LED on Photodiode step response

Hmm. If you have a charge q drifting at constant velocity V between two metal plates, it produces a rectangular current pulse of height qV at the output terminals, until it hits the plates. (I have some pretty pictures of this from back when I was trying to use laser-produced plasmas as air ionizers inside semiconductor process equipment.)

So conceptually each photocarrier produces a pulse roughly t_transit long (depending on where the pair is produced). That does apply a low-pass filter to both the photocurrent and the shot noise. It isn't a particularly steep one, and it obeys the cardinal rule that the shot noise sees exactly the same frequency response as the signal.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
 Click to see the full signature
Reply to
Phil Hobbs
Loading thread data ...

Something like that in a laser printer, quad or hex sided mirror on a motor, and a pickup diode one end of the lens that takes the scanned beam out to the drum. Pickup diodes have quite large surface area, ~4 x 4 mm in one, another was maybe 2 x 4 mm.

Grant.

Reply to
Grant

ope if

rough

s an

a

ons

o

the

rents

ot

ty

by

es

s as

ot

s

=A0No

ay

a

s.

ot

y

ee

se

se.

o
a

t- Hide quoted text -

That's it! 'Much Thanks'. A picture/model that connects the semicondcutor process to the current pulse in the external circuit. It was previously a 'hazy' piece of my understanding.

Now I'm wishing I had a PD that was all diffusion current. (blue light on a 'red sensitive' PD?) I could show a microsecond response time, and noise out beyond that. Oh, how about shot noise from a reveresed biased pn diode. I'll need some hugh area to get any decent current. Do big power diodes have reverse leakage currents in the

10-100 uA range?

George H.

Reply to
George Herold

ward

.

ms

ire

p

of

es

ok'

f I

or,

the

nother

I don't know those. As you sweep faster you need a bigger area for the same signal. (I've never done any rotating mirror stuff, so I'm far from an expert.)

George H.

Reply to
George Herold

I'm no expert, but I like pulling electronics stuff to bits :) See how it works.

Grant.

Reply to
Grant

works.

I like to blow parts up, to see their limits.

John

Reply to
John Larkin

works.

Something I wish more manufacturer designers did, 'cos too much is spec'd right on limits, or just good enough for say 97% (or some percentage) to last out the warranty period.

Grant.

Reply to
Grant

On a sunny day (Wed, 13 Apr 2011 19:31:54 -0700) it happened John Larkin wrote in :

Good thing you do not design airplanes.

Reply to
Jan Panteltje

I do design aerospace stuff. The aerospace people blow up and break everything to test the limits.

formatting link

formatting link

Why would you *not* want to know how close your parts are to failing?

Sometimes we push parts past their datasheet specs, as much as 3:1, based on testing. That delivers performance.

John

Reply to
John Larkin

On a sunny day (Thu, 14 Apr 2011 06:44:51 -0700) it happened John Larkin wrote in :

So what is the legal situation there, say plane crashes, and it turns out your Vce was 2x Vce max as indicated by the manufacturer. Unfortunately that batch was an exception. ?

Reply to
Jan Panteltje

manufacturer.

Hey, all of life is risky. And all electronics has finite failure rates, as all of our manuals point out in the disclaimer. In the field, our gear has MTBFs that are many times better than MIL-HBK-217 or Bellcore calculations predict. Bad designs can be many times worse.

We helped replace a heads-up display an the C130. The old electronics was designed by a big-name British aerospace company. It had a field MTBF of 22 hours.

When you test a part to failure, you know more about the part. Why wouldn't you want to know more?

John

Reply to
John Larkin

On a sunny day (Thu, 14 Apr 2011 08:07:55 -0700) it happened John Larkin wrote in :

the manufacturer.

Calculating MTBF has always been a bit of a mystery to me. I am not even sure the accepted methods make any sense these days. Especially with ever smaller logic and cosmic rays etc.

In the mil the things are a bit different than in civilian life. I mean things have a real short expected life in war time, something like 2 shots for a tank IIRC.

About the war ships I remember that in a combat situation all fuses were bridged, so no blown fuses preventing you to shoot. Of course something else goes ...

Oh, that testing is good, but using parts beyond manufacturer's specs makes you a target of lawyers.

Same with buildings and overloading floors, I remember an architect walking out of a meeting in protest when ever more electronics was installed.

