Maker Pro
Maker Pro

PMT operating voltage, who knew?

G

George Herold

Jan 1, 1970
0
So someone was rewriting our manual for using a PMT.
(Used in photon counting mode.)
There was this confused procedure to set the discriminator. (The
threshold level on the comparator before counter.) So I said let’s
just take the data, and do it right. I took a PMT off the shelf that
had a fair amount of dark current (3nA at 1000V) We looked at count
rate with a light on versus the dark count rate at various
discriminator setting and PMT voltages.
Here’s a plot of the data,
http://bayimg.com/CAlfEaaEb

I must admit to being surprised. I’d always heard/ been told to
operate the PMT near the voltage recommended by the manufacturer
(1000 volts in this case.)

I said to my boss that the best way to find the largest signal to
noise is to set the discriminator to a tad above zero, then reduce the
voltage till the count rate drops to a bit more than 1/2 it maximum
value. I did this the voltage was 537 and the light on/ dark ratio
was 455...more than 5 times better than at 1000V!

George H.
 
G

George Herold

Jan 1, 1970
0
I wonder if the voltage across the first stage or three (near the
photocathode) dominate the s/n. Later stages probably just amplify, so
maybe you could crank them up.

What sort of dark count rates were you seeing?

--

John Larkin         Highland Technology, Inc

jlarkin at highlandtechnology dot comhttp://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro   acquisition and simulation- Hide quoted text -

- Show quoted text -

The power supply is a all in the socket thingie made by Hamamatsu.
It's a Ccokroft Walton ladder so I can't really change the divider
ratio.
The maximum count rate with light one is ~10k/sec. The dark count is
all over the map, 3k to 10's. If I pick the discriminator setting
that gives the best ratio (~3-4k light counts per second.) Then dark
counts are about

1000V 60/sec
800V 30/sec
700V 20/sec
600V ~14/sec
537V 9/sec.

(Like I said I picked a noisy pmt off the shelf.. the range of
reported dark currents was from 5nA to 0.05nA I haven't looked at one
with a low dark current.)

The only thing I could find on the web that hinted at this is a graph
of dark current vs voltage by hamamatsu... It showed a dark current
'turning on' at ~400V.

It's not really a problem or anything, I just never knew... and if
someone was digging for a signal in the noise then it would be
useful.

George H.
 
R

rickman

Jan 1, 1970
0
So someone was rewriting our manual for using a PMT.
(Used in photon counting mode.)
There was this confused procedure to set the discriminator. (The
threshold level on the comparator before counter.) So I said let’s
just take the data, and do it right. I took a PMT off the shelf that
had a fair amount of dark current (3nA at 1000V) We looked at count
rate with a light on versus the dark count rate at various
discriminator setting and PMT voltages.
Here’s a plot of the data,
http://bayimg.com/CAlfEaaEb

I must admit to being surprised. I’d always heard/ been told to
operate the PMT near the voltage recommended by the manufacturer
(1000 volts in this case.)

I said to my boss that the best way to find the largest signal to
noise is to set the discriminator to a tad above zero, then reduce the
voltage till the count rate drops to a bit more than 1/2 it maximum
value. I did this the voltage was 537 and the light on/ dark ratio
was 455...more than 5 times better than at 1000V!

George H.

What that a metric "tad" or an imperial "tad"?
 
T

tm

Jan 1, 1970
0
Phil Hobbs said:
Interesting plot!

Looks like you have a lot of ion events, or maybe light emission being
piped up the walls of the tube to the PC. Reducing the electron energy
and the average dynode current will reduce both.

Cheers

Phil Hobbs

--

He maybe?

Are the tubes old?
 
G

George Herold

Jan 1, 1970
0
Interesting plot!

Looks like you have a lot of ion events,
Yeah maybe... looking at the 'scope and low rate dark counts
there are a number of 'big' events. (maybe 1/3?)
(I can see why you might want an upper window level on the
discriminator.)
I need some MCA...


or maybe light emission being
piped up the walls of the tube to the PC.  Reducing the electron energy
and the average dynode current will reduce both.

I'm not sure what you mean... light from the tube base/socket going up
the side of the PMT? There was a lot of effort put into keeping light
out of the PMT space. (work not done by me, I do know that some BNC
sockets are better than others, and then we paint the ends with
epoxy. And shouldn't stray light 'dark count' look like real light in
terms pulse height?

I'll have to look at a low noise pmt.
(tomorrow should be a quite day.)
George H.
 
G

George Herold

Jan 1, 1970
0
What that a metric "tad" or an imperial "tad"?

