Forums

RF transformer

Started by Robert Baer July 29, 2013
   I understand, that with the proper winding configuration, that a 90 
degree phase shift can be achieved between primary and resonant secondary.
   That is done with a tapped secondary on a FM discriminator transformer.
   What are the guidelines and requirements needed to construct such a 
transformer at a given frequency OTHER than the standard 10.7Mhz?
   Say, on a form in the one inch diameter region, and frequency below 1Mhz?
   Ideal coil diameter, length based on chosen frequency?
   Turns / inductance (no resonating cap)?


"Robert Baer" <robertbaer@localnet.com> wrote in message 
news:jumJt.296288$aW2.143211@fx13.iad...
> I understand, that with the proper winding configuration, that a 90 > degree phase shift can be achieved between primary and resonant > secondary. > That is done with a tapped secondary on a FM discriminator > transformer. > What are the guidelines and requirements needed to construct such a > transformer at a given frequency OTHER than the standard 10.7Mhz? > Say, on a form in the one inch diameter region, and frequency below > 1Mhz?
You need LL high enough (k low enough) that the resonant tank, at whatever Q it ends up at, doesn't draw an excessive amount of current (excessively double-humping that stage's IF response), without being so high that the signal is too small. Just basic IF transformer and bandpass filter stuff.
> Ideal coil diameter, length based on chosen frequency?
Well, you generally get best results with a coil 0.5 to 2" long, pitch twice the wire diameter. Assuming that the form is 1" as stated. That's true of coils in general; were you expecting something specific to this application?
> Turns / inductance
Depends on circuit impedance, but easily figured from the nature of the network.
> (no resonating cap)?
As in, self resonant? At <1MHz? You're going to need more than 1" diameter to get a resonator down there. Especially with any useful Q. Tim -- Deep Friar: a very philosophical monk. Website: http://seventransistorlabs.com
On Sun, 28 Jul 2013 21:48:14 -0800, Robert Baer
<robertbaer@localnet.com> wrote:

> I understand, that with the proper winding configuration, that a 90 > degree phase shift can be achieved between primary and resonant secondary. > That is done with a tapped secondary on a FM discriminator transformer.
That would be a Foster Seeley Discriminator, except there is no 90 phase shift involved: <http://www.tpub.com/neets/book12/51c.htm> A 90 degree phase shift demodulator would be a quadrature demodulator.
> What are the guidelines and requirements needed to construct such a > transformer at a given frequency OTHER than the standard 10.7Mhz?
Make sure the xformer bandwidth (or Q) is sufficiently low that it covers the range of frequency deviation you're expecting. Too wide a range, and you get low detected output voltages. Too high a Q or too narrow a range, and the discriminator can't handle the maximum FM/PM deviation. Note that both the primary and secondary of the discriminator xformer are tuned.
> Say, on a form in the one inch diameter region, and frequency below 1Mhz? > Ideal coil diameter, length based on chosen frequency? > Turns / inductance (no resonating cap)?
Fill in the blanks: 1. Center frequency. 2. Maximum deviation or modulation acceptance bandwidth. 3. Maximum modulation frequency (or modulation index). 4. Linearity required (optional). 5. Do you need AM rejection? Discriminator is also sensitive to AM. 6. What do you have against tuning capacitors? There are too many other specs that have to nailed down (size, tuning range, stability, footprint, solderability, etc) in order to design a proper discriminator coil. I don't want to do that. -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
On Sun, 28 Jul 2013 22:48:14 -0700, Robert Baer wrote:

> I understand, that with the proper winding configuration, that a 90 > degree phase shift can be achieved between primary and resonant > secondary. > That is done with a tapped secondary on a FM discriminator > transformer. > What are the guidelines and requirements needed to construct such a > transformer at a given frequency OTHER than the standard 10.7Mhz? > Say, on a form in the one inch diameter region, and frequency below > 1Mhz? > Ideal coil diameter, length based on chosen frequency? > Turns / inductance (no resonating cap)? > >
use free femm 4.2 to design coil use LTspice to design 'circuit' based upon parameters determined from femm. BOTH user groups, super helpful
On a sunny day (Mon, 29 Jul 2013 05:56:05 -0700) it happened RobertMacy
<robert.a.macy@gmail.com> wrote in <op.w0zh7rng2cx0wh@ajm>:

