Path: ...!Xl.tags.giganews.com!local-2.nntp.ord.giganews.com!nntp.supernews.com!news.supernews.com.POSTED!not-for-mail NNTP-Posting-Date: Sun, 09 Jun 2024 20:51:37 +0000 From: john larkin Newsgroups: sci.electronics.design Subject: Re: RF Connector Type ID Date: Sun, 09 Jun 2024 13:51:36 -0700 Message-ID: <625c6jdqe8f577kh5ik3bkq4p1aqugb4sc@4ax.com> References: <97s96j5fajs26f6eg9odqs48skgghoson5@4ax.com> User-Agent: ForteAgent/8.00.32.1272 MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit Lines: 84 X-Trace: sv3-CejEOl5aqaCKksMD4UuAcMHNgnQYnEolwMFZWTCbdGSzEcMlK6qocJoBOWAXFF5ikzNDdEuQBab6d6W!7XizLizUrzMk5xEAIq7MQ37mrK2WTZ6Yi1Jwz7sUv5Oczl8+1Git0DID8/atyWqHnd0UCjrVnh+g!Rwx6Eg== X-Complaints-To: www.supernews.com/docs/abuse.html X-DMCA-Complaints-To: www.supernews.com/docs/dmca.html X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly X-Postfilter: 1.3.40 Bytes: 4162 On Sun, 9 Jun 2024 22:38:50 +0200, Gerhard Hoffmann wrote: >Am 09.06.24 um 17:03 schrieb john larkin: > >> >> Our boards are mostly 0.062" thick, and I like layer 2 to be a solid >> ground plane. Some internal layers are power planes. If you solder an >> edge-launch SMA or SMB connector to layer 1, the layer 2 ground plane >> is too close to the center pin of the connector, so that bit of the >> signal path is too low impedance. So one does creative shaped cutouts >> in the various layers to keep the impedance uniform and the >> reflections down. We did e-m simulations (with ATLC) and test boards >> to get that right. >> >> The cheaper connectors have a giant round center pin, which requires >> attention. >> >> https://www.dropbox.com/scl/fi/vfi56wy5hkwva51bx77z6/SS_SMA_Edge.JPG?rlkey=nbmbyb0f54uslcpc0nh7rs21p&raw=1 >> >> That pin is around 100 ohms in free space! >> >> There are more expensive conectors with a tiny flat pin that >> theoretically matches a layer 1 trace width directly, but I haven't >> found them to be worth it on FR4 boards and 30 ps edges. >> >> https://www.dropbox.com/scl/fi/rptmh5f4umscqqpegatw6/Mueller_SMA.JPG?rlkey=1eonuljqekvrszyb9tce5s1hr&raw=1 > >That giant center pad gives away the advantage that you could have >from the costly SMA connector. That pic was the connector on a PCB that was intended for the big-pin part. A skinny trace would be appropriate for the Mueller SMA, but in real life the cheaper big-pin part TDRs about the same. > >< >https://www.flickr.com/photos/137684711@N07/51843174578/in/datetaken/ > > >This is nearly as bad as the unpopulated SMA at the right end of the >line. The left SMA is one with a thick center pin. Unsuitable on this >cheap JLCPCB 4 layer board. you even see the amount of solder. We tweak the pad stack to compensate for the big pad. That works pretty well. > >This is the next try: (4 TI LM2594/95 synthesizers + test stuctures) >< >https://www.flickr.com/photos/137684711@N07/51844498620/in/datetaken/ > >red = top, green = bottom, inner layers cut out. > >setup on tdr refl. + transmission: >< >https://www.flickr.com/photos/137684711@N07/53780177186/in/dateposted-public/ > > > >result: >< >https://www.flickr.com/photos/137684711@N07/53780597885/in/dateposted-public/ Big improvement. > > > >The SMAs are at division 7 and 8. Very slightly overcompensated, >it could also mean that I need some more GND vias. > >The microstrip is slightly more than 50 Ohms. JLCPCB's calculator >says 11.5 mil for 50, it took only 10 to see if I can easily escape >a BGA. The funny bends are not really punished. > >The 10 cm of FR4 do not demolish the rise time. Not bad for >€1.40 per 4 layer board. Short traces on cheap boards are pretty good. > >regards, >Gerhard > > > >