Re: Transformer Leakage Inductance
- From: Joerg <notthisjoergsch@xxxxxxxxxxxxxxxxxxxxx>
- Date: Thu, 16 Apr 2009 08:23:43 -0700
Artist wrote:
Joerg wrote:Artist wrote:I am concerned about the output impedance of a transformer with a center tapped primary that will be driven in a class A push pull topology.
Will the part of the secondary that couples to the half of the primary that is not being driven, and is open circuit, appear in the output impedance as leakage inductance? If this does happen are there ways to design the transformer to prevent this?
Class A would always drive a DC component. If this is for your piezo pulser you don't really need class A in a push-pull topology. AB2 would be good enough, as much quiescent current as needed for the desired linearity but not more.
Leakage inductance is the part of the inductance that is "in series" with a winding and doesn't couple to any of the others. In a tightly wound ferrite core transformer at a few hundred kHz it won't be much.
You can get good pointers on how to build this stuff from ham radio literature. Look for articles that describe the construction of a transistorized shortwave power amplifier. The main difference in your case would be that yours would operate a good order of magnitude lower in frequency so the transformer must be changed accordingly. Also, since AFAIR you only need to provide a single cycle once in a while you might be able to provide a fast control for the quiescent current so you can turn it off during most of the time. This will prevent a lot (a whole lot) of power dissipation. But make it slow enough so the transducer doesn't "see" it and maybe highpass-filter it away.
Last but not least mind core saturation. At the power levels you described in your other post you'll need a pretty fat core. A pot core would probably easiest because you can get nice matching bobbin carriers for those. And don't send all that energy of one tiny capacitor anywhere, it might explode. Seriously, I had one turn its white ceramic into greenish glass with a loud bang.
The pulse is 1000 Vp @.5 Ap into a mostly capacitive load. This is 500W of peak power. The output has to be a full cycle with both polarities. A class A amplifer would dissipate too much quiescent power. Class AB outside of push pull would by definition not fully amplify one of the sinewave polarities. It would be clipped.
Yes, you either need push-pull or, with a single-ended stage, a resonant circuit. Push-pull is easier these days where we can buy cheap FETs, this was different in the tube days.
Eliminating the bias current between pulses would cause a flux change in the transformer just like the intended signal would, causing an undesired output from the transformer.
Not really. Jot down the current orientations in the primaries on sketch paper. Ideally they cancel out. But you'd have to remove and re-apply the bias gently in order not to cause a spurious pulse in the transducer because this canceling ain't 100% perfect, no transformer is ideal.
This is an age-old trick in ham radio, for CW (morse code) transmission. During long pauses and receive phases we'd remove the bias and that saved a whole lot of electricity. Plus it stretched the lifetime of the rather expensive tubes a bit.
Ham radio circuits is what I have looked at before deciding on Class B Push Pull. Their 600 meter circuits are close to what I need to do. At this wavelength they use class A using a tetrode or class B push pull with transistors or pentodes. The only difference between my class B push pull and theirs is they have a tank circuit to filter out the higher harmonics and I would use a low pass filter.
If you'd scoot it a tad towards AB you'll see less harmonics.
I have also been looking at avoiding the need for a low pass filter by buffering the output of a high voltage power amplifier such as one of these:
http://apex.cirrus.com/en/products/apex/matrix_all.html
with a mosfet or bipolar power transistor. The problem with power mosfets that can drive the primary current is their input capacitances. The output voltage swing of these opamps declines rapidly with increasing capacitance at 500kHz. They would not be able to drive any power mosfet I can find with enough voltage swing. It is unfortunate because with the opamp's feedback on a drain or collector I could drive the primary with an exact half sinewave. I am aware of the second pole in the feedback loop the output resistance of the opamp and the driven capacitance would cause and how to compensate for it.
To be honest I've never considered the Apex amps. Of course the main reason was price but also that all that nice super-power vanishes at a few hundred kHz when real-life conditions hit.
After Phil Allison's colorful reply I did realize that to the first approximation all the flux from any part of any primary would pass through all the windings of any secondary. I also realized it would pass through the windings of the open circuited other half of the center tapped primary and so this doubles the amount of voltage the pass transistors must withstand.
The FETs are going to be the least of your problems. That's why I always say the good old days are right now. You can buy nice 600-800V FETs that can stomach several amps for less than a buck. When I started out this was unthinkable.
I have designed transformers for flyback power supplies. But I have never had to design a transformer for this much power and voltage. So I will leave the design of this one to Butler Winding:
http://www.butlerwinding.com/
Good choice. Companies like this should know their magnetics rather well, when the ferrite saturates, how hot it's allowed to get and so on. Plus they've got all the materials right there.
However, tell them to oversize the core by 50% or more. RF-transformers that are taylored to a glove-fit will get warm and in a physics or chemistry experiment setup like yours that's not desired because it (very slightly) changes the pulse. The extra cost should be miniscule in the grand scheme of things.
--
Regards, Joerg
http://www.analogconsultants.com/
"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
.
- References:
- Transformer Leakage Inductance
- From: Artist
- Re: Transformer Leakage Inductance
- From: Joerg
- Re: Transformer Leakage Inductance
- From: Artist
- Transformer Leakage Inductance
- Prev by Date: Re: My Tea Party Starts at 3:00pm
- Next by Date: Re: Pattern Recognition / OCR
- Previous by thread: Re: Transformer Leakage Inductance
- Next by thread: Re: Transformer Leakage Inductance
- Index(es):
Relevant Pages
|
Loading
