The AM Forum

THE AM BULLETIN BOARD => Class E Forum => Topic started by: VE3ELQ on November 28, 2016, 07:13:17 PM



Title: 20 Meter GaN FET RF Deck
Post by: VE3ELQ on November 28, 2016, 07:13:17 PM
The Transphorm 17A 600V TPH3206PS GaN FET specs looked attractive for possible 20m operation so a 2 FET test deck was constructed.  The challenge was then how to drive them.  They require at least 6 gate volts to fully saturate with fast rise and fall times at 14.3 mhz.  Two drivers that looked fast enough were considered, the NCP81074 and the EL7104cn.  Since two EL7104s were on hand they were tested but proved to lack sufficient drive current to drive the 760pf gate capacitance. So a bipolar NPN/PNP emitter follower complementary pair were added using the ZXTM07045 and ZXTP07040DFF as current amplifiers built on a small PCB all running at 8 volts.  The resulting gate drive while not stellar, was quite adequate.  The deck consists of two boost converters in co-phase transformer coupled to the output filter and load, the transformer using a single T157-2 powdered iron core with the windings 180 degrees apart as has been done with my other decks.  Input is 50V @ 3.3 Amps output 150 watts giving 90% efficiency in class E and 86% in class D. Modulation tests were not performed.  Some NCP81074 drivers which look a little better are on order so will see how they perform.

So 20 meters is doable with these GaN FETs.  A four FET deck at 40 or 45 Volts carrier should make a nice AM transmitter in the 250 to 300 watt range with sufficient head room for high peak modulation. Some scope traces attached.

73s  Nigel


Title: Re: 20 Meter GaN FET RF Deck
Post by: steve_qix on November 29, 2016, 07:23:05 PM
Hey, that's pretty awesome!  Have any pictures?  Would love to see them.

So, that definitely answers the question as to whether these devices will work at RF frequencies - they obviously will !!

Regards,  Steve


Title: Re: 20 Meter GaN FET RF Deck
Post by: VE3ELQ on November 29, 2016, 09:09:57 PM
Got some NCP81074 drivers today, wow are they fast but lack the needed punch.  Have a couple PCBs worked up in the PC with bipolar current amps, will build them tomorrow and test and tweak. Pretty sure the efficiency can be improved with better gate drive. These common source GaNs are a delight to work with, just bolt em down.
Give me a couple more days to play, then will post pics, schematics, and PCB patterns for the drivers.  BTW these drivers with added bipolars at 12V should do very well with SIC FETs, will test that later.

73s  Nigel


Title: Re: 20 Meter GaN FET RF Deck
Post by: KD6VXI on November 30, 2016, 02:57:41 AM
Schematic of the bipolar gate drive system you are using?   Sounds very interesting....



--Shane
KD6VXI


Title: Re: 20 Meter GaN FET RF Deck
Post by: VE3ELQ on November 30, 2016, 07:08:18 PM
Some interesting experiments today with the  NCP81074 drivers.  Tried 3 configurations, standalone, 2 bipolar amps NPN/PNP and 4 bipolar amps 2NPN/2PNP in parallel.  There was no measurable difference between any of them. The NCP81074 drivers running standalone are slightly better than the EL7104cn with bipolar amps.  At 14.3mhz drive current was 300ma for both phases at 8V regulated and they barely get warm. At a $1.25 these are very good little drivers.

Having a hard time accurately measuring efficiency. My best meters have internal amplifiers and go nuts around RF so I'm limited to analog meters calibrated to my Fluke.  In any case the deck was run today in class E for 20 minutes at 170 watts output as measured with the scope and it barely got warm, I would guess at 5W of heat, so its got to be up there in the mid to high 90s. This is an improvement over the  EL7104cn drivers that were first tested, the deck ran a lot warmer.

Couple pics of the test deck lash up in class E with two different driver configs, phase A has the amplified version and phase B has the standalone. Sitting on the heat sink not connected to anything is the EL7104 driver PCB just for reference.  The inductor is 1.1 uhy, the load cap is 75pf fixed and the variable is a 350pf at about half mesh. The heat sink is 3.5 by 3.5 by3 inch from an old PC AMD CPU.
Will post some schematics and PCB layouts tomorrow after I re-do the driver PCB for stand alone operation.

73s Nigel


Title: Re: 20 Meter GaN FET RF Deck
Post by: VE3ELQ on December 01, 2016, 04:15:43 PM
Schematic attached.  It wont let me upload the driver PCB files, PM me if you want them.

73s  Nigel

Edit:  Forgot to show driver pin 8 tied to ground, please amend, sorry.


Title: Re: 20 Meter GaN FET RF Deck
Post by: N4LTA on January 06, 2017, 01:36:58 PM
Would it be possible to get copies of the driver and PC boards? I emailed you but the email was returned.

Pat Bunn
N4LTA
pbunn@stf-electronics.com


Title: Re: 20 Meter GaN FET RF Deck
Post by: VE3ELQ on January 06, 2017, 02:32:53 PM
Would it be possible to get copies of the driver and PC boards? I emailed you but the email was returned.

Pat Bunn
N4LTA
pbunn@stf-electronics.com

Pat, Sent as requested.  Not sure why Email not working, looks OK here.

73s  Nigel
AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands