Fullscreen HTML5Med Res (??MB)HTML4

Analyser Background image: https://unsplash.com/photos/a-close-up-of-a-circuit-board-H8rKjwyj1Og SEMICONDUCTOR This handy Semiconductor Analyser automatically identifies most two- or three-terminal discrete semiconductor devices, tells you which pin is which and displays some useful information about the device. It is handy for testing if a part is good or damaged, for matching devices, or just for sorting out that bag of mystery semiconductors in your junk box. U sing it could not be simpler. Connect two or three of the test leads to the component, in any order, then press the TEST button. After two or three seconds, the results are displayed on the TFT LCD screen. The displayed information includes the device type, a diagram showing which test lead is connected to which pin, and in most cases, some key electrical parameters. These are shown for 15 seconds, then the analyser switches itself off. While results are displayed, you can press TEST again to initiate another test cycle, or the OFF button to put it to sleep immediately. There is a USB Type-C port on the end of the unit that serves three purposes: • It can be used to charge the internal lithium-polymer (prismatic lithium-­ion) battery. • It provides a simple serial interface that can display the test results in a terminal program running on a computer. • The USB port can also be used for uploading firmware, either programming the chip initially or updating it later. The command-line interface via the serial terminal provides a bit more information about the tests being carried out, so it is useful for debugging or just understanding how the device has come to the decision that it has. Capabilities » Identifies and tests diodes, transistors (bipolar, JFET, Mosfet & IGBT) and thyristors » Diodes supported: standard or schottky (single or pair) plus zeners/TVSs/LEDs up to 10V » Bipolar transistor tests: NPN/PNP, standard or Darlington with or without diodes, hfe: 5-25,000 » Mosfet tests: N/P-channel, enhancement/depletion, Vgs(th) (gate-source threshold voltage) » JFET tests: N/P-channel, pinch-off voltage » IGBT tests: freewheeling diode presence, Vge(th) (gate-emitter threshold voltage) » Thyristor tests: identifies SCRs and Triacs » Other tests: detects short circuits & open circuits in devices » Power supply: 1100mAh rechargeable Li-ion cell giving a runtime of ~24 hours The Semiconductor Analyser works with the following devices: ▶ Diodes – identifies standard diodes, schottky diodes, LEDs and zener (or TVS) diodes with a breakdown voltage between 3V and 10V. It measures the forward voltage of all diodes and the breakdown voltage in the case of zeners/TVSs. The test voltage is limited to 12V, so zener diodes with a breakdown voltage greater than about 10V will be identified as standard diodes. ▶ Diode pairs – identifies common anode, common cathode and series pairs of standard and schottky diodes. It identifies common anode, common cathode and back-to-back LED pairs. The forward voltage of each diode or LED in the pair is displayed along with the test current. ▶ Bipolar transistors – identifies NPN and PNP bipolar transistors and Darlington pairs, with or without integrated freewheel diodes. The Australia's electronics magazine siliconchip.com.au Project by Andrew Levido Features & specifications 28 Silicon Chip base-emitter voltage and the DC gain (hfe) are displayed. DC gains in the range of 5 to 25,000 can be measured. Some Darlington pairs have higher gain than this, and in such cases, the device indicates that the hfe could not be measured. This does not mean the transistor is faulty. ▶ Mosfets – identifies N-channel and P-channel enhancement-mode Mosfets and N-channel depletion-mode Mosfets. (P-channel depletion-mode Mosfets can exist in theory, but nobody makes them). For enhancement-mode Mosfets, the gate-source threshold voltage (Vgs(th)) is displayed. This is the gate-source voltage at which the drain current rises to 5mA. For depletion-­ mode Mosfets, the gatesource pinch-off voltage is measured. This is the gate-source voltage at which the drain current falls to 5µA. ▶ JFETs – identifies N-channel and P-channel junction FETs. The pinch-off voltage is measured as for depletion-­mode Mosfets. The Semiconductor Analyser identifies the gate of a JFET, but the drain and source terminals are usually interchangeable. In many small-signal JFETs, they are electrically symmetrical, so the analyser identifies both as “drain/source”. ▶ IGBTs – identifies IGBTs with and without integrated freewheel diodes. The gate-emitter threshold voltage (Vge(th)) is measured in the same way the gate-source voltage is measured for a Mosfet. ▶ Thyristors and Triacs – identifies thyristors and Triacs with a gate sensitivity of 500µA or lower. This includes most small devices, but some high-power devices may not be correctly recognised because the Semiconductor Analyser cannot source enough current to reliably switch them on. ▶ Short and open circuits – a dead short between any two, or all three leads is detected and flagged as an error. A short-circuit is indicated if the voltage drop between the leads is less than 100mV in both directions. An open-circuit error is flagged if there is no conductivity between any of the three leads. If the device does not check out as one of those mentioned above, it is flagged as faulty or unknown. The voltage applied to all devices is limited to 8V until a preliminary identification is made. This is necessary to avoid damage to logic-level Mosfets, siliconchip.com.au which have a very thin gate