My SSA has undergone some more testing and I have to admit that -- at least partially -- I was wrong with my assumptions, so here we go:
First, before attempting anyting else, I took several screenshots in 50MHz span intervals and 10kHz resolution bandwidth in completely unshielded configuration. Input of the SSA was left open. Since probably not too many readers may be interested in this bunch of screenshots, I packed them all up in a single ZIP file.
After that, reassembly of my SSA commenced. Initially, I only sandwiched the RF board between the shields and put in the long screws, labeled "1" but didn't tighten them at all. The whole assembly was rattling like a can filled half-way with dry beans...
The full span spectrum shows a whole forest of spurs, apparently even stronger than without any shields at all. The 60MHz spur is also much stronger than it was initially and also stronger than without the shields, averaging in at approx. 64dBc. Yet, this figure alone is almost within spec (65dBc) but far off compared to a good instrument (>70dBc). But with only about 25% of the screws and none of them tightened at all, that's probably no surprise. There was approx. 1mm play in the screws until they were seated.
Next step was tightening the long screws but still leaving out the short ones. In that configuration, I unfortunately made a mistake with the SSA settings -- sorry gals and guys -- I left the attenuator at -20dB so it doesn't tell that much about the SSA's performance. The 60MHz spur at least doesn't emerge from the noise floor. Apparently, the SSA performs within specs in this configuration.
In the final run, I had some cans of spinach
(...Popeye...) and inserted and tightened all the screws and took one more screen shot of the 60MHz spur. I guess 72dBc is as good as it will get on my machine...
So my conclusion is -- yes, the shields may have some effect but unless they are machined completely inaccurately or not properly seated at all, they shouldn't push the SSA's performance out of spec. So there has to be something else wrong with the (more recently manufactured) machines that don't meet the specs.
By the way, the shields of my SSA are pretty accurately made and each has got two alignment pins that fit into bores of the RF PCB so the assembly position of the shields is accurate to within approx. one tenth of a millimetre. Not shabby at all I'ld say...
I hope I could shed some light on the shielding situation of the Siglent SSA and its influence of the instrument's performance.
Cheers,
Thomas