Extremely sensitive electrometer for Mass Spectrometry detection

Did you isolate the inputs as suggested in the datasheet?  We have used teflon insulated terminals for OPA128/9 inputs nodes with success.  Are your inputs so low that you could or might need to operate in pulse counting mode?  Mass spectrometers often use electron multipliers to amplify the signal to a better operating range for an electrometer, or if the rates are low operate in pulse mode using something like the Amptek A101 discriminator.

Good luck

Derek;

The OPA129 electrometer-grade op amp has very low input bias current as is, but you can reduce it further by cooling the device with a small Peltier cooler. If the cold side of the cooler is placed on top of the package and bonded in place with a good low-outgassing thermally-conductive epoxy and the hot side is bonded to a copper strip that can connect somewhere to your vacuum chamber to conduct the heat away, the OPA129 can probably be cooled to -40C.

Another device to consider is the INA116   . It has an extremely low input bias current, on the order of 5 femtoamps.

Derek;

I just noticed that the equation in Figure 4 of the OPA129 data sheet is wrong. It should be: Vo = -10V/nA  and Vo = -Iin * Rf (1+ 18k/2k)

The Tee network in the feedback was overlooked.

I did put the ground loop around the input as suggested in the datasheet.  However the Faraday cup is not isolated from the pcb - not sure if this would be significant since there's a ground loop around the input anyway?  It would be possible to isolate the cup from the pcb with a teflon sleeve and run a wire directly to the input pin of the OPA129 without touching the board.  My current design uses a trace on the pcb to connect the input.  I have a relatively small number of ions to work with (hence the desire for better sensitivity and LOD) but the signals are easily visible on an electron multiplier in analog mode.  I would like to replace the multiplier with a solid state detector that doesn't have pressure limitations but it's not a trivial problem when you lose the huge gains of the electron multiplier

The Peltier cooler is an interesting idea but it would make size/weight/power reduction more difficult.  Would cooling the OPA129 make a significant difference in the noise levels or sensitivity?  I just looked at the INA116 and it seems like the circuit would need to be designed differently.  There doesn't seem to be info in the datasheet about a recommended configuration to use the INA116 as an electrometer (vs the OPA129 datasheet which has a diagram)

That will definitely be useful, I guess I was calculating input currents with the wrong equations previously.  I always wondered why the other resistors didn't show up in the equations but assumed they didn't have much of an effect

Derek;

Yes, A Peltier cooler would be somewhat of a PITA but when dealing with Heroic measures for reducing input bias current it is worth looking at. By cooling the INA129, its input bias current is reduced as is its current noise; if you also cool the high value feedback resistor, its thermal noise is also reduced.

An INA116 would function as a voltage follower, measuring the voltage drop across a multi-gigohm resistor. This approach is only practical with an amplifier that has extremely low input bias current.

Hi Neil,

I am working in the field of ion mobility spectrometry, in which ions at ambient pressure are detected with a Faraday plate. Traditionally, people just use electrometer circuit like what you have talked above (OPA129 with large resistor feedback for high gain). But we tried capactive transimpedance amplifier circuit, like burr brown IVC102U with a 1pF capacitor feedback. It seems to work better with smaller noise and better detection sensitivity. 

Since you are a world world renowned experted who have worked in burr brown for so many years. Could you please give me some insights on this topic. By the way, I am a graduate student at Univ of Arizona.

Best Regards,

Hans

Xingzhi,

Neil will likely have more to add on this topic, but I thought I would give some input on the noise advantages of a capacitive feedback transimpedance amplifier such as the IVC102 like you mentioned. The traditionally large feedback resistance of a standard transimpedance amplifier can be a major source of noise. All resistors produce noise due to the random motion of electrons through the bulk resistive material which is commonly referred to as thermal noise. A graph for the noise spectral density versus the value of a resistor is shown below as well as the equation for calculating the noise spectral density from a known resistance (excerpt from a presentation by Art Kay):

You can see from the graph that large resistor values can be the dominant noise source in a system. Some types of resistors (thick film surface mount is one example) also produce additional noise above their theoretical value. The capacitor that is used to integrate the charge on the IVC102 is comparatively noise-free (although most likely not ABSOLUTELY noise free). A very good article series on noise in amplifiers was written by our own Art Kay, the first article in the series can be found here:

http://www.analogzone.com/avt_0904.pdf

Thank you very much for teaching me this. And I have another question with IVC 102 as a capacitive transimpedance amplifier.

For the minimum capacitor I can choose with IVC 102 is the on chip 10pf capacitor. However, my pulse of ion current is so small that 10pf is too large. My measurement is just thousands of charges per 20 ms, so I need high gain through small capacitor and do not need large full well capacity from 10pF capacitor. 

Then it would be really good if I could have a small integrating capacitor ( several hundred fF) for me. Unfortunately, IVC 102 does not have a on chip integrating capacitor on the order of 0.1pF.

So do you have any ideal how can I overcome this problem?

Xingzhi,

Recall that capacitors in series sum reciprocally (like resistors in parallel) so placing a smaller capacitor in series with one of the on-chip feedback capacitors is one possible way to overcome this limitation. For example, by placing a 1 pF capacitor in between pins 3 and 4 and leaving pins 5 and 6 disconnected the total feedback capacitance is now (1 * 10) / (1 + 10) = 909 fF.

There will be some difficulties with this approach however if high precision is required. To start, a tight tolerance will be required on the capacitor placed in series with the on-chip capacitor. For a 1pF capacitor, the tolerance on this value may be at least 50 to 100 fF. A more difficult issue will be the parasitic capacitances introduced by the PCB layout and between the pins of the IVC102 itself. The on-chip switches of the IVC102 will also inject a small quantity of charge when they are opened and this can contribute errors to your measurement.  Rather than try to theoretically calculate the values for all the variables involved, it may be a better approach to develop a calibration routine which measures the voltage response of the integrator to a known quantity of input charge.

I would use a current source to charge a known capacitor to a certain voltage, then connect this capacitor to the input of your integrator and measure the change in the output voltage. You should be able to use Q = C*V to determine the input charge and the feedback capacitance for your integrator.

Hi John,

Thank you for proposing the capacitors in series solution.  So I just wonder if you can just put that 1pF capacitor between terminal 3 and terminal 10, while not connecting pin 4,5,6 at all (not using the on chip capacitors). I think both methods will work, but which one is better in terms of smaller noise and better gain?

For the parasitic capacitance, I would clearly want to minimize it. A on chip 0.1pF capacitor would clearly be the best to minimize stray capacitance and noise. But since ivc102 does not have it, then what would be the best way to minimize the parasitic capacitance if using an external capacitor for integration.

For the calibration, I think it is practical and thank you for your suggestion.

Another question, is there any capacitive trans-impedance amplifier like IVC102 but with a smaller on chip integrating capacitor (<10pF) to you knowledge?

Best Regards,

Xingzhi