Silicon ChipWide-Range Ohmmeter, Part 1 - August 2022 SILICON CHIP
  1. Outer Front Cover
  2. Contents
  3. Publisher's Letter: 100 years of Australian electronics magazines
  4. Feature: IC Fabrication, Part 3 by Dr David Maddison
  5. Subscriptions
  6. Project: Wide-Range Ohmmeter, Part 1 by Phil Prosser
  7. Feature: History of Silicon Chip, Part 1 by Leo Simpson
  8. Product Showcase
  9. Project: isoundBar with Built-in Woofer by Allan Linton-Smith
  10. Review: DH30 MAX Li-ion Spot Welder by Phil Prosser
  11. Project: SPY-DER: a 3D-printed Robot by Arijit Das
  12. PartShop
  13. Serviceman's Log: Spy games and supper-villain gadgets by Dave Thompson
  14. Project: Secure Remote Mains Switch, Part 2 by John Clarke
  15. Vintage Radio: AVO valve testers, part 1 by Ian Batty
  16. Market Centre
  17. Advertising Index
  18. Notes & Errata: Spectral Sound MIDI Synthesiser, June 2022; Digital FX (Effects) Pedal, April & May 2021
  19. Outer Back Cover

This is only a preview of the August 2022 issue of Silicon Chip.

You can view 41 of the 104 pages in the full issue, including the advertisments.

For full access, purchase the issue for $10.00 or subscribe for access to the latest issues.

Articles in this series:
  • IC Fabrication, Part 1 (June 2022)
  • IC Fabrication, Part 1 (June 2022)
  • IC Fabrication, Part 2 (July 2022)
  • IC Fabrication, Part 2 (July 2022)
  • IC Fabrication, Part 3 (August 2022)
  • IC Fabrication, Part 3 (August 2022)
Items relevant to "Wide-Range Ohmmeter, Part 1":
  • Wide-Range Ohmmeter PCB [04109221] (AUD $7.50)
  • PIC24FJ256GA702-I/SS‎ programmed for the Wide Range Ohmmeter (0110922A.HEX) (Programmed Microcontroller, AUD $15.00)
  • 16x2 Alphanumeric module with blue backlight (Component, AUD $10.00)
  • Partial kit for the Wide-Range Ohmmeter (Component, AUD $75.00)
  • Firmware and source code for the Wide-Range Ohmmeter [0110922A.HEX] (Software, Free)
  • Wide-Range Ohmmeter PCB pattern (PDF download) [04109221] (Free)
  • Front panel label for the Wide-Range Ohmmeter (Panel Artwork, Free)
Articles in this series:
  • Wide-Range Ohmmeter, Part 1 (August 2022)
  • Wide-Range Ohmmeter, Part 1 (August 2022)
  • Wide-Range Ohmmeter, Part 2 (September 2022)
  • Wide-Range Ohmmeter, Part 2 (September 2022)
Articles in this series:
  • History of Silicon Chip, Part 1 (August 2022)
  • History of Silicon Chip, Part 1 (August 2022)
  • History of Silicon Chip, Part 2 (September 2022)
  • History of Silicon Chip, Part 2 (September 2022)
  • Electronics Magazines in Aus. (July 2023)
  • Electronics Magazines in Aus. (July 2023)
Items relevant to "isoundBar with Built-in Woofer":
  • Cutting and assembly diagrams for the isoundBar (Panel Artwork, Free)
Items relevant to "SPY-DER: a 3D-printed Robot":
  • Arduino and Raspberry Pi software plus 3D printer STL files for the SPY-DER robot (Free)
Items relevant to "Secure Remote Mains Switch, Part 2":
  • Secure Remote Mains Switch receiver PCB [10109211] (AUD $7.50)
  • Secure Remote Mains Switch transmitter PCB [10109212] (AUD $2.50)
  • PIC16F1459-I/P programmed for the Secure Remote Mains Switch receiver (1010921R.HEX) (Programmed Microcontroller, AUD $10.00)
  • PIC16LF15323-I/SL programmed for the Secure Remote Mains Switch transmitter (1010921A.HEX) (Programmed Microcontroller, AUD $10.00)
  • Firmware and ASM source code for the Secure Remote Mains Switch [1010921A/R] (Software, Free)
  • Secure Remote Mains Switch PCB patterns (PDF download) [10109211/2] (Free)
  • Front panel label and drilling diagrams for the Secure Remote Mains Switch (Panel Artwork, Free)
Articles in this series:
  • Secure Remote Mains Switch, Part 1 (July 2022)
  • Secure Remote Mains Switch, Part 1 (July 2022)
  • Secure Remote Mains Switch, Part 2 (August 2022)
  • Secure Remote Mains Switch, Part 2 (August 2022)
  • Secure Remote Switch, Part 1 (December 2024)
  • Secure Remote Switch, Part 1 (December 2024)
  • Secure Remote Mains Switch, part two (January 2025)
  • Secure Remote Mains Switch, part two (January 2025)
Articles in this series:
  • AVO valve testers, part 1 (August 2022)
  • AVO valve testers, part 1 (August 2022)
  • AVO valve testers, part 2 (September 2022)
  • AVO valve testers, part 2 (September 2022)

Purchase a printed copy of this issue for $11.50.

