[600MRG] Amp

[John Langridge]

Hi Jim,

There is a very important point that I think I  failed to mention here
which changes all of this, for better or worse, and I realized it after you
posted your latest comment.

Purest are going to freak out about what I am about to say but here goes:
I put a very minimal sine wave onto the gate of the FET rather than
squaring it up.  Its not over driven, perhaps a little under driven.  I
cannot tell a change is dissipation to the touch compared to the FET on my
MF Solutions converter that is using  proper squaring, etc.  It gets warm
but not hot  (125w from each amp in a combiner pair) and I do not draw
excessive current while power output is nominal, as measured on the drain.

Note that I have over 10000 hours of actual RF emissions using this
approach and I've not had a FET failure in 2 years and the last failure was
a result of a mismatch prior to starting to use the scope match for tune up.

What about harmonics?  Well, in early 2013, after using this approach for
several months without problems, it was recommended that I look at squaring
that up.  I'm stubborn so instead I had a former coworker bring a spectrum
analyzer over to just look at the surrounding band while transmitting.  You
know what?  it was fine.  no trash, no spurs, nothing.  Clean signal, with
no pops as I key.

During CW QSO's particularly after going to combined amplifier operation I
would ask the other station to listen for clicks - not at minimal signal
levels, but times when the signal was S9+.  No clicks detected.  No
harmonics and no trash as a result.  That was using the HF rig driver -->
MF Solutions converter --> Pi Attenuator -->splitter to each amp
-->combiner --> LPF, etc

So everyone has their approach and thats not a problem.  Everyone has to
decide what works best for them.  I have no plans to change anything here
because it is working like "Grease Lightning".

I used to have similar discussions at the university with the theoreticians
who wanted to argue that their numbers had to be right when the actual
experimental data showed something else.   I'm happy to get on and transmit
a bit if you are able to hear me this time of year.  I'm QRV every night
running something.

73!

John XIQ

On Mon, Aug 3, 2015 at 2:03 PM, Jim Miller <jim at jtmiller.com> wrote:

> The way I'm thinking about this is the drive to the finals (FETS usually)
> needs to be hard and square at 472khz in order to achieve the desired
> minimum switching transition times. This means the drivers need to hit the
> finals with a 472Khz square wave. Anything less than a square wave at this
> point will result in excessive power dissipation if this is a continuous
> occurrence. Obviously not desirable.
>
> OTOH a finely shaped, e.g, raised cosine (ok, I know its not optimal...)
> RF envelope coming into the amp will have around a 5ms period of rising and
> falling signal. How the predriver, driver and final stages deal with this
> ramp will determine whether or not "key clicks" will result.
>
> It seems to me that the amp will necessarily sharpen those transitions in
> order to ensure than switching time at 472khz is minimized and therefore
> produce broader CW signals than if a linear was used.
>
> This should be easy to test on the air. I have a P3 and I'm used to seeing
> "clicky" CW all the time in contests. I usually see a handful of 2-3khz
> signals each time and occasionally a 5khz monster. If someone wants to do
> an A/B comparison on the air let me know and I'll give a look. I don't have
> a experimental license so I can only receive.
>
> 73
>
> jim ab3cv
>
>
>
> On Mon, Aug 3, 2015 at 11:54 AM, John Langridge <kb5njd at gmail.com> wrote:
>
>> >CW is a square wave modulation that is full of harmonic
>> energy which will cause horrible clicks out many KHz if not properly
>> conditioned.
>>
>> I think the key here is "properly conditioned" and specifically related to
>> the RF source.  If the rise and fall times of the *source* CW signal is
>> purely square, then yeah, you are going to have a clicky, harmonic-filled
>> signal.  If, however, the source signal rounds that waveform off at the
>> transitions like it should be, you are not going to generate additional
>> trash (assuming the PS has capacity and is not AMing, for example).  This
>> was the topic of discussion for years on the topband reflector WRT various
>> radio manufacturers and their click problems.  If you are using an HF rig
>> as an exciter, the quality of that signal source is largely
>> going to determine the quality of what comes out of the amp.
>>
>> One of the amps that I built uses a fixed signal source where the amp's
>> drain is keyed directly and, in fact, through RC networks, the rise and
>> fall times are modified accordingly.  The CW waveform is clean, there are
>> no
>> clicks and there is no broadening of the signal because the shape of that
>> input square wave from the driver is rounded appropriately.
>> Neither of the techniques that I use for making a CW signal is clicky or
>> wide because the source signal is "clean" coming into the amp.
>>
>> I think you are correct about linear if we assume that the driver signal
>> is
>> poor.  But it does not have to be and many guys who have been running
>> topband for a long time have either switched to clean radios or modified
>> their existing radios, as I have done.
>>
>> 73,
>> John XIQ
>>
>> On Mon, Aug 3, 2015 at 10:28 AM, Edward R Cole <kl7uw at acsalaska.net>
>> wrote:
>>
>> > John,
>> >
>> > Unfortunately too many hams believe this.  An unswitched carrier is fine
>> > with non-linear modes.
>> >
>> > This was topic was just thoroughly discussed recently on the Elecraft
>> > Reflector.  CW is a square wave modulation that is full of harmonic
>> energy
>> > which will cause horrible clicks out many KHz if not properly
>> conditioned.
>> > A carrier with harmonic energy will have the harmonics amplified
>> greatly by
>> > a non-linear amp;  that is why they are used for triplers and
>> multipliers.
>> > If you want harmonics just overdrive an amp into non-linear operation.
>> >
>> > You will not find any class-C amps offered for hams since CW is now
>> > generated by an exciter and not by direct keying of the amplifier as was
>> > done years ago (of course these amps also are used for SSB, but no mfr
>> > offers a bias adj to place the amp into class-C for CW).  Also, CW
>> > modulation is not zero bandwidth; only a carrier with zero phase
>> modulation
>> > is zero bw.  Even back in pre-SSB years when one used grid-block keying
>> the
>> > CW voltage still had to be shaped to avoid clicks; something that modern
>> > transceiver mfr's seem of have forgotten (except a few wise one's).
>> >
>> > If you run CW thru a non-linear amp the result will be broadening of the
>> > bandwidth and co-channel interference.  FM and some FSK modes are OK.
>> >
>> > Therefore, my amplifiers will be linear; its not that hard to do!
>> Output
>> > filtering is a must to reduce harmonic output even with linear amps.
>> >
>> > 73, Ed - KL7UW
>> >
