SMA Adaptors


Chris Bartram G4DGU
 

Hello Gareth,

I agree with Dave that it's quite unlikely that simply replacing a Gunn with an SMA probe will work properly. This is a question of impedance matching. You shouldn't need to use a screw tuner to get it to work reliably. I can't give you a recipe for a WG16 to SMA adaptor, as I haven't used WG16 for many years: I standardised on WG17 for my own 10GHz systems, and was fortunate enough to come across some lab.standard adaptors. However there are good amateur designs out there, and I'm sure somebody from this group will direct you to them.

73

Chris G4DGU


Neil Smith G4DBN
 

Hi Gareth, I've been making some WR90-SMA transitions milled from solid aluminium, with rounded corners and with the backshort distance and pin length set to give good RL over a few hundred MHz rather than mediocre RL over the 8-12 GHz range. I don't use a tuning screw, but I have to check the RL using a decent directional coupler and a waveguide dummy load.  My cavities have rounded corners, but the backshort distance isn't much affected by that.  I find 5.65 mm is about optimum to balance RL, bandwidth and ease of adjustment of the pin.  I ream the SMA mounting hole to 4.10 mm to take the 4.06 mm PTFE sleeve of the Radiall SMAs, but a cheap 4mm drill should work fine. 

It is a whole lot simpler to fit a fine tuning screw opposite the SMA rather than tuning by filing 0.05mm increments off the pin. My experiments were intended to find how reproducible the design would be and if a no-tune option is viable. Spoiler alert - it is WAY too much hassle unless you leave the PTFE on the pin, as otherwise it is too dependent on how you trim the PTFE back to the guide face. Newer versions will leave the PTFE in place and just machine the end off at a precise distance from the flange, then use a dielectric tuning screw so there are no corrosion or contact issues.

The other issue is that you don't know if you have taken too much off the pin until you are past the sweet spot, so I use a 0.2 mm brass shim under the SMA socket while I'm trimming the pin to length.  Then when I've gone too far, I can remove the shim and do a final adjustment.  Elegant and simple, but way too much work.


http://www.g4dbn.uk/?p=1570

If you want to re-use a gunnplexer, you might have to drill a new hole for the coax socket if it is < 5mm from the backshort, but if it is >6mm, you might be able to file a close-fitting plate from a bit of sheet metal the same as the body of the unit. You could then shove into the bottom of the guide to get the backshort distance to something near 5.5-6.0 mm and glue it in place with a drop of superglue.

Without machine tools, I'd find a bit of brass or copper guide and drill an SMA hole at 5.7 mm or so from the backshort, then solder on a backshort and flange. If you don't have a flange, use a bit of plate drilled and filed to fit. It would then need be aligned with a proper flange using a bit of wood cut to WR90 internal size as mandrel, drilled, then sanded flat using wet/dry paper on a sheet of glass, using WD40 or light oil as lube. I'm assuming 1.27 mm pin size. Fatter pins generally widen the bandwidth, but are fiddly to make.

You could use 22mm water pipe with an end cap soldered on, then file a flat on the side and solder on a 2-hole SMA.  Backshort distance in 22mm pipe should work out around 6.4mm, but I'd do it 7mm and tap an M3 hole in the middle of the cap, then tune that and the pin length to get a decent match. You could force the other end into a rectangular flange or just leave it round, depending on what want to use as a feedhorn.

If you want to make a rectangular transition but have no waveguide, you should be able to anneal a bit or copper pipe and file a bit of aluminium bar to WR90 internal size, then form it into a tapered and rounded mandrel, and use that to force the round tube to become rectangular.

Beware of working out how much to take off the pin to bring it into tune, the relationship is non-linear and it is much too easy to go too far.

If you imagine the pin size and backshort distance behave rather like tuning an L-match. There are infinite pairs of matches within certain limits.  The trick is to get the Q to a reasonable value

Have a look at the W1GHZ papers:

http://www.w1ghz.org/QEX/Rectangular_Waveguide_to_Coax_Transition_Design.pdf

http://www.w1ghz.org/QEX/circular_wg.pdf

and his Antenna Book online: http://www.w1ghz.org/antbook/contents.htm

Neil G4DBN

(Note: I am currently tied up 100% on 24/7 carer duty looking after my wife who has terminal cancer, so only answering emails or twitter messages every few days when a respite carer visits. Life and projects are mostly on hold but I'm doing stuff sporadically)



Gareth G4XAT
 

Thanks Neil, great insight into what I'm trying to do. Made something that 'works' but with just SDR to look at the LNB output, its not exactly precise. Does behave as I expected though, so progress of a sort....
Best wishes
Gareth


Alan Beard
 

Hi Neil,
Are there any broadband techniques as I've done here for
2.3 - 3.5 GHz?