Wide-Range hmMeter Features & specifications Resistance measurement range: 1mΩ to 20MΩ Individual ranges: 1mΩ to 30Ω, 30Ω to 3kΩ, 3kΩ to 100kΩ, 100kΩ to 1MΩ, 1MΩ to 20MΩ Resolution: 0.1mΩ in milliohms range (usable resolution closer to 1mΩ) Accuracy: better than ±1%; typically close to ±0.1% Test current: 50mA up to 30Ω, 0.5mA from 30Ω to 3kΩ, <50μA up to 20MΩ Other features: auto-ranging, battery voltage display Power supply: 6 x AA cells; up to 100mA drawn during tests Battery life: around 24 hours of active use This auto-ranging ohmmeter will measure just about any resistance – from a handful of milliohms to many megohms! T here have been several occasions recently where I have needed to measure low resistances accurately. That includes some speaker projects, where I needed to accurately measure the DC resistance of a voice coil to estimate a driver’s Thiele-Small parameters or determine the resistance of an air-cored power inductor. Another time it was for the Capacitor Discharge Welder project (March & April 2022; siliconchip.com.au/ Series/379), where I wanted to check the resistance of the leads. Theory said they should be 8mW (spoiler alert – with the cables and handles, our welder leads measured 10mW). Your garden variety multimeter won’t measure anywhere near that low. Even my fancy, expensive meter was way off the mark. So what do you do when you want an accurate measurement of a really low resistance? You reach for your trusty old lowohms meter. Like many journeys in life, this design started on one path but ended up somewhere else. The initial plan was to update a previous Milliohm Meter design, adding a digital front end and making it easy to use. But halfway through, somebody said: why 26  Silicon Chip not make it measure up to 20MW? This added a bit of a spin on the design, but we think the result is a very handy and versatile device. So here we have a design for a meter that will measure resistances from a couple of milliohms to 20 megohms, with precision significantly better than 1% across that range. Using 0.1% resistors for calibration (which we recommend), we have seen precision in the region of 0.1% across most of its range. The trouble with multimeters The problem with a standard multimeter is that the lead and banana socket resistance is usually in the 0.2-0.5W range. The variability in these resistances are too high to zero them out. Ohm’s Law is one of the first equations you learn in electronics. It is therefore not surprising that this principal is used in most ohmmeters, with the resistance measured using a constant current source and a voltmeter. A typical multimeter combines these inside the meter and uses two leads, as shown in Fig.1. When measuring a low resistance this way, the constant current needs to flow through the banana plugs, leads Australia's electronics magazine Part 1 by Phil Prosser and from your lead tips into the device you are measuring, then back again. The voltage drops created by their inherent resistances all appear to the multimeter to be part of the measured resistance. This results in significant errors in low-resistance measurements. There are other ways to measure resistance accurately that don’t use this principle. For example, the Wheatstone bridge is a very elegant approach that can be highly accurate. But an automated meter based on one of those would be very complicated. If you are interested in this use for a Wheatstone Bridge, Wikipedia is a good place to find out more. Kelvin connections A four-wire measurement technique can be used to minimise these errors. Two wires deliver a known current through the device under test, while the second pair measure the voltage across the device under test (DUT), as shown in Fig.2. This neatly avoids the majority of errors above. By using a constant current source, even if there are lead and connection resistances, the current is always as expected. The voltmeter is chosen to have a high input resistance, so siliconchip.com.au Measured resistance R=V÷I Measured resistance R=V÷I Measured resistance Rdut = Rref × (V2 ÷ V1) Current = V1 ÷ Rref V2 = Current × Rdut Fig.1: a standard ohmmeter works by passing a known, fixed current through the device under test (DUT), measuring the voltage across it, then using Ohm’s Law to determine its resistance. The problem is that the test lead resistances are in series with the DUT and included in the result. Fig.2: two pairs of leads are used with Kelvin