Help, please, selecting some Electronic Components

[Michael GilliganParticipant @michaelgilligan61133]

I recently found a simple circuit, which forms part of a more complex instrument but should be adaptable to serve my requirements:

.

It is published in this paper:

and is described thus:

.

Some of the components [including the expensive, non-inductive, 100ohm resistor] are specified, but the diode and the capacitors are not. … The noteworthy point being that this runs a very high speed pulse-train.

I am therefore struggling to identify appropriate components for my ‘shopping list’

I wrote to the “corresponding author” and he kindly replied … but he “could not recall” what specific components they actually used.

Would the electronics experts kindly suggest what exact components they might use in this innocent-looking circuit ?

My own application will run at a lower repetition-frequency than theirs, but individual pulses will still be in the region of ‘a few microseconds’.

… I presume that propagation delay and ESR will therefore both be significant, and that high-quality will improve the longevity.

Thanks in advance

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

[John HaineParticipant @johnhaine32865]

I can't imagine what D1 is supposed to do as it's always reverse biased. I doubt omitting it would make the slightest difference.

[VersabossParticipant @versaboss]

Hallo Michael,

I cannot see from the text above who the “corresponding author” is, but I know with high probabilityy who Susan Parker is. She is a (not very active) member of the model engineering mailing list, and also the constructor of a very high end audio amplifier.

If you need to know more, I might be able to find an email address.

Kind regards,
Hans

[Simon Williams 3Participant @simonwilliams3]

… and that 0.1 uF capacitor shorting the input pulse train to ground needs the attention of a pair of snippers.

There's also something a bit odd about having a decoupling capacitor (drawn as an electrolytic) connected reverse polarity across the op amp supply. So I wouldn't take the strictures of component selection too seriously.

R1 could be metal film, they're inherently non inductive and cheap and readily available, also the 3300uf capacitor will have some ESR and inductance at high frequency but that's why it's got a 1 uF capacitor in parallel. I'd go tantalum rather than electrolytic for the 3300 uF though I doubt if it's critical what value this is, and little bead ceramic capacitors for the others.

Don't feel I can offer an opinion on the LED, this seems to be more about the precise definition of the light pulse in duration and spectral purity (can white light have spectral purity?) more than its electrical properties.

HTH, hopefully Joseph Noci will lend us his expertise ere long.

Rgds Simon

[BazyleParticipant @bazyle]

A frequency of one meg is not exactly fast by today's standards. The C3s are to stop the op amp bursting into oscillation and though they will turn the input pulse into a spike that is the aim of the thing anyway. You can get some big electrolytics with a fast frequency response – look for ones with a high ripple rating. I think a 1uf tantalum bead will do and maybe give it a further .1 ceramic too. The extra diode is to protect T1 from reverse polarity from any inductive elements – probably found necessary the hard way.

[Neil WyattModerator @neilwyatt]

It doesn't state duty cycle, but i'm guessing it's less than 10% on?

The discharge capacitors need to be low ESR types for rapid discharge. I guess the 1uF is to help ensure a quick start to the pulse while the larger cap acts as a reservoir. I imagine it could be a fair bit smaller.

All the capacitors are drawn the wrong way round.

If you use CMOS instead of TTL to generate your square wave, you shouldn't need the MOSFET driver as CMOS at 5V guarantees 4.95V output.

In fact, use CMOS and run the whole thing at 9 – 12V and use shorter pulses.

Neil

[S KParticipant @sk20060]

The circuit looks intended to drive very high currents through the LED, maybe 100A or more.

But R1 is 100 Ohms and so that limits the maximum DC current to 0.05 A. Instead, the circuit relies on C2's stored energy to create a short, rapidly-decaying burst of current through the LED and T1.

My guess is that the D1 diode is just a representation of a parasitic reverse-polarity diode inherent in the LED, yes?

T1 has 6,500 pf of input capacitance, which is why U1 is inserted. The latter can drive T1's input capacitance in under 50 ns. T1 has an 80 ns rise / fall time. I wouldn't recommend using a standard CMOS input instead of U1.

