Pulling some things together, prior to designing a complete voltage-controlled multivibrator (VCMV), to assure success in a larger circuit design. By modifying the temperature stable constant current source (CCS) to accept a voltage input, and then inverting the current direction (from current source to current sink), it was possible to construct a voltage-controlled current source (VCCS) that could pull current down from a capacitor tied to a reference voltage. For precision current output ranging for a given control voltage input, a zener reference diode was included. To set the Vin versus Iout range precisely, an adjustment potentiometer is added.
This sub-circuit represents a building block for a precision VCMV, where a precision capacitor is alternatively charged down from a positive reference voltage, first from one side, then the other. The multivibrator thus produces a sawtooth wave (seen differentially, across the capacitor) at 2x the multivibrator square wave output frequency, producing equal-time ramps for a given constant control current (hence control voltage). This idea, and example waveforms, are shown in the block diagram from the notebook page below:
The new VCCS circuit design uses complementary symmetry NPN + PNP differential pair transistors. This was done with my goto parts, the high-ß 2N5088 (NPN) and 2N5087 (PNP). As was shown in the temperature-stable CCS, when the base-emitter junctions are free to move, and if the two transistors are modestly thermally-coupled, temperature-stable base-emitter voltages occur differentially. This provides a relatively constant output current, referencing the VBE of one leg of the differential circuit, across a current-setting resistor, weighted against a constant voltage (like the power supply rail). This circuit allows VBE to change mainly based on the collector current demanded. For this VCMV, only a ≈1.5:1 change in current is needed, because the frequency range needed is very small. The Fout required is only ≈22-33 kHz, covering 0-10 kHz of the audio band, after down-conversion from a 32.768 kHz Local Oscillator (LO), a sinusoidal reference. So, very little ∆VBE occurs from current changes. And the ∆VBE induced by temperature gradient is to the first or self-cancelling: ∆VBE ≈ 0.
An initial VCCS circuit was made to insert a potentiometer to change the ratio of currents between the two legs of the differential pair. The idea was to find a balance point so that for a given precise input voltage, the VCCS IOUT could be calibrated to a specific value. Here's that circuit:
Another thing added to this initial VCCS was a ramp reset circuit that allowed a capacitor to be charged downward against the +15V rail. This circuit used a 2N3904 to accept a positive-going square-wave gate signal input, to then drive a 2N3906 switch, which would shunt the capacitor voltage to nearly the +15V rail (e.g. +15V - VCE(sat)). This allowed testing the VCCS as an actual ramp generator, without having deal with precise current switching circuits for astable operation in a an actual multivibrator.
Adding in the zener reference diode, and designing the current ranges for a specific ramp slope for a design range of ∆V = 5.0V using a precision capacitor, was next. Over the years of collecting parts for the Model III EMS, among the parts I've collected are precision capacitors. For this application, I selected a 0.01216 µF ±0.5% tolerance capacitor I have in abundance. This allows the use modest of charging currents to support a 5 volt sweep at 15.15 µS ≤ ∆T ≤ 22.73 µS. These are the half-periods of the multivibrator, to support 22-33 kHz square wave output frequency. The charging peak current was selected at ≈4 mA, making the minimum current needed ≈2.67 mA. These current ranges are also such that IC/ß or base currents do not offset the current range significantly. This is the final circuit that emerged, and it worked excellently:
Testing was done by using a narrow pulse from a Tektronix FG-502 at ≈5 kHz to drive the ramp reset circuit, for ramp down generation, while using a HP 33120A synthesized signal generator to provide low frequency square wave signals from +0.0 and +5.0 into the VCCS input. The effect of this setup was to allow seeing the differential in IOUT current appear as a different sweep rate on the monitoring oscilloscope. As well, the effects of the adjustment potentiometer could be seen, changing the scale of currents for charging the capacitor. Some scope fotos for testing are next:
Slower ramp down sweep rate for VI = 0.0V.
Faster ramp down sweep rate for VI = 5.0V.
Comments
Post a Comment