Build Your Own Curve Tracer, Part 1: Introduction

This series of articles presents a build-it-yourself tutorial on two-port analyzers (TPAs), historically called curve tracers . A detailed design of a single-board TPA is presented that can be built for about $100 US – or be a “springboard” to your own TPA design.

Curve tracers are measurement instruments used to characterize devices such as transistors and diodes. New ones cost thousands of dollars. Even old Tektronix 575 transistor curve tracers, implemented with electron tubes and found on the surplus market, cost $75 or more. Their huge power dissipation and size, lack of computer interface, and difficulty to maintain make them largely undesirable, even at surplus prices. This article series draws on newer electronics to design a cheaper, simpler laboratory-quality TPA that can be built by a technician or engineer in little time and at low cost. (Note: for more background on TPAs and integration, see the series, 2-Port Analyzers on a Chip? Part 1: What Are TPAs?, by the author during 2Q14 titled “Two-port Analyzers (TPA) on a Chip” on Planet Analog.)

The TPA202

The TPA202 is the first of the TPA200 series and is the only one that is not computerized. It is simplest to build and understand and is a place to start with TPAs. The TPA202 Manual is available. (Request a copy from My one sloppily-built prototype – a kludge – exists but works and meets specs; see photos below.

The parts used in all the 200-series instruments are readily available from the major parts suppliers, and often are chosen because they are enduring “legacy” parts. For instance, low-cost LF353, LF347, LM324, and LM358 op-amps are sometimes used, though where there is a design advantage in using newer “neo-classical” parts (TLV2372, TLC2272A, TLE2022A), they are used instead.

The main difference between a TPA202 and a classic curve tracer is that it has a numerical display instead of a graphic display of a family of v-i curves. Though it lacks the larger view of a curve tracer, it allows for DVM accuracy of the plotted points on the curves, albeit only one at a time. By turning front-panel knobs and watching the display change, curves can be “traced out”. Furthermore, some parametric measurements are faster and more accurate. By setting the input-port (base-emitter) current to a round value such as 100 μA, the BJT β is read on the output-port (collector-emitter) 10 mA current range as the displayed number, and to ≤ 1 % accuracy.

TPA Subsystems

A two-port analyzer consists of two drive-sense or stimulus-measure units (SMUs). Each SMU drives a port.

A port is a pair of interface nodes or terminals of a network (circuit) or device, shown above as a box or block, with a voltage across them and current into or out of them. The + terminal current is defined by port convention as positive going into the block. The terminal currents flowing into or out of the block are equal in magnitude. (Otherwise, a third terminal would need to be included and this defeats the definition of a port.) The unconventional source symbols are generalizations of either a Thevenin or Norton equivalent source, where the x variables are either voltages or currents, written in functional notation, f (x) .

The most general SMU can be configured as either a voltage or current source, and both output port voltage and current are measured. A general SMU block diagram is shown below.

The device to be tested, or device under test (DUT) can have one or two ports. An active device such as a transistor has input and output ports where the input port is across the base-emitter or gate-source terminals and the output port is across the collector-emitter or drain-source terminals. The two ports have a negative terminal in common, the common (COM) terminal. COM can either be grounded to the system ground or be floating. In the TPA202 it is grounded.

SMUs can be less general than the one shown above. Useful functionally-reduced SMUs are V source, I sense and I source, V sense. Abbreviations for the current and voltage drive (DR) and sense (SN) circuits or their output variables for the two ports are given in the following table.

The SMU shown above has a current sense circuit (ISN) that is a high-side circuit (ISN-H) because the sense circuit is in series with the positive or high-side terminal. The TPA202 input-port (INP) SMU has high-side current sensing and the output port (OUT) has low-side current sensing (ISN-L), where the OISN circuit is between the negative port terminal and the amplifier or source ground. The TPA202 output source (ODR) is a voltage source in series with the ISN sense resistor which also functions as a current-limiting resistor. The input source can be switched to be either a voltage or current source – voltage for FET gate-source VGS and current for BJT base IB . It too has the ISN-H sense resistors in series with the drive amplifier. Each drive amplifier consequently has a Thevenín equivalent circuit output consisting of a voltage source in series with a sense resistor, but feedback results in close to an ideal voltage or current source at the port terminals.

