biff44
Advanced Member level 6
- Joined
- Dec 24, 2004
- Messages
- 5,046
- Helped
- 1,376
- Reputation
- 2,748
- Reaction score
- 1,056
- Trophy points
- 1,393
- Location
- New England, USA
- Activity points
- 37,902
Wow, good luck! You are, how we say, up Shit Creek without a paddle!
Here are some of the things I would consider:
How to size the resistors
www.macom.com/Application Notes/pdf/m561.pdf
Here is a picture of a 1:4 splitter/combiner, made up of three 1:2 splitter/combiners. The details are not quite how you will have to do it:
http://www.qsl.net/n9zia/wireless/pics/4way_splitter.jpg
Here are the type of resistors you will need, under "configuration B":
**broken link removed**
1) Use the first article to figure out the wattage you need in the resistors. Pick the next highest size wattage resistor. If you can not find a high enough wattage, use two resistors in series. Note that these resistors screw right into the package floor, so there will be good heat transfer. You will NOT be able to use a simple surface mount chip resistor for this project.
After the resistor leads are soldered onto the printed wiring board, you have to very thoroughly clean off any flux residue, and then inject into the gap between the resistor leads and the ground plane some high breakdown voltage epoxy. This is imporant since there will be a very high RF voltage at these points and a very small gap between the conductor and ground--so there is great danger of the voltage arcing over.
2)The same can be said at the connectors! You will have to use connectors that are rated for this power level, and either use epoxy or dielectric past at the connector to microstrip interface to eliminate arcing. (where ever there is a sharp edge, the arcing is likely to occur). They type of connector shown in the power divider picture is too small to handle 3 KW peak power, since the distance from center conductor to metal ground is too small physically. So expect a much bigger connector there than shown.
3) Circuit design: A standard wilkensen splitter or combiner has an input port, two quarter wavelength transmission lines of 70.7 ohms, and a 100 ohm series resistor between the two outputs. To do a 1:4 divider, you do the same 1:2 divider three times, just like in the picture.
I would suggest you make this imporant modification, though. At the input of the entire 1:4, I would put a quarter wave microstrip transformer to take the input impedance from 50 ohms to 25 ohms (ie the transformer section would have a characteristic impedance of √(50 * 25) ). Then I would design my 1:2 splitters for a 25 ohm impedance. I would interconnect the 3 splitters with 25 ohm characteristic impedance transmission line. Finally at the 4 output ports I would have 4 more quarter wave transformers to get the impedance back to 50 ohms.
This would get you a much lower operating voltage so that arcing will be less important, allow much fatter lines so that copper vaporization will be less of a problem, and make alignment of the resistors less critical since the lines are fatter. The wider lines will mean less resistive microstrip insertion loss.
I would use really thick copper on the printed wiring board (at least "2 ounce" copper, and thicker if you can get it. You might want to even solder coat the lines by hand to make the thickness even more. The thick copper will also help keep the insertion loss low.
Finally I would research the board material a little. You need a thicker material so that the voltage does not arc over at the input/output coaxial connectors, I am going to guess 0.2" thick, certainly at least 0.1" thick. Such a thick board will not let heat out of the copper easily. So look around for a microstrip board material that can conduct heat well (there are some materials that have stuff impregnated into it to get the heat transfer up).
Good luck. I do not know if you will meet the insertion loss goal, but with this approach at least you will not have a smoking black mess of copper and carbon flakes after you test the unit at high power.
4) Oh, and those mounting screws to hold the board down to the chassis! You need the board to touch the chassis to get the heat out, so I would use some conductive thermal goop under the board (possibly silver loaded thermal compound like they use for pentium computer chip heat sinks), and lots of screws. BUT keep the screw heads AWAY from the microstrip lines, or they will arc over too. You might even want to consider plastic screws for any close ones.
Added after 54 minutes:
As far as arcing goes, air arcs over at 29 KV/cm at sea level.
Sharp corners, etc, concentrate the electric field, so I would expect arcing at 2.9 KV/cm (ie allow a 10X safety factor).
ASSUMING no standing waves (well matched source and load), the voltage at the input connectors will be at least
P = (Vrms^2)/50, So the peak voltage will me Vpk = (√(50 * P))/.707 = (√(50 * 3 KW))/.707 = 548 volts
This tells you that any distance between a conductor and ground of less than 0.188 cm will probably arc over.