connections, one to feed the test current to the DUT and one to sense the voltage across it. The voltage drop across the leads supplying current no longer affects the reading, and the voltage drop across the other pair of leads is so tiny that it doesn’t matter. Fig.3: the problem with using the method shown in Fig.2 to measure high resistances is that the test current needs to be really low. So we use this method instead, where the DUT and a fixed resistor form a divider, and we measure the DUT resistance in proportion to the fixed resistor value. when the voltage measurement leads are connected across the DUT, even if the connection is a bit dodgy, we still read the correct voltage, and the R = V ÷ I calculation avoids the majority of errors. There is a bit more effort involved in making really accurate resistance measurements than just adding two wires, but they are necessary to measure values well under 1W accurately. You might wonder why all ohmmeters don’t work this way if it is so effective. Well, using a four-wire ohmmeter is fiddly. There are four wires and most of us only have two hands. Also, the errors are no longer significant above a few hundred ohms. Therefore, all but a few meters (mainly benchtop meters, but some are handheld) use the conventional two-wire approach. The four-wire connection is called a “Kelvin connection” after Lord Kelvin, who invented this to measure low resistances in 1861. While working on this meter, we noticed some nice ‘Kelvin clip leads’ available at reasonable prices. These are essentially crocodile clips with two connections, one for the current source and the other for the sense wire. We found that these worked well over the range of our meter, though for really low resistances, four separate wires will give better accuracy. and, as the voltmeter, an analog-to-­ digital converter (ADC) with a carefully designed voltage reference. These both provide good long-term stability for the meter and the ability to use 0.5mA and 50mA bias currents, which give measurements accurate into the low-milliohm range. Measuring down to about 1mW is practical with a reasonably simple meter. This is about the lower limit before other factors become problematic. Even with higher currents, low resistances mean we need to measure low voltages. Our design uses special very low offset and very low drift operational amplifiers. If we had chosen, say, a common TL074, the worst-case input offset of 4mV would introduce errors of up to 80mW on the low ohms range! The device selected has a worst-case offset of 8uV over its entire operating temperature range, which still could result in an offset error of up to 1.6mW (although we have not seen anything like this sort of error in our testing). This allows our meter to accurately measure a 5mW shunt resistor, which we feel is pretty good. To go beyond this, design approaches that null out these offsets are required – this is usually achieved by switching the current source on and off, allowing subtraction of the nil current offset. By using low-offset parts, we can avoid the need to do this in our design. low current and making the exact same measurement. This is true, provided you can generate a stable current source delivering about 0.1μA with an output resistance much greater than 20MW. But that is not easy to achieve. To avoid this, we use a slightly different technique for measuring higher resistances, as shown in Fig.3. We use a high-value precision resistor to establish the test current. Because this is in series with the DUT, the current flowing will depend on the DUT’s resistance. We do not try to control the current; instead, we measure the voltage across the reference resistor to measure the current flowing for every measurement. By also measuring the voltage across the DUT, we have all the information we need to determine its resistance in proportion to the sense resistor. For the 1MW range, we use a 1MW sense resistor. The current through this will vary. If we measure a 1MW resistor, the current will be Itest = Vsupply ÷ (Rref + Rdut), about 1.5μA. Keep in mind that Itest = V1 ÷ Rref. This relationship is handy, as we will see in a minute. Ohm’s Law tells us that the resistance of the DUT is defined by Rdut = V2 ÷ Itest, where V2 is the voltage across the DUT. Combining this and the previous equation: Rdut = V2 ÷ (V1 ÷ Rref) = Rref × (V2 ÷ V1). Our ADC does not have two channels, but it does have an independent reference (V1) and