oxide layer that can be damaged by voltages above this. Mosfets are never tested at higher voltages, but other components may be tested at up to 12V. The test current is at all times limited to around 10mA, so the Semiconductor Analyser is not likely to damage any component you connect, regardless of the lead connections. How it works It is by no means a simple task to positively identify such a wide range of semiconductors given the enormous variation in parameters, but in the end it all comes down to two basic measurement configurations, as shown in Fig.1. At left is the positive drive configuration, where the drive polarity is positive with respect to the power node, and at right is the negative drive configuration, where the drive polarity is negative with respect to the power node. The power node can therefore either be at ground potential or at 12V. The load node consists of a 1kW resistor in series with a voltage source, Vl. The voltage source can be adjusted to put the appropriate bias between the power and load nodes. When the bias voltage is limited to 8V, Vl would be set to 8V in the positive drive configuration and 4V in the negative drive configuration. The load current, Il, is limited by the 1kW load resistor and the on-­ resistances of the analog switches (more on this below) to a maximum of about 10mA. This current can flow in either direction, depending on the configuration. I have used the convention that current flowing out of the load node is positive and current flowing into it is negative. The third terminal is connected to the drive node. This can be a variable voltage source for voltage-driven devices like Mosfets, or a variable current source/sink for current-driven devices like bipolar transistors. The drive node can also be left open or connected to the load or power nodes if required. A 3 × 3 analog switch matrix (Fig.2) allows each of the three test leads to be connected to any of the test nodes. The voltage at each test lead is measured by the microcontroller. We measure the voltages directly at the terminals of the device under test (DUT) rather than at the power, load and drive Australia's electronics magazine Fig.1: all of the tests conducted by the Semiconductor Analyser are based on these two configurations involving a power node, a load node and a drive node. Fig.2: each of the three test nodes can be connected to any of the three test leads (labelled red, green and blue) by a 3 × 3 analog switch matrix. September 2026  29 nodes because the analog switches each have roughly 40W on-resistances and we do not want the voltage drop across them to introduce measurement errors. Identification process You can probably see how we might use the circuits in Fig.1 to characterise components if we knew what they were and which pin is which, but we know neither of these things at the outset. For this reason, the identification process starts with a set of simple continuity tests between each pair of leads. The continuity is measured twice in each direction, once with the positive and once with the negative drive configuration. During these tests, the third lead is driven by a 500µA current of the appropriate polarity. We need to drive the third lead because we would get unreliable readings if it were left open while connected to a Mosfet or IGBT gate lead. Any stray charge on the open gate lead could put the Mosfet/IGBT in an unknown state. The yes/no results of the 12 continuity measurements (3 pairs of leads × 2 directions × 2 drive polarities) are stored as bits in a test flags variable. The hexadecimal value of this register serves as a ‘signature’ that can be used to help identify the device and its connections. Some devices are unambiguously identified and orientated by just these flags. Single diodes and diode pairs connected anode-to-cathode are two such examples. Most, however, need further disambiguation. It might be best to use an example, since it’s not practical to cover the identification process for each device type. An NPN bipolar junction transistor (BJT) or an N-channel junction field-effect transistor (JFET) with the blue lead connected to the base or gate will have test flags of 0x3C5. The four possibilities are shown at the top of Fig.3. The 0x300 bits indicate conductivity from the blue lead to the red lead, but not from red to blue, for both drive polarities. The 0x0C0 bits indicate the same thing from the blue to the green Fig.3: an NPN BJT (bipolar transistor) and an N-channel JFET have the same continuity signature when connected as shown. Test 1 determines what device it is, while Test 2 identifies the emitter and collector if it is a BJT. 30 Silicon Chip Australia's electronics magazine leads. I have shown this symbolically by the grey diodes in the figure. The 0x05 bits indicate conductivity in either direction, but only when the drive is positive (red diodes). The signature for these same components would be 0xC53 if the red lead was connected to the base/gate and 0x53C if the green lead was connected to the gate. The BJT has this continuity signature because of its base-emitter and base-collector PN junctions and because the transistor is biased on by a positive base current. The collector-emitter path conducts both ways when a transistor is switched on because bipolar transistors have an appreciable reverse gain. The reverse gain is a lot lower than the forward gain; a fact we will use to our advantage below. An N-channel JFET has the same signature as an NPN BJT, but for different reasons. The channel is a single piece of N-doped silicon, with the drain and