--

Rick- Hide quoted text -

- Show quoted text -

The discriminator is set with a ten turn pot and dial. I set it at
~0.01 turns, right at zero you start hitting the noise... Well it's
not really all noise, there's some ~1 MHz interference getting in
too... anyway right at zero the 'dark count' doubles.

George H.
 
G

George Herold

Jan 1, 1970
0
Dark current is supposed to be thermal electrons boiling off the
photocathode, so presumably higher voltage increases the chances
these get away from the photocathode and start into the multiplier
chain.  So, it doesn't surprise me at all that you got this result.
The PMT makers do sometimes show different division ratios to
optimize the tubes for linear vs. photon counting applications.

Jon

Yeah I used 'good' pmt's with coolers built in to keep down the dark
count... I never looked at how the rate changed with voltage.
There are certainly speed advantages that come with higher voltage,
but you don't always care about speed.

George H.

(Turning down the voltage is certainly easier than a cooler.)
 
G

George Herold

Jan 1, 1970
0
He maybe?

Are the tubes old?- Hide quoted text -

- Show quoted text -

No. Dark count data on box had a 2012 date.

George H.
 
B

Bill Sloman

Jan 1, 1970
0
Dark current is supposed to be thermal electrons boiling off the
photocathode, so presumably higher voltage increases the chances
these get away from the photocathode and start into the multiplier
chain. So, it doesn't surprise me at all that you got this result.

Most of it. The first dynodes are reputed to produce a bit, particularly after they've been busy - not every incident electron produces only prompt secondaries.
The PMT makers do sometimes show different division ratios to
optimize the tubes for linear vs. photon counting applications.

There's all sorts of interesting stuff that can go on. I used a zener to fix the voltage from photocathode to first dynode at close to it's maximum tomaximise the signal to noise ratio, and the RCa 8850 went further by usinga GaP semiconductor on the first dynode and some 800V between cathode and first dynode to get about 40 seconary electrons per photo-electron.

http://frank.pocnet.net/sheets/049/8/8850.pdf

It makes sense to use a zener diode between the last dynode and ground - that voltage drop doesn't add to the gain of the tube, and too much anode current can drop the voltage across this stage and increase the voltage acrossthe rest of the tube (which does affect the gain) to give you a perceptible positive non-linearity.

If you are pulling large enough currents out of the anode, you can need to raise the voltage across the last few dynodes to avoid space charge effects, and Sauerbrey thought that the a single photo-electron in the photo-cathode to first dynode space could be shown to have created a significant spacecharge.

Photomultipliers are fun devices.

Sloman, A.W. "Comment on 'Computer aided simulation study of photomultiplier tubes'", IEEE Transactions on Electron Devices, ED-38 679-680 (1991).

lists a few useful references.
 
R

Robert Baer

Jan 1, 1970
0
George said:
So someone was rewriting our manual for using a PMT.
(Used in photon counting mode.)
There was this confused procedure to set the discriminator. (The
threshold level on the comparator before counter.) So I said let’s
just take the data, and do it right. I took a PMT off the shelf that
had a fair amount of dark current (3nA at 1000V) We looked at count
rate with a light on versus the dark count rate at various
discriminator setting and PMT voltages.
Here’s a plot of the data,
http://bayimg.com/CAlfEaaEb

I must admit to being surprised. I’d always heard/ been told to
operate the PMT near the voltage recommended by the manufacturer
(1000 volts in this case.)

I said to my boss that the best way to find the largest signal to
noise is to set the discriminator to a tad above zero, then reduce the
voltage till the count rate drops to a bit more than 1/2 it maximum
value. I did this the voltage was 537 and the light on/ dark ratio
was 455...more than 5 times better than at 1000V!

George H.
Err...the 600V plot is almost two times better...
 
R

Robert Baer

Jan 1, 1970
0
John said:
I wonder if the voltage across the first stage or three (near the
photocathode) dominate the s/n. Later stages probably just amplify, so
maybe you could crank them up.

What sort of dark count rates were you seeing?
Excellent point; treat the PMT as a series of (noisy) amplifiers;
that implies the first stage is the most critical,with the second stage
contributing (to S/N) by about the inverse of its gain (i think).
 
B

Bill Sloman

Jan 1, 1970
0
I'm not sure what you mean... light from the tube base/socket going up
the side of the PMT? There was a lot of effort put into keeping light
out of the PMT space. (work not done by me, I do know that some BNC
sockets are better than others, and then we paint the ends with
epoxy. And shouldn't stray light 'dark count' look like real light in
terms pulse height?

I'll have to look at a low noise pmt.
(tomorrow should be a quite day.)

