Open circuit ground multiplexing?

[thekeystoneclimber]

Hi all.  I am experimenting with electrospray ionization.  The setup has a 4kV power supply connected to an electrode that is separated from a return lug (or ground plate) via an air gap, followed by an inline pico-ammeter and finally ground.  The way this works is that a small drop of solvent is applied to the electrode, the 4kV power supply is energized, and the solvent forms an aerosol electrospray "Taylor cone" between the electrode and return lug.  When this is occurring (it is not especially visible except via laser light) the inline pico-ammeter reads somewhere on the order of 20-100 nA resulting from the ion transfer. 

I would like to test a very large number of electrodes by systematically switching the pico-ammeter & ground between the various return lugs.  My first thought was to use a multiplexer, but I'm not sure if this is even possible, and I have a few concerns.  Specifically, iIs there a mux that can switch a ground reference (through the pico-ammeter) to an open (high impedance) circuit, and has a low enough on resistance as to not substantially effect the 20nA ion transfer current?

Also, every so often the air gap gets a little too small and turns into a 4kV "spark gap" which results in having to change a fuse in the pico-ammeter.  What would be the best practices with regards to circuit protection for preventing this scenario from frying the switching circuitry?

Thanks in advance!

[Kleinstein]

20 nA is not yet a very small current, so one could get away without really special parts.

One possibility would be having a small capacitor (e.g. some 10 nF) for each electrode, a mux to select the electrode and a suitable amplifier to read the charge signals one at a time and reset the capacitor to zero. With large enough caps the voltage will not rise very high and should not effect the electric field too much.
The usual MUX chips (e.g. CD4051) are good for 8 inputs and maybe a few MUX could be combined to a single charge amplifier.
Likely this would need something like a small µC for the control and ADC.

For protection one could use clamping diodes and maybe extra resistors. 20 nA is still high enough to use diodes without much extra effort.
One may need external diodes and the MUX internal diode toward the supply only as a second stage.

[duak]

Do I understand correctly that the multiplexed target electrodes' voltages are individually varied to selectively attract or not attract ions from the emitter electrode?

[thekeystoneclimber]

Individually switched to ground.

[duak]

OK then, so when the MUX channel is off, the electrode will pick up a charge from the ions and its voltage will float toward the emitter electrode's voltage thus reducing the flow of solvent?  What sort of speed is needed?

I can't think of too many semiconductor devices that would have the low leakage current and capacitance needed and yet stand off 4 KVDC.  Thermionic devices might but they are limited to electrons being the current carriers.

Wasn't something done with video camera tubes eg., vidicons and image orthicons that could read electrical charges on a surface with a scanning electron beam?

[thekeystoneclimber]

For clarification, I already have a current measuring system, so I would rather not design one from scratch if it isn't necessary.  I am specifically asking about the identification of multiplexer transistor architectures (if any) that are capable of switching ground to a no load, high Z or floating condition. (like a mechanical switch or relay but much faster and more scalable)  Normally, low side switching with a low Rds On would be the job of an N-Channel Mosfet except they generally won't turn on with an unloaded drain pin. (to my knowledge)  Are there any solid state switching topologies that are capable of this?

[Kleinstein]

A MOSFET will turn on even with the drain not connected. So MOSFETs or CMOS analog switches could be used to switch the electrodes to ground or the current measuring system working towards ground. A switch Chip (like 4051) may be easier than discrete FETs as small fets are rare and more difficult to protect.

However with to current coming from a very high voltage leaving the currently not active electrodes high impedance is not a good option: they would charge up to a relatively high voltage that may be a problem or effect the current flow.  It may still work with diodes to limit the voltage to a relatively low level of a few volts.

A big question is how fast the measurement needs to be. With many electrodes the speed can become an issue.
For a really fast system chances are better using charge measurement, as it would read the current integrated over all the time.
It would also need synchronization between the switching and current measurement.

[duak]

Kleinstein summed it up well.  A drain to source voltage is not needed for a MOSFET to be conductive, but you need some voltage to drive a sensible current through it.  It's like a water faucet that can be open or closed or something in between but unless there is some water pressure on one side, water is not going to flow through it.  MOSFETs also have something called a body diode that is in parallel with the drain to source and in an N-channel device, conducts when the drain is negative relative to the source.  Perhaps this was what you were thinking?  AFAIK, the voltage and/or current has no lower limit ie., no threshold - it may just be swamped by random noise or leakage current from various sources.

