Measure flash duration with a Photodiode and Transimpedance Amp

[nightfire]

After hopefully completing my small home lab by end of the month, my first project will be to build some measuring device geared towards measuring the burn duration of a camera flash.
Some other folks have done this already:

Background: I am a hobby photographer that dives deep in the technology behind all that lighting stuff to get a better and deeper understanding of the limits of the gear used.
(Ok, and I have official certification to work on live electric circuits up to 1000V AC, and also can do security checking of appliances- but the world of electronics is something I want to dive deeper into)
For freezing moving objects the overall burn duration is crucial, also if you do things like high-speed-sync of flash and camera shutter (Supersync/Hypersync) to work around the limitations of the highspeed strobe of HSS you really need to know the characteristics of your gear...

To my understanding, we are talking about 2 crucial parts of this measurement technique:
a) the photodiode itself and its rise time
b) the quality of the (transimpedance) amplifier to present voltage to the oscilloscope, that also has a rise time delay that adds to the delay of the photodiode

Later, when breadboarding is finished and I have solid impression of the tolerances, the whole stuff shall get soldered onto a PCB.

Questions here:
- I stumbled upon some PCB for individual projects: https://www.pollin.de/p/platine-universal-operationsverstaerker-390078  - are those any good, or are there some other (and maybe better) options?
- Are there maybe ready-made PCB available, where I just put in my Photodiode and am ready to go? (Of course, being affordable below the 50 €€€ mark...)


For a begin, I ordered some OPT101 PCB intended to get hooked up to an Arduino to play with and to get some feeling to that whole scenario. I know that the OPT101 is somewhat slow (28 µs rise time) compared to other available Photodiodes like the BPW34, but the OPT101 has the whole amplifier integrated which will save me lots of hassle in the beginning.

[RoGeorge]

Nice project!   

If you want to eliminate any doubt about the rise and fall time, or the saturation level, then eliminate the electronic part.  Use a rotating mirror to project a beam from the flash to a wall, and photograph the wall with an infinite exposure time.

This way the wall is turned into an oscilloscope screen for light.  Turning the rotating mirror slower or faster would be similar to changing the timebase of the oscilloscope.  Should work, but I never tried that.

[StillTrying]

I think he's saturating the photo diode every time in that video, which would mean you can't see the half light width, and the fall time wouldn't be very accurate, you can see a strange kink before it starts to come down.

With the amount of light from a flashgun you can just about get away with just the photo diode and a 180R in parallel, keep the peak <250mV for the best shape.

There's a high bandwidth capture of a 25us flash here, I've done 100s of flash gun captures over the years.
https://www.eevblog.com/forum/chat/20w-halogen-bulb-viewed-by-a-photodiode/msg1318658/#msg1318658

[T3sl4co1l]

I did this a long time ago, which unfortunately I don't have schematics of, wait maybe I'll check, but I verified the photodiode response, using an LED (LEDs are quite fast, some MHz of modulation bandwidth) to be sure the flash was measured correctly.  I also cross-verified it with a current transformer, or inductive probe of some sort.

Let's see here...

This was I think, inductive pickup of a xenon flashlamp discharge.  Probe would've been a common 10x scope probe, tip to ground clip, and the loop held by the capacitor/lamp.



The induced EMF is generally the derivative of current, so, this implies a unipolar pulse on the order of 20us.

Another:



This might've been photodiode, or CT.  Not sure if it was the same setup, or a different capacitor.

I think I had tried several parts, including an axial capacitor (looong pulse), snap-in type (amazingly short pulse, maybe the first image?), and film (in-oil, metal can, 20uF, actually rated for some current not just a motor-run cap, maybe the second image?).

Ah yeah, here's the schematic,



That type of photodiode I think is common in old remote-control receivers, the non-integrated kind?  No idea of specs.  It's black, so it doesn't respond much to visible light, but I got enough response from a red LED to correct the pulse, hence the RCs.

I remember also trying it with a high voltage cap and air spark, with pulses in the 50ns range.  Don't remember if that was attributable to EMI from the spark, or actual optical output...

In any case, I did seem to confirm that conventional xenon simply doesn't switch faster than some microseconds.  Which is indeed one of the problems Edgerton originally had.  And air sparks were a solution, downside being the somewhat bluer color I think, not to mention harmful UV.

