Nothing "specialized" about a plain vanilla PoE flyback transformer.
formatting link
So you get some windings for free. Do you also complain in a restaurant if they give you some free pudding or a free ice cream after a meal?
Learn Digikey. It's the only search engine that really works. The others will never get it.
Everyone uses push-pull CFL or flyback these days. I am surprised that you did not know about all these off-the-shelf transformers. There are tons of versions. Same for inductors with more than one winding. Since years and years (I am designing power converters since 1990).
They don't have to set it up, they've got one.
You can. Money always talks in this market.
If you don't like their deal you can pick out of a large selection of other companies who are then even free to use cores from any vendor. So what was the problem?
If you utilize the energy that is otherwise burned off in the snubber it can be. The main issue is getting EMI under control, that's a pain with flybacks but for low power wide supply range stuff often you don't really have a choice.
If it fits there usually is not a significant penalty and using COTS is the smart thing.
It rather boils down to hiring really smart engineers, the managers are less important (other than not getting in the way too much). Engineers who truly understand the materials properties and trade-offs. And who are most of all keeping in touch with new ferrite materials.
There was an electrolytic on the other side. Also, they ran the uc at the lower voltage limit with a slow clock.
Of course this does not help in some of the markets I have to develop for. Like islands where sometimes Felipe gooses a recalcitrant generator by reaching in and pushing the throttle by hand.
I was amazed how much abuse such a little uC can take.
Sloman is obsessed with the Baxandall thing. He's been working up to actually building one for about a decade now. He claims to have spent thousands of dollars getting a custom transformer wound.
Weird.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
If PoE is what you want. It's not going to be the right turns ratio, voltage or current rating for any other application.
....
They still aren't "tapped inductors" and you've got to couple the windings yourself - using up two pins - to get the reuslt that you get from one pin on a real tapped inductor.
That's useful to know. Their catalogue was always perfectly useless, organised strictly by manufacturer.
The off-the-shelf transformers that I knew about were all pulse transformers, and pretty much always transmission line transformers with high - but unspecified - intervwinding capacitance. PoE seems to have become ubiquitous enough to support off-the-shelf inverter transformers, and I don't - these days - learn the contents of the Farnell catalogue cover-to-cover any more. I never set out to learn their catalogue, but if you spend enough time with it, the content tends to stick.
r
Learn to read more carefully - that sentence doesn't imply that EPCOS hasn't got coil winding facilities, it's a prediction about the customers that can expect to be able to get at it.
it to
But the money that talks is the sum that you can spend on cores and formers.
Getting EPCOS to do it. I was exploiting recommending local coil winders rahter than trying to get wound parts from EPCOS.
t
ss
r"
,
Big if.
"Not getting in the way too much" is the whole art of management, and the only management course that I was ever involved emphasised just that.
Seriously, I think you should read up on flyback converters. The turns ratio in a flyback has a much lesser importance than in pretty much any other architecture.
Or ask yourself this question: How do you think we design switchers that take anything from 80VAC to 260VAC and make a regulated 12VDC or whatever out of that, at EPA mandated efficiencies?
C'mon. All that requires is a soldering iron. I hope you aren't this picky with food :-)
Hint: Nearly all "tapped" inductors have four pins. Guess why ...
I can't even remember the last time I used a paper catalog. Oh wait, it's right in front of me: The AMP Connectors hardcover 1991 edition, used to get my flatscreen monitor to the correct height.
It happened long before PoE. You could always buy flyback transformer for small primary switchers, for a few bucks.
Even in the 60's you could but those were bigger and cost a lot more, as they were for TV sets and thus had a fat profit margin calculated in.
Farnell is not necessarily the best place to learn about this kind of stuff.
Read your own writings more carefully. Quote "Sadly, EPCOS won't wind prototypes for you".
This clearly means "they won't do it for me" and that is incorrect.
Naturally. They have to pay their bills just like any other corporation. You request a quote and then decide whether to accept it or not.
