LMP7721 layout

Hello Paul,

Thank you for your very nice tutorial at:   http://www.edn.com/design/analog/4395651/3/Design-femtoampere-circuits-with-low-leakage---Part-3--Low-current-design-techniques 

In the tutorial you are giving an example of the LMP7721 -femtoamp application- the LMP7721 MultiEvaluation board: http://sva.ti.com/assets/en/boards/LMP7721MA_MultiFunctionEvaluationBoardUserGuide.pdf

 The question would be:

How does the bottom part of the PCB  should be?.   

I suppose that the sensitive area (the smaller guarded area) doesn't have A GND plane on the bottom and no solder mask. Also, I didn't see GND traces or at least I noticed you are using vias for connection to the GND. 

Is this correct?

 If you can post a picture with the bottom PCB that would be great.

 Thank you very much,

Ion Ciobanu

Hello Ion,

Attached is the 3-D view of the LMV7721 board.

For the LMP7721 board, the bottom side board is bare within the area of the input guard trace. The bottom looks just like the top - but without all the input traces (bare). THe "L" and "T" shaped pads are for a tin shield.

1563.LMP7721EVAL_3D_Board.pdf

Actually, I just copied the topside guard traces on the bottom, and removed the copper within the guard area. So there is no copper below the input section. The via connect to the guard ring on the bottom side.

Outside of the bottom guard ring is a ground plane - just like the top.

The thought was to eliminate as much stray capacitance and board soakage as possible. The perimeter of the area was at guard potential due to the guard ring, so there was no need for a lower guard plane.

The assumption is that there will be no noise generating devices or traces below the input circuit on the demo board (it would be tested inside a box or fixture to keep noise out).

If your circuit will have "stuff" below the input circuitry, the plane directly below the input circuitry should be a guard plane - not a ground plane.

You should try to encase your input circuitry within a guarded "cage" or "cocoon". Your input circuitry should only see the "guard" trace potential in all directions. Anywhere you feel the urge to place a "shield", it should be at guard potential. But don't overdo it...leave some space so that the input to guard capacitance is kept low.

Hope this helps!

Regards,

Thanks a lot Paul,

It helps a lot. 

Thanks for reminding me to encase the input part within the guarded cage.

Have a great day!

Dear Paul,

What if the LMP7721 is used as an i to V converter, where the non-inverting input is at ground?

The guard voltage must be at  ground voltage, correct?

Should I use an amplifier to drive that guard ring to ground voltage?

Thanks a lot.

Gabriel

Hi Gabriel,

You want the guard to be at the average voltage that the input would be seeing. In most cases, this would be the same potential as the opposite input. For the I-V converter (transamp), that would be the non-inverting input.

If the input is ground-referenced, then the guard would be at ground potential. If the non-inverting input was biased to some other voltage (say 2.5V), then the guard would be at that voltage (2.5V).

So with ground-referenced inputs, the ground *is* the guard and no separate guard is required. However, I still recommend a "guard" separate from ground to avoid ground noise/currents. Never use the system ground as a return path for tiny signals...for obvious reasons...

Regards

Dear Paul,

Sorry for not really understanding the last two sentences of yours.

My application is a potentiostat where an i to V is used to measure the current in an electrochemical cell. And always in the circuits that I've been taught and constructed myself the non-inverting input of the i to V converter is at ground.

I am a chemist (PhD at the moment), and although I have an electronics background, I still don't get it.

How can I achieve a guard separate from ground? Can I have an example, please?

Unless you mean to bias the non-inverting input...

Gabriel

Hi Gabriel,

Yes... In the classic I/V converter, because of the feedback action of the amplifier through the feedback resistor, the potential between the amplifier inputs is always kept at zero...which means that if you set the non-inverting input to a voltage, the output will move so that the the inverting input (current input node) will be at that same potential. This "set" voltage on the non-inverting node is referred to as the reference voltage. In your case it is 0V, but it could be any voltage within the legal common mode range and output swing of the amplifier.

With no input current, the I/V converter is basically a follower. The negative input and the output will be at the reference voltage with no input current. As current is added or "stolen" away from the inverting input node through the feedback resistor, the output moves to force the inverting terminal to be equal to the non-inverting input. The movement of the output required to offset the input current is what you see as the "output signal' (the difference between the output and the reference voltage is "the signal").

The amplifier does not care if the reference voltage is at zero, or 1V, or 2V. It just wants it's two input terminals at the same potential. Keep in mind that this also sets the "zero" output voltage to the reference voltage, and moves +/- away from that depending on the input current.

This comes in handy if your "sensor" requires a voltage bias. If your sensor needs a bias voltage across it (like a photo diode), you can set the reference voltage to provide that bias.

Google "Transimpedance Amplifier" - that is the formal name of the "I to V" converter.

To have a "grounded guard", this is essentially a separate shield, not shared with any other circuitry. It should only be driven from the amplifier end (so no currents other than leakage flows). It does not hurt to have it separate so that you can bias it at a different voltage if needed should your design change.

I highly recommend reading the Keithley Low-Level Measurements handbook - Particularly chapter 1.5 and all of 2. That should answer most of your questions.

http://www.keithley.com/knowledgecenter/knowledgecenter_pdf/LowLevMsHandbk.pdf

Regards,