RF "Lego" blocks.
The idea behind this work is to build radio-things (TM) in stages which connect to each other.
Note: This HF-PA-v6 is like https://github.com/kholia/HF-PA-v5 which has been
broken out in stages. This makes the designing, and debugging processes much
easier.
October-2022 UPDATE: The PCB manufactured by JLCPCB was successfully validated. We could get 5W+ with IRF510 @ 20v drain @ 28 MHz. Output at 14 MHz can be more than 10W easily too.
PCB bug (in HF-PA-Combined version): Please cut the positive trace going
between C15 and JP2 to enable dynamic PA voltage control using PTT. This bug
is fixed in HF-PA-Combined-v2 version for future production.
NOTE: See https://github.com/kholia/HF-PA-v6/tree/master/HF-PA-Combined for the schematics, design files, and a PCB render screenshot.
IRF510 @ 14 MHz -> 20v @ drain. RF output > 7.5W. No LPF. Clean'ish looking output. No oscillations. BS170 @ 5.5v. DC bias set high for max'ish "Idq".
IRF510 @ 21 MHz -> 15v @ drain. 12.75W DC input. RF output > 3W. No LPF. Clean'ish looking output. No oscillations. BS170 @ 5.5v. DC bias set high for max'ish "Idq". Almost 4W can be achieved with max "Idq".
IRF510 @ 28 MHz -> 14v @ drain. 9W DC input. RF output > 2W (on meter). No LPF. Clean'ish looking output. No oscillations. BS170 @ 5.5v. DC bias set high for max'ish "Idq".
IRF510 @ 28 MHz -> 17v @ drain. RF output > 2.25W (on scope), 3.5W on RF power meter. No LPF. Clean'ish looking output. No oscillations. BS170 @ 6v. DC bias set high for max "Idq" @ 4.2v roughly.
Note: We can even use 30v (and higher+) drain voltages (PA voltages) with IRF510 - with much more care though!
Even 3W can get worldwide DX contacts when propagation is playing along ;)
7W (or less) of HF RF power offers global communications coverage when propagation is good:
See https://github.com/kholia/HF-PA-v5 for the BOM and BOM sources.
L1 can be a simple choke (FT37-43 25 turns of 27 to 29 SWG wire).
L2 needs to be a bifilar transformer on FT50-43 core (10 turns with 24/25/26 SWG wire).
As Vds is increased, the Vgs needs to be increased too - see the following graph.
If the gate DC bias is too less, the MOSFET will not turn on properly, and will generate a lot of distortion - one prominent distorted waveform will be at half-the-operating frequency!
When operating at max "Idq" (for best linearity), the power to the IRF510 should be PTT controlled - otherwise too much (battery) power will turn to heat instead of RF!
While IRF510 can get destroyed due to oscillations, in my case it has been a failed 7805 (or 78L05) regulator which ultimately causes cascading failures.
A 7805 regulator can fail in fail-short mode, where Vout == Vin! This changes
the bias of the driver stage (BS170), and it goes in full conduction
immediately without any RF input because Vgs(th) is exceeded quite easily.
This causes further failures in the PA system!
My 7805s (and 78L05s) of dubious origin were failing (Vout shorted to Vin!)
around 20v to 24v. Here is what bd139 (from https://www.eevblog.com/forum/)
has to say on this topic,
None of the voltage regulators work properly above about 20V if you want 5V
out. If you need to drop more, add a preregulator.
I found this out the hard way as well dropping 28V to 5V DC for logic in an RF
PA driver. Currently waiting for a board spin. I only needed 50mA and didn't
want RF noise so I went with a simple MOSFET/zener/resistor preregulator to
drop it down to 18Vish. Because the current requirements are relatively low I
don't need to heatsink either then as the load is spread evenly across both
MOSFET and 7805.
Also if you value the circuit it's driving, stick a 5W 5.6V zener across the
regulator output. That will smoke the prereg / regulator not the load!
Thank you bd139!
Are you interested in such projects, and spam-free technical discussions?
If yes, you might be interested in joining the 'HAMBREWERS' Telegram group.