In the end it won't matter who does it. It boils down to cost, the quality of their engineers, and (in my line of work) how far they are willing to veer off the trodden paths and embrace something new.
Again, take a look what's out there. Tons. I'd say more than 95% of the larger inductors I use in switcher designs are COTS parts.
Yeah, I wish more people would understand this. Give people the power to make decisions. They'll screw up once in a while. But so have we, and learned from it.
ISDN transformers are cool. We stock one with four independent windings, 1:1:2:2. That's handy for all sorts of things.
I treasure my hardcopy Digikey and Mouser catalogs, even as they age. Searching for a cap or an inductor is way easier in the hardcopy.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
Yup. Although I favor LAN transformers because ISDN is on its way out. It was a fairly short-lived technology, not enough bang for the buck.
What is also very useful are the VersaPac ferrite transformers with six independet windings. I've designed a lot of ad hoc circuits with that, plus some for production.
I keep them as well, just in case. But I've found that the PC is simpler for browsing caps and stuff. Just like the vellum pad gets less and less use. Nowadays almost everything starts right on the screen, sometimes with GoToMeeting and similar collaborative tools. I just wish they had a better whiteboard.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
LEts see, that looks like a TI-30 or an HP calculator on the right and as for the schematic, looks like a digital programmable gain loop for the amp you're using. The programming looks like you're using something like a 4 or more lines input, possibly from a 4 or more row thumb wheel R scalar or some other means like a parallel IO source.
Most likely a unity buffer amp on the output with some filtering (Cap).
3 terminal reg supply down at the lower right and at the top, looks like you have another amp input going to an IC I can't make out, possibly the input is a comparator for the generation of shaped pulses for the following IC instead of an amp? FF maybe for the IC?, etc..
Since my first serious digital project was 10-bit PWM D/A converter, back in 1975, I think I can claim to have known the answer to that question for a few years now. There are always trade-offs in getting voltage step-down or step-up by PWM - getting the right turns ratio minimises them.
......
.
t
Interesting wound components are often pin-limited.
They've been designed by unimaginative engineers to go in square/ rectangular packages ..
rs
But it made sense to specify ityourself, for your application. The off- the-shelf parts weren't cheap enough to be all that interesting, if you could find them.
e
ff.
So where would you suggest?
for
They've done it for you? What was the projected production volume? And did the projection come true. I still doubt that they will wind parts for David Jordan's particular application.
t it to
nt
o
No argument there. Would EPCOS be the place where you'd expect to find innovative engineers?
ch
hat
,
ross
ver"
Hz,
e
fs
it
h
d
So you are designing stuff for a well-established market.
e
e
It's not just that. The engineers at the sharp end have to immerse themselves in the fine details of the project. Educating management to undersand these details takes time - that they usually haven't got. Our rule of thumb was never to write anything longer than two A4 pages for management, and figure that they were only going to pay attention to the first page, so you put your cover-your-back details on the second page.
I just tried your simulation, and by changing C1 and C3 to 1uF ceramics with
25mOhm ESR the efficiency jumped from about 85% to over 97%. There was about
500 mW of heat in the electrolytics. I haven't followed all of this thread but I built a similar boost converter for nominal 12VDC to about 60VDC at nearly 50 watts, on a board about 1" wide by 2.5" long, for a high power
13 LED diving flashlight. I couldn't get much better than 85% efficiency, especially when the battery voltage sagged to 10.5. I was using a PIC16HV616 with PWM up to 200 kHz, but best results were about 50 kHz. I used a single inductor boost topology.
I'm in good company. Jim Williams wrote application notes AN45, AN49, AN51, AN55, AN61, and AN65 about the Baxandall class-D oscillator.
I built one back in 1968 - it's described in my Ph.D. thesis, though I neve r actually built it into the LVDT driver circuit I'd built it for. The UK M ilk Marketing Board got another in 1975, and my particular variant of the c ircuit got shipped with a few GaAs crystal-pullers around 1987. What I'm fu tzing around with at the moment is rather different low-distortion sine wav e oscillator.
