits the USB's output voltage.
  • If properly designed, this circuit should survive "accidents" and protect the laptop and the child against such accidents as: "plugging the probe into the 240VAC wall-outlet".
  • Any supply voltage derived from another source must be less than +5 volts. For example, three 1.5 volt batteries will also work at +4.5 volts -- as long as the batteries are not accidentally reversed !! -- but four 1.5 volt batteries at 6.0 volts present a risk of inputting more than +5 volts into the microphone input.
  • Any externally-derived power voltage (batteries solar cells, etc) directly applied also risks of reveral. If a battery is to be used, add a reverse-protection schottky diode e.g. Avago 1N5711 static-protected by a 0.1 uf capacitor in series with the battery (i.e. a 4.5 volt battery with a 0.4 volt diode drop). A better scheme uses a low-current, +5 volt low-dropout-voltage regulator. Details TBD.

Trick #1 -- "Differential" rather than "single-ended" input: A single-ended amplifer connects one of its two inputs directly to the circuit common -- the metallic part of the USB port, for instance. Any current or voltage on the common goes directly into the laptop. So if somehow such a probe's input-lead is inadvertenly connected to a high voltage, this voltage will appear on the metallic parts of the laptop. If the laptop is electrically isolated, this adverse voltage will present a touch-hazard. If the laptop is electrically grounded, it will be destroyed.

A differential amplifier's two inputs, a "plus" (+) and "minus" (-) are both somewhat isolated from the laptop's common by high-value resistors (true "galvanic" isolation is possible but expensive).

As a properly-designed single-ended amplifier will require use of the same 4 resistors (R1*, R2*, R3**, and R4**) as the "differential" amplifer. Moreover, the differential amplifer subtracts away "common-mode" noise Vn and common-mode voltage Vcm i.e. noise and voltages that appear on both leads (relative to circuit common): i.e.

( V(+) + Vn + Vcm) - ( V(-) + Vn + Vcm ) = V(+) - V(-)

So it makes sense to design the amplifer as a differential amplifer.

Trick #2 -- Artificial "ground": This trick kills two problems with one stone. (i) The laptop accepts only positive voltages: The laptop's analog-to-digital converter (codec: AD1888) converts (works only on) plus-DC voltages between approximately +0.3 volts (minimum, corresponding to bottom of screen with high gain (control near the top)) and approximately +2.0 volts (maximum, corresponding to top of screen with high gain (the slider control near the top)). this means that approximately +1.18 volts puts the trace into the center of the screen.

Thus the probe-circuit must "offsets" the difference between the "+" and "-" inputs with this voltage of approximately +1.18 volts. It derives this from a highly-stable "band-gap reference" (any number of different types are possible -- +1.225 volts +/-1% volt, +1.235 volt +/-1%, 2.50 volts +/-1% etc.) Audio-jack inputs less than this "centering-voltage" will drive the trace lower than center, and inputs greater than this "centering-voltage" will drive the trace higher than center. At the laptop's highest gain the voltages require to do this are quite small -- on the microvolt range.

(ii) Increasing common-mode voltage rejection: Together with the choice of R1 and the gain of the circuit, this +1.18 volt offset dictates the maximum common-mode voltage that the probe can "accept" and still function correctly. For example, with R1* = 1 megohm and an attenuation of 1:50 (i.e. the probe reduces the input to 1/50*( V(+) - V(-) ), a negative "common mode" input will be divided down by a factor of 50. If this common-mode voltage V(+) is negative, it will suck out out of the input approximately Vin/(R1+R2) amps, and this will develop a voltage at the input to the amplifier of Vin*R2/(R1+R2)+1.18 volts.

Example: Common mode voltage is -100 volts, R1* = 1 megohm, R2 = 10K:
-100*20,000/(1,000,000 + 20,000) + 1.18 = -0.96 + 1.18 or -0.22 volts.

Can the amplifier (LM358AN) tolerate this? Yes, it can tolerate up to -0.3 volts at its (+) and (-) pins. What really happens depends on the circuit design. The part can tolerate up to 50 milliamps being sucked from the (+) or (-) pins. And the resistors prevent any more than a few hundred microamps from being pulled from the pins. So the amplifier will survive.

What about the preferred but expensive LT1413N. In this circuit it can withstand voltages on its (+) and (-) pins equal to the supply voltage (e.g. +5 volts) and -5 volts.