source terminals at either end, so it is conductive in both directions when the JFET is unbiased. The gate region is P-doped, so it forms a PN junction with the channel. This diode can be measured from gate-to-source and from gate-to-drain. The channel does not conduct when the gate-channel junction is reverse-­biased, as the depletion region expands to ‘pinch off’ the channel. This is why we see conductivity in both directions when the gate is positively biased and no conductivity when the gate is negatively biased. At this point, we know that we have an N-channel JFET or an NPN BJT, and we know which test lead is connected to the gate or base. It is the blue one in this 0x3C5 example, but depending on the signature, it may be one of the other leads. We need a further set of tests to fully identify the part and its connections. The next test (Test 1 in the figure) is a simple measurement of load current between the two unknown leads (red and green in our example) with the gate/base lead shorted to the power node. A JFET will conduct regardless of the orientation of the drain and source due to the symmetry of the channel. A BJT will not conduct between collector and emitter when there is no base drive. The above statement is technically correct, but we need to add a caveat. In the configuration where the siliconchip.com.au base-emitter junction is reverse-biased (second from the left), there may be some current flow due to reverse breakdown. A reverse-biased emitter-base junction experiences avalanche breakdown (like a zener diode) at around 7-8V, so this could occur with Vl set at 12V. This neatly illustrates the type of challenges encountered in designing this project. I had to choose a current threshold that would reliably discriminate between a BJT with emitter-base breakdown and a JFET with a high-­ resistance channel. A threshold of 5mA means a BJT with a breakdown voltage of ~6V or above will be correctly identified, as will a JFET with a drain-source on-resistance of ~1.2kW or below. This should work for just about every device out there, but the window is narrow and a good understanding of second-order effects is required. If we find the device is a JFET, we have done all we can to identify it unambiguously. As mentioned above, there is no way to tell the drain from the source as they are electrically identical. If we have a BJT, however, we still have to determine which lead is the emitter and which is the collector. To do this, we use the fact that the reverse gain is always lower than the forward gain. Test 2 consists of two parts, as shown at the bottom of Fig.3. The drive node is set to +500µA and the load current is measured in both directions. If the current in the red-to-green direction is greater than 2.5mA and higher than that in the green-to-red direction, we can assume that the green lead is the emitter. If the current in the green-to-red direction is greater than 2.5mA and larger than that in the red-to-green direction, we can assume that the red lead is the emitter. An internal view of the Semiconductor Analyser, showing the PCB and battery. The battery is mounted using double-sided foam tape. If neither of these is true, the forward gain is less than five times, so the device is probably faulty or it is something unknown. This type of discrimination testing is performed in each case of all ambiguous continuity test results. The tests undertaken are specific to the device in question, but there is only ever a handful of possibilities to sort out, so the complexity of each of these discrimination tests is similar to the one described above. Characterising the device After the part and its connections have been identified down to the Screen 1: this serial console log shows the testing process in a bit more detail, in this case for a BC847 transistor and an MMBFJ112 JFET. In both cases, the initial test flags give the same result. siliconchip.com.au Australia's electronics magazine family level, another set of tests is carried out to further characterise it. To continue with our example, if the DUT is a JFET, the pinch-off voltage is measured. If it is a BJT, its base-emitter voltage and hfe are measured. The base-emitter voltage is used to determine if the DUT is an ordinary BJT or a Darlington pair, since the latter has two base-emitter junctions in series and will thus be over 1V. In the same way, forward voltage is used to discriminate between standard diodes (~0.6-0.7V), schottky diodes (~0.30.5V) and LEDs (>1V). Only the final results are displayed on the LCD screen, but a bit more insight into the process is provided via the serial output. Screen 1 shows the serial output when testing a BC847 BJT and an MMBFJ112 JFET, each connected as per Fig.3. Both have initial test flags of 0x3C5 and you can see them going through the tests described above. The first test with the gate/base unbiased gives an IDUT(0) figure of 3.6mA in the case of the bipolar transistor, and around 10mA in the case of the JFET. The 3.6mA value for the BJT suggests an emitter-base reverse September 2026  31 Parts List – Semiconductor Analyser 1 double-sided PCB coded P9062-1-C, 130.5 × 56.5mm 1 front panel label, 55 × 128.5mm 1 Hammond 1593XBK plastic enclosure, 140 × 66 × 28mm 3 test clips; red, blue and green (CON1-3) [Cal Test CT3180-2, -5, -6] 1 1100mAh Li-Po cell, 51 ×34 × 6mm (BAT1) [Core Electronics CE04377] 1 USB4105-GF-A 16-pin USB-C connector (CON5) 1 JST S2B-PH-K-S 2-pin right-angle header, 2mm pitch (CON6) 1 Littelfuse 1210L075/24PR resettable PTC fuse or equivalent (F1) 1 SMD M3225/1210 6.8µH 1A inductor (L1) [Murata 1276AS-H-6R8M=P2] 1 320 × 240. 