In all, the electrical design is pretty easy. It is the thermal and mechanical parts of the design that you have to address to get a good grade.
Here are some of the things I would consider:
How to size the resistors
www.macom.com/Application Notes/pdf/m561.pdf
Here is a picture of a 1:4 splitter/combiner, made up of three 1:2 splitter/combiners. The details are not quite how you will have to do it:
http://www.qsl.net/n9zia/wireless/pics/4way_splitter.jpg
Here are the type of resistors you will need, under "configuration B":
**broken link removed**
1) Use the first article to figure out the wattage you need in the resistors. Pick the next highest size wattage resistor. If you can not find a high enough wattage, use two resistors in series. Note that these resistors screw right into the package floor, so there will be good heat transfer. You will NOT be able to use a simple surface mount chip resistor for this project.
After the resistor leads are soldered onto the printed wiring board, you have to very thoroughly clean off any flux residue, and then inject into the gap between the resistor leads and the ground plane some high breakdown voltage epoxy. This is imporant since there will be a very high RF voltage at these points and a very small gap between the conductor and ground--so there is great danger of the voltage arcing over.
2)The same can be said at the connectors! You will have to use connectors that are rated for this power level, and either use epoxy or dielectric past at the connector to microstrip interface to eliminate arcing. (where ever there is a sharp edge, the arcing is likely to occur). They type of connector shown in the power divider picture is too small to handle 3 KW peak power, since the distance from center conductor to metal ground is too small physically. So expect a much bigger connector there than shown.
3) Circuit design: A standard wilkensen splitter or combiner has an input port, two quarter wavelength transmission lines of 70.7 ohms, and a 100 ohm series resistor between the two outputs. To do a 1:4 divider, you do the same 1:2 divider three times, just like in the picture.
I would suggest you make this imporant modification, though. At the input of the entire 1:4, I would put a quarter wave microstrip transformer to take the input impedance from 50 ohms to 25 ohms (ie the transformer section would have a characteristic impedance of √(50 * 25) ). Then I would design my 1:2 splitters for a 25 ohm impedance. I would interconnect the 3 splitters with 25 ohm characteristic impedance transmission line. Finally at the 4 output ports I would have 4 more quarter wave transformers to get the impedance back to 50 ohms.
This would get you a much lower operating voltage so that arcing will be less important, allow much fatter lines so that copper vaporization will be less of a problem, and make alignment of the resistors less critical since the lines are fatter. The wider lines will mean less resistive microstrip insertion loss.
I would use really thick copper on the printed wiring board (at least "2 ounce" copper, and thicker if you can get it. You might want to even solder coat the lines by hand to make the thickness even more. The thick copper will also help keep the insertion loss low.
Finally I would research the board material a little. You need a thicker material so that the voltage does not arc over at the input/output coaxial connectors, I am going to guess 0.2" thick, certainly at least 0.1" thick. Such a thick board will not let heat out of the copper easily. So look around for a microstrip board material that can conduct heat well (there are some materials that have stuff impregnated into it to get the heat transfer up).
Good luck. I do not know if you will meet the insertion loss goal, but with this approach at least you will not have a smoking black mess of copper and carbon flakes after you test the unit at high power.
4) Oh, and those mounting screws to hold the board down to the chassis! You need the board to touch the chassis to get the heat out, so I would use some conductive thermal goop under the board (possibly silver loaded thermal compound like they use for pentium computer chip heat sinks), and lots of screws. BUT keep the screw heads AWAY from the microstrip lines, or they will arc over too. You might even want to consider plastic screws for any close ones.
Added after 54 minutes:
As far as arcing goes, air arcs over at 29 KV/cm at sea level.
Sharp corners, etc, concentrate the electric field, so I would expect arcing at 2.9 KV/cm (ie allow a 10X safety factor).
ASSUMING no standing waves (well matched source and load), the voltage at the input connectors will be at least
P = (Vrms^2)/50, So the peak voltage will me Vpk = (√(50 * P))/.707 = (√(50 * 3 KW))/.707 = 548 volts
This tells you that any distance between a conductor and ground of less than 0.188 cm will probably arc over.
In all, the electrical design is pretty easy. It is the thermal and mechanical parts of the design that you have to address to get a good grade.