Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: [1]   Go Down
  Print  
Author Topic: How Does An Automatic ATU Work?  (Read 2170 times)
TANAKASAN
Member

Posts: 933




Ignore
« on: January 12, 2013, 01:24:49 PM »

The main topic of discussion around our radio club at the moment seems to be automatic antenna tuners. One member is building a unit from a local design and the club themselves have the new Elecraft 500W tuner on order. So, we were sitting around on Thursday night and someone asked how one worked. None of us had any idea.

We all know the basic circuit, a variable capacitor on the input or output and a variable inductor in series, fiddle with the values until it transforms your antenna and feeder into something that looks like 50 ohms. But, how do you control L and C given that your only four inputs are the voltage and phase of the forward and return signals? Is there a mathematical formula somewhere that says 'if forward power is W and X and reverse power is Y and Z then do THIS to your LC combination'?

Of course, LDG aren't talking Roll Eyes

Tanakasan
Logged
AC5UP
Member

Posts: 3927




Ignore
« Reply #1 on: January 12, 2013, 01:37:05 PM »

How does an LM-723 voltage regulator chip work?

If a relatively simple chip can correct for over / under voltage by comparing the Vout to a reference and generate a +/- error signal, why couldn't you do the same with forward / reflected power?  Sure, there's more involved than tweaking the base voltage of a pass transistor, but the basic concept of better / worse then knowing when to stop is essentially the same.

Throw in a simple CPU with a little RAM and the ability to recall previous trial & error adjustments and you have an ATU with smarts.
Logged

Never change a password on a Friday                
KE3WD
Member

Posts: 5689




Ignore
« Reply #2 on: January 12, 2013, 03:16:07 PM »

Well, it can be derived using Analog or Digital Servo Loop methods, where the loop has to be able to use the reported SWR or Power MAX input to find the best case scenario.  Whether or not variable C and L is used is a matter of design also, some that use continuously variable schemes then have to deal with motor drives to the caps and coil shafts, often these use stepper motors and digital control.  Others will switch fixed C and L in and out of circuit via relays to find resonance.  Since it is a Servo circuit, there must also be Hysteresis applied to prevent constant searching. 

Easy to tell the continuous type from the relay type, continuous type doesn't clatter and chatter as it finds the match, although you may hear the smooth sounds of the motors, the relay types are the chatterboxes that click and clack while finding the match. 


73
Logged
M0HCN
Member

Posts: 473




Ignore
« Reply #3 on: January 12, 2013, 06:24:57 PM »

Ok, suppose I place a sensor at the aerial socket that measures the line voltage and line current and the phase between them (A current and voltage transformer in effect).

I can now compute |V|/|I| which gives me |Z| for the load, and as I know the phase angle I can convert from polar to rectangular, giving me a complex value for the load impedance.

It is then simply a matter of computing a suitable L match given the available range of values in play and the frequency of operation.

I would note that  some systems add things like measurement of the phase shift across the series arm to allow better fine tuning for resonance, and some work with the impedance bridge at the input rather then the output, but the principle is there.

Here is about the best introduction I have seen to how to calculate L match networks http://www.ece.ucsb.edu/~long/ece145a/Notes5_Matching_networks.pdf
 
73, Dan.
Logged
WB6BYU
Member

Posts: 13454




Ignore
« Reply #4 on: January 12, 2013, 07:23:51 PM »

How does an SWR bridge work?

You have a voltage divider across the feedline (often capacitive) that gives an
output voltage that is proportional to the voltage on the feedline.

You have a transformer (or coupled line) that provides an output voltage
proportional to the current on the feedline.

When they are exactly equal in both phase and magnitude, the bridge is
balanced, reflected power is zero, and the SWR is 1 : 1.  We detect this
by taking the difference between the outputs and detecting when it is
zero.

To automate the process, you can digitize the outputs from the two sensors
(magnitude and phase) and from that calculate the actual impedance, or at
least the shift required to match the load to the reference impedance.

The more primitive approaches would look a the ratio of voltage to current
to determine the resistive component:  if it was greater than 50 ohms it
would switch the capacitor to the output of the inductor, otherwise it would
put it on the input side.  Then one the two components is set appropriately
to match that impedance.  The relative phase then tells the circuit which
way to adjust the other component to get to resonance, and the process
continues until the impedance is matched.  This works well with motorized
inductors or capacitors where it takes a finite time to reach the final setting,
so allowing some correction along way as the match is approached.

A more modern approach would be to digitize the magnitude and phase
and from that calculate the complex impedance, then switch in the
required components to get a match.  This works well with relay-switched
binary steps for the components, but may require a couple iterations if the
measurements aren't precise enough to get it first try.
Logged
TANAKASAN
Member

Posts: 933




Ignore
« Reply #5 on: January 13, 2013, 04:14:34 AM »

OK, thanks to WB6BYU and M0HCN for the technical explanations, it looks like I have my answer.

Measure the voltage and current passing through the feeder along with the phase difference between them (if any). From there either use a lookup table or compute the L and C values for a match. What I might do next week is set up a dual beam scope showing the voltage and current waveforms and then experiment to see if L or C has to be increased or decreased for a 1:1 match, if I do this at a low enough frequency the sinewave should be easy to display.

Tanakasan
Logged
AC5UP
Member

Posts: 3927




Ignore
« Reply #6 on: January 13, 2013, 05:59:17 AM »

If you want technical detail, this should scratch your itch:  http://www.g3ynh.info/atu/collins180L.html


1957.  Whooda' Thot?  And how cool is that ribbon inductor............... ?
Logged

Never change a password on a Friday                
WB6BYU
Member

Posts: 13454




Ignore
« Reply #7 on: January 13, 2013, 09:30:35 AM »

One simple approach is to build a phase detector circuit, which will tell whether
the load is inductive or capacitive using a zero-center meter.  There should be
circuits for them online, and I think I have a photocopy of an old one from QST.