TPA202 Specifications

The TPA202 has a 3.3 digit DVM with 7-segment LED display and three units LEDs (nA, μA, mA) to its right. Voltage is measured in units of volts only. For the input port, it is resolvable down to 10 mV and for the output port, to 0.1 V. Port voltage or current can be displayed for either port. Source or drive voltages are varied over a decade range (or 1.5 decades for output voltage) using input and output port front-panel 10-turn potentiometers. They control the voltage of either port or input-port current. Transistor parameters are measured by sequencing through port measurements and calculating the parameter from the displayed numerical values. Some static or quasistatic parameters that are capable of being measured using a TPA202 are β0 , α0 , r0 , VA , rm , and re .

Specified full-scale (fs) values are given in the following table. Measurement resolution decreases linearly for values less than fs. Calibrated accuracy depends on op-amp input characteristics (offset voltage and its drift with temperature). Inaccuracy over a temperature range of 20 o C, +/-15 o C depends on your selection of op-amps, but for the ones on the circuit diagrams (given in future parts of this series) and when calibrated, can range from about 0.2 % to over 1 %. For the design as given, op-amps are selected that result in a measurement inaccuracy of around 0.5 to 2 %. By selecting more precise op-amps or resistors, inaccuracy can be reduced. Other causes for inaccuracy, such as leakage currents, are insignificant for the specified minimum current range.


How did this project originate? Over the years I have developed various microcomputer (μ C) and instrument subsystems and some prototypes of them have been built. Since I am not a “thower-outer”, they languish in my laboratory, awaiting a time of functional fulfillment. Then it finally came to me about three years ago that I can apply all of this shelved technology by using it to build limited-edition (LE) measurement instruments. My instrument quest zealously began. So many seemingly good ideas for mainstream-performance, “workhorse” instruments came to mind that I have summarized them in tabulated form at under Instruments. I am working my way through the list, project by project, and in the spirit of Linux, invite other “hackers” to join the quest of building up a base of open-source, medium-performance instrument designs that do not involve some of the very subtle complications of high-performance instruments and that can be built and made to work by most electronicists without undue grief. The TPA202 design is open-source and these articles are chatty explanations of it.

For total ownership status – that is, buying into the design as well as the parts – building and making an instrument perform puts you in a position of knowing it in depth. You buy not only its parts but also buy into its concepts and can thus calibrate, repair, modify, and even possibly evolve it indefinitely. It becomes yours, both physically and conceptually. These open-source instruments can even be made and sold without anything beyond their Creative Commons license. A cottage instrument industry can possibly spring up just as Linux suppliers have.

Whenever at least three of you indicate that you want a TPA202 (my email address is at, I will finish the board layout and provide blank boards for close to my cost – probably about $40 US each.

The only unusual part of the TPA202 is the 2P7T rotary switch. I have a stock of them that should be adequate in supplying anyone who buys a circuit-board. Ten-turn pots are somewhat expensive ($12 US each) but worth the higher resolution for precise settings. In the next part, we’ll go “under the hood” and look at TPA202 design and operation.

2 comments on “Build Your Own Curve Tracer, Part 1: Introduction

  1. Jimelectr
    January 20, 2017

    Darn it, Dennis, now you've prompted me to undertake yet another project!  At first I was like, pshaw, curve tracer, shmerve tracer, who needs one?!  I haven't used one for 30 years.  But then I started thinking, gee, I'm interested in power amplifiers (PA's), both for audio and radio, so a curve tracer could come in really handy, duh!

    Not like I need another project!  I'm between jobs for about another week and a half now, having been laid off from Broadcom Ltd last November and starting at Raytheon January 30th, and there's no shortage of stuff left to do during this “retirement preview”.  Let's see, I've been concentrating on diagnosing and repairing faulty equipment in my garage lab, while off-and-on working on my homebrew 70 cm band ham radio receiver and streamlining my 25-6000 MHz RF synthesizer box.

    I have a Wavetek model 166 function generator that used to flatline the squarewave, pulse train, and TTL outputs after a few minutes.  I opened it up and found the +5.2 V supply dropped down to about 0.4 V coincident with the flatlines.  But after probing on the +5.2 V rail for a while, it has behaved itself since.  But in the process of troubleshooting the flatline problem, the triangle and sawtooth waveforms got goofed up to close to squarewaves!  I tracked that down to a pot that I must have bumped that shorted the collectors of a diff pair used for injecting triangle or sawtooth current into the preamp.  Tweaked it back to where the paint lines matched up on the trimpot, and now it's good to go.