2.4-inch TFT LCD with ILI9341 driver and 18-pin, 0.8mm pitch flex cable (LCD1) [AliExpress 1005005796800307 “ILI9341-No Touch”] 2 SMT gull-wing tactile switches, 6.6mm, with 8.5mm actuator (S1, S2) [E-Switch TL3301PF160QG] 1 audio transducer (SPK1) [CMT-0525-75-SMT-TR] 4 4G × 6mm panhead self-tapping screws 3 lengths of hookup wire with red, blue and green insulation, each 350mm long Double-sided 3mm-thick adhesive foam tape Semiconductors 3 DG412 quad NO analog switches, SOIC-16 (IC1-IC3) 3 LMC7101 general-purpose op amps, SOT-23-5 (IC4, IC5 & IC7) 4 TLV2186 dual zero-drift op amps, SOIC-8 (IC6 & IC8-IC10) 1 STM32L433CCT6 LQFP-48 microcontroller (IC11) 1 MAX1555 battery charger, SOT-23-5 (IC13) 1 TLV61046 boost converter, SOT-23-6 (REG14) 2 MCP1711T-33 low-dropout linear regulators, SOT-23-5 (REG15 & REG16) 3 AQ4022-01FTG-C bidirectional 12V TVS diodes, SOD-323 (TVS1-3) 1 SMBJ5.0A unidirectional 5.0V TVS diode, DO-214 (ZD4) 1 3mm yellow through-hole LED (LED1) 2 BSS138K N-channel Mosfets, SOT-23 (Q1, Q2) 4 BAV99 series switching diode pairs, SOT-23 (D1-D4) Resistors (all SMD ±1% M2102/0805 unless noted) 7 100kW 1 47kW ±0.1% 1 43kW 4 36kW ±0.1% 1 30kW 5 27kW ±0.1% 1 27kW 4 20kW ±0.1% 1 20kW 1 15kW 7 10kW ±0.1% 1 10kW 2 5.1kW 1 1.2kW ±0.1% A close-up of the test clips when not 1 1kW ±0.1% in use – the plastic colour matches the wire for easy 3 1kW identification. The test clips included in the kit may not be 1 510W exactly the same but will be similar. 2 33W 1 10W Capacitors (all 50V SMD X7R ceramic M2102/0805 unless noted) 9 100nF 7 10nF 6 10µF 16V Optional parts 1 10-pin 1.27mm pitch SMT header (CON4) [CNC Tech 3220-10-0300-00] 4 small self-adhesive rubber feet breakdown voltage of about 8.4V. In the case of the BJT, the forward and reverse gain is measured. IDUT(1-2) is the lesser of the two, so the emitter must be connected to the red lead. Once the primary identification is finished, the relevant measurements are made. The BJT has a base-emitter voltage of 0.74V at 1mA (so it is not a Darlington) and an hfe of 496 at a collector current of 8.7mA. The JFET has a pinch-off voltage of -3.1V. Implementation Kit (SC7725, $95 + P&P): includes an assembled PCB with all top-side components already fitted, plus all the other non-optional parts except the case, battery and label. The hardware is best understood with reference to the block diagram (Fig.4). The input leads are protected by a circuit that limits the voltage between any two leads to a safe level. This protection is mostly there to protect against ESD (electrostatic discharge) and perhaps a fleeting accidental connection to a low-voltage, low-power circuit. The Semiconductor Analyser is not designed to be used as an in-circuit tester; it may be damaged if connected to a powered circuit. The three input lines are connected to the test nodes by nine analog switches in a 3 × 3 matrix, shown as blue-filled circles in the block diagram. As mentioned above, the DUT voltages are measured directly at the test leads. The Power Node is very simple. It has to provide 12V or 0V to whichever lead it’s connected to and must be able to source and sink at least 10mA. The Load Node is a little more complex. A variable voltage source is created using one of the microcontroller’s two digital-to-analog converter (DAC) channels. This drives the Load Node via a 1kW load resistor. The microcontroller measures the voltages on either side of the 1kW series resistor, then subtracts and scales them to calculate a signed value for the load current. The Drive Node comprises three separate sources: two bipolar current sources and a voltage source. Since only one of them is used at a time, they are all driven from a single DAC output. Two current sources, high range and low range, are needed to measure the DC gain of bipolar transistors over a very wide range. At the lower end, we need to measure hfe in the low single digits (say 2) so we need a gate drive current of around ±5mA. The low current range Australia's electronics magazine siliconchip.com.au 32 Silicon Chip Fig.4: the block diagram of the Semiconductor Analyser. The blue-filled circles are analog switches. has a full-scale current of around 120μA. Circuit details The full circuit is shown in Fig.5 and you should be able to see how it relates to the block diagram. The approximate sections that correspond to the Power Node, Load Node and Drive Node are marked in green text. The input leads are protected by three bidirectional 12V TVS diodes connected between the leads (TVS1TVS3) and three diode pairs (D1-D3) that shunt away any voltages above the 12V rail or below ground. The three input lines are connected to the test nodes by nine analog switches out of 12 in three quad packages (IC1-IC3) in a 3 × 3 matrix. The three remaining analog switches are used in the Drive Node to select one of the three sources. I have used low-cost, industry-standard DG412 quad analog switches here. These have an on-resistance of around 40W, so the DUT voltage is measured directly at the test leads. The lead voltages are buffered by op-amp voltage followers (IC10a/b and IC6b) and reduced to a level suitable for the siliconchip.com.au ADC by simple voltage dividers. I have used ±0.1% resistors here because my design goal was to keep errors for all measurements to ±0.5% or better if I could manage it. The 1kW resistors in series with the buffers’ non-inverting inputs provide an extra layer of protection for the op amp’s internal ESD diodes. The Power Node is very simple. I used an LMC7101 op amp in a non-­ inverting amplifier configuration (IC4), driven from a digital output pin on the microcontroller. The gain Screen 2: the display when a 2N7002K N-channel Mosfet is connected. This is a logic-level device, as confirmed by the low Vgs(on) figure. Screen 3: here a 3.3V 1W zener diode is connected. The reverse breakdown voltage is just 2.5V because the 3.3V specification is at a higher current. Australia's electronics magazine September 2026  33 is set at around 3.9× so that any input above 3.1V will drive the op amp to saturation. If the digital output is at logic zero, the op amp’s output will be close to 0V. The absolute voltages at the Power Node are not critical, but we do want them to be reasonably stable with load. The LMC7101 was chosen for this application (and the Load and voltage Drive Nodes) because its common-­ mode input range includes both power rails and because it can drive to within 100mV of either rail while sourcing or sinking 10mA. The Load Node is a little more complex. A variable voltage source is created by a non-inverting amplifier (IC5) driven by one of the microcontroller’s two DAC channels. The gain-setting resistors are ±0.1% types, not because we need this precision for the load voltage, but because these resistors double as a voltage divider to measure the voltage on the op-amp end of the 1kW resistor. The voltage at the DUT end of this resistor is measured via a buffer op amp (IC6a) and another precision divider. The microcontroller subtracts the two unsigned ADC results corresponding to the voltage at either end to get a signed value for the load current. The Drive Node voltage source is a non-inverting amplifier similar to the one used in the Load Node (IC7), except this time there is no need for high-precision components. The two current sources are identical, based around dual op amps IC8 & IC9, except for the 47kW and 1.2kW current-setting resistors. The 12-bit resolution of the DAC means each current step of the high range (IC8) is about 1.2µA, but to maintain a minimum 1% gain precision, the lowest current we can use is 120µA. This corresponds to a DC gain of around 70×, which is why we need a second current range. With a full-scale current of 127µA, the low range has a resolution of around 31nA. The two ranges give us a DC gain resolution of ±1% or better from hfe values of 5 to around 2700, worsening to about ±9% at the upper limit (25,000) that I have imposed. Howland current sources These two circuits are an interesting configuration known as a Howland current source. They are used 34 Silicon Chip Australia's electronics magazine siliconchip.com.au Fig.5: the full circuit of the Semiconductor Analyser. Refer to the text for a complete explanation of how it works. siliconchip.com.au Australia's electronics magazine September 2026  35 measurements and averaging samples over one mains cycle almost completely eliminates mains interference from the measurements. The ADC and DACs are powered from a special 3.3VA analog supply rail that also feeds the Howland sources’ voltage dividers and the LCD driver chip. This supply is switched off entirely (along with the 12V rail) when the Semiconductor Analyser is off. Note the easy to miss cut-out on the bottom The absolute voltage of this analog of the case for the USB-C socket. The LCD supply rail is calculated each time module is mounted to the PCB using doublesided foam tape. the unit starts by reading the voltage of an internal bandgap reference and diagram, we can use Ohm’s Law to using a stored calibration reading. In derive the simple expression for Iout this way, we can convert the ADC and shown below it. If we substitute the DAC codes to absolute voltages with a difference amplifier gain expression known degree of precision. for Vout, we get the expression for outThe LCD screen is a low-cost put current at lower right. 240×320-pixel (QVGA) TFT available In the Semiconductor Analyser, I from the usual Chinese sources. An SPI have set the ratio of R2:R1 to 3.6 for (serial peripheral interface) interface is both circuits to maximise the voltage used to communicate with the LCD’s swing at the op amp’s output. In both ILI9341 driver chip, with a couple of cases, V2 is set to the midpoint of the additional GPIOs (general purpose I/O 3.3V analog supply by 20kW/20kW pins) required for control signals. The dividers. These are the equivalent integral backlight is PWM-controlled of 1.65V sources in series with 10kW via Mosfet Q1. resistors. Another PWM channel is used to The output current of the high-­ drive the audio transducer via Q2. current source will therefore vary Both of these circuits consume relahere because they can provide a vari- from -4.95mA when the DAC voltage tively large spikes of current, so they able bipolar current programmed by a is zero to +4.95mA when it is 3.3V. are both supplied directly from the unipolar control voltage. The circuit When the DAC output is at half-scale unregulated battery voltage. looks complex, but it is pretty easy to (1.65V), the output current will be Two tactile pushbuttons complete follow because it is based on the classic zero. The low-current circuit can like- the user interface. The TEST button difference amplifier, like that shown wise source or sink between ±126.4µA. is wired to a GPIO pin which has an at the top of Fig.6. additional ‘wake up’ function that is The expression for Vout at lower Control circuitry enabled when the device is switched left in Fig.6 is easy to work out using The Semiconductor Analyser is off. superposition. You just calculate the built around an ST32L433CCT6 