Cambridge Instruments bought selected high gain photomultipliers so that they never ended up with more than 1000V across the glass face-plate - between the photocathode and the external metal work.

Above 1000V you got electrical conduction through the glass, which was generating electro-luminescence, which showed up as extra dark current at high gain settings.

Regular insulators between the glass and the metal-work didn't help - they weren't as resistive as the glass, so the voltage drop across the glass stayed at 1000V. We could have done better, but paying a bit extra for selected high-gain tubes was the cheapest option.
 
R

Robert Baer

Jan 1, 1970
0
George said:
The power supply is a all in the socket thingie made by Hamamatsu.
It's a Ccokroft Walton ladder so I can't really change the divider
ratio.
The maximum count rate with light one is ~10k/sec. The dark count is
all over the map, 3k to 10's. If I pick the discriminator setting
that gives the best ratio (~3-4k light counts per second.) Then dark
counts are about

1000V 60/sec
800V 30/sec
700V 20/sec
600V ~14/sec
537V 9/sec.

(Like I said I picked a noisy pmt off the shelf.. the range of
reported dark currents was from 5nA to 0.05nA I haven't looked at one
with a low dark current.)

The only thing I could find on the web that hinted at this is a graph
of dark current vs voltage by hamamatsu... It showed a dark current
'turning on' at ~400V.

It's not really a problem or anything, I just never knew... and if
someone was digging for a signal in the noise then it would be
useful.

George H.
Use a plain socket and put in your own resistor / regulator chain,
just like many did 30-40 years ago.
 
G

George Herold

Jan 1, 1970
0
Most of it. The first dynodes are reputed to produce a bit, particularly after they've been busy - not every incident electron produces only prompt secondaries.


There's all sorts of interesting stuff that can go on. I used a zener to fix the voltage from photocathode to first dynode at close to it's maximum to maximise the signal to noise ratio, and the RCa 8850 went further by using a GaP semiconductor on the first dynode and some 800V between cathode and first dynode to get about 40 seconary electrons per photo-electron.

http://frank.pocnet.net/sheets/049/8/8850.pdf

Nice data sheet, too bad the dark current graph doesn't go to lower
voltages.
figure 9 here shows dark current 'turning on' at 600-700 volts.

www.bo.infn.it/ams/Hamamatsu-PMT.pdf

It makes sense to use a zener diode between the last dynode and ground - that voltage drop doesn't add to the gain of the tube, and too much anode current can drop the voltage across this stage and increase the voltage across the rest of the tube (which does affect the gain) to give you a perceptible positive non-linearity.

If you are pulling large enough currents out of the anode, you can need to raise the voltage across the last few dynodes to avoid space charge effects, and Sauerbrey thought that the a single photo-electron in the photo-cathode to first dynode space could be shown to have created a significant space charge.

Photomultipliers are fun devices.

Did you ever look at the dark current at low voltages?

George H.
 
G

George Herold

Jan 1, 1970
0
   Err...the 600V plot is almost two times better...- Hide quoted text -

- Show quoted text -

Sorry I cut the top off the graph and the single data point at 537
volts was lost.
The light /dark ratio was 455 compared to ~70 at 1000V.
The pulses are only a few mV high basically lost in the noise... I
can't get the 'scope to trigger on them, the comparator finds them
just fine though. Looking at the 'scope display I'd never have
guessed that the light/dark count ratio would be best there.

George H.
 
G

George Herold

Jan 1, 1970
0
So someone was rewriting our manual for using a PMT.
(Used in photon counting mode.)
There was this confused procedure to set the discriminator.  (The
threshold level on the comparator before counter.)  So I said let’s
just take the data, and do it right.  I took a PMT off the shelf that
had a fair amount of dark current (3nA at 1000V) We looked at count
rate with a light on versus the dark count rate at various
discriminator setting and PMT voltages.
Here’s a plot of the data,http://bayimg.com/CAlfEaaEb

I must admit to being surprised.  I’d always heard/ been told to
operate the PMT near the voltage recommended by the manufacturer
(1000 volts in this case.)

I said to my boss that the best way to find the largest signal to
noise is to set the discriminator to a tad above zero, then reduce the
voltage till the count rate drops to a bit more than 1/2 it maximum
value.  I did this the voltage was 537 and the light on/ dark ratio
was 455...more than 5 times better than at 1000V!

George H.

OK I got a low noise pmt off the shelf.
Here's it's plot. Dark count rates are ~ 10 times lower.
http://bayimg.com/cAlkeAaeB
(There should be some big error bars on some of the data points... I'm
getting rates of only 1/sec or so... (ten second counting window.)