Trick #3 -- Matched decent-quality (i.e. not carbon-film or carbon-composition) resistors: The R1/R3, R2/R4 is called a "bridge" circuit. Its "common-mode rejection ratio" -- the ability of the circuit to reject noise that is on both signal-leads simultaneously -- is determined by the precision of matching of the two resistor-pairs R1/R3 and R2/R4 to 0.1%. The exact R1/R3 and R2/R4 ratio (i.e. 20:1, 50:1, 100:1) is not particularly important, but the two ratios should match to 0.1%. (All four resistors can be matched for very precise division, but the resistors themselves can drift with temperature and time, and the amplifier's input offset will drift, as will the voltage reference). Ideally the resistors should not "drift" too much with temperature; but since they will, at about 25-50 parts per million per degree C (the CMF type), they all should be of the same metal-film construction. The recommended type is a tempertaure-time-voltage-stable (50 parts-per-million) metal-film 1/4 watt resistor such as the Vishay CMF55.

One approach is to buy a bunch of decent resistors (e.g. the metal-film industrial Vishay/Dale CMF or IRC RC55LF-D type available from Mouser at www.mouser.com, a quantity of 100 of 10K cost about $7.00 US, a 1000 cost about $25.00 US) and then sort them into 0.1% categories. Unfortunately this requires a digital ohmmeter. But a trial sort of 100 showed extremely tight matching (+/-0.2% around the nominal). Another approach is to just buy precision resistors (at about 8-to-10x the cost):

  • IRC RC55LF-D-10K, 0.1% 25 parts per million per deg C, 1/4 watt at 70C, $0.96 each or $0.64 each in 100's
  • IRC RC55LF-D-20K, 0.1% 25 parts per million per deg C, 1/4 watt at 70C, $0.96 each or $0.64 each in 100's
  • Vishay/Dale PTF5610K000BYBF, 10K 0.1% 10 parts per million per deg C, 1/8 watt at 85C, max 300 volts @ $0.82 ea, or $0.53 each in 100s
  • Vishay/Dale PTF5620K000BYBF, 20K 0.1% 10 parts per million per deg C, 1/8 watt at 85C, max 300 volts @ $0.82 ea, or $0.53 each in 100s

With larger-value resistors R1 = R2 = e.g. 1.0 megohm (the recommendation), measurements with a low-cost ohmeter become problematic and suspect due to phenomena such as test-lead tribo-electric noise. An approach is to sort R3 and R4 to 0.1%, pick an R1 randomly, build a bridge circuit on a plug-in board, and test for R2 matching (the voltmeter goes between R1/R3 and R2/R4. Connect R1 and R2 to a higher voltage supply for more precision -- an acceptable error-reading will be on the order of 0.1 millivolt (100 microvolts).

The recommended procedure, if price is no object, is to buy the precision resistors:

  • IRC RC55LF-D-1.0M, 0.1% 25 parts per million per deg C, 1/4 watt @ $1.08 each or $0.72 in 100's
  • Vishay/Dale PTF651M0000BXBF, 0.1% 15 parts per million per deg C, 1/4 watt, 500 volt @ $1.68 each, or $1.28 in 100's

Trick #4 -- Use higher-voltage input resistors that can handle extreme overloads : Resistors actually have voltage specifications as well as power specifications. An ideal circuit should survive, and protect the laptop against, 10% high line 230VAC, i.e. 358 volts peak, the common household voltage in much of the world. The industrial voltage 400 +10% VAC 50 Hz i.e. 622 volts peak, or 480 +6% VAC 60Hz i.e. 720 volts peak in the Americas will be destructive). If equipped with 1 megohm, higher-voltage (400-500 volt resistors) input-resistors R1* and R2* in both "legs" of the probe should survive this "household" voltage. This corresponds to 0.07 watts of power, 400 microamps in each resistor common mode. But in fact in differential mode the probe divides this power and leakage in half.

Example: Common-mode worst case: the laptop power puts earth-ground potential but the probes are both plugged into 230 +10% VAC and this is referenced to the same earth common: (230 +10% VAC)^2/(1,000,000) = 0.105 watts, and 358 volts peak/1,000,000 = 385 microamps into and out of the probe's (+) and (-) inputs and the laptop's common (0 volts of the USB plug).
Example: Differential-mode worst case: the laptop's power is derived from a "floating" source (e.g. the AC-to-DC adapter provided) but the (-) and (+) are plugged into the 230 + 10% VAC receptacle: (230 +10% VAC)^2/(2,000,000) = 0.05 watts, and 180 volts peak/1,000,000 = 180 microamps into and out of the probe's (+) and (-) inputs.

The tradeoff with larger-value resistors will be more circuit error due to leakage across surfaces, e.g. caused by dust and humidity and their combination. An enclosure together with conformal coating becomes more important. More about this TBD.

Trick #5 -- Pick a decent-quality "centering" adjustment:

As the laptop warms up the internal reference of the AD1888 "drifts" for about 10 minutes until the voltage stabilizes. This causes the centering to move in one direction or the other.

This has proven to be the most difficult and potentially-expensive part of the design. "Trim-pots" can survive only about 200 adjustments (Bourns specification). A 500 full-cycle test of a Vishay/Sfernice trimpot destroyed it. An adjustment potentiometer that can withstand 25 adjustments a day x 200 days a year x 10 years = 50,000 cycles cost as much as the precision resistors and the amplifer. The favored potentiometers -- a Honeywell pc board-mounted pot and a pin equivalent Bourns pot -- P1 are shown on the Bill of Materials. At least one other manufacturer makes a "pot" that is pin-for-pin compatible.

Trick #6 -- Use only parts that can be hand-assembled: Fewer and fewer parts are designed as "through-hole" with leads on 0.10 inch spacings (2.54 mm), and many very nice parts cannot be hand-assembled because of their tiny sizes.

Trick #7 -- Bend the precision resistors' leads correctly (clamp needlenose pliers between resistor body, then bend lead against needlenose0 and do not apply too much solder-heat -- heatsink with needlenose or alligator clips when soldering : The IRC and Vishay/Dale resistors will drift +/-0.03% (typical) due to soldering. They also don't like physical shock (+/-0.03% typical). Precision components require precision treatment.

Trick #8 -- Ruggedize the mechanical design: The pc board itself is rather rugged. But leads into and out of an enclosure will be subject to stress and strain. Potentiometers will suffer blows. Best practices would either "sink" the plane of a control below the surface of the enclosure or as low as possible (a good example is the laptop's controls).

A suggested, alternative design "floats" the pc board inside its enclosure and pushes it (i.e. the potentiometer P1) against the enclosure's lid with a small circle of stiff polyethylene foam. If a child pounds the knob downward, or the enclosure falls on the floor with the knob hitting, there is enough compliance in the potentiometer's shaft to protect it -- the knob hits the lid before the pc board "bottoms out". A sideways blow could be damaging but will probably "flip" the enclosure.

Bill of Materials (BOM)

The following BOM specifies relatively good-quality components. Prices of the resistors assumes a 100-piece buy; all other prices (e.g. bandgap reference, LM358AN, potentiometer) are single-unit. 100-piece buys would cut the cost by at least 1/3 if not more. Substitution of lesser-quality components (e.g. carbon composition resistors, generic LM358N, two 1N4148 signal diodes in series in place of the bandgap reference, single-turn "trimmer" potentiometer) would cut the cost even further but significantly degrade performance.

