Panel Mount SMA Connector Mounting considerations

[themaccabee]

Hi,
May be this is silly .. But i ve a doubt here ..
I need to mount a Panel Mount SMA connector to a chasis, then solder the conductor to a PCB trace.(800MHz)
My doubt is that is there any requirement that needs to be met for the drilling hole that should be put on the chasis for the connector signal conductor to pass through??
Wat exactly decides the minimum or maximum hole diameter??And which performance parameters(like return loss..) can go wrong if we made the diameter smaller but Of course not touching the conductor.. or made higher..??

 

[ckshivaram]

As such there is no hard and fast rule for the width of the hole for connecting the SMA connector.. It should not be tooo far from the place it needs to be soldered... and use co-axial cable if you are sending or receiving signal. I used SMA connector in my Power amolifier upto 10W in GHz range... but width of hole did not matter much for me.

 

[biff44]

That is a good connector if you have a PCB inside of the housing, where the bottom of the PCB is a ground plane and it is firmly attached without any gaps to the housing bottom. The center pin of the sma would stick through the wall and solder to the top of the PCB trace. You would make the hole diameter what is needed to make the coaxial line thru the housing wall be 50 ohm.

If you wanted to make it more robust, you could do a stress relieved short piece of wire between the connector center and the top of the PCB trace.

IF you want to run a coax cable on the inside of the housing between this connector and the PCB, this is the wrong type of connector. There is no connection for the ground shield of the coax line.

 

[enjunear]

biff44 has some very good comments. A similar approach would be to use an SMA connector with an extended dielectric, like this guy.
Digikey - SMA with ext. dielectic

Using that part (or a similar one with a super-long dielectric that you can cut to size), you drill the hole in the chassis just larger than the diameter of the dielectric. Then the E-fields inside the chassis are similar to the fields inside the connector (center-pin to shield). Once the E-fields reach the end of the dielectric, the center conductor is soldered to the PCB, so you get a nice RF launch. This is typically the approach I use on my hardware. Generally, the small amount of mismatch you get below 5 GHz for an unsupported center pin does not degrade your performance significantly. Up past 12 GHz, you need to use stepped holes to maintain the impedance characteristics... they make 2.92mm connectors, and the like, for those millimeterwave applications.

 

[themaccabee]

That is a good connector if you have a PCB inside of the housing, where the bottom of the PCB is a ground plane and it is firmly attached without any gaps to the housing bottom. The center pin of the sma would stick through the wall and solder to the top of the PCB trace.

Thats exaclty what i wanted to do
But i cant understand what you said after that,

You would make the hole diameter what is needed to make the coaxial line thru the housing wall be 50 ohm.

I suppose the coaxial line you are mentioning is the center pin of the SMA plus the grounded hole's wall in the chasis around the center pin,with air as the dielectric in between them .Am i right??If so how to make it 50Ohms?
Is like that being calculated here **broken link removed**
One more thing, is the dielectric constant is a function of frequency ? if so then,shall we need to account the variation in dielctric constant of air for the concerned operating frequency..
Many Thanks..

 

[biff44]

Yes, d=1.27 mm, and er=1, so D=2.92 mm. The relative dielectric constant of air is 1, and is not frequency dependent.

 

[enjunear]

I suppose the coaxial line you are mentioning is the center pin of the SMA plus the grounded hole's wall in the chasis around the center pin,with air as the dielectric in between them .Am i right??If so how to make it 50Ohms?
Is like that being calculated here **broken link removed**
One more thing, is the dielectric constant is a function of frequency ? if so then,shall we need to account the variation in dielctric constant of air for the concerned operating frequency..
Many Thanks..

In practical designs, you want the dielectric of the connector to come right up to the PCB, don't leave the center conductor hanging in open air... that's bad practice at any frequency. Get an SMA with an extended dielectric and cut it down to size for the thickness of your chassis wall. This is less of an issue at lower frequencies, but at 800 MHz, I wouldn't take the chance.

 

[themaccabee]

The connection method i intend to follow is like shown in the picture , im not thinking about hanging the conductor in free air..ie when he conductor comes out of the chasis it lands straight to the PCB track for soldering.The PCB is kept close to the chasis and its bottom is ground plane.

 

[enjunear]

Here is an illustration of "center pin hanging in air". Your image is exactly how I do these launches. As mentioned before, make sure the ground (back-side of PCB) is well-bonded (touching/attached) to the chassis near the RF launch. You need to maintain a short loop-path for the RF return current. If not, it's basically like sticking a paracitic R and L in sereis with your RF output.