Thermally isolating mini-gps from Raspberry Pi 3 without RFI?

[cdev]

For a while I have been using Raspberry Pis with GPSs to provide a time server to my local LAN. But I have been using older models which generate a lot less heat. Even with them I have had problems with RFI so Ive had to keep the leads short.

Now for various reasons I want to use an RPI3.

But even when not enclosed it generates enough heat that I now would like to move the GPS farther away.

But there also is some RFI produced by it- and the longer the distance is the more audible it is. Ive gotten rid of a lot of it.. some ferrite beads got rid of most of it but its still there. broadband hash, and a momentary blit every second.

I am going to try a bunch more SMT caps, much smaller values, and maybe some more beads.

 
Here is the way they are arranged now
(top)
Antenna (passive)
pigtail
GPS-Navspark mini
headers it plugs into
ribbon cable
flat ribbon cable ferrite (I cant find this in the old FairRite catalog I have but judging from its style its likely a Fair-Rite part, likely made of #43 material)
cheap 'hat' ($2) board's floating contact areas
additional ferrite beads on wires to GPIO 18(1pps), Txd+Rxd UART, and +3.3 v positive power rail (not ground) and hat body which consists of the floating contact areas and the GPIO headers all marked to make them easy to connect the wires to.
Bypass capacitors between positive rail and ground there. (If the additional bypassing doesnt cure it just for the knowledge's sake I am going to try slicing the trace for its 5v rail to remove that as a potential source of RFI for now since I currently have no use for it. I can always add it back in by bridging the trace with solder if I need to.) Experimenting now with different bypass caps values and types. Wondering if there is a suggested top limit?
Some kapton tape on hats bottom to reduce (warm) airflow through hat's holes into GPS's area when/if it needs to be vertical. Pi is now being held in a sideways position. Heatsink on CPU fins orient vertically with this tilt. May end up mounting it on a wall.
Pi body itself
(bottom)
No case currently until this gets sorted out.

 ---
Long somewhat useless musing deleted.
---

Are there any good online resources on reducing broadband noise radiation from 12c and SPI control lines, and also good one stop shop resources on the RPI hats and perhaps RFI in that context, in real life usage? (I am not using the hat for external power, although I likely will want to use a power backup in the future, but for now its just a time server, and that involves the Pi, and a GPS and soon one ambient temperature/humidity and barometric pressure sensor - at least one, possibly two of them. (one indoors, one outdoors) on a wire long enough to reach that distance..


I am going to take a little break from working on this for now.

If I cant eliminate the RFI better I may dump the hat for now and go back to attaching everything to a much smaller IDC header by itself.