Maker Pro
Maker Pro

Photodiode amplifier noise

N

Nemo

Jan 1, 1970
0
I'm embarrassed to ask this but -

I have a high gain (transconductance) amp with a photodiode, biased to
be a few hundred pF, on its input. I was tweaking it to optimise the
noise levels and found the noise of the amp was way less than the noise
of (the amp with the diode on the front). So I replaced the diode with a
simple capacitor... and I still get extra noise when I fit it. It does
not appear to be noise from the bias supply; the capacitance can be to
ground and I still see the extra noise appearing.

The texts I've read say there's noise generated by Cin, and it certainly
seems like there is. I can understand it acts as a potential divider on
the amp's gain and attenuates high frequency signals. What I don't get
is why it generates noise without a signal present? What is the physical
mechanism which creates this noise?
 
I'm embarrassed to ask this but -

I have a high gain (transconductance) amp with a photodiode, biased to
be a few hundred pF, on its input. I was tweaking it to optimise the
noise levels and found the noise of the amp was way less than the noise
of (the amp with the diode on the front). So I replaced the diode with a
simple capacitor... and I still get extra noise when I fit it. It does
not appear to be noise from the bias supply; the capacitance can be to
ground and I still see the extra noise appearing.

The texts I've read say there's noise generated by Cin, and it certainly
seems like there is. I can understand it acts as a potential divider on
the amp's gain and attenuates high frequency signals. What I don't get
is why it generates noise without a signal present? What is the physical
mechanism which creates this noise?

It's called thermal noise; or more properly Nyquist noise. The way
I've understood it - it's noise created by electrons 'boiling' off of
surfaces. The effect is actually temperature-dependent, and follows
the same type of plot as thermal radiation. I'm not 100% sure on this
next statement, but I believe the effect is more pronounced on
capacitors (read: capacitors, semiconductor junctions, antennas) than
any other circuit element because electrons boiling off of the
dielectric must always end up in your circuit.
 
Nemo said:
I'm embarrassed to ask this but -

I have a high gain (transconductance) amp with a photodiode, biased to
be a few hundred pF, on its input. I was tweaking it to optimise the
noise levels and found the noise of the amp was way less than the noise
of (the amp with the diode on the front). So I replaced the diode with a
simple capacitor... and I still get extra noise when I fit it. It does
not appear to be noise from the bias supply; the capacitance can be to
ground and I still see the extra noise appearing.

The texts I've read say there's noise generated by Cin, and it certainly
seems like there is. I can understand it acts as a potential divider on
the amp's gain and attenuates high frequency signals. What I don't get
is why it generates noise without a signal present? What is the physical
mechanism which creates this noise?

I've just been bitten by the same bug. (ignoring the input
capacitance to ground in an inverting amplifier.) OK it was a few
months ago, but it led me to sand off the ground plane underneath my
input and improve the response time by an order of magnitude. So the
noise comes from the op-amp and the gain comes from the ratio of the
capacitance from the inverting input to ground and the capacitance
from the inverting input to the output.

George Herold
 
The feedback resistor and input capacitance together, besides
attenuating the output signal also cause a 90 degree phase shift, which
isn't a problem except at the high frequency end of the opamp's useful
gain, where it also starts shifting the phase by close to 90 degrees.
The result is a loop phase shift of nearly 180 degrees, or positive
feedback. This results in a big peak in the gain seen by the input noise
voltage at this high frequency.
  A good place to look is Phil Hobb's websitehttp://www.electrooptical.net/http://www.electrooptical.net/www/frontends/frontends.pdf

You can also look at the data sheets on TI fast FET-input op amps -
they still seem to be being written by the people who were working for
Burr-Brown back when TI took them over.

http://focus.ti.com/lit/ds/symlink/opa657.pdf

who don't yet seem to have been properly indoctrinated about the TI
corporate policy of ripping off the customers.
 
TI makes a lot of kick-ass parts. Their field guys call on us, and
support is excellent.