measurement input (V2). So by connecting our ADC reference across the reference resistor, we Other challenges We need to know the exact current through the DUT and the voltage across it. For DUTs with a low resistance, both of these are easily achieved. We use an LT3092 programmable current source siliconchip.com.au Megohms measurements Adding a megohm range would seem to be a simple matter of setting the constant current source to a very Australia's electronics magazine August 2022  27 can measure the ratio of V1 and V2 with the ADC, and it simply comes out as the measured value! An added bonus of this approach is that we don’t need to care about the exact supply rail voltage or exact current through the DUT. The catch here is that our measurement of the voltages across the reference and DUT resistors has been assumed to be ideal, ie, our ADC has no impact on the current flowing through the DUT. We already know that the current will be in the region of 0.1μA, so the ADC measuring the reference and DUT voltages needs to have very high input resistances and very low bias currents (the current flowing into or out of the input), or else the above assumption will fail. The ADC we have chosen, the MAX11207, only has a bias current of 30nA. The voltage 30nA will develop across a 10MW resistor is 30 × 10-9 × 10 × 106 = 300 × 10-3V, or 0.3V. This is a massive error, given that we will be measuring about 1.5V. So we had to add a buffer amplifier with a super low bias current. Our choice, the MCP6V64, has a typical input bias current of 20pA and a maximum offset current of 200pA (the difference between the bias currents for the + and – inputs). Given the current shortages, we have listed a few alternatives that we have tested in the parts list, but the MCP6V64 is our first choice. This reduces error with a 10MW resistor to 200 × 10-12 × 10 × 106 = 2 × 10-3V or 3mV, a much more manageable error. Circuit description Let’s look at how these decisions come together in our final design. The complete circuit is shown in Fig.4. The heart of this meter is the MAX11207 20-bit ADC. We have also tested this with the similar MAX11210 chip, and the MAX11206 and MAX11200 should also work just fine too. We chose this device as it is very linear, provides great resolution and is available in several pin- and software-compatible forms. It also has fully differential inputs for both the ADC and the reference, which can operate across the entire input range. This means we can pull some tricks and use the reference input in a somewhat unusual manner for high-resistance measurements. This device has a range of settings, the most important ones being internal calibration and internal buffering. The software looks after this, and you should only notice a slight delay at power-on as they are initialised. All the inputs to the ADC are buffered by the MCP6V64 quad operational amplifier. This device provides a very high input impedance, low bias current and low drift buffer for the ADC. All of its inputs and outputs can go close to the supply rails. Its key feature is bias currents in the pA range, and it can operate within 200mV of the rails. When you get to the construction stage, take note that the PCB must be A preview to part two, showing how the batteries and PCB are mounted. 28  Silicon Chip Australia's electronics magazine very clean around this surface-mount IC. Flux and residue from soldering can increase the leakage currents on these extremely high impedance inputs, degrading the performance of your meter. Thoroughly cleaning and coating this area with clear protective lacquer is an essential step in construction. We have included 10kW series protection resistors from the sense inputs to the buffers, and a 10nF capacitor across the sense inputs, providing modest protection to the circuit. That said, we strongly suggest that you do not connect the meter to live circuits, as the application of more than a few volts between the terminals could easily cause damage. On the milliohms range, the reference voltage going to the REFP input (pin 5) of IC1 via buffer IC2a comes from an LM336 2.5V shunt regulator, IC5 (lower left). We’re specifying the LM336B type as it has tighter tolerances. The LM336 is set up with series diodes and a trimpot, which allows us to set it to exactly 2.50V, and the diodes minimise its drift with temperature. The reference input is connected across a resistor of either 100kW, 1MW or 20MW resistors