I need to build a very sensitive current amplifier for an ion detection system. The detector output current is very low, approximately 100 fA. The required bandwidth is 50 kHz.
The bandwidth implies that a typical transimpedance amplifier will not work. So I'm considering a switched integrator amplifier. Something like the IVC102 is what I have in mind but with a smaller feedback capacitor (e.g. 0.1pF). I have not found an integrated solution like this.
The LMP7721 looks appropriate for a discrete solution given its very low input bias current. In addition, I need a means to discharge the feedback capacitor that has very low leakage current and parasitic capacitance.
Can anyone suggest relevant application notes? How about an appropriate discharge circuit?
Has anyone tried modifying the LMP7721 Multi-function Evaluation Board to work in this configuration?http://www.ti.com/lit/ug/snou004/snou004.pdf
Does anyone know of an integrated solution like the IVC102 but with much smaller feedback capacitance?
Thanks.
Andy
Hi,
Slightly off topic. I'm trying to build a very sensitive EMG Amplifier using dry electrodes (capacitive). I think the circuits for that and your ion detector are quite similar considering the front end. In fact some researchers (Prance et. al) have even applied it to both applications.
Though much literature does state the need for capacitor discharge by placing a very high bias resistor, many authors suggest that this just adds noise. (Dry-Contact and Noncontact Biopotential Electrodes: Methodological Review, Chu et. al) and suggest that the ESD protection circuitry may suffice for the leakage. For lower capacitance, everyone suggests guarding techniques.
I have actually been looking at the exact same amplifiers as you mentioned. What do you think of LPV521, LMC6441? In case of LMP7721, what do you think of LMP2012 instead?
Hi Andy,
A similar thread recently popped up asking a similar question (http://e2e.ti.com/support/amplifiers/precision_amplifiers/f/14/p/74262/689819.aspx#689819) one possible solution I proposed was placing a smaller capacitor external to the IVC102 in series with the on-chip capacitor.
The LMP7721 is an incredible product, but as you mention, I believe that the major source of error in a discrete application will be the parasitics of the components and the pcb layout itself. In fact, I think that charge injection from the switching devices will totally dominate the integrated charge in the circuit. Also, as I stated in the above thread, the tolerances on discrete capacitors with values that small will be a major source of error and I recommend a calibration routine.
In order to deal with the issue of creating a precise feedback capacitance, I would build a parallel plate capacitor into the PCB itself. If you know the thickness of your pcb, and its relative permittivity, then it should be possible to calculate the area of copper plans that could be placed on the top and bottom layers of the pcb to form a capacitor.
I will have to ponder awhile on ultra-low leakage methods to discharge the capacitor (perhaps resident guru Bruce Trump can chime in on this one too). Research, by definition, takes place in uncharted waters so it is difficult for us to reach into our pockets and pull out a circuit like this!
John Caldwell
Analog Applications Engineer
PA Linear Apps
John,
Thank you for your reply.
Our application is related to the post you mentioned. Basically, this is time-resolved mass spectrometer. The ion flux is weak and very dynamic. The nature of our instrument limits us to a simple Faraday cup detector which offers no inherent gain. Thus, all of the necessary gain must be provided by the amplifier connected to the Faraday cup.
We're trying to measure currents of approximately 100 fA with a bandwidth of 50 kHz. Given the noise and GBW of the IVC201 it seems that the signal would be lost in the noise even with a small feedback capacitance (e.g. 1 pF).
The instrument runs in bursts. Currents must be measured for 5 ms once every 50 ms. So there is plenty of time to reset the feedback capacitor between bursts. A reset before each burst would not be required given the low bias current of the LMP7721. The integrator's output would be differentiated so I'm not too concerned about baseline drift or the variability of the charge injection.
Thanks again.
First, its awesome that you use the term "ion flux" in your post, that may be a first on e2e!
Thank you for clarifying the details of your application. I still see a discrete implementation of a switched integrator being a difficult task due to the nature of the FETs involved. For example, most FETs will have gate-to-body and drain-to-source leakage currents that will be orders of magnitude greater than the input bias currents of the LMP7721. Also, the output capacitances of these devices can be 10s of picofarads which would reduce the integrator's sensitivity.
I'm guessing that a literature review on this topic hasn't turned up any possible designs?
Update:
The book "The Art of Electronics" by Paul Horowitz has an interesting circuit topology to deal with the switch leakage in this type of application, it is on page 224 in Figure 4.50. However, this does not solve the issue of the output capacitance of the FETs reducing the integrator sensitivity but it might be an interesting place to start.
I'm doing some research into an alternate method of resetting the integrator and will update this thread when I've a solution drawn up.
We like the same books! I found that circuit, as well. Thanks for your help.
John Caldwell I will have to ponder awhile on ultra-low leakage methods to discharge the capacitor (perhaps resident guru Bruce Trump can chime in on this one too). Research, by definition, takes place in uncharted waters so it is difficult for us to reach into our pockets and pull out a circuit like this!
I was wondering if there has been a resoltuion to the reset issue as I have a circuit (designed by someone else, long ago in a far away place) that we have been using on an integrator built around the LMC6001. I think it might work in this case, but I am also interested in getting someone else's opinion of it.
If any interest, let me know!