...okay , maybe I didn't . :-/
:NOTES
About my current gcode.
My process flips the article vertically on the way to the CNC. yours may not. As designed the bias circuitry should be in the upper left corner when translated to the board.
The code was created with
jscut.org. I supplied the .bmp and .svg file if you need/want to play with it.The "full" file just removes the copper. The "cutout" file cuts the parts. The extra transformer end plate is just that , extra.
Your mileage and processes will, not may , vary. This is in no way a finished product. The end result may promote the end of the world as we know it (doubtful but possible). So please exhibit at least rudimentary experimental caution. All the caution necessary around power tools and wildly spinning bits apply.
I didn't bother with GPL license. The fact that it originates with me is enough. It it is modified please let me know. I think enough of you that I know I'll at least get mentioned even if a massive rework is performed.
Now the bad news.
Auto leveling is a must. I can't stress this enough. The path is 4.7 hours JUST for the circuit board. Do you want to do this twice becaus you missed a spot ? The cutout is nearly as long being multiple passes. The .8mm end mills will not survive much past 30 mm/min. I provided a bit of a safety margin but you've been warned.An alternative is to drill the .500" holes and bandsaw the parts off.
The recommended tool diameter for the pocket is .79-.8mm to include all the detail in one pass. I suppose it could be broken up into multiple tool diameters and operations but ...
last but not least DO NOT EXPOSE YOURSELF TO THE FIBERGLASS DUST COMING OFF THE BOARD. PROVIDE A VACUUM SOURCE OR DO IT WET.
JUST DONT...
tem #
4WTH4
This is Muller part from the Grainger catalog .500" ID type "L" hard copper tube
The cores vary in length so some clearance is provided. Recommend measuring the output transformer assembly and matching to the board prior to soldering the end plates.
Number of turns through the transformer is yet to be determined. Feeding the drains through common mode chokes is also a possibility worth investigating.
Input is a common binocular core that can be found in a 75-300 ohm transformer that used to be found on the back of a TV. Barring that find your own . A #43 core is recommended.
Active devices are best sourced from
RFMW
https://www.rfmw.com/products/detail/mrf300an-nxp/618411/
$25 each , MRF300AN and MRF300BN , one each required. Don't mix them up
If anyone is wondering what started this
https://www.rfmw.com/products/detail/mrfx1k80hr5-nxp/607748/
@ $226 for 1800 watts @ 65 Volts vs $50 for 600 watts @ 50 volts it seemed a worthwhile effort.
Then exploring the various pdf available I simplified the board and adjusted dimensions as necessary. The bias circuit is lifted directly from the MRF1K80H board and it's referenced parts are in the .csv file to be uploaded directly to digikey. This will create a BOM of around $50.
It includes the #61 cores as well as solder pasre. From experience chasing grains of sand around a postage stamp with a soldering iron is no fun. This is hot air work however you make it happen.
The Schematic and parts placement are available from NXP.
I'm including the MRF101 , MRF300 and MRFX1K80H pdf for reference.
For those of you that want to take the BIG plunge , go for it . A Canadian on ebay is selling the MRFX1K80H design board and a complimentary 2KW LPF board. The boards are bare but NXP gives you enough information to complete the board and the ebay seller supplies a BOM for the LPF board.You can use the same BOM/.csv file attached for the bias parts if you delete the #61 cores. The transformer will be up to you to complete.
So far it's been fun (yeah right) an education (you can teach an old dog new tricks) and now the real work begins. I have to build a few and let the smoke out.
