[TowerTalk] Fwd: Tower and antenna decisions

[Jim Lux]

On 10/28/13 12:31 AM, Michael Tope wrote:
> On 10/27/2013 11:01 AM, Jim Lux wrote:
>>
>> Whether it will take the SWR to 1:1 is a function of the step size in
>> the tuner design (if it's a switched L and C) and the control algorithm.
>>
>> Most autotuners stop when the SWR is below, say, 1.3:1 or 1.2:1.
>>
>> What you might want is a way to manually configure the L and C, and
>> then store that for the frequency (most tuners can do this now.. the
>> AT200PC can, for sure), or have a computer that knows what frequency
>> you're tuned to, and then set it up.
>>
>> (this is what I was doing with my active phased array.. I used AT200s
>> as essentially computer controlled LC networks)
>>
>
> If you are shooting for really close to 1:1 VSWR to keep an super
> finicky SSPA happy, you probably want something with continuously
> variable matching components in it. RF Concepts (aka Alpha Power) is
> working on one to handle a wide impedance range:
>
> http://www.rfconcepts.com/PRODUCTS/New-Products/Alpha4040
>
> I think they are still struggling with making it work reliably over what
> amounts to a fairly ambitious impedance range, but I am sure it as it
> stands now it would pull in the VSWR of a Tennadyne LPY from 2:1 down to
> 1:1 without breaking a sweat. While it's not exactly the same, the Alpha
> 87A and 9500 as well as the Acom 2000 already do this sort of thing -
> that is - automatically bring in a <3:1 VSWR (<2:1 for the 87A) into
> something close enough to make the finicky ceramic tubes with delicate
> grids happy. The auto tuning of the stepper-motor driven variable caps
> on these products is very fast.
>
> In this case of more forgiving PAs that  don't need super low VSWR, the
> discrete tuning step approach is as you suggest perfectly fine.
>


That's what I was thinking..
I think you could probably adjust the L and C under computer control 
with stepper motors as fast as you can spin the VFO dial.  They do this 
sort of thing on big RF plasma systems (industrial processing) to keep 
the match optimized.

  Big band switches might take a bit longer if you have to make big L 
steps.  OTOH, for a "purpose designed" tuner, you could use relays to 
switch big L in steps, and have a small variable L.

Also, it's really more a matter of how small the LSB is on the relay 
approach.  0.001 microhenry steps (0.37 ohms at 30 MHz)  are doable. 
It's just a cost thing: how many relays do you want to buy.  Narrow band 
is easier (you probably don't need 0.001 uH steps for top band)

The challenge is in calibrating. I've measured 5 AT200PC tuners and
a) they are not the same as each other
b) the steps are not linearly combined.  That is, if you have a 1,2,4 
sequence, turning on the 2 does not always add the same inductance or 
capacitance.  It's, I suspect, coupling between the components, and 
interaction with the wiring.

This isn't a problem in the "autotuner" because the search algorithm can 
always find a minimum, and as long as the steps are always getting 
smaller (i.e. it's monotonic) then it works.

As a result, though, you can't set a particular L and C in an a-priori 
sense.  For the LPDA/Yagi matching scenario, you'd have to make a cal 
run, and store the actual values you need for each frequency.

This is really no different than the SteppIR thing, if you tweak the 
factory default settings.

For the phased array scheme, it's a WHOLE lot more complex because of 
the mutual impedances between elements (that is, changing the match at 
the feedpoint of element A changes the Z at feedpoints of elements B, C, 
and D).  It starts to get very complex, because you now need to measure 
the element current mag and phase in real time. (A pain in the rear 20 
years ago, not so much today, though)

This is why I'm thinking that synthesizing the reactive component with a 
suitable PA, rather than using switchable passive components for "power 
factor correction", is a nice idea. (it also lends itself to very rapid 
frequency changes)