They have the best fast 15-volt analog process on the planet; look up
THS3201.

I started using the their analog parts with the TLC2001 for which the
data sheet didn't mention the anomalously high inut capacitiance (at
around 15pF). When this reared up and bit me I rang TI support, who
didn't have a clue what the input capacitance might be and showed no
willingness to find out.

If they've raised theor game sice then I'll be pleased, but surprised.
Burr-Brown wasn't ever anything like as helpful as Analog Devices, but
they never practised the active deceit that I'd run into with Texas
Instruments in 1972.
And check out THS4303.

And they are *still* shipping SN7400N.

Whose parts do you prefer? Philips? Siemens? Plessey? Ferranti? GEC?
Intermetall?

Analog Devices, Linear Technology, Fairchild, National Semiconductor,
Siemens (now Infineon) and Motorola (now ON=Semiconductor) but I'll
use anything that I can be sure of getting my hands on. Complete and
reliable data sheets are a big plus, but it is the device performance
that is crucial - and good data sheets can save a lot of debugging
time.
 
B

Bob Larter

Jan 1, 1970
0
John said:
TI makes a lot of kick-ass parts. Their field guys call on us, and
support is excellent.

They have the best fast 15-volt analog process on the planet; look up
THS3201.

And check out THS4303.

And they are *still* shipping SN7400N.

Holy shit! Who still uses plain vanilla 74xx parts?
 
G

GregS

Jan 1, 1970
0
I have read around these parts that increasing bias will increase sensitivity
and I would assume increase SNR, but Thats all I know. I would
like to know more.

greg
 
E

Ecnerwal

Jan 1, 1970
0
John Larkin said:
Maxim is the real horror story.

10 Hey look, a neat part!
20 Discontinued.
30 GOTO 10

Insert:

15 Design something that uses it

if feeling really masochistic.
 
J

John Devereux

Jan 1, 1970
0
John Larkin said:
12 non-industry-standard pinout

16 have all your friends and family each request 8 samples

....And all your friends and family are still getting their mailers 10
years later :)
 
I have read around these parts that increasing bias will increase sensitivity
and I would assume increase SNR, but Thats all I know. I would
like to know more.

greg

Get Phil Hobb's book if you want the low down on photo diode preamps.
There is some good stuff on his web site if you can't afford the $200
price tag.
You bias the PD to reduce capacitance (increase bandwidth). No effect
on SNR unless you're running at low currents where the dark current
might be an issue. Bias has no effect on sensitivity. One photon =
one electron

George Herold
 
N

Nemo

Jan 1, 1970
0
Thank you folks, that's helped get it clear in my mind what's really
going on. And to answer some points -

- I'd already pored over Phil Hobbs' front ends article (it was
mentioned here recently)
- I'm already using the OPA657. I noticed it gave much lower noise due,
I think, to its low current noise across the large feedback resistor. It
seems way better than the OPA847 which has <1nV/rt Hz voltage noise but
higher current noise.

So I guess I'm thoroughly indoctrinated in the local groupthink!

(Actually I suspect there's something odd about the OPA847. It seems
very difficult to get stable, even using the Texas recommended circuit
for a mere x20 gain.)

Concerning the "favourite manufacturer" thread:

For high end analog parts, cost no object, I would look first at Texas,
Analog and Linear Tech.

For cost-effective ones I would look at National and OnSemi but they
don't seem to make really leading edge stuff these days.

For cost-effective consumer parts I would also look at SGS Thomson and
Philips. However their product life cycles are only a few years so I
tend to avoid them, despite being European myself. My impression is that
European companies tend to specialise in niches I don't play in, such as
mobile phones, or produce cheap commodity parts in the Far East. I would
like to use more European parts but they just don't suit my
requirements.

Maxim have interesting parts but again, they tend to burn the users with
sudden shortages. I have heard they are fabless (not fab-u-lous) and so
if they run short of something, it can be months before the next batch
is cooked up. Whatever the reason, I found them risky to use in the
past.

