Femtoampere op amp in luggage - will it be degraded by luggage screening xrays?

[Dmytry]

Somewhat unusual question:

Do you think LMP7721MA would get degraded by xrays used for luggage screening? It is an electrometer opamp with extremely low input bias current:
http://www.national.com/pf/LM/LMP7721.html#Overview
Normally the electronics is not affected by the xrays but this chip is far from usual.

 

[yungman]

Femtoamp are mostly CMOS front end, look into whether CMOS is X-ray sensitive. I use femtoamp before, I don't recall any warning. If you cannot find info on CMOS in X-ray and if there is no warning on data sheet, it should not be a problem. If you worry more, call or email to the manufacturer and ask. If you still worry, carry with you on the plane.

 

[jim hardy]

it's nmos that's most sensitive.


generally speaking electronics is tougher than living things.

a computer with cmos will take a couple thousand rads.

i have never tested one of those nmos electrometer ic's

but would be astonished if you were able to measure a degradation in its performance
after any sane x-ray.
surely they get x-rayed in shipment sometimes.

try an experiment - run a roll of photographic film through a scanner and have it developed. if it's not fogged, the dose was in low millirads like <10mr...

i once subjected my TI99 computer to 1000 rads. itr ran fine during and after the exposure.

that's best i can do.
maybe somebody in a nuke plant could test one for you - i put the TI99 in our HP department's calibrator and gave it two hours at 500R/hr, running a memory test loop..

old jim

 

[Antiphon]

It will probably be fine since the low input current is achieved by active means. The typical OpAmp is degraded by thousands or tens of thousands of gamma rads. 1000 gamma rads is 100% fatal dose.

I would be very surprised if this device were degraded in any measurable way.

 

[jim hardy]

1000 gamma rads is 100% fatal dose.

for humans.

most electronics between 10,000 & 100,000.

 

[yungman]

Hey Jim, did you look at the LMP7721 spec.? I must be getting old! This one has low voltage noise down to 30nV/sqrt(Hz) at 10 Hz(below 10nV/sqrt(Hz)) at about 500Hz, GBW of 17MHz, over 10V/uS and less than 2mA supply current! Only draw back if you can call it is low supply voltage limit.

In my days, you'll be lucky to get 10fA input bias current, slow as snail, and quite high noise. This is not for the weak of heart!

It did not say the input transistor. But if it is CMOS, then op won't have a problem. But I doubted because I never seen a CMOS with such low flicker noise at 10Hz...Which again...in my days!

 

[jim hardy]

no question I'm obsolete.
last electrometer opamp i used was OPA128 for integrate/hold stage in analog controller.
http://www.ti.com/lit/ds/symlink/opa128.pdf
which was astounding to me in 2002... it'd hold to a millivolt overnight with just 1uf.
this one looks better by almost an order of magnitude.
(interestingly, the 128 datasheet mentions its use in rad-hardened equipment. it must be pretty stout in that regard. that'd be desirable for something you'd use in a radiation meter for measuring ion chamber current. i'd wager the 7721 is rad-stout also)

the tools available today's youth are amazing.

i guess that's why we're allotted threescore and ten - the world changes so much we'd be a drag on it.

next twenty years will see English language become more precise in its everyday use, to keep up with the computers.

old jim, still clinging to Fortran.

 

[yungman]

When I was doing electronmeter amp, OPAs were still belong to Burr Brown! There goes to show how long has it been for me! I guess I am slightly newer on programming...PASCAL!

I just refused to learn C++ anymore, I just don't have the interest. I had software engine did all the programmings.

Before, we never looked at big manufacturers like National, Motorola and Ti because their stuffs are mostly of lower spec. All the most exotic ones are like PMI, Comlinear ( still remember this? They specialize on high speed op-amps), Burr Brown( low bias current), Apex( high voltage) etc. Now I notice they are got bought out by Analog Devices, National etc. I don't remember who bought out Apex, I know they got bought out by another big company.

 

[Dmytry]

Ya that's some amazing op-amp... my concerns are also for the plastic packaging vs radiation.
Anyhow, I calculated that the background radiation over 10 years should give it a dose of 0.3E-6 *24*365*10 = 0.026 Gy , and i'd reasonably assume that it won't degrade in 10 years...
I can't find the luggage screening doses though. Also, the effects are not precisely same for x-rays and natural background radiation. x-rays fog film a lot more effectively (at doses where the natural radiation will trigger a few grains in the film, x-ray can trigger every grain). I guess hand luggage should be safe enough.

