[Amps] Peter Dahl transformers

[Bill Turner]

Thank you Manfred for some very enlightening thoughts!
I have copied your post in its entirety below for anyone who might have
missed it the first time around. Well worth reading. 

Bill, W6WRT





On Fri, 22 Feb 2013 20:10:59 +0000, Manfred wrote:

>Bill,
>
>> Manfred, in spite of the problems for the homebrewer, I'm still 
>> interested in switching power supplies for high power amps.
>
>It's true that doing the whole development work to build just one unit
>(and possibly publish it) isn't very efficient. But ham radio isn't
>about providing the most cost-effective solution! It's about learning
>radio, electronics, and having fun doing so. In this way, for me it has
>always been very attractive to develop some piece of equipment for ham
>radio use, just if I was developing something for commercial production
>(which I have very rarely done, buy the way, my job of many years having
>been in a research organization, and not in a commercial company).
>
>So I applaud any ham who develops a good design, rather than to just
>copy what others have been doing for decades.
>
>> Do you have any schematics or sources of design information for:
>
>> 1. A 50 VDC 50 amp supply and 2. A 3000 VDC 1 amp supply?
>
>Usually you won't ever find a perfect, ready, complete, bug-free design
>for anything you want to build - specially if you want it to meet YOUR
>specifications! But the datasheets for power supply controller ICs
>contain many referential designs that are useful as starting points.
>Often only small changes are needed to adapt them to your particular needs.
>
>For power supplies of this power class, I would use a full bridge of 4
>MOSFETs. The 50V supply is relatively easy and simple. If you find
>ferrite cores large enough, you could use just a single transformer.
>Otherwise, like if you are bound to what Amidon sells, you will need to
>combine several cores.
>The control could be simple pulse width modulation, like implemented in
>the old and well known LM3524, TL494 or KA7500 chips, or you could use
>fixed 50% duty cycle with phase shift modulation, as implemented in the
>  UC3875. I actually would prefer the latter approach, but it might be 
>necessary to either use a bleeder resistor to impose a minimum load, or 
>else simply switch off the supply during RX. During TX the amplifier's 
>idle current is enough minimum load.
>
>The entire primary side can be the same for both the 50V and 3000V
>versions. But the high voltage version would probably be easiest and
>most reliable if you split it up into about 8 separate, rather small
>transformers delivering a lower voltage each, rectify and filter each
>output separately, and then place them all in series. The primaries 
>would all go in parallel. This allows you to use inexpensive, 
>off-the-shelf diodes and capacitors, and you don't need any 
>voltage-equalizing resistors. The transformers must be all insulated for 
>the FULL high voltage, though. An easy way to do that is winding on a 
>bobbin that has no side walls, leaving a few millimeters on each side 
>without wire but with interlayer insulating sheets (mylar or the like),
>then fill the ends with an insulating material, which could be epoxy
>glue or silicone caulk. In a production setting this would be done with
>vacuum potting.
>
>> I'm looking for designs where the heavy lifting (engineering) has
> > already been done.
>
>The problem is that apparently everybody is doing that! Someone HAS to 
>make that first detailed design, and publish it instead of sitting on 
>it!  I don't have any complete, fully developed designs on hand for 
>either of those specs. Sorry!
>
>The main problem is that people developing circuits for an employer 
>usually have to sign over the rights to that employer, who usually would 
>never publish them. And too few people do such designs just for their 
>own use, and then put them in the public domain.
>
>I have found that I can only very rarely use a published design just as 
>it is. Usually I can't find all the same parts, or the specs are not the 
>exact ones I need, and so on. Instead of modifying existing designs, I 
>usually prefer to start from scratch and roll my own, but of course I do 
>look at all designs I can find, to learn how others have solved specific 
>problems, and copy the good ideas, if I find any!  Then I publish what I 
>cobble together.
>
>The data sheets and application notes by the semiconductor manufacturers 
>are typically the best freely available sources of information and 
>reference designs. I suppose that there must also be engineering books 
>with good designs, but the non-ham books I have seen so far are for the 
>most part either very poor, or purely theoretical, or very expensive, 
>and usually at least two of these! And the ham books rarely contain much 
>modern technology at all, and good ones are becoming ever scarcer.
>
>If you are serious about building either of those power supplies, I 
>suggest that you start by collecting all the data sheets and application 
>notes about suitable controller chips, and read and fully understand 
>them. That should give you the main information you need to do a 
>detailed design yourself. You will probably find some practical circuit 
>examples that are roughly in the correct power class - you can largely 
>use one of them, and just adapt the secondary side to your voltages. And 
>if you have specific questions about the details that are not 
>sufficiently covered in those papers, I can try my best to help you. 
>Just don't ask me to teach a complete course on switching power supplies 
>by e-mail, which is time-consuming! The basics are well covered in many 
>freely available papers.
>
>Manfred
>
>========================
>Visit my hobby homepage!
>http://ludens.cl
>========================
>_______________________________________________
>Amps mailing list
>Amps at contesting.com
>http://lists.contesting.com/mailman/listinfo/amps