microWhen the Semiconductor Analyser expression for Vout for each input sep- controller, IC11. This was chosen is off, the micro is put into its lowest arately (with the other two grounded) because it is a low-cost, low-power power configuration, known as shutand add them together. device but has a powerful M4 Cor- down mode. In this mode, almost With V2 and V3 grounded, the cir- tex core and 256kiB of flash memory. everything in the micro is powered cuit looks like a voltage divider with A good deal of flash is necessary to down, including almost all the periphthe input at V1 followed by a non-in- hold the graphics files to display on erals and even the RAM. The only verting amplifier. With V1 and V3 the LCD screen. things that can wake it are the realgrounded, it looks like an inverting I used two bits per pixel (four alpha/ time clock (which I don’t use) and a amplifier with the input being V2. transparency levels) for the images and few designated ‘wake up’ pins. With V1 and V2 grounded, the circuit font glyphs as a compromise between The measured power consumption looks like another voltage divider, size and image quality. The micro- in this mode is around 1µA, so the with input V3 followed by a non-in- controller comes in a 48-pin leaded battery life when the analyser is off is verting amplifier. Adding these results surface-mounting package, making it virtually unlimited. Because the RAM together gives the expression shown. relatively easy to hand-solder. All but is powered down in shutdown mode, The output is proportional to the two of the pins are used. waking up is the equivalent of restartdifference between V1 and V2, offset The micro includes a 12-bit DAC ing from a reset. by V3. In many difference amplifier and a 12-bit ADC with a built-in overThere is an optional programming circuits, V3 is grounded, and the off- sampler. The ADC oversampler is and debug header (CON4) in case you set term disappears. configured to average 256 individual want to connect a suitable STM32 proIf we add the buffer and sense samples over 20ms for each reading. gramming or debug probe. You don’t resistor, as shown in the right-hand Oversampling reduces noise in the need to load this header if you intend 36 Silicon Chip Australia's electronics magazine siliconchip.com.au to update firmware via the USB port or if you have a pre-programmed microcontroller. The micro’s USB device peripheral is connected to the USB connector via two 33W resistors to provide some basic protection against ESD discharges. The USB control channel pins (CC1 and CC2) are pulled down via 5.1kW resistors to ensure any USB Power Delivery source connected to the unit provides the default 5V bus voltage. The microprocessor senses the bus voltage via a 20kW/30kW voltage divider. If the bus is present, the sleep timer is inhibited, and the device remains on indefinitely unless the OFF button is pressed. TVS diode ZD4 and PPTC fuse F1 protect the device from accidental bus overvoltage or (less likely) reversed polarity. A MAX1555 linear Li-ion charger manages the charging of the single-cell 1100mAh battery. A yellow LED lights up when the cell is charging. The battery voltage is boosted to 12V via switching converter REG14. This is one of the simplest boost converter chips I have ever used, requiring just an inductor and one output capacitor. Two linear regulators provide the 3.3V digital and analog rails. As mentioned above, the 12V rail and the 3.3VA rail are only enabled when the unit is awake. The other 3.3V rail powers only the microcontroller core and is always on. All three regulators were chosen for their very low quiescent currents, which are included in the 1µA off consumption mentioned above. As a happy accident, they all come in hand-soldering-friendly SOT-23-5/6 packages. This is especially pleasing in the case of the boost converter, as most switching regulators seem to come only in tiny leadless packages that are a pain to solder. Device firmware upgrade The STM32L43xxx family of microcontrollers (along with many others from ST) contains a bootloader that can be used to download firmware without the need for a specific programmer or debugger. The bootloader code resides in a special area of flash, separate from the main user memory, and it can’t be modified by the user. Under normal operation, the microcontroller boots into the main flash siliconchip.com.au Fig.6: the Howland current source is a clever circuit that can produce a programmable bipolar current output. It is based on a classic difference amplifier. Fig.7: these overlay diagrams show where parts go on both sides of the PCB. Resistors with red labels are ±0.1% types; the rest can have a ±1% tolerance. The test lead wires loop through the strain-relief holes in the PCB and are soldered from the top side. and executes the user code from there. However, if the main flash is empty or the BOOT0 pin is pulled high when the microcontroller emerges from reset, the bootloader code is executed instead. The bootloader’s role is to download the user code through one of several serial interfaces (USART, CAN, USB, I2C, I3C or SPI) and load it into the flash memory. A specific Australia's electronics magazine communication protocol is defined for each interface. We are only interested in the USB interface, which uses a protocol known as “device firmware upgrade” or DFU. I will explain how to use DFU mode to load or upgrade firmware later. For now, it is enough to know that the BOOT0 pin of the microcontroller is normally pulled down by a resistor, but it can be pulled high by September 2026  37 bridging