Texas USED to be terrible at supporting designers, and I used to avoid
them. However about 10 years ago I realised they were essentially a
different company from the bad old days, much more customer-focused. And
had leading edge parts again. They are a delight to use now. Let's hope
Maxim evolves that way too.

I use plenty of other suppliers but those are the first ones I check for
a new requirement.

I avoid Far Eastern parts because I have been burned several times by
their incredibly short life cycles. However, they are becoming more
interesting. The Japanese have always been worth using for very low
power stuff; I recall using a Hitachi micro once which was withdrawn
without the European distributor telling us. Hitachi were horrified at
this treatment of a customer and bent over backwards to support us and
somehow provided us with the things for the rest of our product's life.
I've also had good experiences with a Korean sensor manufacturer, who
was investing in new product development during the same period the
established players were getting rid of all their own technical staff
and could no longer support their products. (The Korean company was
founded and run by engineer brothers, where the major established
Western ones had been bought by venture capitalists who were trying to
pare down costs... eventually they sold the concern to a major sensor
manufacturer who tried to use their now near-monopoly to force customers
to pay top dollar for increasingly shoddy wares. This simply forced
customers to look for new suppliers, or make their own as the patents
had run out.)
 
G

GregS

Jan 1, 1970
0
Thank you folks, that's helped get it clear in my mind what's really
going on. And to answer some points -

- I'd already pored over Phil Hobbs' front ends article (it was
mentioned here recently)
- I'm already using the OPA657. I noticed it gave much lower noise due,
I think, to its low current noise across the large feedback resistor. It
seems way better than the OPA847 which has <1nV/rt Hz voltage noise but
higher current noise.

I made up some amplifiers using the AD795 after looking around. At
least I thought they were pretty good.

greg
 
G

GregS

Jan 1, 1970
0
On May 1, 3:12=A0am, John Larkin
On Thu, 30 Apr 2009 15:15:46 -0700 (PDT), [email protected] wrote:


Get Phil Hobb's book if you want the low down on photo diode preamps.
There is some good stuff on his web site if you can't afford the $200
price tag.
You bias the PD to reduce capacitance (increase bandwidth). No effect
on SNR unless you're running at low currents where the dark current
might be an issue. Bias has no effect on sensitivity. One photon =3D
one electron

George Herold

I never thought it would help sensitivity but I thought I heard at least on
person mention it did.

I always wanted to jack up the voltage to detect gamma rays.

greg
 
If Cin wasn't specified, why did you assume it was suitable for your
application? That's reckless. I always measure stuff like that when it
matters.




All my dealings with B-B (pre-TI) were great. Their parts worked as
aadvertised, they were available when the parts were first announced,
and support was excellent. THings remained that way after TI acquired
them.

There was a fiasco a year or so ago. The sales of the BB group ramped
up and a lot of old test sets failed and/or couldn't be reproduced to
keep up, or so I've heard. At any rate it became hard to get parts.
I've been told the TI worked very hard to fix it, and it seems to be
fixed. That sort of thing happens in the semi business.

Maxim is the real horror story.





Analog and Linear are excellent lately, although Analog has burned us
a couple of times. National is fair, overall.

I've been trying to buy some parts from Linear Integrated Systems, the
little fet/diode operation in Fremont. So far, they seem downright
hostile to people who want to buy their stuff, sort of like Western
Electric in the old days, reputed to have an anti-sales staff.

John

I've always regarded LIS as a hobby of the founder. I think it had a
different name, or that is Hall's second jfet company. Is there
something they make that Interfet can't provide?
 
If Cin wasn't specified, why did you assume it was suitable for your
application? That's reckless. I always measure stuff like that when it
matters.

