High Voltage RF in Micro-strip circuit

I have to absorb a high voltage signal about 1500 V 3 GHz pulsed 1% duty cycle. I am concerned about voltage gradient and corona. I am using the rule of thumb of 40 V/mil in air to determine my clearances. No sharp bends. Using 80/20 solder. Expect the connectors (TNC) to fail but I have to make them work.

Does anyone know of a good reference for designing to high voltage microwave signals in microstrip? The Reference Data for Radio Engineers (6th edition) barely touches the issue.

Thanks for any responses

Hello,

I don't know a reference (maybe you have to figure it out yourself).
Your 40V/mil equals 1600V/mm. When looking to a coaxial structure with D=6mm and d=1mm, at 1500Vp, E will be about 1700V/m, If you mean 1500Vrms, you will exceed 2300V/mm (close to the inner conductor).

I used E = V/(Rinner * Ln(Router/Rinner))

For atmospheric pressure and 3 GHz, I would say Peak AC breakdown equals DC breakdown value (3kV/mm). Note that at reduces pressure, dielectric strength reduces also.

If you have a dielectric and there are air pockets (or just clearance between the center conductor and the insulation), you will get corona discharge in the pockets. At an air/dielectric transition, Efield in air becomes Er times higher.

When looking to edges of traces on a PCB, E will be higher (because of the small thickness of traces (um range) ), so it is very likely that corona discharge occurs in 1.6mm FR4 50 Ohms microstrips. Coating (reasonable thickness) will reduce the E-field at the coating/air transition.

Copper loss you can guess based on skin thickness (1.2um) and RMS current (3Arms for 1500Vrms, 1% duty cycle in a 50 Ohms system). Take them twice as high to compensate for non-uniform current density distribution (edges will carry more current).

Volumetric dielectric loss you can calculate based on:

Loss/m3 [W] = 2*pi*f*8.854e-12*er*lossfactor*Erms^2
Lossfactor < 1,

In your case Erms averaged over long time is 10 times below Erms during the pulse (due to your 1% duty cycle).
I think you will fry 1.6mm FR4 material.

Maybe others can comment on creepage issues for 3 GHz.

[QUOTE=WimRFP;790167]

I don't know a reference (maybe you have to figure it out yourself).
Your 40V/mil equals 1600V/mm. When looking to a coaxial structure with D=6mm and d=1mm, at 1500Vp, E will be about 1700V/m,
[/QUOTE

The peak voltage is 1500 V.


Quote:
If you have a dielectric and there are air pockets (or just clearance between the center conductor and the insulation), you will get corona discharge in the pockets. At an air/dielectric transition, Efield in air becomes Er times higher.
end quote

So basically there is a real good possiblity that the connectors will fry due to corona. How will I know this occurs? Does the center conductor in this situation destroy the dielectric and therefore the impedance? Will the dielectric melt?

I should have mentioned that my circuit has to survive this condition for 5 seconds. The board material is Rogers 6002 which has a dielectric strength of 700 V/mil. Then I am struggling with a termination that can absorb this power. And o by the way, the customer doesn't want my circuit to reflect the energy.

Do you mean that your termination has to survive 1500Vp in a 50 Ohm environment during 5 seconds (so 23 kW during 5 s)?

I am almost thinking of buying several 100m of coaxial cable (use as attenuator) to store the thermal energy.

Hi,
1500V is peak voltage, the average voltage is 15V, due to 1% duty cycle. So the average power is less than 5W. So you can select Aeroflex/Inmet termination 3112-SF, 25W, 4KWpp@(5us/0.5% duty cycle).

Peak voltage is 1500 V,20 KW, 200 W average. I have been examing multipaction in connectors, clearances and dielectric breakdown, and it seems like to design for this is an iterative process and of course the boss wants to hit a home run on the first try.

40 volts per mil seems really agressive! I think that spec is only for low frequency or dc signals. I would derate that at least 15 times. Electric fields can easily be concentrated by a factor of 10 at sharp corners, like where you attache the board to a connector.

A 70um thick trace, 1.6mm above ground and 1.5kV will experience about 15kV/mm at the edge, so definitely (at least) corona discharge will occur. So it is very likely that a board with transition to air will not work for this voltage level.

I designed several matching circuits for end-fed halve wave antennas (with similar voltage levels) and corona discharge at thin edges was several time the limiting factor.

If he can maintain a coaxial symmetry (or at least round conductors with sufficient diameter), he can keep E-field within acceptable levels.

I did some ballpark calculation for a 4mm wide trace over 1.6mm material, loss factor 0.0012, er=2.94, 1500Vp, 3 GHz. I think dielectric loss will be about 20W/cm (that is about 0.35 dB/m), with concentrations on the edges. That is a lot during 5 seconds. Depending on dielectric properties (max temperature, heat capacity) and thermal resistance to environment, you may fry connectors.