EL7900

[Bill Sloman]

I can't see what either of those sour old hens ever favor. They are
truly hopeless.

John Larkin looks at posts from a rather restricted and narcissist
point of view - if we aren't praising him we are wasting bandwidth.

 

[Marco Trapanese]

Il 28/08/2012 20:52, Phil Hobbs ha scritto:

Why use such a horrible photosensor? You can get tiny PDs for a few cents.

I found these ones:

http://www.farnell.com/datasheets/573494.pdf
http://www.farnell.com/datasheets/1244281.pdf

If I understand correctly:

- the reverse light current is independent of the reverse voltage until
the diode is properly polarized. What is the minimum recommended reverse
voltage?

- the maximum light I can measure is only limited by the thermal
dissipation (Ilight * Vr).

Regards
Marco

 

[George Herold]

Il 28/08/2012 20:52, Phil Hobbs ha scritto:


I found these ones:

http://www.farnell.com/datasheets/573494.pdfhttp://www.farnell.com/datasheets/1244281.pdf

If I understand correctly:

- the reverse light current is independent of the reverse voltage until
the diode is properly polarized. What is the minimum recommended reverse
voltage?

- the maximum light I can measure is only limited by the thermal
dissipation (Ilight * Vr).

Regards
Marco

/a lurker/

If you are going to use a TIA opamp, then reverse biasing only helps
get you a bit more speed (by reducing the PD capacitance) With the
stray C to all the other PD's this is not going to be of much
benefit. So (I think) Phil H. was suggesting to just leave the non-
iverting input grounded and switch PD's in with an open drain or tri-
state output on the shift register (SR). (His original circuit with
the SR just changing the bias condition of the PD's was a (very rare)
mistake. He caught it himself, before we dog- piled on him.)

George H.

 

[George Herold]

This is just another goofy idea, but you could connect a capacitor
across each photodiode. A scanning analog multiplexer would hit each
pd/cap array in sequence, simultaneously measuring how much integrated
current the PD had detected, and recharging the cap. I guess that CMOS
and CCD camera imagers work that way. One 8:1 analog mux per 8 PDs
isn't a lot of overhead.

Well and one cap for each PD. And as long as you don't wait too long
to read out, and let the cap charge up so that it starts to forward
bias the diode. I guess if you found a cap that was 'fully' charged
you could increase the readout rate. A software controlled 'gain',
which is nice.

Reminds me of the difference between Farnsworth's image dissector tube
and Zworykin's iconoscope; the image dissector essentially multiplexed
the photocurrents of a pixel array, and the iconoscope multiplexed the
integrals of the pixel currents. The s/n advantage of integration was
huge.

Yeah, seems a shame to only use the photons that are hitting the PD
while you are looking at it.

George H.

 

[Marco Trapanese]

Il 29/08/2012 18:13, George Herold ha scritto:

Yeah, seems a shame to only use the photons that are hitting the PD
while you are looking at it.

On my board I have plenty of photons, no problem to lose some

Marco

 

[[email protected]]

It's certainly a very odd-looking circuit, if the block diagram on

page 4 of the datasheet is to be believed.



The two P-MOSFETs seem to form a current mirror, with what's drawn as

an N-channel MOSFET as the current source. It would make more sense if

that MOSFET at the top were another P-channel device. It would then -

mostly - be a resistor when enabled with more than 2.7V of gate-to-

source voltage - not less than 2.4V according to the data sheet - and

guaranteed to not act as a resistor with less than 1.2V of gate-to-

source voltage



My guess would be that if the photo-diode draws too much current, the

MOSFET at the top comes out of saturation (stops looking like a

resistor and starts looking like a constant current source) and the

voltage at its drain and the common source connection for the two

MOSFETs forming the current mirror would then collapse down to

something very close to the negative rail.



It could be that the top MOSFET in your particular device comes out of

saturation at a relatively low current when it's only got 3V of gate-

to-source voltage.



See what happens if you boost the device supply voltage closer to 5.5V

- perhaps with a battery.

If you look at that inordinately low pA dark current spec, low for a MOSFETanyway, it means they're doing something unusual. My guess is this is achieved by keeping the output FET operating sub-threshold all the time (achieved by doping for a high Vthreshold), and in this region the IDS vs VGS operation is exponential, not at all like the resistive-triode characteristic. So an abrupt turn-off in the vicinity of compliance voltage vs Iout would not be all that strange, just a few 10's millivolts will do it.

 

[Marco Trapanese]

Il 29/08/2012 20:25, Phil Hobbs ha scritto:

The OP probably wants to avoid adding another 400 parts to his board,
which is a big hit even by my standards.

Yep, you're right

Marco

 

[Marco Trapanese]

Il 30/08/2012 05:06, John Larkin ha scritto:

And, apparently, much
sooner than would be expected from the data sheet.

Exactly.

Marco

 

[Bill Sloman]

It's not behaving like a saturating amplifier; it's folding back down
to zero output as the light input increases. And, apparently, much
sooner than would be expected from the data sheet.

The pfet current mirror model explains the effect, but not why it does
it at low output currents.

It's not the output current - as such - that's the problem, but the
light level.

The P-channel MOSFET current mirror model explains the effect, but
without data on the specific P-channel MOSFET involved it can't say
anything about the current level where the tail MOSFET switches from
being a resistor to being a constant current source.

All we know is that the P-channel MOSFET's in the production batch
from which Marco Trapanese bought his parts switched at a rather lower
current than the batch the manufacturer tested to generate the data-
sheet.

