Flyback vs half-bridge

They do it because they can get away with it. It takes more work to organis e the information in a way that leads the reader through it, and if there's not a lot of competition they don't have to bother. My impression is that it doesn't take much - if any - extra text.

Ralph Morrison's book "Grounding and Shielding Techniques" - I've now got t he 4th edition, ISBN 0-471-24518-6 - was remarkably easy to read, and relat ively slim.

I've spent a career doing state of the art stuff in a couple of different a reas. Getting proficient enough involved a lot of reading, and most of the books I had to read were poorly organised and hard to plow through. William s and Taylor's "Electronic Filter Design Handbook" 0-07-070430-9 is a fairl y horrible example, which doesn't stop it from being an extremely useful bo ok.

Hell - at one point I read a Ph.D. thesis in German "Electronenstrahl Tests ystem fur die funktionscontrolle und fehleranalyse hochstintegrieter schalt kreise" by Eberhard Menzel (which is actually pretty well organised). I've still got it - a two page thesis copied onto a hundred pages of A4 at the r ate of two pages of thesis for one page of A4. Not an easy read.

That is actually one of the books I used the most in my career. Until affordable computer tools showed up, now I hardly use it anymore, thinking about donating it. IMHO it is a good book, and lots of engineers at clients thought so as well. Most already had it on the shelf and I probably triggered another half dozen book sales. Without getting a cut, hurumph ...

I've had not-so-easy reads as well. Mostly in the form of source code that needed to be review where all comments were in Spanish, and then some Portuguese. Also Italian. Every time I got to a power supply section (they call that alimentazione) I became hungry.

He's not the first. The 1997 Linux gEDA project insisted on human-readable files from the start. Back around 1988 I started on electronic CAD on a Uni x-based Metheus system where you could use grep and awk to edit your design files - great for shifting large blocks of circuit around quickly.

Thanks, I was wondering if that was the case.

That is strange. Was that on the bench or in LTspice? It's not clear how that could happen.

To protect against polarity reversal, I add diodes at each voltage output to prevent damage in the event of an inadvertent short, for example when probing. All it takes is one incident.

I think it's more than that:)

I never paid much attention to switchers due to the RF noise they spew out. However, I started getting interested when you began posting your work on switching circuits. Knowing your sensitivity to nanovolts of noise in most of your circuits, I started thinking if you can make them work, then maybe there's some hope I might be able to get some success also. So you have definitely changed my direction.

I should add that nobody else could have done that!

Thanks,

JK

On a sunny day (Wed, 17 Jul 2013 20:47:22 -0400) it happened Phil Hobbs wrote in :

That is completely insane. It is a worse design than Chim Pansee made after being told she could not go back to Affrica, and being fed Bananas in a cage in the zoo.

This is how it SHOULD be done:

only 10 components, better output regulation, better efficiency, less RFI, better reliability, better MTBF, less size, cheaper, etc etc. and as old as the world:

I have the same circuit in an other programmer I designed, still working after

On a sunny day (Thu, 18 Jul 2013 03:01:08 GMT) it happened John K wrote in :

For Gods sake stop feeding the man's ego, you know how dangerous bubbles can be.

Glad to see that your ego remains unimpaired as well. ;)

Having grown up as a discrete-circuit designer, I like the resonant converter, and if I didn't need a higher-current negative supply as well, I might well do it that way, if I could find a standard part that would work for the transformer.

My gizmo piggybacks on an existing -15V supply, and requires two dual diodes, one quad resistor pack, and four cheap caps.

Yours requires a custom transformer, four caps, one resistor, and one transistor, and a dual diode.

I'd need the -15V switcher anyway, so my scheme is considerably simpler and cheaper than yours. What's insane about that?

Cheers

Phil Hobbs

On a sunny day (Fri, 19 Jul 2013 08:34:08 -0400) it happened Phil Hobbs wrote in :

Yes I am glad too:-)

If your designs require 'standard parts', well in short, US would never have made it to the moon. Poeple have fobia for coils and transformers, especially winding a wire around some former it seems. Should be a training requirement.

Your load regulation sucks, always does with R3 R6 C5 C6 and diodes in series. Your efficiency sucks for the same reason, you waste power in resistors and diode drops. Where does V2 come from, it is floating? You are driving R5 and R1 as load I presume, hardly any current at all. This type of resonant converter I did draw can easily power some relais, a very big industrial plant has boards with it I designed with that converter to isolate relays... Probably no switcher needed, just an extra diode and capacitor (depends on how stable supply is). Anyways now I know you have a fetish for Schotkeys on top of one for Ferry High value resistors..

