Visible Light Oscilloscope Probe

[graybeard]

I describe the design and breadboard construction of a visible light probe for an oscilloscope that is useful for measuring light waveforms such as flicker from LEDs, florescent, and incandescent lights.  It can measure fast switching light waveforms from LED lights and photo strobes (flashes).  It as also useful for measuring the output waveforms of infrared remote controls.

I go over my initial design specifications, how I chose the key components, and the details of how the circuit works.  I go over the input offset sampling noise  of zero-drift operational amplifiers (op-amps) and show how that effects circuit performance.

I demonstrate measurement of the fast switching light waveform from an underwater flashlight, the measurement of a IR remote control, and the flicker waveforms from a LED and florescent light.

You can download a copy of the slides here

[NiHaoMike]

So basically a higher bandwidth version of the "light to sound" kit.

I wonder if a USB version could be made with a direct sampling RTLSDR. Could be useful when combined with a smartphone to evaluate flicker of lighting. (A really cheap version would just be a phototransistor or photodiode connected to a headphone jack, but smartphones are slowly phasing that out...)

[james_s]

That's neat! I think I might build one of those. In the past I have just used a small photovoltaic panel connected directly to the scope and it worked well to look at the flicker waveforms of various lamps but I have no idea what the frequency response is like or how linear it is. Sure would be nice if that noise could be reduced without needing to filter the output though.

[not1xor1]

You can download a copy of the slides here

interesting project, but I can't imagine anything more printer unfriendly than a black baground PDF...
You must be paid by the printer ink lobby 

[graybeard]

That's neat! I think I might build one of those. In the past I have just used a small photovoltaic panel connected directly to the scope and it worked well to look at the flicker waveforms of various lamps but I have no idea what the frequency response is like or how linear it is. Sure would be nice if that noise could be reduced without needing to filter the output though.

You can use a more standard high speed op-amp instead of a zero-drift op-amp, but then you will need to add an offset adjustment to compensate for the input voltage and current offset of the op-amp.  You may also need to add some resistors in series with the non-inverting inputs to compensate for input bias currents of bipolar op-amps.   The only reason I did not do that was my desire to build a unit that did not require offset adjustment.

[tggzzz]

I'm glad you provided slides rather than a video

Why didn't you use a reverse-biassed PIN photodiode as the sensor? They are very fast and linear.

What was the frequency response you required and obtained?

You mention "...slower decay at low levels probably due to phototransistor...". You could prove or disprove that by simply modulating the current through any LED that doesn't have phosphors; I'm sure you have a plain red LED somewhere in your box!

For more information on photodetectors, see TAoE3 section 8.11.

[graybeard]

Why didn't you use a reverse-biassed PIN photodiode as the sensor? They are very fast and linear.

I chose the phototransistor because I wanted the peak response to be in the visible light range, the peak response of PIN diodes are in the infrared.
You can replace the phototransistor with a reverse biased PIN detector and the circuit will work.  However I tuned the gains for the phototransistor and they might need to be adjusted for a PIN depending it's sensitivity to light.

[tggzzz]

Why didn't you use a reverse-biassed PIN photodiode as the sensor? They are very fast and linear.

I chose the phototransistor because I wanted the peak response to be in the visible light range, the peak response of PIN diodes are in the infrared.
You can replace the phototransistor with a reverse biased PIN detector and the circuit will work.  However I tuned the gains for the phototransistor and they might need to be adjusted for a PIN depending it's sensitivity to light.

The peak response depends principally on the material.

Some devices have suppressed IR response. Some devices have a response that mimics the eye. That tailoring is achieved by optical filters that reduce the peak sensitivity, which usually doesn't matter since gain is cheap.

If you don't need speed, then a reverse-biassed ordinary photodiode can be a good choice. The reverse bias minimises device capacitance and makes the photon-to-electron conversion ratio stable and predictable.

There will be a DC leakage current depending on temperature, area, and bias voltage. Frequently that is small enough to be neglected, or if it cannot be distinguished from background illumination, it has to be dealt with by other means.

A quick google indicates Hammatsu has a wide range of photodiodes: https://www.hamamatsu.com/resources/pdf/ssd/si_pd_kspd0001e.pdf

[graybeard]

Why didn't you use a reverse-biassed PIN photodiode as the sensor? They are very fast and linear.

I chose the phototransistor because I wanted the peak response to be in the visible light range, the peak response of PIN diodes are in the infrared.
You can replace the phototransistor with a reverse biased PIN detector and the circuit will work.  However I tuned the gains for the phototransistor and they might need to be adjusted for a PIN depending it's sensitivity to light.

The peak response depends principally on the material.

Some devices have suppressed IR response. Some devices have a response that mimics the eye. That tailoring is achieved by optical filters that reduce the peak sensitivity, which usually doesn't matter since gain is cheap.

If you don't need speed, then a reverse-biassed ordinary photodiode can be a good choice. The reverse bias minimises device capacitance and makes the photon-to-electron conversion ratio stable and predictable.

There will be a DC leakage current depending on temperature, area, and bias voltage. Frequently that is small enough to be neglected, or if it cannot be distinguished from background illumination, it has to be dealt with by other means.

A quick google indicates Hammatsu has a wide range of photodiodes: https://www.hamamatsu.com/resources/pdf/ssd/si_pd_kspd0001e.pdf

I have use the hsmamatsu parts for other projects, but they are ten to fifty times the price of the $0.80 phototransistor I chose.

[tggzzz]

Why didn't you use a reverse-biassed PIN photodiode as the sensor? They are very fast and linear.