I am indeed planning to make a radiation detector with it, but only meant to be used near background radiation level (or i'd use something waaay less sensitive)

 

[jim hardy]

Ya that's some amazing op-amp... my concerns are also for the plastic packaging vs radiation.
Anyhow, I calculated that the background radiation over 10 years should give it a dose of 0.3E-6 *24*365*10 = 0.026 Gy , and i'd reasonably assume that it won't degrade in 10 years...
I can't find the luggage screening doses though. Also, the effects are not precisely same for x-rays and natural background radiation. x-rays fog film a lot more effectively (at doses where the natural radiation will trigger a few grains in the film, x-ray can trigger every grain). I guess hand luggage should be safe enough.

I am indeed planning to make a radiation detector with it, but only meant to be used near background radiation level (or i'd use something waaay less sensitive)

i think that were airport scanners high level there'd be lots more shielding around them than i have noticed.

i paid careful attention to PCB layout, guarding my input tracks as advised in datasheet appnotes and keeping them short and far away from power traces even on other layers. Polypropylene capacitors work well for low leakage applications and are cheaper than polycarbonate, or at least they were a few years ago.

good luck with your project. What are you using for a detector? ion chamber? GM tube? something new?

old jim

 

[Dmytry]

i think that were airport scanners high level there'd be lots more shielding around them than i have noticed.

i paid careful attention to PCB layout, guarding my input tracks as advised in datasheet appnotes and keeping them short and far away from power traces even on other layers. Polypropylene capacitors work well for low leakage applications and are cheaper than polycarbonate, or at least they were a few years ago.

good luck with your project. What are you using for a detector? ion chamber? GM tube? something new?

old jim

Just an ion chamber, I'm hoping to be able to detect individual cosmic ray tracks. Yes, the guard line is very important there... i'll connect guard to two n/c wires on the package as well. Also I'm thinking of bending the input pins upwards and hanging the sensitive circuitry in the air. The circuit will be in a small metal box inside the ionization chamber, with wire sticking out. I'll use the capacitance of the ionisation chamber as the capacitor.
I actually have 4 of those chips so I plan to build several different circuits. In one i want to use very high value resistor, in other a mechanical switch to discharge the capacitor. I can use a tiny galvanometer as actuator for the switch. I'll be discharging actively (connecting the negating input of amplifier to the amplifier's output).

One really sweet thing about this opamp is that it got back-to-back diodes between inputs. I.e. you can have this ultra sensitive input hanging in the air and use it as a fairly robust electroscope, and it won't get zapped.

 

[yungman]

Just an ion chamber, I'm hoping to be able to detect individual cosmic ray tracks. Yes, the guard line is very important there... i'll connect guard to two n/c wires on the package as well. Also I'm thinking of bending the input pins upwardsGood idea. Guard trace only put the potential close to the input, but any residue of flux will still create a conduction path. We also use ultra sound freon bath for cleaning the boards. and hanging the sensitive circuitry in the air. The circuit will be in a small metal box inside the ionization chamber, with wire sticking out. I'll use the capacitance of the ionisation chamber as the capacitor.
I actually have 4 of those chips so I plan to build several different circuits. In one i want to use very high value resistor, in other a mechanical switch to discharge the capacitor. I can use a tiny galvanometer as actuator for the switch. I'll be discharging actively (connecting the negating input of amplifier to the amplifier's output).

One really sweet thing about this opamp is that it got back-to-back diodes between inputs. I.e. you can have this ultra sensitive input hanging in the air and use it as a fairly robust electroscope, and it won't get zapped.

Don't trust the back to back diode inside the opamp that much. They are very weak.

 

[jim hardy]

air insulation works and is suggested in some application notes.

sounds like a fun project.

I can use a tiny galvanometer as actuator for the switch...

darn - i just looked at a mil-style sealed -50 0 +50 microamp galvanometer in a surplus store yesterday... had two contact outputs and was only ten bucks. It was at "Pappy's" surplus near Goldsboro NC... if i'd known you needed one i'd have got it.

 

[Dmytry]

Don't trust the back to back diode inside the opamp that much. They are very weak.

I'll connect some 100K..1M resistor in series. That should do it. With the ionisation chamber the issue is that the input may get shorted onto 100v supply.

The guards are really neat idea. I'm going to make guarded posts inside the ion chamber and suspend inner electrode using a fishing mono-filament tied to those posts... if i end up using micro-controller in this meter, I can make it vary voltage on the guard to measure guard-to-input resistance (and contact voltage) and compensate for it.

Re: galvanometer, i was thinking of the kind that used to be common as audio level indicators. I have various old soviet electronics for really cheap. It's not critical, I can make electromagnet that'd short the connection, or reed switch. Thinking about it, with those back to back diodes it is not important to keep input shorted when the device is turned off, so i would use normally-open switch that i'd close momentarily to zero the input. So the power consumption of the switch will be low in any case. I'll put the switch between the output and negating input (which will be connected to the inner electrode in ion chamber). Eventually I want to use microcontroller to process the data and do the switching as needed, and send the data to computer for logging, as well as display digital readout (and the standard deviation for the noise so i know if something is statistically significant).