Whoever wrote the data sheet has got egg all over their face, but
that's human error, and it happens from time to time.

 

[Bill Sloman]

You would think that the parts would be tested in production.

Obviously not at high light intensities ...

 

[josephkk]

It's not behaving like a saturating amplifier; it's folding back down
to zero output as the light input increases. And, apparently, much
sooner than would be expected from the data sheet.

The pfet current mirror model explains the effect, but not why it does
it at low output currents.

The data sheet actually explains that this collapse will happen somewhere
past 6 mA. The devices Marco has on board are collapsing at a clearly
lower current (~2.6 to 3.6 mA) and are thus not conforming to datasheeet
specification. It may be he got some "out of spec" parts or possibly
"counterfeits". Does that sound reasonable to anyone?

?-)

 

[Bill Sloman]

(1) It's ALL P-channel, mirror AND enable devices
(2) The mirror is 1:M
(3) Crank it up to Vcc=+5V and see what happens
(4) Refer to:

Jim hasn't been paying attention. Marco Trapanese has already done
that, and told us about it.

Using his original 3.3V supply, the output went up to 2.7mA with
increasing light levels before collapsing on further increase. When he
raised the supply close to the absolute maximum 5.5V, the output made
it to 3.5mA before collapsing, well below the 6mA mentioned in the
data sheet.

 

[[email protected]]

(1) It's ALL P-channel, mirror AND enable devices
(2) The mirror is 1:M
(3) Crank it up to Vcc=+5V and see what happens
(4) Refer to:

Date: Tue, 28 Aug 2012 11:36:15 -0700
Message-ID: <[email protected]>

I believe what John is saying that the schematic explains why it
will happen, but not why it happens at a much lower current than
it should be able to output according to the datasheet

-Lasse

 

[Bill Sloman]

You could easily have done without the first sentence of your reply.

I've been exposed to your posting habits for too long, and in any
event this is the same Jim Thompson who claims to have reported me to
the FBI for having dangerously anti-American ideas. His other
imbecilities need to be advertised - people might otherwise take him
seriously

That would have moved your position from "fatheaded jerk" to
"helpful."

Really? Jim Thompson is the fat-headed jerk around here - and I'm one
of the people who's had more than enough of his silly ideas. Giving
him a hard time doesn't make me either fatheaded or a jerk.

And who would I having been helping when I pointed out something that
would have been obvious to anybody who'd actually followed the thread?

Enjoy your retirement.

I can think of better things to do with it than correcting narcissist
jerks like you, but I've haven't been able to make any of them work
yet.

 

[Marco Trapanese]

Il 30/08/2012 17:00, Bill Sloman ha scritto:

Obviously not at high light intensities ...

"High intensities"... just a desktop lamp! Or the led of my smartphone.

Marco

 

[Marco Trapanese]

Il 30/08/2012 18:20, Jim Thompson ha scritto:

The OP should define his set-up more accurately... right now, nothing
computes.

What other information should I provide?
Even if the actual board has a chain of 74HC595 to enable one EL7900 at
time, the tests I reported was conducted with only one sensor soldered
on an empty PCB.

Vsupply: 3.3V (and 5V as described)
Enable: tied to GND
Rload: 562 ohm (I tried other values)

I think such a setup should be very close to the one the Intersil's guys
used to characterize the pd.

I use a desktop lamp to illuminate the sensor and I have both a
voltmeter and a ammeter to measure the output current and the voltage
across the Rload.

Marco

 

[Marco Trapanese]

Il 30/08/2012 18:01, Jim Thompson ha scritto:


A whole batch, indeed. I have bought 1000 pcs and they all have this
behavior.

Crank Vcc up to +5V and re-test.

I'm sorry I've already written about this.
At 5V it collapses around 3.5-3.6 mA.

Marco

 

[Marco Trapanese]

Il 30/08/2012 18:09, Jim Thompson ha scritto:

Aha! It finally dawns after staring at the data sheet forever!

I apologize. I cannot understand the meaning of your sentence. It's my
fault, English is not my primary language.

Where are you measuring the current??

Between the pin "output" of EL7900 and one end of the rload. The other
end is tied to gnd.

Marco

 

[[email protected]]

Il 30/08/2012 18:09, Jim Thompson ha scritto:


I apologize. I cannot understand the meaning of your sentence. It's my
fault, English is not my primary language.


Between the pin "output" of EL7900 and one end of the rload. The other
end is tied to gnd.

Marco

what the resistance of the ammeter?

-Lasse

 

[George Herold]

Il 30/08/2012 18:20, Jim Thompson ha scritto:


What other information should I provide?
Even if the actual board has a chain of 74HC595 to enable one EL7900 at
time, the tests I reported was conducted with only one sensor soldered
on an empty PCB.

Vsupply: 3.3V (and 5V as described)
Enable: tied to GND
Rload: 562 ohm (I tried other values)

I think such a setup should be very close to the one the Intersil's guys
used to characterize the pd.

I use a desktop lamp to illuminate the sensor and I have both a
voltmeter and a ammeter to measure the output current and the voltage
across the Rload.

Marco

Ahh, Hey as Phil pointed out that thing has a weird spectral
response*.
If your desktop lamp has an incandescent bulb then it can be putting
out a lot of photons in the near IR.


George H.

*(I wonder how they managed to get a 'hole' near 630 nm... what they
don't like HeNe's?)