In any case, for the photodiode -- expect quite large photocurrents, even for very short pulses.  Flashes are intense!  You won't need much transimpedance to see it, heck even a low impedance (biased photodiode direct-driving 50 ohms?) may be good enough!

Tim

[bdunham7]

To my understanding, we are talking about 2 crucial parts of this measurement technique:
a) the photodiode itself and its rise time
b) the quality of the (transimpedance) amplifier to present voltage to the oscilloscope, that also has a rise time delay that adds to the delay of the photodiode

You need to consider the trailing edge response as well.  Saturation can cause extended fall times.  You should aim for a response time measured in nanoseconds, PIN photodiodes with 50ns response times are easily available.

https://www.mouser.com/datasheet/2/308/QSB34-D-1814747.pdf

It is an interesting project, but you should consider the possibility of measuring the flash speed in another way for comparison.  RoGeorge suggested one, I would suggest just photographing a fast-moving (probably rotating) object with a reflective point and seeing how fast you have to move it to get a motion smear.

[tggzzz]

Since you have control over the flash and measurement, I would try the simplest fastest circuit first.

Reverse bias a PIN diode, and have a series resistor to measure the current. Higher resistor will be more sensitive, but slower (RC time constant with diode's capacitance).

Since you are interested in nanosecond risetimes, you will need good high frequency construction techniques. Keep wires very short and have a groundplane. Forget using solderless breadboard.

[S. Petrukhin]

What equipment do you use for the flash? I don't remember my old SB-800 anymore, but if I'm not mistaken, the glow intervals are normalized and indicated there. And you may also find a pre-flash, which is done for measuring.

[Kleinstein]

As the Photo-flash has reasonable higher power, one can get away without an amplifier. Just photodiode, 50 Ohms resitor to ground and coax cable to the sope should be sensitive enough. For higher speed add reverse bias (e.g. 9 V battery) for the photodiode and a cap to bybass the battery for the pulse.

For a test one could try some LED with a high current pulse (e.g. some 100-1000 mA for some 10 µs). Some red / IR leds are relatively slow with a decay time of a few µs, other LEDs can be quite fast, down to the ns.

[Peabody]

My understanding is that setting the flash to full power discharges the capacitor fully, but for lower power settings it cuts off the discharge increasingly early.  So setting it to low power gives you a shorter duration flash, which is better for capturing moving things, and a quicker recharge time.  It would be interesting to see if your measurements confirm that they work that way.

[mawyatt]

Back in 2016 we did some measurements on various DSLR camera speedlights and strobes in support of our ongoing chip imaging efforts. At first just used a 1N4148 diode since any PN junction is light sensor. We weren't interested in really fast or accurate edges, just doing comparisons.

https://www.photomacrography.net/forum/viewtopic.php?f=25&t=29729&hilit=Strobe+Duration

Later when we started the LED strobe developement we also used LEDs as sensors.

https://www.photomacrography.net/forum/viewtopic.php?f=25&t=40999&hilit=LED+Strobe

Modified Godox Video LED for strobe use.

https://www.photomacrography.net/forum/viewtopic.php?f=25&t=41353&hilit=LED+Strobe

Also modified IKEA Jansjo for more optical power and strobe use.

https://www.photomacrography.net/forum/viewtopic.php?f=25&t=41464&p=260928&hilit=LED+Strobe#p260928

If you are interested in the optical edge speed & fidelity, then consider using the photo diode sensor in a manner which removes the junction capacitance effect from the output. This method employs utilizing a fast op-amp and placing the photo diode from the inverting input to a fixed (clean Vbias) negative voltage, the non-inverting input is analog ground. A resistor in the feedback from op-amp output to inverting input keeps the diode cathode at virtual ground and keeps a constant reverse voltage across the diode of Vbias. Since the diode voltage generally remains constant unless the opamp output saturates, then there is no dv/dt across the diode to cause displacement current in the diode junction capacitance, thus no junction capacitance current. This effectively removes the diode junction capacitance to a 1st order. Op-amp stability can be an issue due to the feedback time lag caused by the feedback resistor and diode capacitance, and must be addressed, usually a small feedback capacitance shunted in parallel with the feedback resistor suffices. For very high speeds a more complex setup is required and likely beyond the scope of this discussion for strobe use. There are some methods of using a single supply voltage and a bipolar transistor for an approximation to the above op-amp solution of keeping a constant voltage across the photo diode.