 

[yungman]

I'll connect some 100K..1M resistor in series. That will do it. Use a longer body resistor so static HV less likely to creep over the body surface.That should do it. With the ionisation chamber the issue is that the input may get shorted onto 100v supply.

The guards are really neat idea. I'm going to make guarded posts inside the ion chamber and suspend inner electrode using a fishing mono-filament tied to those posts... if i end up using micro-controller in this meter, I can make it vary voltage on the guard to measure guard-to-input resistance (and contact voltage) and compensate for it.

Re: galvanometer, i was thinking of the kind that used to be common as audio level indicators. I have various old soviet electronics for really cheap. It's not critical, I can make electromagnet that'd short the connection, or reed switch. Thinking about it, with those back to back diodes it is not important to keep input shorted when the device is turned off, so i would use normally-open switch that i'd close momentarily to zero the input. So the power consumption of the switch will be low in any case. I'll put the switch between the output and negating input (which will be connected to the inner electrode in ion chamber). Eventually I want to use microcontroller to process the data and do the switching as needed, and send the data to computer for logging, as well as display digital readout (and the standard deviation for the noise so i know if something is statistically significant).

One other thing you did not mention whether the input go through coax like some of the electronmeters. You need to use rigid coax if possible. Even double shield coax will create noise from vibration even if you don't move it. In our spectrometer, the vibration of the vacuum pumps causing problem and we had to switch to rigid coax.

 

[jim hardy]

""I'll put the switch between the output and negating input... ."

as yungman said, input is not bulletproof so put a resistor to limit current when close switch.

old jim

 

[Dmytry]

One other thing you did not mention whether the input go through coax like some of the electronmeters. You need to use rigid coax if possible. Even double shield coax will create noise from vibration even if you don't move it. In our spectrometer, the vibration of the vacuum pumps causing problem and we had to switch to rigid coax.

I'll be placing the circuit inside the ion chamber can... i want to keep it simple and i won't be measuring high dose rates so I don't need to keep electronics safely away from the chamber.
edit: A question though: how do you deal with huge input capacitance when you have coax? Sounds like it would massively increase the time constant.

resistor: yes, of course. I'll not even have anything connect directly to input I think, not interested in high frequency operation, and it'll be convenient to just solder the resistor straight to the input leg in the air.
Though this chip specifies ESD protection for 2000v human body model and 200v machine model, which is fairly amazing for something like this.

 

[yungman]

I'll be placing the circuit inside the ion chamber can... i want to keep it simple and i won't be measuring high dose rates so I don't need to keep electronics safely away from the chamber.
edit: A question though: how do you deal with huge input capacitance when you have coax? Sounds like it would massively increase the time constant.

resistor: yes, of course. I'll not even have anything connect directly to input I think, not interested in high frequency operation, and it'll be convenient to just solder the resistor straight to the input leg in the air.
Though this chip specifies ESD protection for 2000v human body model and 200v machine model, which is fairly amazing for something like this.

They are all slow, your feedback resistor is going to be in 10EE11 Ω. The input cap not just slow things down, it create more noise as the apparent gain of the stage is $\frac {Z_f}{Z_{in}} \;\hbox{where }\; Z_{in} \;$ is from the input capacitance and decrease with frequency. As frequency going up, the closed loop gain of the op-amp rise and the output noise increase. Depend on the the speed requirement. The main way of dealing with this is a small cap across the feedback resistor. With this Z_f will decrease with frequency to keep the closed loop gain constant to keep the noise down. But yes, this will slow down the stage. Consider yourself lucky that this amp has 17MHz of GBW. We dealt with slow amp in old days.

One way is to put a voltage devider using lower value resistor at the output of the op-amp and the divider output connect the large feedback resistor to the summing junction. Same if you have a 1/10 divider, your feedback resistor can be decrease to 1/10. The problem with this is your offset error increase by 10 times and noise increase by `10 times at low frequency.

 

[DragonPetter]

It will probably be fine since the low input current is achieved by active means.

Could you or someone else in here explain a little more about achieving the low input current by active means? How is this done more specifically?

All the different transitor and diodes I can find have leakage currents in the nano to micro amp range, so I am curious how it can be cut down to 1 or 2 femtoamps. Also, since this is done actively, wouldn't it reduce the bandwidth of the amplifier?

I was thinking about somehow using a current mirror to reduce the leakage current on a critical component, is this the method they use?

 

[yungman]

Don't know this specific amp, but they did use current source to compensate for bias current.

CMOS input bias current are below 1pA easily.