Part Description Preferred Manuf Preferred Part Number Comments Qty each, USD 100 piece price 1000 piece price 1-piece price total 100-piece price total 1000 piece total
C1 Capacitor: 0.1 microfarad (uf), 10%, 63 V, metallized polyester, 0.20 inch lead spacing, box style Epcos B32529C334K Visay/ Roederstein MKT1817410065 at $0.25/0.18/0.14; AVX BQ014D01044K at $0.11/0.06/0.036 1 0.17 0.066 0.046 0.17 0.07 0.046
C2 Capacitor: 0.33 microfarad (uf), 10%, 63 V, metallized polyester, 0.20 inch lead spacing, box style Epcos B32529C334K Visay/ Roederstein MKT1817433065 at $0.26/0.138/0.077; AVX BQ024D03344K at $0.21/0.1125/0.0675 1 0.26 0.138 0.077 0.26 0.14 0.077
IC1 IC: Dual op-amp, single-supply 3 volt min, 3 mv max input offset, 30 nA input offset current max, 8-pin DIP Fairchild Semi, National Semi LM358AN or LM258AN ST Micro, Texas Instruments, etc. 1 0.34 0.24 0.138 0.34 0.24 0.138
(IC1) IC: Precision, dual op amp, single supply 3 volt min, 0.5 mv max input offset, 23 na max input current, 8 pin DIP Linear Technology LT1413CN8 1 2.50 2.50 2.00
LED1 LED: green clear, 20 millicandela, 2.2V forward voltage, T-1 3/4, 0.1 inch lead spacing Kingbright WP7113SGC Any color, any manufacturer: Avago HLMP-3507 1 0.13 0.10 0.07 0.13 0.10 0.07
(P1) Potentiometer, single turn, conductive plastic, PC mount, 10K ohm +/-20%, 50000 cycle life Honeywell 574SX1M48F103SD PC-mount hole pattern is same as Bourns and other manuf. 1 3.10 1.68 1.445
P1 Potentiometer, single turn, conductive plastic, PC mount, 10K ohm +/-20%, 50000 cycle life Bourns PCW1JC24BAB103L IP 40 rating, so could be potted 1 5.39 2.77 2.16 5.39 2.77 2.16
P2 Potentiometer, trimmer, sealed, single turn, 1/4 inch cermet, 1K, 10%, 100 ppm/deg C, vertical mount Bourns 3362X-1-102LF Or equivalent staggered pinout e.g. 3/8 inch 3386X-1-102LF or BI Technologies equivalents, Vishay/Sfernice equivalents 1 0.83 0.65 0.555 0.83 0.65 0.555
R1*, R2* Resistor: 1.00 meg, 1%, 1/2 watt commercial, metal film, matched to 0.1% Vishay/Dale CMF601M0000FHEK Generic metal film 1%, 100ppm/deg C: RN60D1002F 2 0.18 0.15 0.13 0.36 0.30 0.26
(R1*,R2*) Resistor: 1.00 meg, 0.1%, 1/2 watt precision, metal film, 10-25 ppm/deg C Vishay/Dale PTF651M0000BYBF also: IRC RC55LF-D-1.0M, 25 ppm, 200 volt,L $0.96/$0.64 2 1.87 1.21 0.94
R3**, R4** Resistor: 20.0K, 1%, 1/4 watt, metal film, matched to 0.1%, 50 ppm/deg C or better Vishay CMF5520K000FHEK Generic metal film 1%, 100ppm/deg C: RN55D2002F, matched to 0.1% 2 0.14 0.07 0.055 0.28 0.14 0.11
( R3**, R4** ) Resistor: 20.0K, 0.1%, 1/4 watt precision, metal film, 10-25 ppm/deg C Vishay PTF5620K000BYBF also: IRC RC55LF-D-20K, 0.1%, 25 ppm, 200 volt: $0.96/$0.64 2 0.88 0.685 0.55
R5, R6, R11 Resistor: 10.0K, 1%, 1/4 watt commercial, metal film, 50 ppm/deg C Vishay CMF5510K000FHEK Generic metal film 1%: RN55D1002F 3 0.14 0.07 0.055 0.42 0.21 0.165
R7 Resistor: 8.66K, 1%, 1/4 watt commercial, metal film, 50 ppm/deg C Vishay CMF558K660FHEK Generic metal film 1%: RN55D8661F 1 0.14 0.07 0.058 0.14 0.07 0.058
R8, R9 Resistor: 20.0K, 1%, 1/4 watt commercial, metal film, 50 ppm/deg C matched to 0.1% Vishay CMF5520K000FHEK Generic metal film 1%, 100ppm/deg C: RN55D2002F, matched to 0.1% 2 0.14 0.07 0.055 0.28 0.14 0.11
R10 Resistor: 499 ohm, 1%, 1/4 watt commercial, metal film Vishay CMF55499R00FHEK Generic metal film 1%: RN55D4990F 1 0.19 0.15 0.11 0.19 0.15 0.11
R12 Resistor: 1K, 1%, 1/4 watt commercial, metal film Vishay CMF5510K00FHEK Generic: RN55D1001F 1 0.14 0.10 0.058 0.14 0.10 0.058
TB1 Terminal block: screw-type, 5.08mm pitch, 2 terminal, vertical mount, 10mm height, with clamp Molex 39880-0302 2 0.55 0.29 0.26 1.10 0.58 0.52
(TB1) Terminal block: screw-type, 5.08mm pitch, 2 terminal, vertical mount, 10mm height, with clamp Phoenix Contact MKDSN 1.5/4-5.08 1 1.44 1.03 0.92
( Z1 ) IC: Bandgap reference, 1.225 volt fixed, 1%, TO-92 package ST Micro TS821AIZ Obsolete? 1 0.64 0.40
Z1 1.235 volt fixed shunt reference, 1%, 80ppm/deg C, TO-92 ON Semiconductor LM385BZ-1.2RA or RAG Prices vary, as low as $0.29 1-piece 1 0.62 0.40 0.275 0.62 0.40 0.275