It was certainly reckless, but lots of cheap op amps don't specify Cin
and it's usually close enough to a couple of pF. I'd actually checked
to see if I'd need a feedback capacitor for an input capacitance in
that ball-park. Since the TLC2001 was in a non-critical bit of the
circuit, I didn't have any trouble solving the problem with a 4.7pF
capacitor from output to inverting input.

It pissed me off because it was - in fact - quite a hairy circuit
which I'd designed and laid out without any protoptyping, and all the
tricky bits of the circuit worked exactly as I'd intended, except for
this boring little integrator wrapped around the TLC2001. Most of the
components were SMD so the 4.7pF stuck out like a sore thumb.

<snip>
 
J

James Arthur

Jan 1, 1970
0
John said:
John Larkin wrote:

[...]
This weekend I'm going to buy a tin of Danish sugar cookies, eat all
the cookies, and build a pA range diode-curve measuring setup into it.
I have a bunch of LMC6001s (Ib is 10 fA typ) to use. The datasheet
hints somewhere in one of the app circuits that Cin is big. But it's
not specified anywhere.
It's "Danish Butter cookies" :)

Make sure to _throughly_ clean it before you build anything into it,
until wiping shows no grease residue. Else you might experience an
impressive meeting of ants in there. BTDT.

The rule seems to be that if Cin is large, don't mention it on the
datasheet.
Cin is probably largish and usually not used directly:

http://www.postech.ac.kr/ee/paust/p...a Beam-Profile Monitor for Cyclotron MC50.pdf


Interesting; that's vaguely like what I'm tying to do. But I can't
find a datasheet on the JU421, even on the Vishay site, so I assume
it's another FormerFet.

www.vishay.com/docs/70248/70248.pdf

Cheers,
James Arthur
 
S

Spehro Pefhany

Jan 1, 1970
0
John said:
On Sat, 02 May 2009 21:29:47 GMT, James Arthur

John Larkin wrote:
On Sat, 02 May 2009 13:14:40 -0700, Joerg

John Larkin wrote:

[...]

This weekend I'm going to buy a tin of Danish sugar cookies, eat all
the cookies, and build a pA range diode-curve measuring setup into it.
I have a bunch of LMC6001s (Ib is 10 fA typ) to use. The datasheet
hints somewhere in one of the app circuits that Cin is big. But it's
not specified anywhere.

It's "Danish Butter cookies" :)

Make sure to _throughly_ clean it before you build anything into it,
until wiping shows no grease residue. Else you might experience an
impressive meeting of ants in there. BTDT.


The rule seems to be that if Cin is large, don't mention it on the
datasheet.

Cin is probably largish and usually not used directly:

http://www.postech.ac.kr/ee/paust/p...a Beam-Profile Monitor for Cyclotron MC50.pdf

Interesting; that's vaguely like what I'm tying to do. But I can't
find a datasheet on the JU421, even on the Vishay site, so I assume
it's another FormerFet.
www.vishay.com/docs/70248/70248.pdf

Cheers,
James Arthur

Cool. Vishay didn't recognize "JU421".

Nice part for leakage, but with En=30 nv/rthz and Gm=200 uS, it's sort
of the Marching Band of Noise. This guy adds a second one in
differential mode to add bonus noise and lower the gain to boot.

Well, good luck finding one :)

If you do it'll probably have "DoD pricing".

Why do these scientific types so love differential jfet front-ends?

In electronics they tend to hang on to stuff much longer than we do.

[...]

I hadn't paid much attention to discrete jfets in, well, some decades.
They have low transconductance, high capacitances, miserable matching,
and astounding datasheet parameter spreads, like 10:1 Idss limits.

But for really low noise (like, under 1 nV/rthz) amplification of
high-impedance signals, nothing can touch them. The best fet opamps
are, like, 5x as noisy as the best jfet.
True.

Pity that most people, at
least the scientists, trend to use them wrong.

John

What wrongness have you seen? The commercial stuff for research
applications looks pretty decent to me.



