The other day I took the top cover off of my D and put some tape on the inside of it and that’s definitely done the trick. I used 3m VHB tape, because that’s what I had, but any foam-core double-sided tape should do the trick; I didn’t remove the backing from the side that’s in contact with the PCB. The VHB tape that I have is about 1mm thick, and reinstalling the top cover was a firm fit, so it doesn’t need to be anywhere near as thick as what Radu attached as an example. The PCB is 100% stable now, I don’t have any issues with the USB port anymore, and the power-on button only protrudes slightly.

Please post some pics, this is of major interest.

I can’t be 100% sure, because I don’t have any photos, but I think the top plate might have been installed back to front on mine – i.e. with the holes for the Geiger tube (and GPS LED) on the wrong side. It’s probably worth everyone checking to make sure that the holes for the Geiger tube are on the left, when the display is facing you.

Yes, the circular holes must be in front of the Geiger tube. Please correct this if wrong.

I had an issue just now where I arrived home and the D connected to the wireless, after which I started a sync of the offline data from the drive home. The samples uploaded at a rate of about 2 samples per second, at which point I initially thought that the API must have had a dramatic performance improvement, however I discovered that none of the data uploaded and the dashboard was still reporting that the last data upload was as I left work. I disabled the wireless and re-enabled it and it connected this time but, of course, the offline data from the drive home is gone. Nothing of note happened, and I didn’t go out of my way to record any specific data, so I’m not too worried but there was obviously an issue with the wireless.

There is a hard-copy of the data, in the CSV files, this never gets deleted. I’m saying it just in case, but it’s becoming obvious to me that the sync feature must be improved.
I’ll reply to the other messages here:

Like Kurt, I’m finding the GPS to be unpredictable. I’m unable to predict when it might work or figure out how to orient it for greatest reliability; it’s basically fluke if I can position it so that it can achieve a lock.
Sitting in a cup holder that’s mounted quite high on the dash of my car, next to the centre vents, I can have 12 satellites on an overcast day and no signal on a completely clear day. Sometimes, when walking around the house, I’ll take it out of my jacket pocket to find that it has a lock with anywhere up to about five satellites but sitting on a counter in the kitchen, with the front facing outside, it has no lock – although standing on the kitchen table, a few metres from the counter and facing into the house, it will often lock on to 3-4 satellites.
A similar situation occurred just now when I was doing the grocery shopping. A few minutes after entering the store I took it out of my jacket pocket and saw that it had four of five satellites; I sat it in the bottom of the trolley (metal tubes with wire baskets in Australia) sitting on its back and it lost sync. It’s highly unlikely that that the trolley was behaving as a Faraday cage, since the wire spacing is quite large – about half the width of the D (when standing) and about the same depth.

We might try something here. Get one of these antennas:
https://www.aliexpress.com/item/6-20-6-6x20x6mm-thin-small-active-GPS-antenna-VSWR-Active-antenna-high-signal-top-performance/32686638729.html
And replace the current ceramic antenna, under the plastic cover. It uses an IPEX connector, so it will snap in place. You might get better signal, and this would solve the problem.

I hope something can be done about the offline sync performance. On each of the last two work days I haven’t been able to upload live, which means that I’ve had 2840 and 3092 samples to upload by the time I arrived home; the 2840 samples took 5 hours and 20 minutes to upload last night, so I’m guessing that it’s going to take nearly 5 hours and 45 minutes tonight with a corresponding gap in samples. It actually started phantom-uploading yesterday (uploading quickly but not actually uploading at all), however poking the reset button and trying again after powering back up saw things happen correctly; fortunately, in this case, interrupting the upload caused it to start again and not lose any samples.

I will investigate it further, but it will take some time. We will need a new firmware to improve sync, hopefully the speed too.