Guard ring - please critique

[Jester]

I'm doing a 2 layer PCB with a somewhat sensitive analog circuit using a LTC1966 RMS converter. The data sheet emphasizes that the output signal from the converter should be surrounded with a guard ring. In the data sheet example the destination IC has a ground pin adjacent to the signal input pin, however the closest ground pin on the destination IC I'm using is two over from the input pin (see schematic below). So there is another signal within the guard ring, is this okay?

In the sketch I "painted":

• 
  the sensitive signal pink (actual top layer)
  the ground ring green (also top layer)

Bottom of board is also a ground plane

Does this look okay?

 

[barry]

It looks like they want you to run copper between the two pins (the signal pin and the one just below it on U2). Do you have enough clearance to do that? It looks tight. It looks like your destination IC has a small footprint. Do they make it in a larger package where you might be able to squeeze some copper through there?

 

[Jester]

It looks like they want you to run copper between the two pins (the signal pin and the one just below it on U2). Do you have enough clearance to do that? It looks tight. It looks like your destination IC has a small footprint. Do they make it in a larger package where you might be able to squeeze some copper through there?

0.5mm pin spacing, so < 8mils between pads, so too tight for even a 4mil trace

 

[BigBoss]

I recommend a separate GND copper pour that is connected to main GND through multiple vias.
Because if this pin is sensitive, you should protect it by covering a clean GND otherwise current which are coming from different sources such as power and signal returns etc. will disturb your closed ring.

 

[KlausST]

Hi,

The left upper picture shows a guard on a signal line between an amplifier and an ADC.

The left lower picture shows an amplifier and an U-I converter.

==> what is your configuration?

Gard ring: usally is used on high ohmic signal lines. But your signal line is not high ohmic.
It is used to avoid that leakage currents on the PCB don't change voltage on high ohmic signal lines. To do so, the guard must be driven from an amplifier and the guard must have the same voltage as the signal line. This usually is done by an unity gain amplifier (buffer) ... often at he receiver side.
XTR111 has a built in buffer that could be used as guard buffer, LTC2420 has no buffer.
In your circuit the guard is ground voltage (zero) instead of signal line voltage. Therefor ther will be leakage currents..

So i'm a bit confused.

Please clarify:
Why do you need a guard ring?
Where is the left top picture from?

Klaus

 

[Jester]

Hi,

The left upper picture shows a guard on a signal line between an amplifier and an ADC.

The left lower picture shows an amplifier and an U-I converter.

==> what is your configuration?

Gard ring: usally is used on high ohmic signal lines. But your signal line is not high ohmic.
It is used to avoid that leakage currents on the PCB don't change voltage on high ohmic signal lines. To do so, the guard must be driven from an amplifier and the guard must have the same voltage as the signal line. This usually is done by an unity gain amplifier (buffer) ... often at he receiver side.
XTR111 has a built in buffer that could be used as guard buffer, LTC2420 has no buffer.
In your circuit the guard is ground voltage (zero) instead of signal line voltage. Therefor ther will be leakage currents..

So i'm a bit confused.

Please clarify:
Why do you need a guard ring?
Where is the left top picture from?

Klaus

Klaus,

The picture on the top left is an example guard rail from the LTC1966 data sheet, in the example the destination IC is an LTC2480, in my case the destination IC is a XTR111.

The LTC1966 does have significant output impedance (85kΩ)

Pin 6 of the LTC1966 is grounded, and as shown in the the top left picture, this is what the guard rail is connected to in LT's example.

The data sheet recommends the guard rail "Guard Ringing the Output - The LTC1966’s combination of precision and high output impedance can present challenges that make the use of a guard ring around the output a good idea for many applications."

Thanks for looking

 

[KlausST]

Hi,

OK, now i see the high impedance source. Maybe a was tired when i wrote the last post, because additionally i recognized a buffer in XTR111, but there is no. :-(

****
The pictures in the datasheets show a GROUND ring around the signal line. But between SIGNAL and GND there is voltage and any finite resistance causes a current.
So i see the only benifit of the GROUND ring is that there is a known and stable voltage around the signal. But it does not avoid leakage current to flow. And thus it does not avoid voltage drop.

To truely avoid current the ring voltage has to be the same as the signal voltage. Usually form a buffer at the receiver.
Then you have no voltage difference between signal and guard ring. And thus no current and no voltage drop.

Klaus

 

[Jester]

Understood,
Thank you