[Michael GilliganParticipant @michaelgilligan61133]

Thanks for the thoughts, everyone

As their hardware project is ’open source’ I have just realised that this link will probably take you to a readable/downloadable version of the full document: **LINK**

https://link.springer.com/article/10.1007/s43630-021-00127-6

… I have also just noticed that this statement in my opening post is wrong:

[quote] My own application will run at a lower repetition-frequency than theirs [/quote]

… because they actually run the ‘white-light probe’ at 20Hz.

In fact, I need to run at a higher repetition rate … but this is probably irrelevant, because it is the brevity of the individual pulses that dictates the performance-requirement of the circuit components.

MichaelG.

.

P.S.

As I originally mentioned, the little circuit at Fig. 4(a) is the only part of this system that I wish to use.

Their specific choice of LED is not important to my requirement [they used it for its spectral content]

 

 

 

 

Edited By Michael Gilligan on 17/03/2023 04:14:03

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Bazyle on 16/03/2023 23:23:09:

A frequency of one meg is not exactly fast by today's standards. The C3s are to stop the op amp bursting into oscillation …

.

Thanks for that thought … But I should mention that U1 is a specific ‘High-Speed MOSFET Driver’ not a regular op-amp

**LINK**

https://uk.farnell.com/microchip/tc4420cpa/ic-driver-mosfet-6a-4420-dip8/dp/9762566

MichaelG.

Edited By Michael Gilligan on 17/03/2023 04:31:39

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Simon Williams 3 on 16/03/2023 22:43:32:

[…]

R1 could be metal film, they're inherently non inductive and cheap and readily available,
[…]

.

Available a little cheaper at some sources … but here it is at RS **LINK**

https://uk.rs-online.com/web/p/through-hole-resistors/3205028

Listed correctly, but with a photo ‘representative of the range’ 

MichaelG.

Edited By Michael Gilligan on 17/03/2023 05:07:11

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Versaboss on 16/03/2023 22:33:51:

.

Thank you, Hans … I have sent you a P.M.

MichaelG.

[Joseph Noci 1Participant @josephnoci1]

Posted by Michael Gilligan on 16/03/2023 21:53:37:

I recently found a simple circuit, which forms part of a more complex instrument but should be adaptable to serve my requirements:

.

 

Some of the components [including the expensive, non-inductive, 100ohm resistor] are specified, but the diode and the capacitors are not. … The noteworthy point being that this runs a very high speed pulse-train.

I am therefore struggling to identify appropriate components for my ‘shopping list’

I wrote to the “corresponding author” and he kindly replied … but he “could not recall” what specific components they actually used.

Would the electronics experts kindly suggest what exact components they might use in this innocent-looking circuit ?

My own application will run at a lower repetition-frequency than theirs, but individual pulses will still be in the region of ‘a few microseconds’.

… I presume that propagation delay and ESR will therefore both be significant, and that high-quality will improve the longevity.

Thanks in advance

MichaelG.

Even for a 1MHz rep rate, your do not need fancy resistors, etc…The Resistor is only 100 ohms, so any carbon film 1/2 watt will be just fine. Even a 1/2watt wirewound resistor of 100 ohms will do – it will have an inductance of around 150nh = 1ohm @ 1MHz , which is lost in the noise of the 100 ohms, and way above the pulse rate you might work at.

Make C1 a ceramic 1uf, nice and fast, and C2 can be electrolytic. Use 3 x 1000uf and that will reduce the ESR nicely. Ditch C3. That is a typo…U1 is essential.

However, running this circuit at 20Hz is nonsense. The RC time constant of R1-C2 is around 300millisec. 63% of the 5V supply is 3V, which is around the threshold voltage of a typical white led. So the voltage at R1/C2 junction will reduce and stabilise at the LED junction voltage and stay there as the pulse rate is increased. The current through the LED will then only be limited by R1, ie 50mA. Hi current pulse will only be evident below the RC time constant period…This will be modified by the period of pulse to some extent – a very short pulse – 10 of ns, won't discharge C2 completely, so the regulation point will be higher.