( Z1 ) 2.50 volt fixed shunt reference, 1%, 80ppm/deg C, TO-92 ON Semiconductor LM385BZ-2.5RA or RAG etc. Requires resistor changes: R6--> 20.0K, R9 --> not used, R7 --> 8.45K 1 0.62 0.40 0.275
( Z1 ) 2.50 volt 1% shunt regulator/reference, TO-92 Texas Instruments TL431ACLP Pinout different & requires resistor changes: R6--> 20.0K, R9 --> not used, R7 --> 8.45K 1 0.40 0.27 0.135
Total Total Total 10.65 6.05 4.71
PCB PC Board: 0.062 inch glass-epoxy FR4, 1 oz copper two sides, plated-through holes, 2.913 x 1.230 inch, 67 holes standard sizes, no solder masks or component identifiers AP Circuits --- For budgetary (estimation) purposes 1 8.60 4.97 3.34 8.60 4.97 3.34
Case & lid Plastic, translucent blue: 3.150 x 1.575 x 0.787 inch Hammond Manufacturing 1551KTBU 1 1.84 1.32 1.19 1.84 1.32 1.19
knob Knob diameter TBD -- requires "skirt recess" of 0.11 inch, 0.5 inch shaft depth, knob height approx 0.65-0.75 inch Rogan RB67-2 ? Radio Shack knob 274-433 (1 inch dia) or 274-403 (1/2 inch dia) require rework (use 0.25 inch drill + end mill to increase shaft-depth to 0.5 inch) 1 2.00 1.00 1.00 2.00 1.00 1.00
Sub-total Sub-total Sub-total 12.44 7.29 5.53
PL1 Plug, mono audio 3.5 mm (1/8 inch), black plastic barrel and strain relief CUI Inc MP3-3501 Has somewhat flexible strain relief 1 0.76 0.51 0.38 0.76 0.51 0.38
(PL1) PL1: Plug, mono audio 3.5mm (1/8 inch), black plastic barrel Kobiconn Mouser 171-PA-3191-1-E Generic 1 0.71 0.50 0.42
PL2 USB plug: "silver/white", A-type 4 pin, "naked" (no attached wires) Kobiconn Mouser 154-UAW16-E Assmann A-USBPA-R: $0.40/0.24/? 1 0.68 0.42 0.35 0.68 0.42 0.35
PL2 USB plug-hood: for Kobiconn 154-UAW-16-E Kobiconn Mouser 154-UAC16-E Acts as a strain relief. Assmann A-USBPA-HOOD-R $0.36/0.22/? 1 0.54 0.33 0.27 0.54 0.33 0.27
test-lead wire RED PVC insulation, 28 AWG 19x40 stranding, 0.018 inch insulation, 0.047 inch total diameter. www. e-z-hook.com 9501-100-red; 9501-1000-red Do not use solid wire. Units in feet. 1000 foot roll is $160 US. Alpha UL1007 may work at $105/1000 ft. 3 0.228 0.228 0.16 0.684 0.68 0.48
test-lead wire BLACK PVC insulation, 28 AWG 19x40 stranding, 0.018 inch insulation, 0.047 inch total diameter. www. e-z-hook.com 9501-100-black; 9501-1000-black Do not use solid wire. Units in feet. Alpha UL1007 28 AWG 7x36 may work at $105/1000 ft. 3 0.228 0.228 0.16 0.684 0.68 0.48
heatshrink Very-flexible polyolefin 1/4 inch final diameter, black 3M VFP-876 1/8" x length, black $2.49 per 5', $28.53 per 100', $103.01 per 500' 1.8 0.498 0.2853 0.103 0.90 0.51 0.1854
heatshrink Very-flexible polyolefin 3/32 inch final diameter, black 3M VFP-876 3/32" x length, black $2.31 per 5', $26.45 per 100', $95.48 per 500' 0.5 0.46 0.265 0.191 0.23 0.13 0.0955
heatshrink Very-flexible polyolefin 3/16 inch final diameter, black 3M VFP-876 3/16" x length, black $3.03 per 5', $30.70 per 100' 0.1 0.606 0.307 0.307 0.06 0.03 0.0307
heatshrink Flexible polyolefin 3/32 inch final diameter, RED 3M FP332R x length $1.92 per 5', $22.03 per 100' 0.15 0.384 0.220 0.220 0.06 0.03 0.033045
strain relief C-ring Crimp-ring: 0.125 inch inner-diameter, 14 AWG (0.062) copper or aluminum wire Handmade --- Wind e.g. 16 turns on .125 inch steel mandrel, grind a slot through all turns simultaneously (use 2 abrasive Dremel grinding wheels on Dremel arbor) -- makes 16 C-rings 2 0.01 0.01 0.01 0.02 0.02 0.02
Foam cushion 0.125 thick, approx 0.5 inch diameter Seattle Fabrics TBD -- stiff polyethene foam For knob compliance (impact protection) -- glue beneath pot P1 1 0.01 0.01 0.01 0.01 0.01 0.01
Sub-total Sub-total Sub-total 4.61 3.35 2.32
Total Total Total 27.70 16.70 12.57
Optional
minihook Standard construction, square-hole tip, "minihook" to minihook, 36 inches red 22 AWG 66x40 www. e-z-hook.com 204-36W-red $4.32/3.90/3.25 divided by 2 1 2.16 1.95 1.62 2.16 1.95 1.62
minihook Standard construction, square-hole tip, "minihook" to minihook, 36 inches black 22 AWG 66x40 www. e-z-hook.com 204-36W-BLK $4.32/3.90/3.25 divided by 2 1 2.16 1.95 1.62 2.16 1.95 1.62