I chose the phototransistor because I wanted the peak response to be in the visible light range, the peak response of PIN diodes are in the infrared.
You can replace the phototransistor with a reverse biased PIN detector and the circuit will work.  However I tuned the gains for the phototransistor and they might need to be adjusted for a PIN depending it's sensitivity to light.

The peak response depends principally on the material.

Some devices have suppressed IR response. Some devices have a response that mimics the eye. That tailoring is achieved by optical filters that reduce the peak sensitivity, which usually doesn't matter since gain is cheap.

If you don't need speed, then a reverse-biassed ordinary photodiode can be a good choice. The reverse bias minimises device capacitance and makes the photon-to-electron conversion ratio stable and predictable.

There will be a DC leakage current depending on temperature, area, and bias voltage. Frequently that is small enough to be neglected, or if it cannot be distinguished from background illumination, it has to be dealt with by other means.

A quick google indicates Hammatsu has a wide range of photodiodes: https://www.hamamatsu.com/resources/pdf/ssd/si_pd_kspd0001e.pdf

I have use the hsmamatsu parts for other projects, but they are ten to fifty times the price of the $0.80 phototransistor I chose.

That was an example of the range of characteristics that is available. No doubt there are cheaper ones available.

The more interesting points are about how you use the photodetectors.

[StillTrying]

Looking at the light shape of light sources is an interesting idea, I wish I'd thought of it.

[Zero999]

For high speed, high intensity applications, such as monitoring very bright pulses from a flash lamp, just use a photodiode and resistor. I just used a 50 Ohm terminator, at the end of a co-axial cable connected to an oscilloscope. Here's a schematic of the set up I used to drive a high power LED, with very high current pulses and sense them with a photodiode.

https://www.eevblog.com/forum/projects/ultra-short-ultra-fast-led-flash/msg2247069/#msg2247069

[graybeard]

[not1xor1]

I'm not an EE and have not much experience, so I wonder if there any particular reason to put the switch on the negative of the battery and to connect capacitors from the virtual 2.5V ground to both the positive and the negative supply rails.

In single supply circuits I'm used to put a switch on the positive rail and capacitors between the virtual ground and the negative and between the positive and the negative of the battery rails (close to the supply terminal of the opamps).
I realize your circuit replicates the usual schematic of real ground circuits. So I wonder if I'm wrong or if that makes any difference. 

I think batteries have low noise, but with a noisy single supply voltage, wouldn't capacitors from the positive rail to the shunt regulator increase the virtual ground noise?

[graybeard]

In single supply circuits I'm used to put a switch on the positive rail and capacitors between the virtual ground and the negative and between the positive and the negative of the battery rails (close to the supply terminal of the opamps).
I realize your circuit replicates the usual schematic of real ground circuits. So I wonder if I'm wrong or if that makes any difference. 

The battery and the switch are in series, the order of components in a series circuit does not matter, only the orientation (of some components) needs to be maintained.  Thus it makes no difference where the switch is, if you like it better on the positive battery terminal you can put it there and it will function the same.

I think batteries have low noise, but with a noisy single supply voltage, wouldn't capacitors from the positive rail to the shunt regulator increase the virtual ground noise?

The capacitors are there to provide a low impedance return path at high frequencies for the signals flowing onto the voltage reference node through R3 & R5 since the impedance of the shunt bandgap reference U3 increases with frequency do to the roll of of it's internal feedback.  The battery starts as a low impedance, but that increases as it discharges.

I did not really pay that much attention to the decoupling capacitors.  I just put them in and only paid attention to the 10µf limit of capacitance in parallel with U3.  I suspect the circuit would work fine C1 through C4 removed.   If I was going to mass produce this I might look into removing them to decrease the parts cost.

The simple answer to your noise question is yes, but the real answer is more subtle.  You would need to look at the impedance of U3 vs frequency and factor that with capacitors, the sensitivity of the resistivity of Q1 to V_CE changes, the impedance of the battery vs frequency and discharge state, and the power supply rejection and balance that against the high frequency current return paths.  I did not do that analysis since I did not think it was important in this case.  However it might be in another case.


Both of your questions are good questions, thank you for asking.

[not1xor1]

In single supply circuits I'm used to put a switch on the positive rail and capacitors between the virtual ground and the negative and between the positive and the negative of the battery rails (close to the supply terminal of the opamps).
I realize your circuit replicates the usual schematic of real ground circuits. So I wonder if I'm wrong or if that makes any difference. 

The battery and the switch are in series, the order of components in a series circuit does not matter, only the orientation (of some components) needs to be maintained.  Thus it makes no difference where the switch is, if you like it better on the positive battery terminal you can put it there and it will function the same.

I meant that I thought the contact resistance of a real switch on the negative rail might increase the noise at the terminals of the TL431 while if it were on the positive rail that would just be in series to the resistance. But I guess I'm just paranoid ... 

[tggzzz]

The battery and the switch are in series, the order of components in a series circuit does not matter, only the orientation (of some components) needs to be maintained.  Thus it makes no difference where the switch is, if you like it better on the positive battery terminal you can put it there and it will function the same.

You have to include all the components for that statement to be valid. In some circuits the stray capacitance is critical, and in others a wire's inductance is important (rule of thumb for wires is 1nH/mm). Higher voltages and/or higher currents make those points more important.

Whether that is relevant to this circuit is left as an exercise for the student

[Yansi]

What is the bandwidth of such probe?  Guessing not much...

I'd use a PIN photodiode in photovoltaic mode (shorted, zero volts across it) with a transconductance amp instead of this.

[graybeard]

What is the bandwidth of such probe?

The bandwidth is ~500KHz (700ns rise time) which is fast enough for the target application.

[Yansi]

That seems quite good. Have expected about 100k.