et
e e a m t.net-Hide quoted text -
snip
some numbers in this:
-Lasse
et
e e a m t.net-Hide quoted text -
snip
some numbers in this:
-Lasse
et
e e a m t.net-Hide quoted text -
Yeah, I noticed that.. Thanks to Grant too.
I didn't save the screeen shot on my flash. (I should hit the save button more often.) IR led's are slow too.
Bring out the big guns ehh? I've never looked at the step response from our laser diode, but we've modulated it up to ~150 MHz. There's a reported change in the modulation sensitivity, at the microsecond range and longer its due to heating, at shorter times it drops by an order of magnitude and is due to (index of refraction?) carrier density effects. At very high frequency you can get an enhancement due to relaxation oscialltion in the LD... about 6 GHz, in the LD's I know.
Yeah sorry I didn't save all the crummy 'scope shots. But I'm certainly open to some silly measurement error. There's a TIA with OPA134 (GBW~8MHz, 1-3pF Cin) and 10kohm feedback resistor, with
3.3pF parallel capacitance (plus some extra stray... call it ~2 pF, perhaps a bit less).I think I'm seeing a LED temperature effect. Does the light output go up as an LED gets hotter? If it's linear with delta T, I can't change drive level and make it go away.
George H.
ng
ty.)
he top
get
icture
the
ati,
-nd
=A0The
they
ave
have a
t I am
expect
)ee if
cal.net-Hidequoted text -
?rn-
ere
Once!... no Twice!!
George H.
pping
y.)
=A0(The top
et
icture
Hmm, The circuit allows students to muck about with the input. (They are measuring noise.) We didn't want soldering. We're pretty much stuck with terminal blocks and switches. I figure the opamp is a $2 fuse, so the socket is a must. (It's nice to be able to change all sorts of bits too!)
George H.
net
Hide quoted text -
Interesting.
The fastest LEDs I know of are around 900 nm. They aren't exactly fast, about 30 MHz. Have a look at the SFH4550 (12 ns, 850 nm) and VSLB3940 (15 ns, 940 nm)
LEDs have long tails on their falling edges due to the relatively long carrier lifetime. That rising edge was when the light turned on, right? Because the falling edges always look like that.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal
What would happen if they were reverse biased by about a volt or so at turn-off time? Would that help?
John
It might. Trying to speed up a LED is a bit like lipstick on a pig, of course. Lasers are two orders of magnitude faster, some nearly 3 orders.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal
net
LEDs are less efficient when they're hot. But the forward drop goes down. So there is usually some source resistance where the effects cancel.
John
Great! Now I have to clean coffee spray off my monitor screen.
John
g
.)
e top
get
cture
he
ti,
d=A0The
hey
ve
ave a
I am
xpect
e if
al.net-Hidequoted text -
n-
ou
at
ed
book,
E2L
.net-Hide quoted text -
, ?Yeah the rising edge, but I just looked and the falling edge looks identical! Is there a reason the rising and falling edges should look different?
I've been trying to convince myself it's just an LED effect. Perhaps a temperature change, or the above carrier lifetime in the LED? The reason I don't think it's a Photodiode effect is that when I look at the high frequency (BW ~ 1.5 MHz) noise with a DC light source and several hundred mV of bias 'current' across the feedback resistor, (noise signal dominated by shot noise) I see exactly the same amount of noise regardless of the light source.
George H.
t- Hide quoted text -
nt
) po get
ture
ei,
and
eey
have
ve a
I am
pect
.)
if
. dcal.net-Hidequoted text -
n-
Thanks John, If it's due to more current because of a drop in forward voltage, then I should see that in the current. (Which I don't... sigh)
OK let me try a totally different idea. Can the stray capacitance of 'stuff' have a long tail like that. Charges moving around on pieces of plastic/fiber glass or something?
I could air wire the circuit and see what that does...
George H.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Is there any hidden connection between the insistence of these LEDs on continuing to emit light long after they've been turned off, and the insistence of the participants in this thread in continuing to re-emit screenfuls and screenfuls of previous posts, before finally adding a one-sentence comment at the end?!?!?!
net
net
Hide quoted text -
There's a tail on the falling edge of the output of a LED, at least when you hit it really hard. In the high-level injection regime, the light output goes like the carrier density squared, so it drops very fast initially, but then as the carrier density drops, it slows down to the usual minority carrier lifetime, which is somewhere between 20 and 200 ns.
I'm sure that's why LED speed is always specified with some really gross amount of drive.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal
Modern newsreaders often run on machines with LED backlights? ;)
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal
net
What's the LED driver circuit like?
Probably not the tail you're seeing. PCB 'capacitors' can have 'hook' behavior, but your waveforms don't look like that.
Post (or email me) your LED driver and pd circuits, and I'll see if I can spot anything.
Time constants in the microseconds probably aren't strays.
John
On a sunny day (Tue, 12 Apr 2011 10:06:32 -0400) it happened Phil Hobbs wrote in :
Samsung, the idiots, these days gives customers the choice between LCD and LED TVs.
net
How about posting the schematic? If the edges are symmetrical, it's probably the TIA--the LED would be cooling down when it wasn't emitting light, so you wouldn't see the thermal tail.
Op amps have really terrible supply rejection when it comes to fast transients--depending on the feedback, you can easily get _gain_ from the supply pin to the output. (PSR is specified with respect to the amplifier _inputs_.)
Another possibility is phase funnies in the open loop gain. Composite amps are notorious for that sort of thing. The long-time settling behaviour is dominated by the lowest-frequency pole or zero in the open loop transfer function--lead-lag networks can cause this sort of thing even though the closed-loop response looks nice on a spectrum analyzer.
Try touching the summing junction with your finger and see what happens to the step response.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal
ng
rent
y.)
top
to get
icture
the
ati,
-, and
The
they
e have
have a
t I am
expect
ht.)
ree if
...
sed
tical.net-Hidequotedtext -
d?
urn-
e she
IA
go
ge
I just tried a laser diode (lamda=3D 785nm). Exactly the same tail... It's some circuit thing?
No, A current pulse through a resistor gives a nice square output.... Grrr.
I'm going to order a blue LED and set this aside for a bit. (Unless someone has some idea?)
Thanks for all the input, George H.
Rereading, it looks like you have something like a 20 pF summing junction capacitance and a 2 pF feedback cap. That means you have an open-loop pole/zero pair separated by a factor of 10 in frequency, up near the GBW of the op amp. That will lead to some entertaining settling behaviour, which at this point would be my best guess as to what you're seeing.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal
Hi George,
You may find VCSELs easier to use than laster diodes. They can usually be safely driven at constant current and have ~circular beams, some of them have built-in collimator lenses too, They are noisy compared to LEDs if that is an issue, but are designed to be modulated very quickly (GHz),
-- John Devereux
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.