More detail to follow

Construction Details

Because of 50-60 Hz AC interference, prototype the pc board on a double-sided laminate -- e.g. 1 oz copper on 0.062 inch thick glass-epoxy laminate. Glass-epoxy boards are hard on drills -- use a hardened-steel or carbide drill (e.g. 0.038 inch). Place all interconnecting traces on the top (component) side and leave as much copper as possible while maintaining around the components' pads. Leave the underside all copper excepting the components' pads and a few traces (required only for a hand-built version -- a fabricated pc board with plated-through holes would require no bottom traces. The exception to the 'as much copper as possible rule' is under resistors R1 and R2. For the "hot" end of these components use wide clearances and remove component-side copper -- as a precaution in case someone plugs the input into 240VAC.
This is drawn from the USB 2.0 "A" type plug specifications to be found at www.usb.org. The particular insert shown is the Kobiconn 154-UAW16-E plug. A "hood" is also available: 154-UAC16-E. Drawings are available at www.mouser.com.

This circuit was actually prototyped -- without use of an oscilloscope -- on a "plug board" purchased from Radio Shack. Components were purchased from www.mouser.com and www.digikey.com. A homemade cable from the pc board to a "naked" USB 2.0 plug a "pigtail" for the audio plug is described below.

Construction method: Crude point-to-point wiring on a piece of "perf" board (unplated PC board material perforated on 0.1 inch spacings) exhibits "AC pickup" (capacitively-coupled interference between the student's hand and the components and traces of the pc board). For this reason, this method should be abandoned in favor of a two-sided pc board either hand-fabricated with top- and bottom-soldering, or a commercially-fabricated as a double-sided plated through pc board.

To hand-fabricate such a board, tools required include a digital voltmeter, a steel dial calipers (to act as a ruling scribe as well as a measuring device), a Dremel tool mounted in a "drill press" and quipped with a 0.032 inch drill etc., safety glasses (required for careful observation while drilling and soldering), soldering iron with pointy tip, thin solder with rosin or other activating agents in the solder, solderwick, needlenose pliers, miniature flush cutters, a wire stripper, etc. Tape and a small vise or other holding device (e.g. for fishermans' fly-tying) will be very useful.

Use the dial calipers' sharp points to rule a 0.10 inch grid over the surface of the board. Carefully locate and mark each hole with a prick-punch (a sharpened steel nail will do). Then drill the holes at a high speed-setting. (The copper will be on the bottom, the components on the top). Carefully mark and then remove etch with a Dremel tool -- the trick is to remove, with a tiny Dremel burr (#106 0.032 burr is best, a #107 0.062 burr will work in a pinch), as little etch as possible. Use a cork-backed, stainless steel ruler as a guide for both the marking pen and when cutting the etch away with the Dremel tool. If the copper is tinned, so much the better -- bare copper will tarnish; although this does not affect the circuit, soldering is more difficult.

Kits exist to etch pc boards. See www.mouser.com or www.digikey.com. A commercially-fabricated pc board is preferred. (Details TBD). The following book is excellent -- Al Williams, 2004, Build Your Own Printed Circuit Board, McGraw-Hill, New York, ISBN0-07-142783-X. The book includes a free-ware CD (Eagle software) for Windows or Linux that permits schematic capture and double-sided pc board layout.

Solder overheating, cold-solder joints, etc: Use rosin-core solder (expensive). Practice your technique. The components can take a few seconds of soldering heat. But prolonged application of heat, or the use of huge soldering tips (e.g. don't use a soldering "gun"), and monster blobs of solder will overheat the matched resistors R1 and R2, R3 and R4, the voltage reference Z1 and the operational amplifer IC, melt away the PVC insulation of the lead-wires, etc. Be on the lookout for "cold-solder" joints -- places where the solder doesn't flow nicely and leaves the connection "unmade".

Conformal coating: Manufacturers use sprays (expensive) to "conformally coat" their boards for water-, dust- and fungus-resistance. Use multiple coats. See www.mouser.com or www.digikey.com. Be careful to tape over the output connector and any "entry points" of potentimeter P1.

Assembling into a box, Potting: The assembly should be put into a little plastic box. "Potting" (casting the board in a compatible 2-part RTV rubber) would be ideal excepting that the potentiometer P1 may "leak", and once potted the circuit may not be repairable, and the compounds are expensive. The better alternative is to conformally-coat the board and enclose in the suggested box (more measuring and drilling is required). Details TBD.

  • Tools required: wire-stripping tool, Exacto knife, prototyping plug-board, little insulated jumper-wires (solid wire okay), ball clips, digital voltmeter, mold and molding materials if used.
  • Soldering tools: rosin-core solder (NOTE: country-specific regulations may not allow use of lead-tin solder), soldering iron with small tip, solder-wick.