The Xeon strobes used in cameras are an interesting device that produce enormous optical bursts for short durations. They work on the principle of a initial ignition voltage burst of several KV derived from the strobe trigger pulse and voltage scaled with a small transformer to the KV range. This voltage is impressed across the Xeon gas by a small wire (or small thin conducting plate) wound around the Xeon tube, this voltage starts the Xeon gas ionizing process which is regenerative. A high voltage (usually ~400VDC) stored in a large capacitor is connected across the main Xeon tube terminals. Once the trigger ionizing process starts, this voltage sustains the complete gas tube ionizating which happens quickly. The capacitor discharges into the Xeon gas tube with an energy of (CV^2)/2, which is how the strobes are rated (Joules or Watt*Seconds).

BTW watch out for photo diode saturation, which can easily happen with these high energy strobes.

Anyway, looks like a fun project and keep us informed of your progress.

Best,

[Kleinstein]

The transimpedance amplifier are the way to go for good speed at high sensitivity. The speed is still limited by the OP - usually quite a bit slower than the GBW of the OP.

With the very high intensity of a flash one can go old style with a relatively low value resistor, like 50 Ohms. Some 5 pF (a relatively high value for a small photodiode, with bias) from a photodiode and 50 Ohms give only 250 nV time constant - one needs a pretty fast OP to make a TIA that fast. As a nice side effect the resistor can also be used as termination for the cable.  If there is a termination at the scope side too, the phodiode sees 25 Ohms and this a even larger speed.  The photflash should be powerfull enough to get a visible signal even with such a low resistance.

[mawyatt]

That low value resistance should work if the photo diode doesn't saturate. Back 2015~6 when we were evaluating camera strobes using a simple diode required a much higher resistance than 50 ohms to get a good scope trace. Using the LED as a sensor provides a higher output but also has a higher capacitance to deal with.

Another viable option that is simple to implement is using a pnp (2N3906) configured as a common base stage and connect the diode from the emitter to ground. The scope is connected across the collector to ground resistor. The emitter voltage doesn't move much (~18mv/octave) with diode induced current, so the junction capacitance displacement current should be small compared to the total diode current, and no feedback stability issues.

Anyway, fun project for the OP, and lots of options!!

Best,

Edit: Added schematic.

[jonpaul]

Most consumer flash have a duration depend on the power, ~ 10 uS to 1 mS.

As a historic note,

Dr Harold Eddgerton (EGG, inventor of flash) could achieve 1 nS risetime for the AES.

https://en.wikipedia.org/wiki/Doc_Edgerton

See the book Electronic Flash by Harold Edgerton.

The fastest sensing is with photomultipliers


Enjoy,


Jon

[mawyatt]

Most of the older photographic type studio strobes used the capacitor charge voltage as a means to control the optical flash energy. Controlling the flash energy by way of pulse width was generally with lower energy camera speedlights since this involves interrupting the very high currents flowing thur the Xeon tube during ionization. Lower energy levels produced lower currents which could be handled by IGBT technology that was moderate cost, as this technology became more affordable some of the higher end studio strobes began to use pulse width as a control, and now has trickled down to moderate cost studio strobes. The lower end cost studio strobes still use capacitor voltage control, with some clever circuits involving the voltage doublers and triplers that are direct mains connections. The Adorama Studio Strobe 300 is an example. Note used a Lecroy DSO Later found out it was a rebranded Siglent, which is what I have now (SDS2102X Plus) 

https://www.photomacrography.net/forum/viewtopic.php?f=25&t=38830&hilit=Adorama+Studio+Strobe+300 

Best,

Edit: Gas ionization is an extremely fast effect as it's an avalanche type function, and sort of feeds off itself. This is why these Xeon tubes can be "ignited" with a very low energy HV pulse. One of the old clever means to gate the Xeon tube off was use another gas tube in parallel with the main optical tube, but this second tube was not optically exposed and hidden. This tube was ignited to extinguish the main tube by sucking the capacitor energy away and allowing the capacitor voltage to quickly drop below the Xeon gas sustaining voltage. So in use the main tube was ignited to start the exposure and the second "hidden" tube ignited to extinguish the main tube and terminate the exposure. Believe this was used until moderate cost solid state switches became available.

