I'm playing with 600V ex-PSU MOSFETs (with very high Ciss) as HF
switching finals, with some success.

I'd like to put together a driver that takes sinusoidal input into a
fast voltage comparator, and produce a variable pulse-width low-Z
output that could be used to efficiently trim the output power.

With such a driver, a suitable slow data stream into a DAC could be
used to generate a variable control voltage that could in turn cause
the RF output to vary as to conveniently simulate ionospheric fading.
(Just kidding. I promise).

I read specs of many possible ICs types but I am at loss as to which
common and inexpensive types may be suitable, and found no examples of
specific types used in that specific role.

I had klutzed with DC-offsetting a sinusoidal signal into an ugly
nonlinear driver to that effect, but the idea of using a comparator is
from Paul Harden N5AN (who is CCed):

http://www.aoc.nrao.edu/~pharden/hobby/_ClassDEF2.pdf

Any suggestions?

----------------------------------------------

Also, did anyone press into HF service a standard suite of PWM and
MOSFET driver ICs used in automotive applications?

I saw operating frequencies in the order of 400kHz, but with driver
transition times in the 10-20ns range, which is already promising -
although perhaps I would not try them on a MOSFET totempole at 10MHz,
or at least not right away...

----------------------------------------------

TIA!

Filippo N1JPR/I2

Filippo -
You should reconsider trying to PWM your RF switching amplifier. If
you are running Class-E, changing the duty of the drive signal will
not affect your output power significantly, but will destroy your
efficiency if you deviate from 50% duty.

With Class-D power amp, you can PWM the drive to control output power,
but it is very stressful to the MOSFET. However, there is a way to do
it, see US patent 7092269.

Good luck, keep us updated.

73 - Dave NM0S

On Jun 5, 11:32*am, spamhog wrote:
I'm playing with 600V ex-PSU MOSFETs (with very high Ciss) as HF
switching finals, with some success.

I'd like to put together a driver that takes sinusoidal input into a
fast voltage comparator, and produce a variable pulse-width low-Z
output that could be used to efficiently trim the output power.

With such a driver, a suitable slow data stream into a DAC could be
used to generate a variable control voltage that could in turn cause
the RF output to vary as to conveniently simulate ionospheric fading.
(Just kidding. I promise).

I read specs of many possible ICs types but I am at loss as to which
common and inexpensive types may be suitable, and found no examples of
specific types used in that specific role.

I had klutzed with DC-offsetting a sinusoidal signal into an ugly
nonlinear driver to that effect, but the idea of using a comparator is
from Paul Harden N5AN (who is CCed):

http://www.aoc.nrao.edu/~pharden/hobby/_ClassDEF2.pdf

Any suggestions?

----------------------------------------------

Also, did anyone press into HF service a standard suite of PWM and
MOSFET driver ICs used in automotive applications?

I saw operating frequencies in the order of 400kHz, but with driver
transition times in the 10-20ns range, which is already promising -
although perhaps I would not try them on a MOSFET totempole at 10MHz,
or at least not right away...

----------------------------------------------

TIA!

Filippo N1JPR/I2

On Jun 6, 2:00*pm, NM0S
wrote:

If you are running Class-E, changing the
duty of the drive signal will not affect your
output power significantly, but will destroy
your efficiency if you deviate from 50% duty.


Thank you, Dave!

My understanding was that duty cycle would set
output even in class E.


I like to gather silicon from the recycling heap,
and although I have IRF510's, my current victim is
a SK2545, wrung out of a PSU, Pd=40W, Vdss=600V,
and Ciss=1.3nF (!) not easy to stress - actually,
that's exactly why I opted for that class of
device, only partly fit for purpose, but
forgiving, and easy to find.

I already wrung several watts of RF out of at
10MHz it with decidedly ugly and bad drivers which
I did not manage to run in switch mode. They
heated up a lot, but the final got sufficiently
"overdriven" to partly go switchmode, and did not
heat up that much.

I have a totempole ready with complementary BJTs
from a printer driver (2SA2037 / 2SC5694, 60V, 7A,
300MHz), they need .85Vbe to go into saturation.

Ample current is needed to quickly charge and
discharge the MOSFET gate, I reckon it'd take ~
250mA, hence the driver'd better be switching too
rather than overheat in linear mode. So, duty
cycle aside, I need a clean square wave.

MOSFET gate must be swung from 3V to 10V, driver
bases need 8.5Vpp.

I could use a 74HC240 line driver for VXO and amp,
enough current for a IFR510, surely not for Ciss
1000pF, but OK for the complementary pair.

Yet a comparator would indeed be ideal. With it I
could one day also futz with duty and see by
myself what happens... [read John 20:29].

Now, off to read US patent 7092269! :-) After
all, I am shooting for class D, not E. I want to
design a frequency flexible matching and filter
network to cover many bands with few components,
perhaps resorting to old fashioned variable
capacitors, E would be way too complicated anyway,
and efficient output throttling would be nice to
have.



Does anyone have a suggestion for a common and
inexpensive type that works well throughout HF?

Now I see my error w.r.t. class E:
by moving the switching actions along the cycle
you can't possibly time them for the [I=0, V=0] points.

Still ruminating on class D.