# Amplifying pulse signals causing pulse lengthening problems

I am currently trying to design a circuit that would sense a voltage change at its input, convert that change into rapid "on"/"off" change at output and amplify same switching voltage to high current (up to 10A or so), which would flow through through low ohmic inductor.

WORKING PRINCIPLE: (using well-known circuit)

First things first, I am using IR sensor for input device. It is basically an IR diode, which is receiving IR light from another IR diode that is few centimeters away from sensing IR diode. As and obstacle moves into the light path between two lined-up diodes, the sensing diode drops voltage across its terminals to few millivolts or so. As obstacle moves away, light path between two diodes is established and sensing diode drops few hundred millivolts or so. Sensing diode is therefore connected to comparator, which compares IR diode's voltage drop with reference voltage (defined by voltage divider). Operational amplifier compares two voltage inputs and defines the output state (high/low or on/off). Its low current capability output is driving high current capability Darlington BJT configuration power stage. Inductor (load) is connected in series between power supply and Darlington's collector. Reverse biased diodes are used to protect transistors junctions from voltage breakdown due to high voltage spike caused by inductor.

NOTE: I am controlling output current with operational amplifier rather than just with IR sensing diode connected to input of BJTs because diode's "on" state only produces only few hundreds of millivolts, while operational amplifier outputs "on" state that rises up to rail voltage (at unloaded condition), which turns on BJTs hard (which is desired, in order to minimize Q3's and Q4's $$\ V_{CE(sat)} \$$ and allow sufficiently high voltage drop across the load and therefore desired current flow).

simulate this circuit – Schematic created using CircuitLab

C1 and C2 are decoupling capacitors, R5 and R6 are used for current sharing, P1 is used for reference voltage setting adjustment and R4 somehow decreases oscillations at higher frequencies (as output pulse signal changes rapidly, overshoot and ringing occur). Note that this is what actual circuit looks like with actual components that were breadboarded. Also, I didn't use conventional resistor from each BJT's base to another BJT's base for all transistors, because that somehow decreases high frequency performance of this circuit (R4 was chosen and put there experimentally).

This circuit was meant to be designed in such manner that it would work properly only up to few kilohertz. If a pulse of specified length occurs at input of comparator, then a pulse of same length should be seen at output but amplified in current. However, this circuit does not act exactly as desired.

PULSE LENGTHENING:

First of all, to simulate object passing through (and interrupting IR light emitting path), a function generator was used (connected to node $$\A\$$), producing pulse signals.

If P1 is set correctly, the circuit behaves as desired (with no ringing and overshooting "on"/"off" state) up to 2 kHz. However, beyond that value, pulse signal measured at node $$\B\$$ starts to progressively lengthen "off" state (which results in longer $$\V_{CE}\$$ and shorter $$\V_{LOAD}\$$ "on" pulse). At approx. 10 kHz distortion of output signal progresses so far that $$\V_{CE}\$$ stays constantly "on", meaning $$\V_{LOAD}\$$ stays constantly "off". Such behavior is completely undesired.

From provided pictures you can see that at 2 kHz pulses are all good but at 3 kHz you can already notice slight pulse lengthening. I wonder, why is this happening at such relatively low frequencies. I firstly assumed that this would start occurring after few tens of kilohertz. Also, the idea was for this circuit to work as desired up to 10 kHz at least.

CONCLUSION:

As I haven't gained any experience with any kind of switching circuit so far I cannot tell what is going on here. Or how to approach to such problem. Is some kind of compensation needed for desired performance of such circuit? Or should I be choosing different components (such as different operational amplifier)? Slew rate of operational amplifier (which usually presents an obstacle for high frequency signal) isn't so low that it would present an obstacle in such relatively low frequency circuit. The slight overshoot followed by ringing (which might not be seen here, but is presented) is not the issue, it is the pulse lengthening, which is obvious from provided pictures. This circuit was breadboarded with minimum possible usage of jumper wires and conductive paths between individual components were as short as possible. And I am positive that making an actual PCB for such circuit wouldn't solve mentioned issue, as this is not some ultra high frequency circuit.

• Another thing to consider is whether you have problems with the power supply. I suggest using a separate supply for the high-current load, to isolate its effects from the supply for the op amp. – Elliot Alderson Jan 25 at 20:12
• I suggest using a comparator instead of an op amp. Op amps are slow to recover from saturation. – Hearth Jan 25 at 20:12
• @Hearth is not being facetious. There's a category of parts called comparators. Superficially they appear to be op-amps, their circuit symbol is the same, and much of the internal circuitry is the same. But where an op-amp is designed as a linear amplification device, a comparator is designed as an on-off device. A cheap old jelly-bean example is the LM393. Note that comparators often have open-collector outputs; you need to decide if you need open collector or push-pull, and choose appropriately. – TimWescott Jan 25 at 20:46
• Suspect the diode, too. You have it in photodiode mode; I would expect it to ramp up to 0.5V or so rapidly, and then have that voltage decay slowly. It would be much better to keep it reverse-biased and set a threshold on the current actually flowing in it. This could either be done in a single stage with a comparator only, or a bog-standard photodiode amplifier circuit (using an op-amp) followed by a comparator. – TimWescott Jan 25 at 20:54
• Move collectors of Q1,2 to V+in add 1 Ohm to emitter , change D4 to a 20A power diode or several 1N400x or change design to a power FET. IR detector will have a slow Off decay time due to this high impedance on D2 (PD) many other improvements to denoise comparator – Sunnyskyguy EE75 Jan 25 at 21:14