LTSpice transformer mystery

Actually 10 MHz is not the real resonance frequency. When I used the actual resonance frequency, the voltage at the primary drops to 70mV! I was guessing it's the inductance series resistance but I set those to 0 and still get the same result.

The primary side will "see" an inductance in series with a cap, or an inductance in series with a parallel combination of cap and inductor.

Google around for the equivalent circuit of a transformer -- a transformer with both coils grounded looks like a pi arrangement of three coils, with the coupling coil's value determined by the mutual inductance. The details escape me, but the equivalent circuit should make things more clear.

--
www.wescottdesign.com

There are a couple of things going on here:

1. The default Spice time step is wrong, whatever it is. LT Spice goes for IC analysis speed, and does a bad job on resonant circuits. Force the time step down, 1 ns or less.
2. This resonant circuit has a high Q, like 15K or so. You'll need to sim it for a long time until it settles. Combine that with a small time step, and things get slow.

Better to an an AC analysis.

I'm playing with a coaxial ceramic resonator Colpitts oscillator in LT Spice, and I am skeptical about the analysis. It does things that shouldn't be possible.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

is?

to get reflected to the primary as 1K. Now with 50 Ohms of series resistanc e, the voltage at the primary should be pretty close to 1 V. However SPICE shows about 700mV. Any idea why this happens?

Yes, I think there is some kind of bug in SPICE. As evidenced by the fact t hat I would get some result, then close the waveform window, and resimulate and I get a different result.

I also think because of the high Q, any minor deviation from the resonance frequency causes extreme attenuation of the output. Problem is that the res onance frequency isn't what would be calculated from L and C. It seems a bi t different.

Finally, AC analysis seems to show the expected result, but at a third reso nance frequency.

Well, all simulators have bugs.

Again, the transient sim time step will affect the apparent resonant frequency.

That's probably the right one! Use a lot of frequency steps, over a narrow sweep range, to nail the frequency. (Or use a calculator.)

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Looks like you're talking how much the mutual inductance affects two tuned circuits that are tuned to different frequencies. Or am I wrong ?

If i am right, seems like that might be tough on a simulator.

[now see in the old days we didn'tt use no ocnfouded simulators, we just blew shit up until it worked - then it was "years in development" - a selling point]

M=1 here, and there's one cap, so there's only one resonance.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

On Saturday, January 10, 2015 at 10:42:17 PM UTC-7, snipped-for-privacy@gmail.com wrote :

d circuits that are tuned to different frequencies. Or am I wrong ?

blew shit up until it worked - then it was "years in development" - a selli ng point]

I think my main Gripe with the simulator is that results have changed just by closing and reopening the waveform window. On the other hand, I now real ize the Q may have been so high that what seemed to be minor deviations fro m the resonance frequency may have had extreme effects.

So I decided to try a similar circuit live. I wound 10 turns of #18 enamel on a ferrite donut, and another 10 turns on top of them. Inductance L came to about 48uH. I chose a 600pF to resonate around 1 MHz. Turn ration here s hould be 1 since both primary and secondary are 10 turns. I connected 2.2K at the load and 2.2K as the source resistance. I applied about 4V pk-pk and was expecting 2V pk-pk (at resonance since resonance will cancel the windi ng inductance).

Instead, I got about 1V pk-pk which meant the reflected impedance wasn't th e 2.2K I'm expecting (ZL/n^2), it was closer to 1K!

After a few agonizing hours, a lot of head scratching, and numerous LTSpice simulation + Scilab calculations, I finally figured it out...

Yes, they call that doing real bench work, using real equipment and maybe a cak-culator or some sort, abacus, fingers, slide rule, what ever..

Jamie

I didn't look at your sim however, the default model assumes the ratio of L to be the ratio of turns.. You may want to fiddle with that one.

There is another model you can use that allows you to specify closer to your real world circuit. Open the Help and look at the "L" and you'll see the other model.

Jamie

Try this:

Run it, probe the LC, and zoom the pulse tops. There is AM modulation, entirely a Spice artifact.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I see about 4 mV variance.

Try the Gear algorithm!

The frequency is off by roughly 1%, with any of the integration methods, using the default (automatic) time step. Speed kills!

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

One of my guys, a lot younger than me, still uses a slide rule now and then.

I sure don't miss slide rules!

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com