Re: WR90-SMA transition update
Thanks Neil, fascinating read – never even seen it done, let alone read the process.
Colin de G8LBS.
Sent from Mail for Windows 10
From: Neil Smith G4DBN
Sent: 10 September 2020 07:49
Subject: Re: [UKMicrowaves] WR90-SMA transition update
The flange face was fly-cut using a 7-insert 80mm end mill with a single polished carbide insert installed, 650 rpm, 35 mm/minute, 0.15 mm DOC finish pass, WD40 lube. The insert has about 1 mm corner radius. Quick lap with Brasso to get a mirror finish so I can check for surface imperfections. Normally I would lap them on 2000 grit paper on a surface plate with WD40, like I do for the 122 GHz flanges, but the flycutter does a decent enough job to make that unnecessary at 10 GHz. Other surfaces are done with a 20 mm two-flute YG1 slot drill at 1200 rpm, 72 mm/min. That has sharp corners so there are some tool marks where the end of the cutter was used, but it does create a nice finish on the faces. Inside is done with an HSS-Co 8mm slot drill which is centre-cutting. I plunged a few holes to 0.5 mm short of full depth, then did a conventional milling cut to 0.2 mm from the dimension, then a climb cut for the finish. The cutter has two flutes and chatter was a problem at 24 mm deep, so I ran the finish cut at 500 rpm with a good splash of SAE 30 hydraulic oil mixed with WD40. I’m doing to try a 3-flute carbide end mill for the next ones to see if I can avoid some of the chatter in the corners which I got during roughing. It would be way easier doing a trochoidal cut on a CNC.
SMA position was done using an edge-finder from the flange face, 4.1mm centre hole reamed to size. I drilled through into the guide to tap the threaded M2.5 holes, but there is 6mm of metal that side, so I’ll try doing them 5.5mm and bind using a spiral-flute tap next time.
I deburred the edges using a miniature Noga tool and a Noga hand countersink. It really needs anodising or painting, but for a quick and dirty solution, it will do.
The OpenEMS model suggests that the bandwidth is good enough for the amateur band just using the standard 1.27 mm probe on the Radiall SMAs. I did consider making a bobbin or tapered probe, but I’ll see what bandwidth I get. The rounded corners appear to have little effect other than making the guide wavelength a bit longer, so the backshort distance needs to be a few percent more than with square corners.
Modelling it without the mating WR-90 section just needs the probe length to be trimmed a little, and changing the dimensions other than the backshort and probe length/diameter doesn’t show any dimensions being critical.
Once I’ve got this one trimmed to best match at 10368, I’ll check the match from 9 to 12 GHz and see if it looks anything like the simulation. I’ll publish the drawings once I’m happy the dimensions are optimal.