RF shield for 32MHz Xtal + caps

Thread Starter

Tera-Scale

Joined Jan 1, 2011
164
I need to shield a 32Mhz quartz crystal with an RF shield. I have 0.1mm copper shield, is this enough? I have a ground plane on the PCB which is slightly larger than the shield.

Should I connect the shield directly to ground plane because a friend told me I have to add picoF from the shield to the ground plane. How can I calculate values for these small ceramic caps? thanks
 

Papabravo

Joined Feb 24, 2006
21,225
Ask your little friend what the caps are doing. I don't see the point and given the low power of crystals I'm confused about the purpose of shielding. What do you have that is so sensitive?
 

kubeek

Joined Sep 20, 2005
5,795
Picture is worth a thousand words. Show us the actual board, best along with a schematic and description what frequencies / data rates are involved.
Also, shield is usually a box or plate, this sound more like a guard ring.
 

Thread Starter

Tera-Scale

Joined Jan 1, 2011
164
A peripheral called ctmu working with currents as low as 55μA. And there is a calibration of this current by passing it through a 40.2k resistor to measure voltage. Also I have some analog stages that are near.

But is this technique implemented in practice? - to put caps from shield to ground?
 

Thread Starter

Tera-Scale

Joined Jan 1, 2011
164
Picture is worth a thousand words. Show us the actual board, best along with a schematic and description what frequencies / data rates are involved.
Also, shield is usually a box or plate, this sound more like a guard ring.
Here is a view from the 3D visualization. It has very few components left to add..
 

Attachments

kubeek

Joined Sep 20, 2005
5,795
I think you should rotate that big ic 90° to the right, so that the oscillator is as far as possible from the other components.
Other than that, without a scematic it is hard to say, but I think the ground plane should be everywhere on both sides, not just around the crystal.
Also the two parallel traces seem that they could be quite important as far as noise pickup goes, so I think shortening them couldn´t hurt.
Generally you should have three areas with reasonable distances between them, one for analog stuff, second for digital, and third for supply. This board looks like it has everything mixed together.
 

Papabravo

Joined Feb 24, 2006
21,225
I would try to get the crystal leads and the load capacitors as close to the IC as possible, and put the resistor on the other side of the crystal package.

If the noise you are worried about is radiated then a shield will help. Thing is, there is nothing like an efficient antenna to radiate at that frequency and as I said before there is not much power in a crystal oscillator. The 7th harmonic of 32 MHz is only at 224 MHz which is between VHF and UHF with a wavelength of 1.25 Meters.

If the noise is conducted then shielding won't help a bit since the noise will be conducted along the traces to the IC and couple to parallel traces -- if there are any.

First thing to try is separating the sections and make sure that ANALOG ground and DIGITAL ground are joined at one point only on the PCB where the power enters the board. Make sure that each chip has parallel bypass capacitors for bulk current demand and HF bypass.

Last thing is to rent the equipment to measure the RF, but only after you have a reasonable chance of getting the result you want. A general coverage communications receiver might do in a pinch for a qualitative indication of RF field strength. Listen at the fundamental and the harmonics through the 7th or 9th and record the readings on the S-meter. Each S-unit is 6 dB which corresponds to a voltage ratio of 2 or a power input ratio of 4. S9 is equivalent to 50 uV into 50 ohms for the HF bands.
 
Last edited:
Top