>On Sun, 28 Jul 2013 22:48:14 -0700, Robert Baer wrote: > >> I understand, that with the proper winding configuration, that a 90 >> degree phase shift can be achieved between primary and resonant >> secondary. >> That is done with a tapped secondary on a FM discriminator >> transformer. >> What are the guidelines and requirements needed to construct such a >> transformer at a given frequency OTHER than the standard 10.7Mhz? >> Say, on a form in the one inch diameter region, and frequency below >> 1Mhz? >> Ideal coil diameter, length based on chosen frequency? >> Turns / inductance (no resonating cap)? >> >> > >use free femm 4.2 to design coil
It is actually really simple: Find a nice C value, 220 pF is my favorite. Use online calculator to find L for resonance, or estimate from Xperience. Use standard coil formers (8 mm diameter, or 6 mm diameter, whatever), with adjustable cores. Add some turns (how about 100? for 1MHz), apply your F from source or generator. Tune for maximum. Now make 2, put them, 1 cm next to each other, wind 5 to 10 coupling turns from the first one, your LF comes from that. do that diode thingy. I have done it twice, for TV audio (was 5.5 MHz FM here, thing then needs to be very precise as else you get video rattle in the audio. (the 5.5 MHz came from the video detector). Did not take more than an hour to build and adjust. Later they had TBA120 quadrature detector, ebay... At 1MHz 4046 PLL should demodulate FM too, did an audio link with that. Is it narrow band? Then that 2 coil thing is likely no good, and you MUST use a quadrature detector, Motorola once had some nice chips. So need more info. In the same time you look that up you can try it out. LTspice is only for self-justification and to show customers ;-)
On Mon, 29 Jul 2013 06:31:45 -0700, Jan Panteltje  
<pNaonStpealmtje@yahoo.com> wrote:

> On a sunny day (Mon, 29 Jul 2013 05:56:05 -0700) it happened RobertMacy >> ...snip.... >> use free femm 4.2 to design coil > > It is actually really simple: > ..excellent construction directives snipped... > So need more info. > In the same time you look that up you can try it out.
I like your, "JUST DO IT!" approach. Tens to result in working breadboards.
> LTspice is only for self-justification and to show customers ;-)
When LTspice first started out, seemed a bit like that. But now, LTspice has mutated into a generic super tool and has essentially scooped the SPICE simulation market for ease of use, accuracy, access to support, access to models [thanks to an intrepid individual in Russia, Alex Bordodynov] I use LTspice extensively for modeling EVERYTHING, from electromechanics to cabling to EMC/shielding to ESD protection, even antenna designs, etc. which contain NO Linear components. And now with Mike's new updated version... On a particularly mind boggling topology that I've been working with since 1989! LTspice predicted at 100Hz there would be 35nV/rtHz noise and the breadboard had an actual MEASURED noise of 36nV/rtHz!!! Considering that S/N is the next frontier for improvement in my circuitry, LTspice is USEFUL! I used to be reduced to soldering/resoldering, head scratching until hair was gone, and then repair repair repair the !@#$#!@#$ overworked breadboard; now, can sit and explore all kinds of 'improvements' to the topology ...using LTspice. and even better, keep a record of the attempts. Breadboards tend to get lost/destroyed along the way. If you're goinng to pick on simulation tools that are self serving, you should go after National and their online tool for SMPS design, with so many 'unknown to you' approximations you're likely;, as happened to me, to get wrong results. And, Analog Devices who likes to make horrific, 'unstable' models, or TI who likes to make TINA-only models. I'll stick with any company that has the courage to be 'opensource'.
On Sun, 28 Jul 2013 23:54:03 -0700, Jeff Liebermann <jeffl@cruzio.com>
wrote:

There's a detailed explanation of how to design an FM discriminator
for AFC in the Radiotron Designers Handbook 4th edition (1953) Pg
1152-1156.  
<http://www.paleoelectronics.com/RDH4/CHAPTR29.PDF>
Note the concern for controlling the Q of both the primary and
secondary.


-- 
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558
On a sunny day (Mon, 29 Jul 2013 07:04:15 -0700) it happened RobertMacy
<robert.a.macy@gmail.com> wrote in <op.w0zlddr02cx0wh@ajm>:

>On Mon, 29 Jul 2013 06:31:45 -0700, Jan Panteltje ><pNaonStpealmtje@yahoo.com> wrote: > >> On a sunny day (Mon, 29 Jul 2013 05:56:05 -0700) it happened RobertMacy >>> ...snip.... >>> use free femm 4.2 to design coil >> >> It is actually really simple: >> ..excellent construction directives snipped... >> So need more info. >> In the same time you look that up you can try it out. > >I like your, "JUST DO IT!" approach. Tens to result in working breadboards. > >> LTspice is only for self-justification and to show customers ;-) > >When LTspice first started out, seemed a bit like that. But now, LTspice >has mutated into a generic super tool and has essentially scooped the >SPICE simulation market for ease of use, accuracy, access to support, >access to models [thanks to an intrepid individual in Russia, Alex >Bordodynov] > >I use LTspice extensively for modeling EVERYTHING, from electromechanics >to cabling to EMC/shielding to ESD protection, even antenna designs, etc. >which contain NO Linear components. > >And now with Mike's new updated version... On a particularly mind boggling >topology that I've been working with since 1989! LTspice predicted at >100Hz there would be 35nV/rtHz noise and the breadboard had an actual >MEASURED noise of 36nV/rtHz!!! Considering that S/N is the next frontier >for improvement in my circuitry, LTspice is USEFUL! I used to be reduced >to soldering/resoldering, head scratching until hair was gone, and then >repair repair repair the !@#$#!@#$ overworked breadboard; now, can sit and >explore all kinds of 'improvements' to the topology ...using LTspice. and >even better, keep a record of the attempts. Breadboards tend to get >lost/destroyed along the way. > >If you're goinng to pick on simulation tools that are self serving, you >should go after National and their online tool for SMPS design, with so >many 'unknown to you' approximations you're likely;, as happened to me, to >get wrong results. And, Analog Devices who likes to make horrific, >'unstable' models, or TI who likes to make TINA-only models. I'll stick >with any company that has the courage to be 'opensource'.
Oh, I use LTspice, usually for small sub-circuits, but hey, human mind has a trillion ? of neurons (or it should be ;-)), and more than 60 years E design experience in my case. I just think differently. For example in case of 'coil', if you get ONE reference coil say 220 pF at 6 turns = 25 MHz, then, as w^2.L.C = 1, and L goes up as the square of turns.... you could estimate that for 1 MHz and 220 pF, 6 x 25 turns = 150. But some will be over-wound, closer space, 100 would be enough, and that would be with core half in. No LTspice, no online calculator, no complicated math, just some weapons of math destruction, wind the 100, and you are done. I wind 100 faster than I can write this. And 100 you know, as the grocer says when weighting the cheese: Can it be a little more? It is easy to remove a few turns... if you end up too high. Now, now it works, you can ask the online calecujilator what L is for 1 MHz, and put the in LTspice, and show it to customer, he (the custardtomer) now feels much more secure after seeing 'mathematical proof'. Hey I just had a discussion at CNN.com. and found out how to get rid of the US deficit: Advertising on dollar bills, on banknotes, how much does google make ... FED can do it too. Made available to all for free. (And in that discussion I invented gambling on banknotes too). So there you are, I like LTspice, in runs in wine (that is a windows emulator for the microsof users), so I can actually use it. It is the best I have seen (LTspice, not wine).
On Mon, 29 Jul 2013 08:23:20 -0700, Jan Panteltje  
<pNaonStpealmtje@yahoo.com> wrote:

>> ...snip... > > Oh, I use LTspice, usually for small sub-circuits, > but hey, human mind has a trillion ? of neurons (or it should be ;-)), > and more than 60 years E design experience in my case.
I know. I saw the 'smiley' I just used the opportunity to vent my spleen on three entities that should know better. Using my own brain? I pride myself on looking at a schematic and actually view it 'moving' Instantly see what it does, where it's weak, etc. But, this topology has driven me NUTS! Using my intuition to increase bandwidth, the change reduces bandwidth. Increase gain, nothing happens - constant gain! I sat with a tablet of equations for two days solid to no avail, even with super simplifications. Now even with LTspice this !@#$#@! circuit doesn't 'act' right! where the circuit should null and be the same voltage, it's not! I'm starting to simply take the attitude, it works, because it works.
> I wind 100 faster than I can write this. > And 100 you know, as the grocer says when weighting the cheese: > Can it be a little more? > It is easy to remove a few turns... if you end up too high. > > Now, now it works, you can ask the online calecujilator what L is for 1 > MHz, > and put the in LTspice, and show it to customer, > he (the custardtomer) now feels much more secure after seeing > 'mathematical proof'. >
I carry 'data points too, especially the 'stacking factor' and coil cross section size.
> Hey I just had a discussion at CNN.com. and found out how to get rid of > the US deficit: > Advertising on dollar bills, on banknotes, > how much does google make ... FED can do it too. > Made available to all for free. > (And in that discussion I invented gambling on banknotes too). > So there you are, I like LTspice, in runs in wine (that is a windows > emulator for the microsof users), > so I can actually use it. > It is the best I have seen (LTspice, not wine).
Great ideas!! If Facebook can turn a profit of $480M in this economy then the govt should be able to do the same thing! Uh, who paid that $480M? and, did THEY make a profit? Who's spending this money?
On Mon, 29 Jul 2013 15:23:20 GMT, Jan Panteltje
<pNaonStpealmtje@yahoo.com> wrote:

>Hey I just had a discussion at CNN.com. and found out how to get rid of the US deficit: >Advertising on dollar bills, on banknotes,
That's already being done, sorta: <http://www.tosscards.com/million.htm> About $0.05/ea. One of the local Pizza houses was passing them out recently. However, it might be possible to do what the various baseball and football stadiums have done for many years. Like renaming the stadium after major sponsors, we should rename government buildings, programs, and monuments after corporate sponsors. Instead of the white house, it should be the Verizon house, or something equally disgusting. Instead of naming airports and freeways after famous politicians, they should be named after corporate sponsors. For example, the Starbucks freeway. -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558