Automatic tuner implementation
Locked
 
|
"It's all about "pruning the search tree," to borrow a term from game theory."
That always struck me as being a bit like trying to whack a pinata while blindfolded. Why is it not possible to measure the complex impedance and compute the solution? That should be a lot faster than any iterative method. Sure, it needs a better detector but isn't Elecraft about elegant solutions? I'll probably find the answer for myself in a while as it's on the list of things to try with my Kenwood/Elecraft interface (I have LP-100A complex load impedance available). Pointers to any failed or successful attempts to do this would be appreciated. 73, Andy, k3wyc ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html |
Re: Automatic tuner implementation
|
|
Wayne is correct here. If you don't use a principled approach to finding
the the best tuning solution, there are excellent binary searches that have roots in game theory that are extremely good. The optimum is to marry them both. The most successful commercial (militairy) algroithms start with a phase-amplitude transfer function inversion to get very close to a solution followed by a binary search to the final solution. -----Original Message----- From: [hidden email] [mailto:[hidden email]] On Behalf Of Andy Durbin Sent: Sunday, April 7, 2019 8:20 PM To: [hidden email] Subject: [Elecraft] Automatic tuner implementation "It's all about "pruning the search tree," to borrow a term from game theory." That always struck me as being a bit like trying to whack a pinata while blindfolded. Why is it not possible to measure the complex impedance and compute the solution? That should be a lot faster than any iterative method. Sure, it needs a better detector but isn't Elecraft about elegant solutions? I'll probably find the answer for myself in a while as it's on the list of things to try with my Kenwood/Elecraft interface (I have LP-100A complex load impedance available). Pointers to any failed or successful attempts to do this would be appreciated. 73, Andy, k3wyc ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html --- This email has been checked for viruses by AVG. https://www.avg.com ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html |
Re: Automatic tuner implementation
|
|
In reply to this post by ANDY DURBIN
My XYL (Sue, N2LBR) was also interested to try this. She was watching
closely as I was doing alpha testing for the KPA1500 firmware, and was wondering if using the network solution would provide guidance for an iteration scheme. Her solution works even if the complex load impedance is not available. Based on the measured SWR straight through, the minimum and maximum real parts (load resistance) can be calculated, along with the corresponding (plus and minus) imaginary parts (load reactance). This defines the "space" in which the solution exists. A binary search proceeds by first solving the "exact" solution for the given real part (and the corresponding plus and minus imaginary parts), and then using those tuner settings (for L and C) to get a SWR with the actual load (which is modeled in her solution). The algorithm proceeds by zeroing in on the iterated settings that have the lowest SWR. It usually converges in less than 5 iterations to get a SWR less than 1.1:1, even for very large values of (initial, unmatched) SWR. Walt, WA1KKM On 4/8/19 1:20 AM, Andy Durbin wrote: > "It's all about "pruning the search tree," to borrow a term from game theory." > > That always struck me as being a bit like trying to whack a pinata while blindfolded. Why is it not possible to measure the complex impedance and compute the solution? That should be a lot faster than any iterative method. Sure, it needs a better detector but isn't Elecraft about elegant solutions? > > I'll probably find the answer for myself in a while as it's on the list of things to try with my Kenwood/Elecraft interface (I have LP-100A complex load impedance available). > > Pointers to any failed or successful attempts to do this would be appreciated. > > 73, > Andy, k3wyc > ______________________________________________________________ > Elecraft mailing list > Home: http://mailman.qth.net/mailman/listinfo/elecraft > Help: http://mailman.qth.net/mmfaq.htm > Post: mailto:[hidden email] > > This list hosted by: http://www.qsl.net > Please help support this email list: http://www.qsl.net/donate.html > Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html |
|
|
In reply to this post by ANDY DURBIN
Finding a tuner solution need not be an iterative process --- there is a
simple two-step approach if you measure the right parameter as you tune. See QEX, Nov/Dec 2016, p. 3. Chuck MacCluer w8mqw ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html |
Re: Automatic tuner implementation
|
|
Chuck,
I believe that is quite possible, but requires a more sophisticated measurement of the transmission line parameters that the Forward power and reverse power provided by a wattmeter. Measure the parameters of the transmission line and calculate the network needed to bring those parameters to a 50 ohm non-reactive load. It took a great deal of time to do the math when I was in college, but we did not have computers back in the early 1960s. The principle is the same. 73, Don W3FPR On 4/8/2019 10:46 AM, Chuck MacCluer wrote: > Finding a tuner solution need not be an iterative process --- there is a > simple two-step approach if you measure the right parameter as you tune. > See QEX, Nov/Dec 2016, p. 3. Chuck MacCluer w8mqw > ______________________________________________________________ > Elecraft mailing list > Home: http://mailman.qth.net/mailman/listinfo/elecraft > Help: http://mailman.qth.net/mmfaq.htm > Post: mailto:[hidden email] > > This list hosted by: http://www.qsl.net > Please help support this email list: http://www.qsl.net/donate.html > Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html |
Re: Automatic tuner implementation
|
|
In reply to this post by CR
Ideally... yes, practically... no. Given the load impedance and the ideal
impedance, you can always calculate a transfer function that is exact (simply the path of movement on a Smith chart). However, the question then becomes how good are your measurements? A high performance line section and detector (log amp or other) has at best an error of +/- 1db in linearity and resolution. The step size of the tuning elements adds more uncertainty. Bottom line here is that what seems easy can be difficult. -----Original Message----- From: [hidden email] [mailto:[hidden email]] On Behalf Of Chuck MacCluer Sent: Monday, April 8, 2019 9:47 AM To: [hidden email] Subject: Re: [Elecraft] Automatic tuner implementation Finding a tuner solution need not be an iterative process --- there is a simple two-step approach if you measure the right parameter as you tune. See QEX, Nov/Dec 2016, p. 3. Chuck MacCluer w8mqw ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html --- This email has been checked for viruses by AVG. https://www.avg.com ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html |
|
|
In reply to this post by CR
I should have added that QEX failed to print the third page of my article
"How to tune an L-network matchbox" until "Letters to the Editor" in the Jan/Feb 2017 issue. Moreover, my careless dismissing of the low Z case, |Z|< 50, was corrected by K6JCA in "Letters," Mar/Apr issue. If these QEX issues are unavailable, feel free to write me for preprints. Chuck MacCluer, [hidden email] On Mon, Apr 8, 2019 at 2:46 PM Chuck MacCluer <[hidden email]> wrote: > Finding a tuner solution need not be an iterative process --- there is a > simple two-step approach if you measure the right parameter as you tune. > See QEX, Nov/Dec 2016, p. 3. Chuck MacCluer w8mqw > ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[hidden email] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html |
«
Return to Elecraft
|
1 view|%1 views
| Free forum by Nabble | Edit this page |