>> > At 03:33 AM 8/3/2015, John Langridge wrote:
>> >
>> >> >Linear is needed for CW
>> >>
>> >> Linear is NOT required for CW.. only when amplitude and/phase varies is
>> >> linear required.  It would be needed for PSK, which has been rarely
>> used
>> >> in
>> >> the last few years.
>> >>
>> >> 73!
>> >>
>> >> John XIQ
>> >>
>> >> On Sun, Aug 2, 2015 at 11:25 PM, Edward R Cole <kl7uw at acsalaska.net>
>> >> wrote:
>> >>
>> >> > Jim, and all:
>> >> >
>> >> > The K3 can be upgraded to operate down as low as 100-KHz by upgrading
>> >> the
>> >> > Synth board for $219.95.  Reception is improved by a minor change in
>> the
>> >> > General Coverage band filter and one component on the main board.
>> >> >
>> >> > So it is capable at this time by incorporating those changes.
>> Elecraft
>> >> > has just rolled out a new model called the K3S which comes standard
>> with
>> >> > these.
>> >> >
>> >> > 100w seems a good target for 5w ERP (depends on your antenna
>> >> efficiency).
>> >> >
>> >> > I think one should just figure on using a linear amp.  Linear is
>> needed
>> >> > for CW and psk modes and gain is not difficult to achieve at
>> 500-KHz.  I
>> >> > would guess 60-65% efficiency in the high power stage so you need
>> >> > dissipation of say 50% of RF output as a minimum to have a safe
>> margin.
>> >> > 100w dissipation would ensure almost bullet proof running of any
>> mode.
>> >> >
>> >> > Not sure you saw my recent post on the Elecraft reflector regarding
>> 630m
>> >> > amplifiers?  I will be looking at making up a 100w linear amp that
>> runs
>> >> > with 1mw drive sometime this winter when outdoor stuff is not as
>> nice to
>> >> > do.  I am giving thought to offering it as a kit or partial-kit for
>> >> sale.
>> >> > I was once considering transverters but others have beat me to that
>> and
>> >> > with the K3 and TS590S both able to run 630m an amplifier is the
>> needed
>> >> > item (probably a preamp too).
>> >> >
>> >> > Don't ask for particulars at this point - just an idea that needs
>> some
>> >> > research and building a proof of concept.  No idea of cost.  Most
>> likely
>> >> > will use FETs and run with 13.6v for simplicity, though 28v and 50v
>> >> > transistor specs are better.  Driving with 1mw makes such an amp
>> usable
>> >> > with simple mixer schemes, as well.
>> >> >
>> >> > I discussed this with Wayne-N6KR of Elecraft and the market is too
>> small
>> >> > for them to consider making an amp.
>> >> >
>> >> > 73, Ed - KL7UW
>> >> > http://www.kl7uw.com/kits.htm
>> >> >
>> >> > At 05:10 PM 8/2/2015, Jim Miller wrote:
>> >> >
>> >> >> Now that the K3 is getting closer to being a 630 capable machine and
>> >> the
>> >> >> FCC is lumbering toward a band decision I'm thinking more about an
>> amp.
>> >> >> I'm
>> >> >> assuming the market is too small for an Elecraft offering.
>> >> >>
>> >> >> The K3 puts out 1mW on the transverter output. I'm assuming that
>> I'll
>> >> need
>> >> >> 50-100w to get to 5w ERP.
>> >> >>
>> >> >> Assuming 100w for a minute and assuming 10-15db for the final amp
>> stage
>> >> >> then you only need 5-10w before then. That should mean that even
>> with
>> >> poor
>> >> >> efficiency not a lot of power will be dissipated until the finals.
>> That
>> >> >> means that I could keep the chain to that point linear and make the
>> >> output
>> >> >> stage decision based on what modes are most prevalent.
>> >> >>
>> >> >> I assume CW for historical reasons and JTXX for practical reasons,
>> >> WSPR as
>> >> >> well. Not sure about PSK or other modes which would require a more
>> >> linear
>> >> >> final.
>> >> >>
>> >> >> So I could build a 1mW to 10W linear then wait and see about the
>> >> finals.
>> >> >>
>> >> >> Thoughts?
>> >> >>
>> >> >> 73
>> >> >>
>> >> >> jim ab3cv
>> >> >> _______________________________________________
>> >> >> 600MRG mailing list
>> >> >> 600MRG at w7ekb.com
>> >> >> http://w7ekb.com/mailman/listinfo/600mrg_w7ekb.com
>> >> >>
>> >> >
>> >> > 73, Ed - KL7UW
>> >> > http://www.kl7uw.com
>> >> >     "Kits made by KL7UW"
>> >> > Dubus Mag business:
>> >> >     dubususa at gmail.com
>> >> >
>> >> >
>> >> > _______________________________________________
>> >> > 600MRG mailing list
>> >> > 600MRG at w7ekb.com
>> >> > http://w7ekb.com/mailman/listinfo/600mrg_w7ekb.com
>> >> >
>> >> _______________________________________________
>> >> 600MRG mailing list
>> >> 600MRG at w7ekb.com
>> >> http://w7ekb.com/mailman/listinfo/600mrg_w7ekb.com
>> >>
>> >
>> > 73, Ed - KL7UW
>> > http://www.kl7uw.com
>> >     "Kits made by KL7UW"
>> > Dubus Mag business:
>> >     dubususa at gmail.com
>> >
>> >
>> _______________________________________________
>> 600MRG mailing list
>> 600MRG at w7ekb.com
>> http://w7ekb.com/mailman/listinfo/600mrg_w7ekb.com
>>
>
>