a pair of pads on the PCB (JP2) to enter DFU mode. Construction All the components except for the battery are mounted on a PCB coded 9062-1-C that measures 56.5 × 130.5mm. Most of the components are mounted on the top side of the board, with the exceptions being the LCD screen, pushbuttons, beeper and the battery charge LED. Start construction by mounting all the components on the top of the board. Make sure you observe polarity where relevant and watch where the ±0.1% resistors are placed. They are marked in red on the overlay diagram (Fig.7). While the parts are all spaced out fairly well and so could be fitted in any order, we suggest you start with the finer-pitch devices or those with many leads, such as the regulators and ICs, Fig.8: drill the front of the enclosure and the end panels according to this diagram. Don’t forget to snip out the two bosses inside the case as shown. then move on to the discrete semiconductors and passives. The advantage of doing it that way is that you have a little more room to work on the more difficult devices. Various soldering methods could be used, including an IR reflow oven, hotplate, hot air wand, or regular soldering iron. If using a regular soldering iron, apply a little flux paste to the pads before placing each part, then check its orientation carefully after placing it and tack-solder one pin. Check the orientation and positioning again and adjust it if necessary. Once you are happy that all leads are centred over the correct pads, tack another pin, then add a little more flux paste on top of the leads. You can then either drag-solder the remaining leads or solder them one at a time with a small amount of solder on the tip of a clean iron. Make sure you don’t touch the two initial joints holding the part in place until more joints have solidified. Also be sure to refresh those initial joints with a little flux paste and some extra heat from the iron to ensure they have flowed correctly. If any pins are bridged during soldering, simply add more flux paste and use a little solder-wicking braid pressed down by the tip of a hot iron to draw the excess solder away. Flux is your friend when soldering finepitch devices. The USB connector has both SMT and through-hole pads. If you solder the through-hole pins first, it helps to locate the SMT pins, making it much easier to solder. It is pretty easy to accidentally create a solder bridge in the microcontroller of USB connector leads, so check these carefully under magnification and clean up any possible bridges with solder wick and plenty of flux. It’s easier to see possible bridges after soldering if you clean away the flux residue (eg, using isopropyl alcohol and a lint-free cloth or nylon brush). Once you have fitted all the components to the top of the board, it’s time to prepare the case. Case preparation Next, prepare the case according to Fig.8. Mark out the top of the case carefully, then remove the two bosses on the inside of the case top as shown 38 Silicon Chip Australia's electronics magazine siliconchip.com.au in the figure. These coincide almost exactly with the locations of the tactile switches, so will interfere with drilling their holes if not removed. It is sufficient to just snip them off with side cutters so they are more-or-less flush with the inside of the case. Drill the holes and cut out the display window. I added a chamfer to the edges of the display opening, but that is purely cosmetic, so it’s entirely optional. The enclosure is supplied with both flat and profiled end plates – we only use the flat ones. I made the three lead holes 2mm in diameter because that suited the test leads I was planning to use. You may have to adjust this to suit the leads you intend to use. The USB slot is best made by drilling two holes as shown and cutting out the material between them with a sharp blade. Once the case is ready, you can fit the components on the bottom side of the PCB. It is best to fit the LCD screen first. Place the LCD face-down on the bottom of the board with the end of the flat flex aligned vertically with the horizontal lines on the overlay. Make sure the contacts are aligned with the pads, and the LCD is parallel with the edges of the PCB. Use a small piece of tape to temporarily hold the flat flex in position, then carefully solder each connection. Once you have finished that, fold the display up into its final position and temporarily secure it to the board with sticky tape. The display will be finally fixed down with double-sided foam-core tape, but I suggest testing everything first. Accessing LCD terminations is difficult to impossible once it is permanently affixed. Now you can add the pushbuttons, making sure to use the marks on the PCB silkscreen to centre the switches’ actuators so they line up with the holes in the enclosure. Solder in the audio transducer next, then thread the LED through the appropriate holes, taking care to get the polarity right (the cathode goes toward the USB connector), but don’t solder it just yet. Fit the PCB assembly into the case and secure with self-tapping screws. You can now push the LED through the hole in the case just far enough that the domed part at the top is just proud of the front panel. When you are happy, you can solder the leads siliconchip.com.au Screen 4: the free STM32CubeProgrammer app can be used to download the firmware to the Semiconductor Analyser via its USB port. Fig.9: a simplified 3D view of the finished board, showing how the wires, LCD screen, LED and other parts are fitted. on the top side of the board and trim off the