Alan VK2ZIW

On Wed, 9 Jun 2021 13:02:16 +0100, Neil Smith G4DBN wrote
Hi Gareth, I've been making some WR90-SMA transitions milled from
solid aluminium, with rounded corners and with the backshort
distance and pin length set to give good RL over a few hundred MHz
rather than mediocre RL over the 8-12 GHz range. I don't use a
tuning screw, but I have to check the RL using a decent directional
coupler and a waveguide dummy load.  My cavities have rounded
corners, but the backshort distance isn't much affected by that.  I
find 5.65 mm is about optimum to balance RL, bandwidth and ease of
adjustment of the pin.  I ream the SMA mounting hole to 4.10 mm to
take the 4.06 mm PTFE sleeve of the Radiall SMAs, but a cheap 4mm
drill should work fine.

It is a whole lot simpler to fit a fine tuning screw opposite the
SMA rather than tuning by filing 0.05mm increments off the pin. My
experiments were intended to find how reproducible the design would
be and if a no-tune option is viable. Spoiler alert - it is WAY too
much hassle unless you leave the PTFE on the pin, as otherwise it is
too dependent on how you trim the PTFE back to the guide face. Newer
versions will leave the PTFE in place and just machine the end off
at a precise distance from the flange, then use a dielectric tuning
screw so there are no corrosion or contact issues.

The other issue is that you don't know if you have taken too much
off the pin until you are past the sweet spot, so I use a 0.2 mm
brass shim under the SMA socket while I'm trimming the pin to
length.  Then when I've gone too far, I can remove the shim and do
a final adjustment.  Elegant and simple, but way too much work.

http://www.g4dbn.uk/?p=1570

If you want to re-use a gunnplexer, you might have to drill a new
hole for the coax socket if it is < 5mm from the backshort, but if
it is >6mm, you might be able to file a close-fitting plate from a
bit of sheet metal the same as the body of the unit. You could then
shove into the bottom of the guide to get the backshort distance to
something near
5.5-6.0 mm and glue it in place with a drop of superglue.

Without machine tools, I'd find a bit of brass or copper guide and
drill an SMA hole at 5.7 mm or so from the backshort, then solder on
a backshort and flange. If you don't have a flange, use a bit of
plate drilled and filed to fit. It would then need be aligned with a
proper flange using a bit of wood cut to WR90 internal size as
mandrel, drilled, then sanded flat using wet/dry paper on a sheet of
glass, using WD40 or light oil as lube. I'm assuming 1.27 mm pin
size. Fatter pins generally widen the bandwidth, but are fiddly to make.

You could use 22mm water pipe with an end cap soldered on, then file
a flat on the side and solder on a 2-hole SMA.  Backshort distance
in 22mm pipe should work out around 6.4mm, but I'd do it 7mm and tap
an M3 hole in the middle of the cap, then tune that and the pin
length to get a decent match. You could force the other end into a
rectangular flange or just leave it round, depending on what want to
use as a feedhorn.

If you want to make a rectangular transition but have no waveguide,
you should be able to anneal a bit or copper pipe and file a bit of
aluminium bar to WR90 internal size, then form it into a tapered and
rounded mandrel, and use that to force the round tube to become rectangular.

Beware of working out how much to take off the pin to bring it into
tune, the relationship is non-linear and it is much too easy to go
too far.

If you imagine the pin size and backshort distance behave rather
like tuning an L-match. There are infinite pairs of matches within
certain limits.  The trick is to get the Q to a reasonable value

Have a look at the W1GHZ papers:

http://www.w1ghz.org/QEX/Rectangular_Waveguide_to_Coax_Transition_Design.pdf

http://www.w1ghz.org/QEX/circular_wg.pdf

and his Antenna Book online: http://www.w1ghz.org/antbook/contents.htm

Neil G4DBN

(Note: I am currently tied up 100% on 24/7 carer duty looking after
my wife who has terminal cancer, so only answering emails or twitter
messages every few days when a respite carer visits. Life and
projects are mostly on hold but I'm doing stuff sporadically)


---------------------------------------------------
Alan VK2ZIW
Before the Big Bang, God, Sela.
OpenWebMail 2.53, nothing in the cloud.