You said you want to increase the pulse rate – this will make it worse. For 50Hz rate, you need an RC constant of at least half that, 10ms or so, which would make R1 = 3ohms , making lower pulse rates instantly destructive…

What rate do you want to run at? This sort of circuit is remarkably non-linear – the LED current reduces fast with increase in pulse rate, and is highly dependant on source pulse width.

I think a short description of your final application might suggest a better circuit, or allow tailoring of this one to suite..(avalanche circuits are commonly used for this)

 

Edited By Joseph Noci 1 on 17/03/2023 06:11:00

[Michael GilliganParticipant @michaelgilligan61133]

Thanks for joining the discussion, Joe

Essentially: I want a short duration ‘strobe light’ to illuminate living specimens under the microscope

I currently have two different global-shutter cameras available for the microscope, and will want to use the highest realistic frame-rate [perhaps 80 frames/second] with the shortest-possible duration of flash.

Obviously there are trade-offs between these requirements, the intensity of light available, the magnification, etc. etc. and building a strobe is the first step in a long journey … This circuit appeared to be simple enough, and to originate from an authoritative source, so I felt that I should be able to develop it to suit my needs.

The discussion so far seems to cast doubt on even my first assumptions

It’s of no relevance, but, for anyone sufficiently curious … I want to image the detailed motion of the cilia on ciliate protists … which we usually see only as a blur, or with rolling-shutter artefacts dominant.

MichaelG.

.

Ref: **LINK** https://en.wikipedia.org/wiki/Ciliate

[Sonic EscapeParticipant @sonicescape38234]

I think that the 100nF (C3) capacitor to ground at the input of the circuit is wrong. At 10MHz that is equivalent to 0.16Ω.

Now I saw the previous message, for 80Hz it should be fine

Edited By Sonic Escape on 17/03/2023 07:05:00

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Michael Gilligan on 17/03/2023 06:50:26:

[…]

… which we usually see only as a blur, or with rolling-shutter artefacts dominant.

.

This is about as good as it gets: **LINK**

https://www.nikonsmallworld.com/galleries/2018-small-world-in-motion-competition/stephanoceros-fimbriatus-rotifer-feeding

… but there is so much more detail to observe, if only we have the tools.

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Sonic Escape on 17/03/2023 07:01:05:

I think that the 100nF (C3) capacitor to ground at the input of the circuit is wrong. At 10MHz that is equivalent to 0.16Ω.

Now I saw the previous message, for 80Hz it should be fine

Edited By Sonic Escape on 17/03/2023 07:05:00

.

I admit to struggling a bit with the detail in the paper, but I think 10MHz is stated only as a bandwidth requirement [for the purpose of pulse-shape] rather than as a frequency.

MichaelG.

[Joseph Noci 1Participant @josephnoci1]

Posted by Sonic Escape on 17/03/2023 07:01:05:

I think that the 100nF (C3) capacitor to ground at the input of the circuit is wrong. At 10MHz that is equivalent to 0.16Ω.

Now I saw the previous message, for 80Hz it should be fine

Edited By Sonic Escape on 17/03/2023 07:05:00

Not really – it depends on the duration of the pulse as well. A 1us pulse will be suppressed by that cap.

[Michael GilliganParticipant @michaelgilligan61133]

Regarding the much-disputed capacitor at the input … this is from the data-sheet for U1

.

.

MichaelG.

[Martin KyteParticipant @martinkyte99762]

So C3 not connected to the input then.

The limit on repetition rate is going to be dependent on the time constant of the 100 ohm resistor R1 and the capacitor C2. Which with the values listed is 0.33s giving you about 3 Hz.

You could split C2 into two or more components which would shorten the time constant of each.
I think Joseph Noci said as much above.
Regards Martin .

[Joseph Noci 1Participant @josephnoci1]

Posted by Martin Kyte on 17/03/2023 09:37:15:

So C3 not connected to the input then.

