Electronics > Projects, Designs, and Technical Stuff

GaN's datasheets missing td(on)/td(off)/tr/tf

<< < (4/4)

Weston:
Plasma / induction heating applications can have pretty widely varying loads, which can cause high efficiency zero voltage switching amplifiers, like class E, to lose zero voltage switching and cause the switching device to dissipate a lot more heat.

Given that GaN devices have a small die and relatively high thermal impedance they are not the most suited to experimenting with this stuff, as they are easy to blow up with a load mismatch unless you have a really good heatsink setup and some sort of protection.

Up to ~13MHz or so you can get away with smaller die size silicon MOSFETs intended for SMPS. Modern devices are a lot higher performance than the IRF510, just stay away from any "superjunction" devices, their output capacitance makes them unsuitable for class E applications.

A few years ago I used two STP17NF25 devices in push-pull class E at 6.78MHz to supply ~250W of RF. If you want a specific part number to use, you could try that device. It has about ~3x the capacitance and 1/3x the on resistance of the IRF510 though, so it would be best for higher power levels.

For higher frequencies you could use silicon LDMOS devices, they are a lot more robust than GaN devices and with careful design you can get comparable efficiency. They are decently expensive though. The MRF101AN would be similar to the IRF510 but with lower capacitance. I used it up to 40W at 40.68MHz in class E before with decent efficiency http://superlab.stanford.edu/rfchallenge.html

OwO:
Maybe consider class F as well.

Class E: output is a high Q resonator; switch is turned on at voltage peaks and "drags" the waveform lengthwise to add energy to the resonator. Narrowband (<1% BW).
Class F: output network is designed to have high impedance around the 3rd and 5th harmonic, and short circuit at even harmonics. Switch sees "soft" resistive load rather than a capacitive one, with device capacitance absorbed into the network. Can be designed for up to 10% BW.

Using an ordinary LDMOS transistor (recycled cell tower ones will do) it's possible to get 70-80% efficiency at 100MHz with class F, probably higher at lower frequencies. Btw most 800/900MHz band transistors will work fine because the internal impedance matching is invisible below 100MHz. I personally use MRFE6S9125N a lot (~$5 refurbished).

Navigation

[0] Message Index

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod