response time

[girishji]

I am using a PT at low current. I was able to achieve impressive rise time (<10us). But I cannot do this at any reasonable frequency. If I switch IR on and off at frequency >1000hz the square wave misses in regularity (see last pic for general idea). It is unusable.
Any suggestion of how I can increase frequency while keeping rise time low?

[Ian.M]

That's nearly as good as you'll ever get with that phototranstor as it doesn't have an accessible base lead - the phototransistor 'sees' an almost constant voltage across it (varying from Vcc-Vbe_(TR1) to Vcc) so isn't externally slowed down by slewing a significant load capacitance.   A genuine 2N3904 has an f_T of >250 MHz so you shouldn't be running into any limitations of the transistor (TR1) till above 1Mhz.

A very small improvement could be gained by using the phototransistor as a switched  current source into a virtual ground, namely the inverting input of a very fast OPAMP, but its not worth the expense to gild the turd!

[jonpaul]

fast photo  diodes,PIN, etc can reach nS réponse but very costly.

Cheap consumer photodiodes are inherently slow.

[wraper]

If you need speed, better use something like this https://www.ttelectronics.com/TTElectronics/media/ProductFiles/Datasheet/OPL530-550-560.pdf or make a photodiode based circuit.

[Ian.M]

What do you get if you use a 50 ohm resistor as a load for the phototransistor, with no buffer stage (reducing the supply voltage if necessary to avoid excessive dissipation in the phototransistor)?

If you aren't happy with the rise and fall times in that test setup (even though the amplitude isn't what you want), there's no hope of improving the situation without causing light level dependent pulse width distortion.

Are you *SURE* this is a simple phototransistor not a more complex IR detector IC?  Do you have a part number or datasheet for it? 
1KHz does seem to bit low - unless its a very large area device, you shouldn't be running into trouble till approx. an order of magnitude higher pulse frequency.

How are you switching your IR source?

[schmitt trigger]

The schematic doesn’t show the R1, R2 values, nor does it mention the photo transistor's part number, nor how are you actually illuminating it, the optical coupling, etc. Several unknowns here.

[Kleinstein]

The circuit shown help to get a fast fall, but there can be quite some delay from the rising part. The rise time can still be good but there can be some delay.  A problem is driving the transistor or even worse the phototransistor into saturation (all the way on).
There are special gold doped transistors (mainly the old 2N2369 or as PN2369 MMBF2369) for a short storage time. A high ft is for small signal operation only. The time it takes to get out of saturation (storage time) is a different parameter. A high ft alone is not good enough when driven into saturation.
The alternative is to have some shottky clamping similar to the TLL 74LSxx ot 74Sxx gates (to rather small diode can be tricky to find).

[Ian.M]

Yes, as the junction capacitance of the clamping diode directly limits the slew rate by the Miller effect most general purpose small signal Schottky diodes are likely to make the risetime worse.

[BillyO]

You don't need a faster transistor.  You might be over driving the 2N3904 and causing to go deep into saturation.  You might want to try using a base resistor.  Try a value of 10 times R2 and go from there.

[T3sl4co1l]

I can't see anything... those screenshots are zoomed very far out and the aliasing is terrible.

Please show the input waveform in comparison (both on screen simultaneously, trigger on input).  Is the rising or falling propagation delay the problem?  This should be evident at any frequency, and as you raise frequency you'll clearly see the propagation delays overlap and then the output disappears.

How much does it depend on distance (between emitter and detector), or drive current?  Should it be tolerant of such variations?  Can you tune it for a particular condition?

What about ambient light, should it be tolerant of variations?  Or will it be in a black box, well shielded from that?

And what kind of signals are you looking to communicate here?

Tim

[T3sl4co1l]

Why assume when you can measure

Tim

[Kleinstein]

The circuit should not saturate the phototransistor: extra current just flows to the transistor base.
If one has more supply voltage available it can help to have the transistor in base configuration and ideally a smaller resistor at the output. So less signal and if needed add another amplifier stage.

Additional difficulties come up if the signal amplitude varies as one than needs saturation.

The the added capacitance from a BAT54 is quite large and this way slow down the amplifier.

[Damianos]

The schematic in the OP shows two abused transistors!
The phototransistor is driving almost a short (a p-n junction)... The base current of the second one is "limited" only by the power supply (when the first is saturated)! In the pictures it is shown an around 300mVpp output, that is probably the result of the shorted power rails through the transistors.

To start playing with it (if the transistors are yet in good health): Use as R₁ around 2KΩ; add a resistor in series with the base of the second transistor (I think something between 2 and 5KΩ will be good); Reduce the R₂ to 5KΩ or less, to reduce the effects of parasitics.

[Infraviolet]

You can get pretty fast photodiodes cheaply if you're able to sacrifice on sensitivity:
https://uk.farnell.com/osram-opto-semiconductors/sfh203-fa/photodiode-ir-filtered/dp/1212743
That sort of thing.

I'd then recommend using an op amp arrangement rather than a transistor for amplification of the photodiode's signal in to something useful.

I have used this sort of thing with appropriate op amp bandpass filtering circuits after it, for reading square waves of >100KHz without trouble. 1000Hz will be very easy, a decent phototransistor should be able to to that too (I've found phototransistors often to struggle to keep up above 10KHz).

[Zero999]

The schematic in the OP shows two abused transistors!
The phototransistor is driving almost a short (a p-n junction)... The base current of the second one is "limited" only by the power supply (when the first is saturated)! In the pictures it is shown an around 300mVpp output, that is probably the result of the shorted power rails through the transistors.

To start playing with it (if the transistors are yet in good health): Use as R₁ around 2KΩ; add a resistor in series with the base of the second transistor (I think something between 2 and 5KΩ will be good); Reduce the R₂ to 5KΩ or less, to reduce the effects of parasitics.

It depends on how much current is flowing through the transistors and the power dissipaiton. If the phototransistor's power dissipation and base current of Tr₁.

As the phototransistor never saturates, it's most likely Tr₁ which is saturating and being slow to turn off. A Baker clamp might help.

[BrianHG]

IR Photodiodes are cheap, like 0.50$ or so and you can get response times in the 5ns to 50ns range.
They are also very sensitive, so, it is recommended to add an external IR lens filter.
IR Phototransistors tend to be slow by comparison.

[BrianHG]

I used to use these: https://www.digikey.com/en/products/detail/liteon/LTR-516AD/121725

But at 50ns, the best I could achieve was around 5 megabaud with a bit of distance.  1 megabaud across the room using fairly strong  and narrow beam 940nm IR LEDs.  (This was 15 years ago...)

I had a linear 3 transistor amp.  The first transistor was a MPSH10 or MPSH81 for sensitivity and better speed.

Far cheaper, newer, 10x faster IR photodiodes shown here: (some below 5ns)
https://www.digikey.com/en/products/filter/optical-sensors-photodiodes/543?s=N4IgjCBcoMwCwyqAxlAZgQwDYGcCmANCAPZQDaIATDABwBsYERYlADAOyXsjN01wBOOjyqtqcYUUqUajSVTg0hAkZTrsOAVlWywcbVIES6iIjFaaYMJiHZwW3IgMpxpIsKytH3CTWJHsYEqscCKuspQqzPz2plRgmgny0tKCIAC6RAAOAC5QIADKOQBOAJYAdgDmIAC%2BRAC0lEggqJAlAK6EJOQgBiAq6XUg9SrQLVAdXaSQFBCDQ00zIKXFAARZABbEOcQAJqV7eBlEwmOlu-n1Hja5%2BSI5AJ5ZR5AgGDioNTVAA

[Infraviolet]

Bandpass is only for when you're looking for a modulated signal, I assumed you were doing that, but as you mention use as a light level etecting switch I'm not sure if it would apply any more.

Figures 3-29 and 3-30 here https://www.ti.com/lit/an/snla140d/snla140d.pdf have some exmaple cofngiruations.

Horowitz and Hill's Art of Electronics has some too.

And there's a good set of slides here https://ece.northeastern.edu/courses/eceu401/2007fa/lab5.htm about using op amps to process phototransistor signals, the techniques can work for photodiodes too.

My advice for giving TTL level output voltages would be to get a first stage, or several stages, of amplification to give you a voltage which rises to a clear level, any level, for light on, and drops to another level for light off (or vice versa), then use a comparator to give the TTL output.

[Doctorandus_P]

You can try the following circuit.

[Damianos]

The schematic in the OP shows two abused transistors!
The phototransistor is driving almost a short (a p-n junction)... The base current of the second one is "limited" only by the power supply (when the first is saturated)! In the pictures it is shown an around 300mVpp output, that is probably the result of the shorted power rails through the transistors.

To start playing with it (if the transistors are yet in good health): Use as R₁ around 2KΩ; add a resistor in series with the base of the second transistor (I think something between 2 and 5KΩ will be good); Reduce the R₂ to 5KΩ or less, to reduce the effects of parasitics.

It depends on how much current is flowing through the transistors and the power dissipaiton. If the phototransistor's power dissipation and base current of Tr₁.

As the phototransistor never saturates, it's most likely Tr₁ which is saturating and being slow to turn off. A Baker clamp might help.

You are correct (one more time)! It seems that I have mixed up irrelevant things...

[Zero999]

Try this.

EDIT:
If the BJT is just being slow and the phototransistor has enough gain, then why not omit the BJT?

[Terry Bites]

How about bit of positive feedback?