excess. Testing and programming We are just about ready to fire up the unit and get it working. The best way to start is to connect a current-limited power supply to CON6, set to 4V DC. The correct polarity is marked on the overlay diagram in Fig.7. A current limit of 150mA is a good place to start. Fire up the supply and use a multimeter to measure the 3.3V rail, across C18 (the ceramic chip capacitor between REG16 and REG15) or somewhere else convenient. If the 3.3V rail is good, you can go ahead and connect the battery. Check that the battery charge light (LED1) comes on when the USB port is connected to a power source. If you have a pre-programmed microcontroller, the device should come to life, and you can skip the next section. If you are programming your unit via Australia's electronics magazine the debug header, now is the time to do it. If you plan to use DFU mode, you can follow the steps outlined below. 1. Download the STM32CubeProgrammer application. This is a free download (registration required, unfortunately) from the ST website (www.st.com/en/development-tools/ stm32cubeprog.html). It is available for Windows, macOS and Linux. 2. Unplug the Semiconductor Analyser from the USB cable and disconnect the battery. Solder a short piece of wire or an M2012/0805 0W resistor across the pads labelled “DFU” on the board. This forces the microcontroller to boot into DFU mode. You can skip this if you have an unprogrammed chip – it should boot into DFU mode by default. 3. Reconnect the battery and connect the USB cable to your computer. Open the Programmer application and select USB from the dropdown at the top right of the screen (see Screen 4). September 2026  39 Screen 5 (left): testing a TTC004B 1.5A NPN bipolar junction transistor (BJT). The hfe is specified as 140-280 at 100mA and this one falls close to the middle of that range at nearly 9mA. Screen 6 (middle): the LCD screen shows this display while tests are being run. They take a second or two. Screen 7 (right): this splash screen appears when you press the “TEST” button. Click “Connect”. The programmer should connect and read the device info into the “Target information” panel at the bottom right. 4. Click on the “Open File” tab and navigate to the ELF file containing the code (available as part of the download package from siliconchip.au/ Shop/6/3644). Click on the “Download” button on the top right of this tab and the flash will be erased and programmed. You may get the error messages “Failed to Download Sector[0]” and “failed to Download the File”, but you can ignore them. They appear to be caused by a bug in the programming software that occurs when the code to be flashed occupies more than 50% of the available space. I have not been able to find a fix online, although there are a few posts describing this same problem. 5. Close the programmer, unplug the USB cable and the battery, then remove the shorting link if you used it. Once you plug the battery back in, the device should boot into the uploaded firmware. That’s all there is to it. Finalising assembly Once the device reboots, it should display the version screen briefly, then commence a test cycle. The test should result in an error message saying that all three test leads are open circuit. If everything is satisfactory, you can finish the assembly. Remove the PCB from the case and permanently affix the LCD screen to the PCB using a few pieces of double-­sided adhesive foam tape. The stuff I used is about 3mm thick, and this positions the display firmly up against the front panel. Don’t forget to remove the protective film from the face of the LCD. You can now connect the test leads to the board. These are fed through the end panel, then from the top side of the board, down through the strain-relief holes and soldered in from the top side of the PCB as shown in Fig.9. I made my test leads about 350mm long, although the leads that come pre-attached to the clips supplied in our kits will be closer to 200mm long, which is still plenty. Apply the label to the front panel (Fig.10). The artwork is available for download from siliconchip.au/ Shop/11/3645 I printed mine on self-adhesive glossy paper and covered it with transparent adhesive vinyl film. Use a sharp blade and a straightedge to cut out the display window before applying the label. The label has two rectangles for the display window. Use the outer one if you chamfered the display opening, or the inner one if you did not. The holes for the pushbutton actuators and LED are best made with the label in place. You do not need to cut the label over the hole between the two pushbuttons – the hole in the case is enough for the beeper sound to escape. Finally, you can fix the Li-Po cell to the inside base of the case using double-sided tape. Once you’ve put the case together your Semiconductor Analyser is ready to use. If you run into trouble, you can use the command-line interface for troubleshooting. Connect the USB cable to your computer and fire up a serial terminal program, then connect to the port associated with the USB device. If you type “help” at the “>” prompt, you will see a list of commands you can use to manually control the switch matrix, Power, Load and Drive Nodes, and read the load current or test lead voltages. The CLI is basic, but it has autocomplete for the command (press the tab key) and you can use backspace SC to correct mistakes. Fig.10: the front panel label artwork. This can also be downloaded as a PDF from the Silicon Chip website (the link is in the text). 40 Silicon Chip Australia's electronics magazine siliconchip.com.au