George H.
 
T

tm

Jan 1, 1970
0
So someone was rewriting our manual for using a PMT.
(Used in photon counting mode.)
There was this confused procedure to set the discriminator. (The
threshold level on the comparator before counter.) So I said let’s
just take the data, and do it right. I took a PMT off the shelf that
had a fair amount of dark current (3nA at 1000V) We looked at count
rate with a light on versus the dark count rate at various
discriminator setting and PMT voltages.
Here’s a plot of the data,http://bayimg.com/CAlfEaaEb

I must admit to being surprised. I’d always heard/ been told to
operate the PMT near the voltage recommended by the manufacturer
(1000 volts in this case.)

I said to my boss that the best way to find the largest signal to
noise is to set the discriminator to a tad above zero, then reduce the
voltage till the count rate drops to a bit more than 1/2 it maximum
value. I did this the voltage was 537 and the light on/ dark ratio
was 455...more than 5 times better than at 1000V!

George H.

OK I got a low noise pmt off the shelf.
Here's it's plot. Dark count rates are ~ 10 times lower.
http://bayimg.com/cAlkeAaeB
(There should be some big error bars on some of the data points... I'm
getting rates of only 1/sec or so... (ten second counting window.)

George H.
+++++++++++++++++++++++++++++++++++++++

Are you operating the PC at ground potential? I.e. positive HV?

I have found that having all the shielding at the same potential as the PC
helps lower the noise.

tm
 
G

George Herold

Jan 1, 1970
0
One other thing that occurred to me--the number of light counts is
decreasing as well, so ISTM that either
(a) you're effectively losing photocathode area due to regions where the
field is too weak to make the primaries hit the first dynode,
(b) the first dynode gain has dropped low enough that there's a
significant chance of getting zero secondaries, or
(c) the field at the PC surface is too weak to extract all the primaries
from the PC.

If (c) is the case, it might suppress thermionic emission more than
photoelectrons, provided that the photoelectrons haven't entirely
thermalized by the time they reach the PC surface, which I'm pretty sure
they haven't.

Hmm the number of light counts is prety much constant except at the
lowest voltage (537 V) With the discriminator set at zero (probably
about 1mV) I see,

voltage light counts dark counts
(in 10 seconds)
537 40.5k 89
600 78.8k 1k
700 92.1k 4.4k
800 100k 9.7k
1000V 135k 31k

I get the best light to dark ratio when the discriminator is set such
that the light count rate is down to 1/2 to 1/3 it's maximum
value.

Does this tube have a NEA photocathode?

NEA photocathode?
Here's a data sheet...
javascript:eek:penreq('http://www.datasheetcatalog.org/datasheet/
hamamatsu/R212.pdf')
(well maybe not sure that will work)

Don't spend any skull sweat on this on my account. It's pretty much a
done deal.. just getting data for a manual re-write. (And scratching
my head.)

It doesn't seem that photoexcited electron's from the photocathode can
have that much more energy than the thermal ones... all green photons
so a maximum energy of a volt or two. Would that make such a big
difference?

George H.
 
G

George Herold

Jan 1, 1970
0
OK I got a low noise pmt off the shelf.
Here's it's plot.  Dark count rates are ~ 10 times lower.http://bayimg.com/cAlkeAaeB
(There should be some big error bars on some of the data points... I'm
getting rates of only 1/sec or so... (ten second counting window.)

George H.
+++++++++++++++++++++++++++++++++++++++

Are you operating the PC at ground potential? I.e. positive HV?

I have found that having all the shielding at the same potential as the PC
helps lower the noise.

tm- Hide quoted text -

- Show quoted text -

No the anode is near ground... that's a lot easier for the pulse amp
electronics.
(It's a CFA run as a current to voltage converter.... I didn't quite
understand how my circuit was working till I tried to get a bit more
gain out of it, then I had to give myself a dope slap.)

George H.
 
B

Bill Sloman

Jan 1, 1970
0
No the anode is near ground... that's a lot easier for the pulse amp
electronics.
(It's a CFA run as a current to voltage converter.... I didn't quite
understand how my circuit was working till I tried to get a bit more
gain out of it, then I had to give myself a dope slap.)

Note my second contribution to this thread yesterday. The PMT where we
started getting extra dark current when the photocathode was more than
1KV negative than the metalwork on the othere side of the glass was a
EMI tube, not Hamamatsu - so the leakage current through the glass
might not be generating electroluminescence or might start to generate
it at different voltage difference, depending on the transition metal
impurities in the glass used to make the tube envelope, the thickness
of the glass and the phase of the moon.
 
Top