Best regards,
Spehro Pefhany
 
S

Spehro Pefhany

Jan 1, 1970
0
John said:
John Larkin wrote:
On Sat, 02 May 2009 13:14:40 -0700, Joerg

John Larkin wrote:

[...]

This weekend I'm going to buy a tin of Danish sugar cookies, eat all
the cookies, and build a pA range diode-curve measuring setup into it.
I have a bunch of LMC6001s (Ib is 10 fA typ) to use. The datasheet
hints somewhere in one of the app circuits that Cin is big. But it's
not specified anywhere.

It's "Danish Butter cookies" :)

Make sure to _throughly_ clean it before you build anything into it,
until wiping shows no grease residue. Else you might experience an
impressive meeting of ants in there. BTDT.


The rule seems to be that if Cin is large, don't mention it on the
datasheet.

Cin is probably largish and usually not used directly:

http://www.postech.ac.kr/ee/paust/p...a Beam-Profile Monitor for Cyclotron MC50.pdf

Interesting; that's vaguely like what I'm tying to do. But I can't
find a datasheet on the JU421, even on the Vishay site, so I assume
it's another FormerFet.
www.vishay.com/docs/70248/70248.pdf

Cheers,
James Arthur

Cool. Vishay didn't recognize "JU421".

Nice part for leakage, but with En=30 nv/rthz and Gm=200 uS, it's sort
of the Marching Band of Noise. This guy adds a second one in
differential mode to add bonus noise and lower the gain to boot.

Well, good luck finding one :)

If you do it'll probably have "DoD pricing".

Why do these scientific types so love differential jfet front-ends?

In electronics they tend to hang on to stuff much longer than we do.

[...]

Nothing can touch them on noise performance, although differential
input costs noise and power consumption.


Best regards,
Spehro Pefhany
 
J

James Arthur

Jan 1, 1970
0
John said:
John said:
On Sat, 02 May 2009 21:29:47 GMT, James Arthur

John Larkin wrote:
On Sat, 02 May 2009 13:14:40 -0700, Joerg

John Larkin wrote:

[...]

This weekend I'm going to buy a tin of Danish sugar cookies, eat all
the cookies, and build a pA range diode-curve measuring setup into it.
I have a bunch of LMC6001s (Ib is 10 fA typ) to use. The datasheet
hints somewhere in one of the app circuits that Cin is big. But it's
not specified anywhere.

It's "Danish Butter cookies" :)

Make sure to _throughly_ clean it before you build anything into it,
until wiping shows no grease residue. Else you might experience an
impressive meeting of ants in there. BTDT.


The rule seems to be that if Cin is large, don't mention it on the
datasheet.

Cin is probably largish and usually not used directly:

http://www.postech.ac.kr/ee/paust/p...a Beam-Profile Monitor for Cyclotron MC50.pdf
Interesting; that's vaguely like what I'm tying to do. But I can't
find a datasheet on the JU421, even on the Vishay site, so I assume
it's another FormerFet.
www.vishay.com/docs/70248/70248.pdf

Cheers,
James Arthur
Cool. Vishay didn't recognize "JU421".

Nice part for leakage, but with En=30 nv/rthz and Gm=200 uS, it's sort
of the Marching Band of Noise. This guy adds a second one in
differential mode to add bonus noise and lower the gain to boot.
Well, good luck finding one :)

If you do it'll probably have "DoD pricing".

Why do these scientific types so love differential jfet front-ends?
In electronics they tend to hang on to stuff much longer than we do.

[...]

I hadn't paid much attention to discrete jfets in, well, some decades.
They have low transconductance, high capacitances, miserable matching,
and astounding datasheet parameter spreads, like 10:1 Idss limits.

But for really low noise (like, under 1 nV/rthz) amplification of
high-impedance signals, nothing can touch them. The best fet opamps
are, like, 5x as noisy as the best jfet. Pity that most people, at
least the scientists, trend to use them wrong.

John

I may have a wacky idea. What's your signal bandwidth (if you can say)?

Cheers,
James Arthur
 
Top