    And there's my TLA711 logic analyzer that wouldn't even boot up because of a dead battery in the realtime clock module.  Unfortunately that RTC module is obsolete and unavailable, but there is a cross to it, without the battery and 32.768 kHz crystal.  So I made my own, and jsut by chance, a battery holder fit perfectly between the backup power pin and the VSS (GND) pin on the rather large 24-pin DIP.  Unfortunately, it still complains about not having a hard disk connected, so I have to open it up again.

    Then there's my Iwatsu SS-5710D 60 MHz scope with the horizontal circuit problem.  It only sweeps about 90% of the way across the screen and puts a bright dot at the end of the trace if I keep rotating the Horizontal Position knob clockwise.  I got a schematic off the Internet and tracked it down to the horizontal sweep signals saturating at about 100 V differential, but I still don't know why.  I'll have to ask for help on one of the forums.

    And in troubleshooting the Iwatsu, my go-to scope, a Tek 7904 with a 7A26 vertical amp and 7B85 timebase, now shows some funniness.  Channel 1 of the 7A26 goofs up a few minutes after power-up when volts per div is greater than 50 mV.  The gain goes way high and the edges of the calibrator squarewave round to about 100 us transitions instead of <1 us.  From the 7A26 schematic (Internet, of course!), this leads me to suspect the second 10X attenuator from the input, as that one is clicked in for >50 mV/div and out for 50 mV and less.  The trace is unstable after the first few minutes of working properly, so I suspect a poor connection in the contacts clicking the 10X attenuator in and out, but they look so delicate that I don't want to touch them.  Again, I'm appealing to those knowledgeable folks on the Internet to help me out when I've reached the end of my knowledge.

    The synthesizer works OK, but it leaves a few things to be desired.  I dropped +15 V from a wall-wart power supply to +5.5 V to go to the Hittite/ADI HMC833 synthesizer evaluation board and post-amplifier board through an old-time LM317 linear regulator, so it warms up the whole aluminum box!  I'd like to build a PCB with a switching regulator and the power-on reset circuitry to clean that all up.  And the output amplitude drops like a rock up at high frequencies because both the HMC833 synthesizer chip itself and the post-amplifier roll off, so I'd like to add an RF PCB to level the output.  And in order to get minimum attenuation in a PIN diode attenuator in the leveling circuit, I need to bring in greater than the +15 V that the 10 MHz reference OCXO needs, so I plan to make it work with a ubiquitous +19.5 V laptop computer power supply.

    Beyond that, I have a couple of semi-working computer monitors that somebody in my neighborhood left at the curb around Christmastime.  One appeared to be working, but now that I have it all set up in my computer-music-scrapbooking room, of course it has crapped out!  I suspect electrolytic caps, as we had numerous monitors at Broadcom die due to nothing but failing electrolytics.  Wish me luck!

    Sorry for the long post, but that's just me.  Have a good one!


  2. D Feucht
    January 22, 2017

    Hi Jim,

    I can relate to your list of broken instruments. I have my own pile of them.

    “Tweaked it back to where the paint lines matched up on the trimpot, and now it's good to go.”

    It is always a real boost to have a broken instrument working again. I've fixed my ESI 253 RLC meter about 3 times, here in the humid air of the Belize jungle. (Each time, it was a CMOS 405x part; the glass passivation layer must not have held out the moisture.)

    Your Iwatsu scope: if you run into a roadblock on it, send me a copy of the circuit diagram for the horizontal amplifier and we'll figure it out. As for the 7A26 and 7B85, the designers of both of those plug-ins are friends of mine. Electromechanical parts – even Tek ones – are prone to have a higher failure rate than electronic parts – something we have to live with for now.

    I have tried to minimize the electromechanics of the TPA202, though without a microcomputer, it has more manual switches than otherwise. I too did not think I would use it much, but it turns out to be a much more versatile instrument than originally supposed. Programmable supplies are always in demand on the bench.

Leave a Reply

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