Reply to
Jan Panteltje

formatting link

There is this long tail on the step response. Here?s the same picture with the timebase slowed down.

formatting link

Looks something like a 2-2.5 us tail on the step.

What?s it caused by? I first thought it was perhaps heating of the LED during the step. But I was reading ?Photodetectors? by S. Donati, over the weekend. In Chapter 5 he talks about two sets of photo- generated carriers the majority are made in the depletion layer, and travel with the drift speed, (times in the nano-second range.) The second smaller set are those generated in the doped region. If they are within a diffusion length of the depletion region then these have a chance of adding to the photocurrent. Donati says that these have a time constant in the microsecond range! I wonder if this is what I am seeing?

I looked at the step response with a different color LED. (you expect more absorption in the doped region with shorter wavelength light.) So here is the pulse response from a 635 nm red and 594 nm Amber LED.

formatting link
The red LED is the trace that is a bit higher. I?m looking to see if I can dig up any blue or white LED?s.

Any other ideas how I might determine if that's what I'm seeing...

Oh, the PD is a OSI-optoelectroncs PIN-3CD (3.2mm^2 area) reversed biased at ~12V.

George H.

------------------------

Hi George,

This same problem plagued me some 10 years ago and was solved here. Some of the regulars here helped me getting the proper ideas. Unfortunately the thread is not archived anymore (can't find it).

The problem was caused by photons falling OUTSIDE the ring that surrounds the active region. The electrons there drift slowly into the active region and then get measured. The same timing and tail shape was seen there.

The problem was solved by hacksawing my $100 PD and painting the rim black. Result: a perfect square response (I was using it to test the squareness of laser pulses). Later we redesigned the optics so the light only fell in the centre.

Happy hacksawing! Arie de Muynck

Reply to
Arie de Muynck

top

e

ti,

That's interesting, Thanks Arie! I could try putting some mask over the the outside portions. But perhaps you are right ... Time for the dremel.

Hey that's it! I put a mask over it with a hole in the center. (The mask is not perfect so a little is leaking around the edges still.)

Here's the 'scope shot.

formatting link

Thank you again!

George H.

Reply to
George Herold

Ten points!

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
 Click to see the full signature
Reply to
Phil Hobbs

2003, it was. Google on "faster than a perfect transconductance amp". Good catch!

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
 Click to see the full signature
Reply to
Phil Hobbs

For my Ph.D-thesis work I needed a fast, but lossless shutter (to create pulses of squeezed light) and used a 25µm slit glued onto the rim of 3.5" hard disk drive overdriven to up to 300 rpm.

That gave pulses < 0.5 µs. The big advantage compared with the polygon mirror approach which we tried before was that during the pulse duration the phase fronts of the beam stayed surprisingly nice -- the squeezed light was measured with a homodyne detector and therefore had to interfere with a reference beam. The interference visibility was still very good, even though the spinning disk moved quite a lot of air inside its case and was not exactly noiseless.

That's very impressive and quite a bit faster than the slit. Balancing a polygon mirror for 100 000 rpms probably is no simple engineering task either...

--
GPG key: 
http://pgp.mit.edu:11371/pks/lookup?search=J%FCrgen+Appel&op=get
Reply to
=?UTF-8?B?SsO8cmdlbg==?= Appel

"Phil Hobbs" ...

Thanks, that info pointed me to my original thread from September 1997 "Problem with photocell turnoff". I found three messages, but the thread was much longer.

Sic transit gloria data...

Arie

Reply to
Arie de Muynck

This is great:

formatting link

He uses a slow single-pole filter to lowpass a PWM output, then adds four opamps to make the output jump instantly to the correct value.

John

Reply to
John Larkin

Smoking began at age ten, 1970. It was Camel and Pall Mall non-filter. REAL cigarettes. I smoked those and Marlboro for a few years, and my habit got to be two packs a day, and I smoked 'em!

I tried Kool for a short time, but I switched to Winston and never stopped smoking those at a rate of 2 packs a day all the way until 1987. A full 17 years. I was getting sick of them. 50% of all smokers are allergic to it, and fought through it not even knowing.

I quit, cold turkey that week. I have never picked them up since. Tobacco free since '87. Nearly a quarter of a century, and soon to be so. I cycle a large number of miles every day, Lots of fun, and hardly all that tiring, but I wont mention the number so it doesn't blow John's pathetic numer0logical position.

Reply to
AtTheEndofMyRope

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.