Best practices: The most difficult part of the design is how to assemble it into a robust object suitable for use by children. (1) The current recommendation is to have the board loose (somewhat -- 0.015 inch on each side) in the box but pushed up against the cover by a couple circles of stiff polyethylene foam. The knob should have just enough clearance so if a child pushes hard on the knob the whole board is pushed into the box, but then springs back and pushes the board up into position. (2) Wires into the USB and audio plugs must be strain-relieved by bending-over of the "prongs" onto heatshrink tubing over the wires (the heatshrink tubing protects the wires). (3) A strain relief for the wires into the box must also be provided. Details TBD.

USB 2.0 Connection for +5 volt power

A detailed description of the cable between the circuit and the laptop: (Drawings to follow):

  • 1 Form the USB twisted pair and audio twisted pair:
    • 1.1 Cut red wire at 18 inches, cut black wire at 18 inches. Twist together at approximately 0.5 turn per inch. These form the +5 volt/USB pair.
    • 1.2 Cut red wire at 23 inches, cut black wire black at 23 inches. Twist together at approximately 0.5 turn per inch. These form the audio-plug pair. Put tape around the left end of this pair to identify it.
  • 2 Form a single cable from the two pairs. Start with all four wires flush at the left end. Then cable the two twisted pairs together into a smooth bundle of four wires. (The right end will have the audio-plug pair sticking out about 5 inches longer than the USB pair).
  • 3 Place a 15.5 inch "jacket" over the cabled wires: and a second jacket over the audio-plug pair:
  • 3.1 Cut a piece of 3/16 inch (beginning diameter -- it will shrink to 1/8 inch diameter) heatshrink tubing 18 - 1.0 - 1.5 = 15.5 inches long and thread it over the two pairs from the left end. Start with the tubing 1.5 inches from the left ends. Start at the left end and use a heat gun to shrink the tube over the cabled wires (air from a hair-drying gun may work, but is not quite hot enough). Or carefully rub the barrel of a hot soldering iron (not the tip) over the length of tubing (takes practice, dwelling too long will melt the wires and/or destroy the heatshrink tubing).
  • 4 Place a 4.0 inch "jacket" over the audio-plug pair: Cut a piece of 1/8 inch heatshrink tubing 4.0 inches long and slip over the audio plug pair. Important: Space this tube about 0.25 inch away from the end of the first heatshrink tubing (about 5 - 0.25 - 4.0 = 0.75 inch) from the end of the audio-plug pair. Heatshrink this tube onto the audio-plug pair.
  • 5 Form the "pigtail" or "knot": Bend the audio-plug pair backward along the two-pair bundle. Cut a 0.75 inch piece of 0.25 heatshrink tubing and place over the bend, covering it. Heatshrink this tubing in place. This "knob" or "knot" will receive the crimp of the USB connector -- i.e. the crimping of the jaws will occur over this "knot", thereby strain-relieving both connectors.
  • 6 Slip the USB plastic "boot" over the right end (over the 4-wire bundle and the 2-wire bundle), wide-end to the right.
  • 7 Mount the USB connector to the USB pair of wires (in the following, the mechanical equivalent of an extra pair of hands such as fly-tying holder -- will be very helpful).
* 7.1 The "knot" must be about 1/8 inch inside the crimping zone so that the crimps will fall solidly over the "knot". From this criming zone measure out 0.5 inch and cut the USB wires flush. Strip back 3/16 inch (0.187 inch, 4.8 mm). Tin the wires.
  • 7.2 Tin the two outer-most contacts on the USB connector's white insert.
  • 7.3 Solder the red wire to the uppermost and the black wire to the bottommost contact of the white insert.
  • 7.4 Inspect your workmanship carefully. If the solder is blobby and ugly, or any loose solder is on the insert, or little wires are loose, trim them away with a sharp knife (e.g. an Xacto tool).
  • 7.5 With careful attention to orientation, snap the little black clip in place. If it doesn't fit, you may have put too much solder on the terminals. Fix any problems.
  • 7.6 Insert the white insert "upside down" in the larger, outer metal shell. The insert should slide easily into the shell; if not you're inserting it upside down.
  • 7.7 Snap the smaller metal shell/cover over the larger metal shell to cover up the wires .
  • 7.8 Carefully (workmanship counts here) crimp the "teeth" of the jaws over the "knot". Be sure not to crimp too hard, but make the crimping neat and tidy and firm, round as possible, and make sure the jaws are closed and the teeth mesh. Test with tug.
  • 7.9 Rotate the boot so that its little USB symbol is on the unjointed side of the metal connector, then slide it forward over the knot (may be a bit of a tight fit) and continue sliding it forward until it clicks in place.
  • 8 Mount the audio plug to the audio-plug pair of wires (again, the fly-tying holder will be a help here):
  • 8.1 Disassemble the audio plug's boot from the plug and slip it, narrow end first, over the audio wires.
  • 8.2 Cut a piece of red 3/32 inch heatshrink 1.25 inches long and slip this over the wires.
  • 8.3 With the red heatshrink flush to the black heatshrink, heatshrink this red tube in place. This red tube will be acting as both a marker -- red plug into red jack of laptop -- and as a wire protector.
  • 8.4 Cut the red wire 3/8 inch from the end of the two heatshrink tubes, and strip it back 1/8 inch or a bit longer if using a stereo plug.
  • 8.5 Cut the black wire 1/8 or less from the end of the two heatshrink tubes, and strip it back 1.8 inch.
  • 8.6 Tin both wire-ends. Thin the connector's terminals being sure to keep the holes open -- the black-wire solder joint will be underneath the "tang".
  • 8.7 Solder the red wire through the terminal(s) (Through both terminals if the plug is a stereo plug).
  • 8.8 Pass the tinned black wire through the hole in the long "tang" and solder it to the underside tinned area.
  • 8.9 Inspect your workmanship for cold solder joints, blobs, messiness, loose strands of wire, etc.
  • 8.10 Carefully (again, workmanship counts here) crimp the "teeth" of the jaws over the heatshrink. If the jaws are over the wires, redo the job. Be sure not to crimp too hard, but make the crimping neat and tidy and firm, relatively round, and make sure the jaws are closed and the teeth mesh. Test with a tug.
  • 8.11 Screw the "boot" over the plug and tighten firmly but not so tight that you strip the threads.
  • 9 Strain-relieve the circuit-board end (left end) of the cable:
  • 9.1 Place a C-crimp 1/4 inch from the end of the heatshrink tubing. As before, carefully crimp the C closed. Be sure it is firmly grabbing onto the wire -- test by tugging.
  • 9.2 Place a second C-crimp 1/8 inch from the first one. (Measure this, it's important). Crimp this C closed. test by tugging.
  • 9.3 Cut a piece of 1/4 heatshrink to 1.0 inch and slip over the two C-crimps so that that it is flush with the heatshrink that is already there.
  • 9.4 Heatshrink the tube in place.
  • 10 Test the cable and prepare it for attachment to the pc board: strip back the four ends 3/16 inch and tin, but keep track of which pair is which (with tape etc). Test continuity with an ohmmeter (i.e. a digital voltmeter in the "Ohms" setting).