The limit on repetition rate is going to be dependent on the time constant of the 100 ohm resistor R1 and the capacitor C2. Which with the values listed is 0.33s giving you about 3 Hz.

You could split C2 into two or more components which would shorten the time constant of each.
I think Joseph Noci said as much above.
Regards Martin .

Caps in parallel add, so final RC remains the same.

[Clive SteerParticipant @clivesteer55943]

White LEDs are usually constructed using a blue/UV led covered with a white fluorescent material and the shade of white depends on the materials recipe. I suspect that the material may have a degree of persistence that would moderate 10Mhz operating.

I used a similar circuit to make an LED strobe light to visually check the operation and timing of watch escapements but these operate at relatively low frequency.

CS

[SillyOldDufferModerator @sillyoldduffer]

I looked at the circuit last night, but didn't comment because it was past my bedtime. Just as well perhaps.

I assumed the circuit to be rather simple, seeing a fast chip on the input, which I guessed was to sharpen up incoming pulses, a fast FET connecting an LED to ground, and powered by a small fast electrolytic, backed up by a big slow one.

The blurb explains the circuit can be used at 5, 10 or 20V, so I took R1 as intended to be protective rather than a RC filter with R1/C1+C2. It also says the input pulses are about 1 microsecond.

In my limited experience the C1+C2 arrangement is used to speed up decoupling. It makes use of the fact that small electrolytics charge and discharge much faster than big ones, and of course all capacitors are much faster than a battery or DC power supply.

Is it possible in this circuit that flashing the LED with wide spaced 1uS pulses doesn't fully discharge C1 + C2? Instead the pulse empties C1 whilst C2 is still slowly waking up, and finishes before C2 has had time to discharge much energy. Tjhen, after the pulse stops, C2 quickly recharges C1, whilst both are slowly topped up by the supply. I note the duty cycle is very low, 0.00005:1 at 50Hz. In this scenario the R1/C1+C2 filter maths would have to be modified because 1 microsecond pulses mean the C2 to C1 time constant dominates. In comparison to C1/C2, I think R1/C2 is dozing in the slow lane. Putting a scope on the circuit would reveal all.

Anyway, C3 are ceramics – dreadful tolerances, but being cheap and wonderfully fast makes them excellent decouplers. I think the diode is belt and braces protection and almost anything would do, provided it's faster than the LED. How about a 1N4148?

Curiously I was thinking of trying a circuit like this after a pendulum discussion about determining amplitude with a camera. I was going to set a pendulum swinging in a blacked out room and photograph it with my camera on the 'B' setting (shutter always open), whilst strobing the bob with a bright white LED. The photo should show the position of the bob at each accurately timed flash, with a scale behind graduated in degrees.

Too many ideas, too little time! It's past 11am and I haven't started any of today's jobs.

Dave

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Martin Kyte on 17/03/2023 09:37:15:

So C3 not connected to the input then.

[…]

.

So it would appear, Martin … and it wouldn’t be the first time that a modern schematic had been mis-interpreted

I really need to check with the authors in case it’s a typo in their Fig. 4(a)

… I mourn the passing of those helpful little ‘hump-back bridges’ that we had on old drawings !

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by SillyOldDuffer on 17/03/2023 11:06:57:

[…]

In my limited experience the C1+C2 arrangement is used to speed up decoupling. It makes use of the fact that small electrolytics charge and discharge much faster than big ones, […]

.

That was my understanding too, Dave, and pretty-much the extent of it.
… which is why I came here asking advice about which specific components to buy.

Unfortunately I seem to opened the proverbial ‘can of worms’ instead.

Regarding the Diode … I still don’t understand whether it needs to be something ultra-fast

I have a bag-full of 1N4006 available, but am concerned that might throttle the performance of the transistor.

.

If we can’t write a shopping-list between us all, then I will just have to build it with what I can find, and have a look on the oscilloscope … a potentially long-winded series of iterations.

MichaelG.

[Author]

Posts