Correction to the title:  It is 200 Watts as I stated in the body of the post.

TR tube ? ( gas discharge tube )

<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

It must be solenoid operated.  A GDT is not an option.  Customer seems to think it will work.  I'm not so sure.

Your description is vague.  What exactly do you mean by "open" or "shorted"?  Is it totally reflective in both cases? How many ports, 1, SPST, SP2T?  Do you have to hot switch, ie from one state to the other with 200W applied.  Is the 200 Watts CW or pulsed?  What is the bandwidth?  Why does it have to be a solenoid, why not PIN diode?

If you really can't make it, then split the power 2x or 4x, put 2 or 4 switches in parallel, then add another 2x or 4x combiner to sum the energy again. Takes up alot of real estate, but cuts the power down to 100 or 50 watts.
Heck, make a 1x8 split if needed to use smaller units.

Most 1xN RF switches at high power are "cold switched", meaning the RF source is turned off temporarily, then the switch moves to another port, then the RF is turned on again.

I'd guess you need to interupt the RF power, then turn it back on, i.e. SPST.

kch

It is 100% reflective in both cases.  It is as described.  Normally Open.  Then, 24 VDC is applied to the solenoid, and a shorting bar is actuated across the NO contact to ground (inside the case).  The power is 200 watts at 2 Ghz.  I am told by the customer that 2.77 A Peak is the max current in the shorted state.  It will not be hot switched.

My main concern is the heating that will occur in the shorted state.  The customer believes it will work, and I also believe that.  The main concern is accurately determining the temperature rise within the switch so that the components (contact diameters, contact plate size, dielectric materials) can be specified properly.

I have done DC and 60 Hz experiments applying varying currents into a prototype, but correlating to 2 GHz is not clear to me.  In those experiments, the switch contacts did not heat up very much.  I applied as much as 5 A at 60 Hz.  I have no high power test equipment.

I hope this clears up the vague points.

That does help.  Let me think on it and get back to you.

My initial thoughts:

The pin that makes shorting contact will have some inductance associated with it.  Because of this, the reflection will not look exactly like a short circuit (ie. not be at exactly -180 degrees reflection coefficient).  If this is not a problem, then the design is simplified.

When the "pin" contacts the microstrip, there has to be a good contact to the microstrip or the contact point will have high resistance and overheat.  Also, further back in the solenoid, the connection from the pin to the ground plane must be good, or heating will occur there also.

At 2 GHz, most of the RF current will flow in the outer part, not the center, of the shorting pin.  You have to size the pin a little larger.

When the device is "open circuited", you have to worry about the RF energy at the open circuited end arcing over to the pin in the solenoid.  There will be a large voltage at the open circuited end, and you will need to insure that the pin is far away when open circuited.  It will also help to reduce arcing if the pin does not have any sharp edges on it.

Thank you biff44.  We have condiered all of these parameters.  Our main concern is the actual temperature to which this device will heat up.  My simulations indicate that this will get to approximately 120C with a convection coefficient of 10.  The customer tells us there is no concection within the enclosure and it must be able to operate up to +85C.  This concerns me and I am not sure this is a practical design.  

I know it is dependent on good solder joints and low contact resistance, however, I was hoping there might be someone with some actual design experience with this type of device.

The difference in phase angle between the open and short is not a concern.  Aside from the inductance, there is an actual physical difference between the location of the short and open.  Arcing should not present any problem to us as the clearances should handle the peaks of the voltage maxima.  All pins will be radiused.

Why are you using solder at all?  Have you ever used a gap welder?

By the way, I wouldn't go with an air gap of less than 1 mm.

We weld the solenoid wires to the contacts on our other switches and will do the same here.  As to soldering the switch to the board, we will be having discussions with the customer about that, but usually for a surface mount it will some sort of solder reflow mmethod as on our other switches.  

The switch will be able to withstand 500V RMS as a hipot test between any conductor and ground, so arcing should not be a concern.

Thanks you all for your tips.

You should expect 200 V RMS, 282 V Peak near the open circuit (assuming 50 ohm line).  Microwaves behave differently than low frequency, so being able to withstand 500V RMS does not necessarily mean you will not arc over--depends on the geometry of the pin relative to the wavelength.