Tuneful track-tracing

Another day, another dodgy device. This time, it was the continuity beeper on my second-best DMM. Being bored with just open/short indications, I pondered making something a little more informative.

Perhaps it could have an input stage to amplify the voltage, if any, across current-driven probes, followed by a voltage-controlled tone generator to indicate its magnitude, and thus the probed resistance. Easy! . . . or maybe not, if we want to do it right.

Wow the engineering world with your unique design: Design Ideas Submission Guide

Figure 1 shows the (more or less) final result, which uses a carefully-tweaked amplifying stage feeding a pitch-linear VCO (PLVCO). It also senses when contact has been made, and so draws no power when inactive.

Most importantly, it produces a tone whose musical pitch is linearly related to the sensed resistance: you can hear the difference between fat power traces and long, thin signal ones while probing for continuity or shorts on a PCB without needing to look at a meter.

Figure 1 A power switch, an amplifying stage with some careful offsets, and a pitch-linear VCO driving an output transducer make a good continuity tester. The musical pitch of the tone produced is proportional to the resistance across the probe tips.

This is simpler than it initially looks, so let’s dismantle it. R1 feeds the test probes. If they are open-circuited, p-MOSFET Q1 will be held off, cutting the circuit’s power (ignoring <10 nA leakage).

Any current flowing through the probes will bring Q1.G low to turn it on, powering the main circuit. That also turns Q2 on to couple the probe voltage to A1a.IN+ via R2. Without Q2, A1a’s input protection diodes would draw current when power was switched off.

R1 is shown as 43k for an indication span of 0 to ~24 Ω, or 24 semitones. Other values will change the range, so, for example, 4k3 will indicate up to 2.4 Ω with 0.1-Ω semitones. Adding a switch gave both ranges. (The actual span is up to ~30 Ω—or 3.0 Ω—but accuracy suffers.) Any other values can be used for different scales; the probe current will, of course, change.

A1a amplifies the probe voltage by 1001-ish, determined by R3 and R4. We are working right down to 0 V, which can be tricky. R5 offsets A2a.IN- by ~5 mV, which is more than the MCP6002’s quoted maximum input offset of 3.5 mV. R2 and R6–8 help to add a slightly greater bias to A1a.IN+ that both null out any offset and set the operating point. This scheme may avert the need for a negative rail in other applications.

Tuning the tones

The A1b section is yet another variant on my basic pitch-linear VCO, the reset pulse being generated by Q4/C3/R13. (For more informative details of the circuit’s general operation, see the original Design Idea.) The ’scope traces in Figure 2 should clarify matters.

Figure 2. Waveforms within the circuit to show its operation while probing different resistances.

This type of PLVCO works best with a control voltage centered between the supply rails and swinging by ±20% about that datum, giving a bipolar range of ~±1 octave. Here, we need unipolar operation, starting around that -20% lowest-frequency point.

Therefore, 0 Ω on the input must give ~0.3 Vcc to generate a ~250 Hz tone; 12 Ω, 0.5 Vcc (for ~500 Hz); and 24 Ω, ~0.7 Vcc (~1 kHz). Anything above ~0.8 Vcc will be out of range—and progressively less accurate—and must be ignored.

The output is now a tone whose pitch corresponds to the resistance across the probes, scaled as one semitone per ohm and spanning two octaves for a 24 Ω range (if R1 is 43k).

The modified exponential ramp on C2 is now sliced by A2b, using a suitable fraction of the control voltage as a reference, to give a “square” wave at its output—truly square at one point only, but it sounds OK, and this approach keeps the circuit simple. A2a inverts A2b’s output, so they form a simple balanced (or bridge-tied load) driver for an earpiece. (There are problems here, but they can wait.)

R9 and R10 reduce A1a’s output a little as high resistances at the input cause it to saturate, which would otherwise stop A1b’s oscillation. This scheme means that out-of-range resistances still produce an audio output, which is maxed out at ~1.6 kHz, or ~30 Ω. Depending on Q1’s threshold voltage, several tens of kΩs across the probes are enough to switch it on—a tad outside our indication range.

Loud is allowed

Now for that earpiece, and those potential problems. Figure 1’s circuit worked well enough with an old but sensitive ~250-Ω balanced-armature mic/’phone but was fairly hopeless when trying to drive (mostly ~32 Ω) earphones or speakers.

For decent volume, try Figure 4, which is beyond crude, but functional. Note the separate battery, whose use avoids excessive drain on the main one while isolating the main circuit from the speaker’s highish currents.

Again, no power is drawn when the unit is inactive. (Reused batteries—strictly, cells—from disposed-of vapes are often still half-full, and great for this sort of thing! And free.) A2a is now spare . . .

Figure 3 A simple, if rather nasty, way of driving a loudspeaker.

Setting-up is necessary, because offsets are unpredictable, but simple. With a 12-Ω resistance across the probes, adjust R7 to give Vcc/2 at A1b.5. Done!

Comments on the components

The MCP6002 dual op-amp is cheap and adequate. (The ’6022 has a much lower offset but a far higher price, as well as drawing more current. “Zero-offset” devices are yet more expensive, and trimmer R7 would probably still be needed.)

Q3, and especially Q1, must have a low R_DS(on) and V_GS(th); my usual standby ZVP3306As failed on both counts, though ZVN3306As worked well for Q2/4/5. (You probably have your own favorite MOSFETs and low-voltage RRIO op-amps.) To alter the frequency range, change C2. Nothing else is critical.

As noted above, R1 sets the unit’s sensitivity and can be scaled to suit without affecting anything else. With 43k, the probe current is ~70 µA, which should avoid any possible damage to components on a board-under-test.

(Some ICs’ protection diodes are rated at a hopefully-conservative 100 µA, though most should handle at least 10 mA.) R2 helps guard against external voltage insults, as well as being part of the biasing network.

And that newly-spare half of A2? We can use it to make an active clamp (thanks, Bob Dobkin) to limit the swing from A1a rather than just attenuating it. R1 must be increased—51k instead of 43k—because we no longer need extra gain.

Figure 4 shows the circuit. When A2a’s inverting input tries to rise higher than its non-inverting one—the reference point—D1 clamps it to that reference voltage.

Figure 4. An active clamp is a better way of limiting the maximum control voltage fed to the PLVCO.

The slight frequency changes with supply voltage can be ignored; a 20°C temperature rise gave an upward shift of about a semitone. Shame: with some careful tuning, this could otherwise also have done duty as a tuning fork.

“Pitch-perfect” would be an overstatement, but just like the original PLVCO, this can be used to play tunes! A length of suitable resistance wire stretched between a couple of drawing pins should be a good start . . . now, where’s that half-dead wire-wound pot? Trying to pick out a seasonal “Jingle Bells” could keep me amused for hours (and leave the neighbors enraged for weeks).

 —Nick Cornford built his first crystal set at 10, and since then has designed professional audio equipment, many datacomm products, and technical security kit. He has at last retired. Mostly. Sort of.

 Related Content

• Power amplifiers that oscillate— Part 2: A crafty conclusion.
• Revealing the infrasonic underworld cheaply, Part 2
• A pitch-linear VCO, part 2: taking it further
• 5-V ovens (some assembly required)—part 2

The post Tuneful track-tracing appeared first on EDN.