on the higher ranges. These can be found near IC5. The stability of these resistors is important for the accuracy of these ranges. Again, we will be calibrating the device, so initial precision is less critical than stability for these parts. The MCP6V64 buffers for the ADC (IC2b & IC2c) can drive to within a few millivolts of the rails, but not quite to the rails. To accommodate this, the 2.50V voltage reference and reference resistors connect to ground through D8, a BAT85 schottky diode. Similarly, the DUT connects to the positive rail through D4, a 1N5819 schottky diode. These drop about 0.3V at the currents we operate them. We use a constant-current device (IC3) to pass either 0.5mA or 50mA through the DUT on the milliohms and ohms ranges. The stability of the voltage and current references is essential to the accuracy of these ranges. But because we calibrate this meter against known resistors, absolute precision is less of an issue. With a 3.6V supply rail, the maximum voltage that we can handle across the DUT is 1.7V. This is calculated ...continued page 31 siliconchip.com.au Parts List – Wide-Range Ohmmeter 1 double-sided PCB coded 04109221, 90.5 × 117.5mm 1 189 × 134 × 55 sloping ABS instrument case [Altronics H0401] 2 3 AA cell battery holders with leads [Altronics S5033 + P0455] 1 backlit 16×2 character alphanumeric LCD screen with HD44780-compatible controller (LCD1) [SC5759] 2 4-pin tactile switches (S1, S2) 1 subminiature DPDT solder tag slide switch with mounting screws (S3) [Altronics S2010 + S2014] 3 Omron G6H-5V or G6S-5V telecom relays or equivalent (RLY1-RLY3) [eg, Altronics S4128B] 1 10kW top-adjust multi-turn trimpot (VR1) 1 10kW top-adjust mini trimpot (VR2) 1 2-pin header with jumper shunt (JP1) (optional; only needed for in-circuit programming) 2 2-way vertical polarised headers with matching plugs (CON1, CON2) [Altronics P5492 + P5472 + 2 x P5470A] 1 16-pin header (CON3; for mounting the LCD) 1 6-pin header (CON4) (optional; only needed for in-circuit programming) 1 2-pin right-angle polarised header with matching plug (CON5) [Altronics P5512 + P5472 + 2 x P5470A] 1 5-pin header (CON6) (optional; for monitoring SPI) 2 red captive head binding/banana posts (CON7, CON8) [Altronics P9252] 2 black captive head binding/banana posts (CON9, CON10) [Altronics P9254] various lengths of light-duty hook-up wire 1 pre-made set of Kelvin clip leads [www.ebay.com.au/ itm/263861879033] OR 1 DIY set of Kelvin clip leads (see section below) Hardware 4 M3 × 10mm tapped metal spacers 4 M3 × 6mm panhead machine screws 4 M3 × 6mm countersunk head machine screws 8 M3 shakeproof washers 1 small tube of clear neutral-cure silicone sealant 1 can of PCB conformal coating/protective lacquer Kelvin clip leads (if not using pre-made leads) 2 Kelvin alligator clips [Mouser 485-3313 or 510-CTM75K; Digi-Key 1528-2279-ND] 1 2m length of 17AWG (1.0mm2) black figure-8 cable [Altronics W4146] OR 1 2m length of two-core heavy-duty microphone cable [Altronics W3028] 1 1m length of 18AWG (0.75mm2) red silicone hightemperature hook-up wire [Altronics W2400] 1 1m length of 18AWG (0.75mm2) black silicone hightemperature hook-up wire [Altronics W2401] Semiconductors 1 MAX11207EEE+ 20-bit ADC, QSOP-16 (IC1) ● (alternatives exist – see text) 1 MCP6V64-E/ST quad low-drift rail-to-rail op amp, TSSOP-14 (IC2) ● ■ 1 LT3092EST or LT3092IST programmable current source, SOT-223 (IC3) ● siliconchip.com.au 1 PIC24FJ256GA702-I/SS 16-bit microcontroller programmed with 0410922A.HEX, SSOP-28 (IC4) ● 1 LM336BZ-2.5/NOPB voltage reference, TO-92 (IC5) ● 1 555 timer, DIP-8 (IC6) ● 2 AZ1117H-ADJTRG1, AMS1117 or equivalent adjustable 1A LDO regulators, SOT-223 (REG1, REG2) ● 4 BC547 100mA NPN transistors, TO-92 (Q1, Q3, Q5, Q6) 2 IRLML0030TRPBF N-channel Mosfets, SOT-23 (Q2, Q4) ● 7 1N4148 75V 250mA signal diodes (D1, D2, D5-D7, D10, D11) 2 1N5819 40V 1A schottky diodes (D3, D4) 1 BAT85 30V 200mA schottky diode (D8) 1 1N4004 400V 1A diode (D9) Capacitors 7 10μF 50V radial electrolytic 5 10μF 16V X7R SMD M3216/1206-size ceramic ● 5 100nF 50V X7R through-hole ceramic 5 100nF 50V X7R SMD M2012/0805-size ceramic ● 2 10nF 100V PPS [Kemet SMR5103J100J01L16.5C] ● 4 10nF 50V X7R through-hole ceramic Resistors (all axial 1/4W 1% metal film unless noted) 2 10MW 0.1% 25ppm SMD M3216/1206-size ● 1 1.5MW 1 1MW 0.1% 25ppm SMD M3216/1206-size ● 2 1MW 1% SMD M2012/0805-size ● 1 100kW 0.1% 25ppm SMD M3216/1206-size ● 1 47kW 1 33kW 1 22kW 1 10kW 