[nightfire]

Wow! Many thanks to all comments here- Now I can choose between several setups to get some feeling for the specs of different diodes and some practice in getting my new scope going ;-)

To answer a few questions: I have a Nikon setup, so two SB-800 hanging around. And then some el cheapo Yongnuo YN465, an old, but trusty Metz 54MZ4i, and as big iron a Godox Witstro with 360 Ws barebulb flash.
The regulation of flashes generally is divided by two basic principles: Studio flashes mostly (at least in the past) were charged up to a certain voltage and released all of it whilst burning. If you wanted to turn them a bit down, you would turn them down and press the test button to discharge the capacitors (in German: "abblitzen").
On-camera flashes with less power could use modern electronics to cut off the discharge at any convenient point. Nowadays, some studio flashes (and/or with big flash generators) are also manufactured in a more modern way since the necessary components have become available and reliable. I think Broncolor was one of the first to offer that with their highend (read: expensive) line...

In my experience, flashes that were designed to run cool, have a long burning duration. My Metz 54MZ does  around 1/200s on full power, some cheap chinese studio flashheads can sometimes burn even longer... An SB-800 at full power is around 1/900s (if i remember that correctly from some datasheet)- this is also the reason why after 20 flashes on full power some cooldown is specified in the manual...

Why is that important? If you want to photograph moving things, you usually will want a short burn duration. For other stuff, like doing extremely short exposure, that is shorter that the flash sync time of your camera, you can do tricks like having the flash on full power (=long burn time) and then do a very short exposure around 1/1000s or shorter.
Attention; due to the working of the mechanical shutter of a camera ((D)SLR etc.) the shutter needs 4ms or longer to move from top to bottom, which also goes for the 2nd curtain that closes after the first one has opened. This means, that on very short exposure time the 2nd curtain will begin to run very shortly after the first has begun, and not yet finished. So even a "1/1000s exposure" will need 4ms (=1/250s) to get finished...

Attention: With mirrorless cameras and their electronic shutters it is far worse: 20ms for reading the whole sensor is not uncommon for them, so that is the reason why flash photography will usually not work with electronic shutters...

[mawyatt]

We developed a bypass for the Nikon Electronic Curtain cameras for use in our image stacking endeavors, these can involve taking 20,000 images to render one final gigapixel like macro chip image. Nikon blocks the hot shoe flash trigger when utilizing the Electronic Rear curtain to terminate, in the all electronic curtain mode. Believe they do this because during the long image readout without the benefit of a mechanical curtain to shield the sensor an intense optical input could corrupt or even possibly damage the sensor. Not sure of the damage mechanism but this could be related to large photon burst causing a PN junction to become forward biased which could in turn forward bias another PN causing a latch up condition on the sensor chip. This was called SCR latchup in early CMOS where an over voltage or photon induced effect caused the chip to latchup and draw high currents from the supplies destroying the chip. Also high density radiation effects caused this and Flash X-Ray was used to test chips against this effect for use in radiation hardened applications. The sensor exposure to a flash type event (when using the fully electronic mode) during readout may be reasoning behind Nikon's decision to block the hot shoe.

If you spend some time studying over at the Photomacrography sites you can find all sorts of threads related to strobes, speedlights and LEDs for macro use. You can search using the author as mawyatt as I've done some of this work, but many others have contributed.

Best,

[nightfire]

Update: In the meantime my new (and first DSO) scope has arrived- a Rigol  DS1102Z-E that currently can be obtained for some low money...

I was able to do some breadboard fiddling with an OPT101 (integrated 1MOhm resistor and opamp) and was able to see times for flash durations of an embedded in-camera flash of about 1/4000s, which would fit the bill and be plausible.
Also, I ran into probably some capacitive effects or swinging with the opamp, but as discussed before, this behaviour in such setups is not unexpected.

The next weekend I will probably spend exploring my scope with the trigger options, and put together my basket to order some stuff like some photodiodes and connectors...

[StillTrying]

some photodiodes

I didn't realise you didn't already have a photo diode and scope to play with.

I use the clear visible light SFH 213  because 2 or 5 of them are usually cheap,   for the looking at pulsed LED's I've done they're near enough electrically the same as the much more expensive BPX65, haven't tried BPV10.

I'll try capturing a 200us wide flash with just a photo diode + resistor when/if I get a round to it.