Way back in a galaxy distant (really, the identical galaxy however the Milky Means is transferring so it should logically be distant from the place it was manner again then), I sketched up a easy circuit for a DC coupled amplifier that I believed (in my youthful naivety) seemed very cool. See Determine 1.
Determine 1 It’s a much better (temperature sensing) oscillator than an amplifier. Transistor bias is particularly chosen so: d(Von – Vsat)/dT = 1/300 at 300oOkay.
Wow the engineering world along with your distinctive design: Design Concepts Submission Information
It consists of three direct-coupled common-emitter amplifers (Q1 by way of Q3) for a cascaded achieve of maybe 120 dB (β3) with inverse suggestions from Q3 to Q1. Capacitor C1 was supposed to supply a single dominant loop bandwidth-limiting stability-promoting pole.
I couldn’t wait to wire it up and see the way it labored. Spoiler alert: it (very a lot) didn’t—at the very least not as an amplifier.
To say it “oscillated persistently” could be (very a lot) an understatement. In truth, “oscillation suppression” C1 might be elevated seemingly with out restrict, with out lowering oscillation amplitude one iota. Oscillation frequency decreased (inversely proportional to C1), however oscillation amplitude remained stubbornly unattenuated. Determine 2 reveals why.
Determine 2 C1 gives oscillation-provoking constructive suggestions by bypassing Q2, as revealed by Q1 and Q2 collector waveforms.
C1’s connection between Q1 and Q2 collectors successfully bypasses Q2 and its stage of sign inversion, making the web polarity of loop suggestions round Q1 and Q3 constructive moderately than unfavourable as I initially supposed. C1 thus ensures oscillation, as a substitute of stopping it!
The 4-step sequence of every oscillation cycle turned out as follows:
- Vc3: The cycle of oscillation begins with Q3 turn-off, (inexperienced hint) driving Q3’s collector Vc3 to ~4 V.
- Vc1: This activates Q1, driving its collector from Von = 700 mV to Vsat = 100 mV (pink hint), which turns Q2 off.
- Vc2: Concurrently, C1 {couples} the Von – Vsat = 600 mV transition at Vc1 to Vc2 (yellow hint), holding Q3 off. This state persists till Vc2 ramps again as much as Q3 Von, which drives Vc3 low (inexperienced hint).
- This turns Q1 off, permitting Vc1 (pink hint) to ramp the left aspect of C1 as much as Von. This activates Q2, driving Vc2 to Vsat (yellow hint), turning off Q3, returning Vc3 to 4 V (inexperienced hint), beginning the following oscillation cycle at step 1.
What’s of specific curiosity right here is that the oscillation interval is the same as the sum of the Vc1 ramp interval (T1) and Vc2 ramp interval (T2), and the oscillation frequency equals the sum’s reciprocal.
T1 = T2 = T
T = R1C1(Von – Vsat)/(5 – (Von + Vsat)/2) + Td (transistor-delay instances)
Fout = 1/(2T + 2Td)
Research of the 2N3904 datasheet reveals that Vsat = 100 mV unbiased of temperature and collector present (as much as Ic = ~10 mA), whereas Von varies at a predictable charge with each temperature and present. If bias present is chosen rigorously, a temperature response carefully approximating Fout = 1000Tabs = (1 kHz/ oOkay) might be achieved over a helpful vary of temperatures. The element values proven in Determine 1 achieve this from 270 oOkay to 330 oOkay = -3 oC to +57 oC = 27 oF to 135 oF, that is plotted because the realized response (pink curve) in Determine 3 in opposition to the perfect 1 kHz/ oOkay (blue line).
Determine 3 Determine 1’s output frequency versus absolute temperature (Tabs). Fout = 1 kHz(Tabs +/-2 oOkay).
So, as a substitute of an amplifier, Determine 1 turned out to be the premise for an attention-grabbing digital temperature sensor.
A few concluding feedback observe.
Self-heating: Of concern with all energetic temperature sensors is self-heating error. The 2N3904 in TO-92 warms itself by ~200 oC/Watt = 0.2 oOkay/mW in nonetheless air (worst case, much less in transferring air). Q3 dissipates < 2 mW whereas Q2 conducts for such a small fraction of the cycle that its warmth dissipation and temperature rise are negligible.
Q3 contributes half of the oscillation interval so whole max self-heating error of the Determine 1 circuit is subsequently 1 kHz*0.2/2 = ~100 Hz = 0.1 oOkay. That is acceptable in most functions (notice, nonetheless, that Q2 and Q3 ought to be properly remoted from the ~65 mW dissipated by the resistors).
Value/efficiency: Evaluating worth and efficiency of discrete element circuits to competing ICs seldom goes properly for the multi-part system, however Determine 1 doesn’t fare too badly. Its accuracy (+/- o2 over a helpful vary of temperatures) and determination (e.g., ~16bits for a ¼ second counting interval) measure up properly in opposition to typical digital temperature measurement chips (e.g., TMP125) as does the price of its elements (~$2).
All issues thought of; it was a really cool comfort from the frustration of a failed amplifier design.
Stephen Woodward’s relationship with EDN’s DI column goes again fairly a good distance. Almost 100 submissions have been accepted since his first contribution again in 1974.
Associated Content material
