[hpsdr] Linear IMD

[FRANCIS CARCIA]

HNY All,
Last week I finished my HD 48VDC/24VDC power supply to power my Erbtec MRI amplifier.
3 push pull stages MRF136, MRF148 and MRF150 finals. I am also able to hit -40 dB IMD3 at 50 watts. At 100 watts it drops to about 32dB. The bias setting does not effect IMD3 that much but the spec says the MRF150 will do -50dB IP3 at 100 watts out if you bias the pair at 6 amps. I'm presently looking at the BB transformers. I had added extra turns to the DC shunt choke inductor of the final to move it from 63MHz to 2 MHz. This hurt the IMD a lot so replaced it with 3 stacked .82 inch toroids of type 61 material to avoid saturation at current peak. The other transformers have been modified to shift the frequency range down to 160meters. It works 160 through 6 meters. I'm trying different combinations of core material and turns to dial it in.This change helped me by 5dB. I notice newer applications the shunt feed was deleted and DC fed into the CT of the output transformer. The only MOT app that has good IMD is the 20 watt driver everything else is
 around -30dB. I guess that was considered good back then. My goal is to have two of these strips as a table top final. Penny easily drives one strip with a 10dB pad on the input.   
A friend built a pair of EB104s and was very unhappy with the IMD performance. The tiny shund feed core gets very hot so I suspect it is saturating. Freescale sent me some APP notes that say their 1200 watt FET looks quite good at 600 watts It was around -40dB IP3. They feed DC into the transformer CT. The new Juma amp does the same thing but they have not published any IMD numbers yet. 
I have never seen an IMD spec on the ALS600 or ALS1300, they are a copy of EB104 so not surprised they don't. My point is I think the transformer design is just as critical as the FETs. 
I'll be glad to share the final design when I get there. frank WA1GFZ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20120104/3e6b99f4/attachment-0002.htm>