0.1% 15ppm ● 7 10kW 4 4.7kW 3 3.3kW 1 2.2kW 2 1.2kW 1 820W 1 205W 0.1% 15ppm ● 2 100W 1 47W 2 1W 1% 50ppm ● Calibration resistors (not required if another highprecision ohmmeter is available) 1 27.4W 1/4W 0.1% 15ppm axial [YR1B27R4CC] ● 1 2.94kW 1/4W 0.1% 15ppm axial [YR1B2K94CC] ● 1 97.6kW 1/4W 0.1% 15ppm axial [YR1B97K6CC] ● 1 976kW 1/4W 0.1% 15ppm axial [YR1B976KCC] ● 1 10MW 1/4W 1% 50ppm axial [MF0204FTE52-10M] ● ● all these parts (with IC4 pre-programmed) are available in a set (Cat SC4663) for $75.00. ■ compatible op amps need to be rail to rail, unitygain stable with very low input offset voltages and input bias currents in a TSSOP-14 package. Good alternatives are the MCP6V79, MCP6V34 and OPA4317. Australia's electronics magazine August 2022  29 Fig.4: all measurements are made by IC1, the ADC, controlled by microcontroller IC4. IC4 switches relays RLY1-RLY3 to select the appropriate range and displays readings on the 16x2 LCD module. Voltage reference IC5 is used in the lower (milliohms & ohms) ranges while IC3 regulates the test current, with Mosfets Q2 & Q4 switching it between 0.5mA & 50mA. In ratiometric (high-range) mode, IC3 and IC5 are not used, and precision resistors of 100kW, 1MW or 20MW are connected in series with the DUT. 30  Silicon Chip Australia's electronics magazine siliconchip.com.au by subtracting the voltage drops from the supply rail due to diode D4 (0.3V) and IC3 (1.6V). Let’s say we can allow up to 1.5V across the DUT to be safe. This means a maximum reading of 1.5V ÷ 50mA = 30W on the milliohms range and 1.5V ÷ 0.5mA = 3kW on the ohms range. The maximum readings on the other ranges are limited by the values of siliconchip.com.au the 100kW, 1MW and 20MW reference resistors. The current regulator For the higher current (lower resistance) ranges (milliohms and ohms), we use IC3, an LT3092 constant current source. We have chosen this for its long term stability and ease of use. This device sources a constant 10µA Australia's electronics magazine from its SET pin, and the OUT pin is maintained at the same voltage as the SET pin. With a 10kW resistor from the SET pin to GND, there will be 0.1V across it (10kW × 10μA). The parallel combination of 205W, 47kW and 1.5MW resistors results in 204.08W between the OUT pin and ground, giving a current of 490μA. Therefore, the IN pin August 2022  31 sinks 490μA + 10μA = 500μA for these two currents combined, which is our goal (0.5mA). For the milliohms range, parallel Mosfets Q2 & Q4 switch on, so the two series 1W resistors are connected in parallel with the 204.08W resistance. But note that the on-resistance of the Mosfets (40mW || 40mW = 20mW) adds to the 2W from the resistors. With 2.02W in parallel with 204.08W, we get 2.0002W. Thus the current from the OUT pin will be 49.99mA + 0.01mA or 50mA. This way, the software can switch the constant current source between 0.5mA and 50mA to suit the resistance detected on the meter by controlling the gates of Q2 and Q4. We recommend using 0.1% 15ppm resistors for the 10kW and 205W parts, as specified in the parts list. We found 1W 0.1% resistors too expensive, so we used 1% parts instead. These are MF0207FRE52-1R, which have a 50ppm temperature coefficient, so they should be pretty stable. We have provided the current source with a good heatsink in the form of a large copper fill on the top layer of the PCB. The keen-eyed will also note that we have placed a guard track around the SET pin, which has an extremely low current flowing from it. This will reduce leakage currents interfering with our carefully-designed current source. The reference resistors We measure resistances in three ranges above 3kW: 100kW, 1MW and 20MW. Our measurement technique uses reference resistors at each of these values. We have specified parts that should provide a low temperature coefficient and long-term stability. We again recommend 0.1% parts where reasonable. 