Revision History, simple differential amplifier

The design is subject to revisions as it evolves.

  • REV 0: Prototype-test phase, redraw "layout" drawings, add info re E-Z-Hook to schematic & remove top-view symbols (2 Jan 08)
  • REV 1: Correct "layout" drawing -- change potentiometer symbol from bottom to top view (3 Jan 08)
  • REV 2: Change gain from 1:10 to 1:20, show optional components on schematic with . Add 2nd layout drawing for single-turn potentiometer implementation, plus an LED power-on indicator. Update Bill Of Material. (5 Jan 08)
  • REV 3: Extensive revision of Bill of Materials with added 1000-piece price estimates, modified circuit schematic with terminal block and second "rough trim" potentiometer, modified assembly and layout drawing, new cable-assembly method, etc based on experience of two hand-built prototypes (12 Feb 08).

Other ideas

  1. A low cost (possibly $2) probe that would increase the range of voltage that can be applied to the input.
  2. An ultrasonic distance measurement system that connects to the XO and allows the XO to log data. This has immense applications in water level monitoring in villages and also in robotics applications.
  3. A general purpose hardware kit that interacts with the measure activity. Something on similar lines to http://www.create.ucsb.edu/~dano/CUI/
  4. Sensors could be distributed along with the XO as peripherals. The advantage of this would be that the measure activity can be calibrated against the known specs of the sensors.
  5. Being able to measure resistance
  6. A low cost version of LEGO mindstorms system built around the XO. The XO has good hardware and software capabilities to achieve this.
  7. RS232 or serial interface would allow a host of electronic devices to interact and communicate with the XO.
  8. Build a sensor network using a sensor connected to each XO and utilize the mesh networking capability of the XO. This would give us a highly powerful, robust and reconfigurable sensor network.
  9. Build medical applications using sensors and the XO. See TeleHealth_Module
  10. Do some Fablab projects using the XO. See http://fab.cba.mit.edu/labs/vigyan/
  11. Collaborative music creation over the mesh using the music software on the XOs and giving input from an array of different sensors.
  12. Integrate sensor input into LOGO / Turtle activity
  13. Milk purity in rural areas. I remember reading about it in some FabLab's website - I think it requires an ADC, which we have.