Shielding SDRs

[KaneTW]

I've noticed that most low-cost (<$1k) SDRs appear to not have been designed with RF shields in mind. How much does that affect performance? I'm looking at the LimeSDR layout (https://myriadrf.org/app/uploads/2016/06/LimeSDR_transp_3.jpg, Altium files are on a Github repo) and properly shielding the RF portion from both interference from the digital portion and radiated interference looks like a giant pain.

Does it just not affect things that much or do these low-cost devices plain don't care?

[kruszi]

Shielding is kind of a difficult topic. First of all, shielding is only addressing one special problem. Tow use-cases come into my mind when thinking about shielding:
1. You want to prevent unwanted emissions from leaving your system. Very often seen with WiFi-cards, Modules, Bluetooth, ... They have pretty nasty PLL synthesizers on board which create a whole lot of harmonics you cannot tolerate (if you're doing a wireless certification). Does not apply to low-cost (wide frequency range) SDRs since you're not going to get any certification at all and you do not care about these problems in any way.
2. You want to prevent nearby electronic to disturb your sensitive receiver input and thus degrade SNR. Mostly the case on RF measurement equipment in general and e.g. GPS/GNSS modules. Biggest problem: You need to filter all electrical connections from an to the shielded part in the relevant frequency ranges. Your shielding can be absolutely useless if you don't do it properly. In most measurement equipment shielding by solid metal is done in a preventive way. It simply does not really matter in terms of total system cost. Furthermore, it provides a very solid ground path which is vital for any RF. On an SDR again this is not really a topic of interest. Your SDR receiver has an enormous frequency range and you do not have any filter at the input. Thus you have a horribly bad sensitivity due to bad large-signal immunity. The effects of shielding can be very impressive if done the right way. On most cheap SDR not much care is taken about shielding due to the missing input filter. This offers disturbances an excellent entry point. Shielding would not be very helpful at all. Furthermore, shielding is not an all-purpose answer for all disturbances. It will only work properly in a limited frequency range and it will only address a fraction of problems you probably have. In any way, there are much better ways to improve the performance of your SDR than blindly adding a shielding.

In many cases, you can see the "option" for a shielding (as seen on the LimeSDR PCB you linked above). Cutouts in the solder mask allow you to put/solder shielding there. If it would help ... depends on your problem. Generally shielding is kind of an expensive thing to do in high volume manufacturing (requires extra steps in manufacturing and assembly). If you're trying to cut down cost, that is probably the place to start saving cost. If you already bought an SDR I'd simply take a shielding and solder it in place. It won't do any harm to the performance of your SDR (*). Just make sure to have a "tight fit" and no cavities. (*) Depending on the frequency of interest and size of the metal can it is possible for standing waves to occur within the cavity. RF absorbing foam inside the cavity is helpful if it can be tolerated for thermal reasons. (Btw. next reason to avoid shielding -> thermal issues due to missing radiation under the hood). But I would not expect too much of an improvement.

Do you have a specific problem with the the LimeSDR or is it just a general question?

[9aplus]

Later version have shielding....


On the first hand you must accept that LimeSDR USB is one building brick in development
of proper end user application. Suitable shielding, cooling, filtering and amplification is part
of fun. 

73 Djani

PS here have 2 x LimeSDR USB v1.4 and LimeNet Micro on order... together with 3 other cheap SDR RX

[mark03]

So a shield is not so useful for keeping digital noise from one part of the SDR, out of the quiet front-end part?

I have wondered about this in the context of a simple, low-frequency SDR, the ~ 100 MSPS "direct sampling" type for shortwave reception.  One of these days I want to design my own and I think I have a decent grasp of layout guidelines for the ADC, FPGA, etc.  but I've heard it is very difficult to get rid of spurious signals due to digital stuff coupling back into the front end.  I'll be limited to four layers so I can't just bury all of the digital signals between ground planes.

[xyrtek]

So a shield is not so useful for keeping digital noise from one part of the SDR, out of the quiet front-end part?

I have wondered about this in the context of a simple, low-frequency SDR, the ~ 100 MSPS "direct sampling" type for shortwave reception.  One of these days I want to design my own and I think I have a decent grasp of layout guidelines for the ADC, FPGA, etc.  but I've heard it is very difficult to get rid of spurious signals due to digital stuff coupling back into the front end.  I'll be limited to four layers so I can't just bury all of the digital signals between ground planes.

Shielding can be very useful, but as mentioned already shielding must be done carefully. (check Dave's SA tear-down videos for good examples of shielding)

SDRs like any device that mixes analog/digital circuits can benefit from shielding but the PCB layout will be where you want to spend the time.

As mentioned, the LimeSDR has no shielding but I have found that it meets its purpose without one. On the other hand when I did build a HackRF, shielding was worth the extra bucks.

Look at the pics, an eval board for 16 bits/105 Msps with no shielding at all and I assure you it meets the specs.

I worked on a SDR based on the eval board for the AD9467-250 and used no shielding at all on any of the prototypes.

Do some googling and you can find something similar, I'll link it if I find the bookmark.