Conceptually, If current is flowing in all electrodes at the same time and you just want to measure one of the currents, there are a number of MUXs available to do what you want.  Assuming the HV source is positive relative to ground, the selected channel will connect the electrode to the picoammeter.  The unselected channels' voltages will rise until the protection diodes in the MUX inputs conduct and divert the current into the Vdd power supply.  Providing the current is limited to under a few microamps, this should be well within the operating conditions of the MUX.  However, during a "spark gap" event, the current can reach damaging levels.  To protect the MUXs, fast, high pulse current, low leakage diodes have to be connected to the electode inputs to divert the high currents to a voltage a volt or two above ground.   Series resistors should also be added between the electrode signals and the MUXs to limit the current into the MUXs during a spark gap event.

Some caveats about this scheme: the electrode voltage would be either essentially ground, which is the burden voltage of the picoammeter plus the voltage drops through the protection resistors and MUX on resistance or it will be what the electrode voltage will be clamped to by the protection circuitry.  From the use of the term "high impedance" the impression I get is that the MUX is not only selecting the electrode current to be measured but also selecting which electrode is attracting the solvent.  Sort of like an inkjet printer when the ink is selectively placed.  This would imply that the MUX would have to be able to switch, or at least stand off 4 KV which I don't think is feasible with semiconductors, at least inexpensively.

I mentioned the thermionic devices because as far as I know, low leakage, high voltage stuff is done in a vacuum.

BTW, when I looked up Electrospray and Taylor Cones I was thinking they are one of those "Wow, would you look at that" effects.

[thekeystoneclimber]

Kleinstein & duak thank you both for the insight.  With regards to... "Conceptually, If current is flowing in all electrodes at the same time"  It is not.  Potential (4kV) is present in all of the electrodes simultaneously. 

When you said "From the use of the term "high impedance" the impression I get is that the MUX is not only selecting the electrode current to be measured but also selecting which electrode is attracting the solvent." that is correct.

To alleviate some confusion, I should probably call the "ground plate" a return plate because it isn't always grounded, rather, it is usually floating with no potential until it is connected to ground to induce the ion transfer.  Current only flows through the single return plate that is presently pulled low to ground potential.  All of the other return plates are open circuit. 

When the return plate is connected to ground, ~ 20uA of current flows through the ammeter as a result of the electrospray.  This only occurs until all of the solvent is evaporated, at which point the current drops to zero.  Simply energizing the 4kV supply on an electrode over a return plate that is connected to ground is not enough to induce a current. 

The electrospray acts like a giant resistor effectively limiting the current and creates a huge voltage drop for the 4kV.  If I did my math right, the circuitry on the mux / ammeter side of the gap should only see about 64uV.  Obviously this changes if the air gap gets too small and turns into a spark gap because the plasma that the spark creates has a much lower resistance.

I'm not sure I understand how a floating return plate would develop a charge without a ground return path.

Regarding what "really fast" means, I have about 1 minute before the solvent is consumed and the electrospray stops.  With 96 electrodes to test, this leaves me 625mS for each sample. 

[NiHaoMike]

Could you consider switching the high voltage? Maybe using a servo to move a plastic arm holding a carbon brush against contact pads?

[duak]

Quoting above: "... I should probably call the "ground plate" a return plate because it isn't always grounded, rather, it is usually floating with no potential until it is connected to ground to induce the ion transfer.  Current only flows through the single return plate that is presently pulled low to ground potential.  All of the other return plates are open circuit."

Unless there is something I don't know about, I don't think that things work this way - there won't be any plates that won't have a voltage on them in the range of 0 V to 4 KV.  The solvent exits the HV plate with a charge - you mentioned ion transfer in the first post.  The charged  solvent is drawn towards whatever objects have some voltage that is lower the HV plate and when they reach it, deposit the charge from the ions.  If that plate is grounded, the circuit is complete and a current flows.  The other plates initially have a voltage - they can't avoid it.  If they are high impedance,  the solvent will deposit a charge until its voltage reaches the same as that of the HV plate.

The problem is that this voltage will be up 4 KV and that any switch or other device or component has to be able to withstand this voltage and not conduct significant current.  Any conduction will drain off the charge and reduce the voltage until an equilibrium is reached.

There may be an out here.  What is the relationship between plate current  and plate voltage? Is it linear or exponential?  Is there a threshold voltage needed before any current can flow?  Is a minimum voltage required to sustain current flow?  If so, then it may be possible to use lower voltage devices.

When I google "electrospray ionization" I get something that has only one receiving plate and an aperture for the ionized samples to pass through.  I don't see multiple receiving plates each with a selective attractive voltage.

Perhaps the two functions could be separated.  An outer ring that is grounded to attract the ions or charged to 4 KV to not attract them and an inner electrode that is essentially grounded by the multiplexer and attracts and collects any ions that get by the outer ring.