Then you can take an SWR bridge with the current and voltage sensors and
rectify each individually (rather than combining them at RF) and feed them to
another zero center meter, which would then show whether the impedance
magnitude is greater or less than 50 ohms.

Then you just need to develop an appropriate algorithm for your matching
circuit to determine how to adjust the components accordingly.


One of the slickest circuits I've seen was by W6/SM6MOM  in RF Design as an
Honorary Mention in one of their design competitions.  It was a very simple circuit
using 1/8 wave lengths of coax and diode detectors that gave outputs relative to R
and X.  He fed the outputs to the X and Y plates of a scope and could draw Smith
Charts on the display.  I built a version using a pair of zero-center meters and used
it to check the impedance of a commercial antenna we were converting to 2m in the
days before VHF SWR analyzers:  we calculated a specific length of 75 ohm coax to
match the antenna, and got 1.2 : 1 with no further trimming.  I've been looking back
through my collection of magazines and haven't found it again, but I it shouldn't be
difficult to recreate.  It only works on one band, of course, and is more convenient
at VHF/UHF where the lines are shorter, but it shows how a simple circuit can
resolve the impedance into 4 quadrants based on the sign of two numbers, and if
each output drove a control one way or the other it should arrive at a match.
Logged
K5LXP
Member

Posts: 4521


WWW

Ignore
« Reply #8 on: January 13, 2013, 10:53:03 AM »

Chapter 2 in this Agilent reference gives the various means one can measure an unknown impedance:

http://tinyurl.com/b2gvqhc

Add a micro to crunch the numbers using one of the methods to measure R and X, and the proper reactance to match the load can be selected without having to iteratively making measurements and changes.


Mark K5LXP
Albuquerque, NM
Logged
N4CR
Member

Posts: 1688




Ignore
« Reply #9 on: January 13, 2013, 09:46:36 PM »

They start stepping the inductor (Clunk) and start trying different capacitance's (tink tink tink ...)

Clunk, tink tink tink tink tink
Clunk, tink tink tink tink tink
Clunk, tink tink tink tink tink
Clunk, tink tink tink  (went too far, previous sweep was the one)

Clunk tink (set it)

This is how a lot of them sound including the expensive Palstar. I suspect it's trial and error.

Each sweep plots a curve. When they find something under their threshold, they stop.

Once they have it, they have the correction factor and if there's a frequency counter in there, they can save the frequency and the setting in memory. Next time, they see that frequency they jump to it and test it. If it's bad, they delete it from the memory and sweep. If it's good, they stop. Saves all that hunting when it's a repeat frequency.

Only a few of them I've heard don't sound like this on their first tune although some of them are faster and some of them take more time.

If it was a formula, I'd expect they'd take a few samples, do a calculation and one more move to the final position. This I have yet to hear.
Logged

73 de N4CR, Phil

Never believe an atom. They make up everything.
W4DRR
Member

Posts: 82




Ignore
« Reply #10 on: January 14, 2013, 09:24:01 AM »

Many of the schematics I have looked at for the low-end autotuners only have a simple SWR measuring circuit.  There is no attempt at doing any kind of complex impedance measurement, and then picking the precise L and C combination to achieve a match on the first try.  They tune the same way you would with a manual tuner....tune for minimum SWR.  If increasing L (or C) causes the SWR to go up, then decrease L (or C).

73, Bob
Logged

73,
Bob
WB6BYU
Member

Posts: 13454




Ignore
« Reply #11 on: January 14, 2013, 09:30:45 AM »

After thinking about it over the weekend, here is how the SM6MOM approach works:

You need 4 detector diodes, 4 bypass capacitors, pair of calibration pots, and either
a scope with X-Y inputs or a pair of zero-center meters.

Put one of the detector diodes at the output port with a capacitor from the output
to ground, so it generates a positive voltage relative to the voltage at that point.
Put another one 1/4 wave back up the coax.  If the impedance at the output port
is higher than the coax impedance, the first one will have a higher voltage than
the second.  To get a convenient indication, reverse the diode on the one 1/4 wave
back up the coax so it gives a negative voltage, then connect the two voltages
to the end terminals of one of the pots.  You can then monitor the voltage on the
common terminal and see it go above or below zero.  (You need a DC load across
the coax for the detectors to work properly:  the simplest solution is to put a 3dB
resistive attenuator on one end or the other.

Now, to measure reactance, we build an identical circuit but spaced 1/8 wave from
the first.  In practice, this would involve 3 pieces of coax each 1/8 wavelength long.
Or 4 pieces, actually, since it may be more convenient to use an extra 1/8 wave
at VHF/UHF from circuit to the actual measurement port.

The circuit is calibrated by putting a good dummy load on the output and adjusting
both pots for zero reading (the center of the scope display.)


Now, if you want to build your own auto-tuner to show the club how it works,
you could make such a circuit for, say, 10m, then assemble a tuner using a
parallel-tuned circuit, basically one side of the balanced Johnson Matchbox, with
a capacitor across the coil for setting the reactance and a differential capacitor
to set the output resistance.  These would be motor driven based on the +/-
outputs from the metering system, though as a demo for the club you could just
place one meter above each control and use the club members as the servo
mechanism to rotate the capacitor one way or the other based on the meter
reading.  (An alternate tuner circuit would be a "T" with fixed coil and two
variable capacitors, though you'd have to experiment with which capacitor
worked best with which meter.)  That's a very simple circuit, and limited to one
band, but it should convey the basic principle of how it works.
Logged
Pages: [1]   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!