[TowerTalk] trying 3 different T-tuner suggester programs: TLW seems to lead an unwary user (me) astray.

[David Gilbert]

That's why when I use TLW I initially specify large value capacitors so 
I can see what's going on first.  I do the same thing with the W9CF 
online tuner app.  I don't think there are many cases for an antenna 
tuner where you'd want to specify a Q since generally speaking you 
simply want it to be low ... you don't want to have to adjust the 
setting as move around the band.  Most of the time (not always), an 
L-Network is going to give you the best efficiency, and the best 
impersonation of an L-Network by a T-Network is when either the input 
capacitance or the output capacitance is as high as you can make it.

Generally speaking, there are only three reasons to use a T-Network in 
the first place:

1.  L-Networks only match in one direction, and it makes a difference 
whether the load is less than or greater than the source as far as which 
side of the inductor (assuming a shunt inductor) the capacitor is on.  A 
T-network is capable of pretending it is an L-Network in either direction.

2.  The L-Network capacitance and inductance values get almost 
impossible to achieve for loads that are close to the source impedance.  
Try some samples and you'll see what I mean.  In those situations a 
T-network essentially uses the extra capacitor to shift either the 
source or the load impedance so that the L-Network portion can 
accomplish a match.  An older generally well regarded commercial tuner 
billed itself as a more efficient L-Network tuner, but they included 
provision for switching in a fixed capacitor (or two, if I remember 
correctly) on the other side of the coil in order to match loads near 
the source.

3.  An L-network only has two degrees of freedom ... the coil and the 
capacitor ... and both have to be continuously variable to achieve a 
match.  That means you have to use something like a rotary inductor, 
which also requires a turns counter if you want to be able to quickly go 
back to predetermined settings.  Both are expensive.  A T-Network allows 
you to find a match for an inductance that is merely close to the 
desired value ... i.e., a fixed coil with taps ... even if the resultant 
efficiency might not be optimum.

73,
Dave   AB7E



On 12/1/2010 1:20 AM, Kevin Normoyle wrote:
> Steve G3TXQ said "If you want to get meaningful loss results with that simulator, be
> prepared to manually change the default Q value as you change bands and
> change inductance values."
>
> Yes, if you want to just see the result for a known combination. I ran into a surprise assuming I understood how some things out there create a "suggested" result, not even considering the constant vs varying component Q issue.
>
>
> I was comparing the suggested values produced by 3 T-Tuner simulations, and was
> surprised to see the difference in results. Be interested in feedback about what
> other people have used/done.
>
>
> 1) 4nec2 which allows you to specify min network Q (lower network Q = lower
> loss). Don't need filtering, so high Q result isn't useful.
>
> 2) ARRL Antenna Book TLW (n6bv 2004)
>
> 3) The very nice t_tuner.exe from G4FGQ (free) which allows you control L and C
> ratios, and max C so you can get different results  (varying L/C ratio and
> comparing different loss/volts/current results)
> http://www.smeter.net/feeding/t-network-antenna-tuner.php
>
> 4) ARRL Antenna Book ATT. I just loaded this, and see it constrains things
> differently. Am playing with it running in DOSbox.
>
> t_tuner.exe and 4nec2 quickly let you see the effects of varying L and C for a
> given match.
> 4nec2 lets you specify a min network Q for the result.
> t_tuner  allows you to control the L/C ratio with an artificial parameter "A"
> with small values given large C and lowest loss.
>
> (G4FGQ also has l_tuner.exe, (L network) but I focused on T)
>
>
> I got really confused looking at the TLW results. For my test case, they
> suggested a result, which was very bad for power loss?
>
> I played with both 4nec2 and t_tuner.exe, trying to get the same predicted
> result. I get them to create the one result TLW suggests, but it's clearly not
> optimum. It's a high Q result, when I want a low Q result.  (low loss) (I'm
> using TLW high-pass Tee Network Tuner selection).
>
> At first I thought "This seems like a bug in TLW".  I found the problem below,
> which I am calling a program design flaw..i.e. you have to have some knowledge
> to realize why it's a bad result. It's easy to believe there is only one
> possible result, and not tune things to get something good.
>
> t_tuner is best for getting different results and comparing.
> Even if I constrain t_tuner to use C of<350pf, a nice low Q result is suggested.
>
> The TLW suggested poor result, seems to have a Q of about 60 (reported when same
> values used in the other programs)
> The TLW result is a correct match. (verified elsewhere)
> The suggested result in TLW doesn't seem to vary, if the unloaded component Q is
> changed.
> Those unloaded Qs just affect the final loss calculations.
>
> Here's the problem:
> TLW allows you to vary results by controlling one of the C's (Tee Network Series
> Capacitor).  You have to "know" that it needs to be increased to get a good
> result in my test case (and may need to be small to get a good result for a
> different test case..or intermediate).
>
> If you keep that small, in my case, you get a bad result. (this goes to the
> question of the goodness/badness of different L/C match combinations)
>
> It would be nice if they all reported final network Q like 4nec2, but seeing the
> power loss is related, so okay I guess.
>
> So what TLW is really doing is giving you a good result for what you specified,
> but doesn't give you enough info to realize you need to tweak the input you gave
> it, to see different possible results.
>
> It is much nicer to specify a min-max range for C, or L, and maybe a final
> network Q, or have it find the lowest loss result.
>
> I haven't find a program that gives a single suggested optimal result using the
> known Q of the unloaded components? Even 4nec2 seems to not use component Q's
> when reporting a  "min network q" of the result, although he modifies the info
> used later, with component Qs.
>
> Are there some other good ways other there of getting comparative calculated
> results that I'm missing?
>
> It seems like this could all be easily automated. Computers are fast enough now
> so that all the constraints could be described, and walk over all possibilities
> and draw the very nice info "map"  like  G3TXQ posted. (that represents "what's
> doable" with the particular set of constraints in a single real tuner).
>
> Like an excel spreadsheet would be nice.
>
> -kevin
> ad6z
>
>
>
>
> _______________________________________________
>
>
>
> _______________________________________________
> TowerTalk mailing list
> TowerTalk at contesting.com
> http://lists.contesting.com/mailman/listinfo/towertalk
>