No problem for me.

What is 'simpler?'

??????????????

On a sunny day (Fri, 19 Jul 2013 12:57:30 GMT) it happened Jan Panteltje wrote in :

And CHEAPER? ????????????????????????????

Not even in one of, or in milions.

Last night, even before seeing that LT spice thing, I was remembering a watch I won at a fair. It had a potcore in it the size of a pin-head (the type you get when you buy a new shirt). I think the watch price was around 3 $ (new including battery). The potcore was part of a resonant circuit to drive the alarm beeper.

I wind my own coils all the time for one-offs. Getting hundreds of special coils is a pain, however.

Check out the efficiency meter. It ignores the losses in the switching regulator chip, but otherwise it's the real deal. Including the -15, it comes out at almost 90% once the turn-on transient is over.

And this is a photodiode bias supply, so it's going to need cap multipliers anyway. It's very simple to add shunt regulation to a cap multiplier, so that's what I usually do.

And you were the guy who said that efficiency didn't matter in a 20 mA supply. ;)

Mine needs seven parts, and costs a lot less than yours. The others are part of the -15V regulator. And no custom transformer. Did I mention that it doesn't need a custom transformer?

In the immortal words of Zaphod Beeblebrox, "Count the heads." ;)

Cheers

Phil Hobbs

On a sunny day (Fri, 19 Jul 2013 09:16:35 -0400) it happened Phil Hobbs wrote in :

Sounds OK, a question, here it was 30 degrees C today indoors in the shadow, and I just went upstairs it it reads 35 C, just 2.5 C below normal body temp, so how would the Schottky's leakage be when say in the sun at 90 C? Would the leakage current be higher than the load current? ;-) There is a reason why I wrote 1N4148... But if you insist I can use a double Schottkey too, and now I have 9 parts, including the supply decoupling, that you cheated on by using a 'perfect zero Ri' voltage source. without it I have would have 8 parts, but hey, use finer wire and some more turns, and / or higher Al, and no need for the tuning cap, now I have 7 parts.

OK, how abouy this then:

Yes, but ruining it is an other matter altogether ;-)

I don't think it is cheaper. In the time you did the spice I wind 3 transformers... Time is money blah blah.

Never met that guy, heads roll easily :-)

Hey!

Do much work in nanovolts? Mix switchers and nanoamps? Phil does.

He changed my long-held belief that switchers and nanovolts do not mix.

This is a very big deal for me, since I no longer have to put up with huge heatsinks and impossible thermal problems.

I suspect some others may have come to similar conclusions.

I believe he is uplifting the entire spirit of this newsgroup, in contrast to the interminal battles and insults that prevailed until recently.

Please allow the man credit for what he has accomplished.

JK

That's a few more parts though. ;)

Yup, Hamamatsu has been selling that basic idea for decades. I expect they use an op amp for the voltage regulation.

The resistors hardly cost any efficiency. They're there to protect the diodes from the turn-on surge, but a BAV99 is specified to take 1A for 1 ms nonrepetitive, 500 mA repetitive, so it doesn't need much. The current waveform is basically square, i.e. the duty cycle per arm is near 50%. Thus the load current sees an average of just about exactly

For a one-off, I agree. For hundreds to thousands, not so.

The Hitchhiker's Guide to the Galaxy. Literary comedy unsurpassed since the days of P. G. Wodehouse.

Cheers

Phil Hobbs

Cheers

Phil Hobbs

On a sunny day (Fri, 19 Jul 2013 14:34:57 GMT) it happened John K wrote in :

Actually I am way a the nanovolts right now :-) It is not really that little... And a lot of work still needs to happen.

On a sunny day (Fri, 19 Jul 2013 10:55:10 -0400) it happened Phil Hobbs wrote in :

I have that on disk somewhere, but I think the answer to your question was 42.

I, in retrospect think it is not really that good at all. Snoopy is more fun.

I expect it helps to be a native English speaker, and to have read a lot of old science fiction, because the humour is so atmospheric, like Wodehouse in his way.

Cheers

Phil Hobbs

On a sunny day (Fri, 19 Jul 2013 11:01:26 -0400) it happened Phil Hobbs wrote in :

No problem spend an hour coding some standard from some official released pdf this morning, got stuck, started googling, realized that pdf has a typo, and I already wrote the right code some time ago. I can still use the new code for decoding SETI signals perhaps. Will leave it in :-)

I did say: s*it.

On a sunny day (Fri, 19 Jul 2013 13:24:05 -0400) it happened Phil Hobbs wrote in :

Yes possible, I did not know English had humour?