20MW tight-tolerance resistors are both expensive and uncommon, so we use two 10MW resistors in series. Stability is probably more important than actual precision, as the meter will be calibrated. Again, cleaning off all flux and residue around these is very important, as is coating it with a protective lacquer to optimise long-term stability. We used 3.2 × 1.6mm SMD parts here (M3216/1206) as our survey of suppliers found that 0.1% parts are more available and less expensive in these packages than in through-hole. 32  Silicon Chip Switching the ADC inputs Because we have five different ranges and can’t handle any additional bias currents, we need to do some switching, and that’s done with relays. The resulting switching arrangement might initially look complicated but there isn’t too much to it. Regardless, the auto-ranging feature means that the user doesn’t need to know the details. One relay, RLY1, switches the reference input between the fixed 2.50V reference and the three reference resistors. The other two relays, RLY2 and RLY3, connect either the constant current device (IC3) or one of the three reference resistors to the lower pin on the Force connector, CON1. The PCB has been laid out to handle two of the most common types of signal relays, conforming to the Omron G6H and G6S layouts. These are available from a range of electronic outlets. Just make sure you use 5V non-­ latching versions. Microcontroller and display We have kept the display and control circuitry simple. We see this as a utilitarian device, so it should put function over form, and seek to ‘do what it says on the box’ as simply, cheaply and reliably as possible. The LCD screen operates from the VDD rail of about 3.4V, but these displays are almost always powered from 5V. It turns out that the LCD bias between the VDD and VO pins on the LCD module needs to be about 5V, but the actual controller is specified to operate from 2.7V. Therefore, we can generate a negative voltage of about -2V for the VO bias reference and power the LCD from the same VDD rail used for the PIC micro. We need to do this because some LCDs are incompatible with the 3.3V CMOS outputs from microcontrollers. Annoyingly, it is very difficult to tell which LCDs work with 3.3V logic and which don’t. To avoid this frustration, we have arranged the circuit so that all LCDs should work. The negative VO bias is generated by 555 timer IC6, which oscillates at a couple of kilohertz. This drives a switched-capacitor voltage inverter comprising two 10μF capacitors and two 1N4148 diodes. This runs off the relay 5V rail and generates -2V or so. By using the 5V rail, we avoid running this ‘noisy’ Australia's electronics magazine circuit from a rail used for the sensitive current sources and ADC. User interface The goal of simplicity has led us to remove all buttons from the front panel and implement an auto-range function. There are two buttons on the PCB which are only used for calibration; we will discuss them later. Upon initial connection, the Meter will first check the DUT on the 100kW range. Depending on the result, it will increase or decrease the range appropriately until the optimal measurement range is found. We start with the 100kW range as most of the resistors we measure seem to be less than this resistance. The way the meter does auto-ranging means it will generally jump from the 100kW range straight to the final measurement range. The initial test current will be 30μA or less, and this will increase to 500μA for resistances between 30W and 100kW, or 50mA for resistances below 30W. The highest possible power delivered is 75mW for a 30W resistor. This should be safe for all bar the most sensitive devices. The microcontroller used is a PIC24FJ256GA702-I/SS. This is just right for the job in terms of pin count, though we also use four ‘free’ digital I/O pins provided on the ADC, as they were too convenient to ignore! We have used a simple schottky diode to drop the 3.6V rail to something closer to 3.3V for the ADC and the microcontroller, as 3.6V is right at their upper limits. The micro drives a 16 column, two-line alphanumeric LCD with an HD44780-compatible controller. These are bog-standard but, as a result, come in a bewildering variety of layouts. We have included two very common footprints on the PCB, which gives you some options for selecting a display. When you purchase the display, check the pin-out, as the LED backlight, in particular, seems to change around a lot. There are two headers that you probably won’t need. The first is the ICSP header, CON4. This allows the microcontroller to be reprogrammed on the board, which we used in development, but many readers will build the device with a pre-programmed PIC. There is also a footprint for CON6, siliconchip.com.au SPI_MON. You should definitely not need this unless you want to look at the SPI activity between the microcontroller and ADC. This sort of facility is super helpful when developing a project like this. We also have pads for an external 8MHz crystal and associated 22pF and 100W resistors, although these components are not required in this design as we use the PIC’s internal oscillator instead. The ADC, buffer op amp and microcontroller are all surface-mount parts. They are simply not available in through-hole packages in the first two cases. We also had a desire to fit this project into a handy instrument case. Power supply The circuit operates from six AA cells. We chose this approach to ensure the meter would have a good runtime and that the 5V rail stays up as the batteries discharge. The meter can draw close to 100mA when measuring low resistances. This should provide over 24 hours of runtime on a set of batteries, which will be fine provided you do not forget to switch it off overnight! There are two linear low-dropout regulators. One has a 5V output to power the relay coils, LED backlighting on the LCD screen and the -2V generator (REG3). The other has a 3.6V output (REG2) to power the ADC, buffer op amps and micro. Both regulators are specified as the AZ1117 type, but there are many pin-compatible LDO regulators (usually with 1117 in their part code) that will work fine too. We’ve provided all the components to allow two identical adjustable regulators to be used for REG2 & REG3. Still, you could use a fixed 5.0V output regulator for REG3, omitting the resistor between the OUT and ADJ pin and its series capacitor, and replacing the resistor between ADJ and GND with a 0W resistor (or a short piece of wire across the pads). You could theoretically do that for REG2 as well, but unfortunately, 3.6V is not available as a fixed output option on this type of regulator. So stick with the adjustable type for REG2. Kelvin leads We used Adafruit 3313 Kelvin clips leads with the prototype, which are amazingly cost-­ effective; certainly less expensive than a double espresso (let alone that smashed avo!). Availability from the usual suppliers is mixed. We also tried Mouser Cat 510-CTM-75K, which is a delight to use but rather more expensive. These are simple to wire up, as shown in the adjacent photo. All you need to do is wire the Force+ and Sense+ wires to either side of the “+” Kelvin clip (with the red wire) and the other two terminals to the remaining black wires of the “-” Kelvin clip. Keep in mind that the force and sense wires only contact either side of the DUT lead. Where you measure larger or more fiddly items, separate force and sense test leads might be better. Again, the force current must run through the whole item you wish to measure the resistance of, and the sense lines are connected to measure the part you desire, as shown in Fig.5. We made two sets of leads for our meter. One set had separate sense and force leads, and these are essentially conventional multimeter leads. We made them using 18AWG silicone-coated high-temperature hook-up cable (Altronics W240X), which is very flexible. We connected these wires to clips for the force and probes for the sense lines. We did not use these much in the end, as the Kelvin clips are excellent right down into the low-milliohm region. We used Altronics Cat W4146 sheathed figure-8 flex for our Kelvin Clips, though we feel that a lighter gauge would be easier to use if you can find it. We used coloured heatshrink tubing to clarify which wires are + and – (although this generally isn’t important when making measurements). One Kelvin clip connects to “Force -” and “Sense -” while the other goes to the “Force +” and “Sense +” sockets on the meter. The length of leads should not matter as the conductors are close, so any EMI picked up should mostly cancel out. We felt that 600mm was about right, but that is a matter of preference. If you don’t want to make up your own set of Kelvin clip leads, they are available to buy pre-made at reasonably low prices at sites like eBay. Search for “LCR clip leads”. For example, www.ebay.com.au/itm/263861879033 Next month We don't have space in this issue for all the construction, testing and set-up details, so they will be in a follow-up article next month. SC siliconchip.com.au Fig.5: when working with Kelvin probes, it doesn’t matter whether you connect the ‘sense’ leads closer to the DUT than the ‘force’ leads or not. Regardless, the section between the two connections on either side is not measured because there is no current flowing through it or the measurement point is further along. Australia's electronics magazine August 2022  33