John-out-of-touch-with-reality-Larkin gets it wrong again. The inductors an d transformer I got wound for the low-distortion oscillator didn't cost any thing like that much - unless of course the US dollar has fallen precipitat ely against the euro in the last few hours. And it's certainly not any kind of Baxandall oscillator.
Very weird. John Larkin isn't usually quite this demented.
Not sure what a D/A converter has to do with this. But if you claim to have understood flybacks since 1975 I really do not understand why you said "It's not going to be the right turns ratio, voltage or current rating for any other application".
The turns ratio only really matters to optimize the thing so the lowest cost parts possible can be used.
formatting link
Nope.
They have been designed by smart engineers who wanted to maximize markets for this part. For example, as John said, you can also use them as iso transformers. That use would be precluded it it had three pins, so less pins is not smart.
The other reason is one I regularly use on designs: I often need a negative supply, in other cases I need a higher voltage supply above logic level for amps and stuff. I can do both with the same part. With your three-pin device that ain't possible.
I have found them and shown you. You still don't see it? And no, they won't be in a 1975 Farnell catalog :-)
I already said that: Digikey.
Epcos has not done it for me personally. But I had a talk with them and they would have. Volume was several k/month. The reason we didn't go ahead was that we needed another company's ferrite material.
Before you actually ask, such doubts are meaningless. Epcos is a pretty good company.
Most companies have them. But this wasn't about that. The fact is, Epcos can make custom parts.
Generally not.
In my experience out-of-the-box thinkers are mostly found in small business, that's where I go first for difficult projects.
My strategy for proposals and such is this: The first section contains an executive summary of my ideas. 1/2 to 3/4 page. Then 3-5 more pages with details. So if a busy executive gets curious and wants to no "No how on earth is this supposed to work?" he or she can dig deeper before having to call or email. But they don't have to.
Yes, but the 4051 is decoding three inputs, so it's doing logic, too. It's an analog LUT.
Come on, Jim, point out the bug.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
It's the fact that it used PWM to produce the output voltage we wanted that should have been the clue.
You optimise transformer design for the application - you don't buy more core or more copper than you have to - unless you insist on designing your power-supply with the nearest transformer/inductor that you can buy off the shelf that won't actually burn out.
Lowest parts cost is one aspect of optimisation.
t
Then you haven't designed the kind of stuff I did, and taken full advantage of your wound components.
Unlikely. Off-the-shelf parts are optimised for one specific application, which buys almost all of the parts produced. Anything else is a boutique part, and costs a lot more.
Fewer pins than four wouldn't be smart. My last transformer got stalled when Farnell sent me 10-pin formers when I'd ordered 12-pin formers.
With a four pin device it often isn't possible too. Twelve pins can be handy.
-
IIRR the 1975 Farnell catalogue did include pulse transformers - but they were transmission line tranformers with horrible interwinding capacitances. I'd found them all right.
as
the
stuff.
A catalogue that's rigidly organised by manufacturer isn't really a good place to browse, and while Farnell wasn't good on printing extra parametric data, Digikey was worse.
Several k/month is bigger than my 10k per year volume estimate of minimum interesting volume. I've never work on anything that was expected to sell more than 1k per year.
An excellent company, but big. And my doubts aren't meaningless - realistic is the word you hould have chosen. David Jordan could always ask for quote from them, but he'd be mad not to explore more realistic options at the same time.
mit it to
d
n.
But you don't hang around waiting to find out if they will give you a quote you can live with, and you go for quotes from other suppliers.
But they are very unlikely to do it for you if you can't realistically project 10k per year volume.
e
How often do you do a parallel design for a a custom transformer and see how the costs would have compared? Time-to-market and short product life-cycles do tend to discourage this, but presumably you do design stuff that stays in production for long enough to make a bit of cost-engineering feasible.
EMI Central Research was full of them, and Bell Labs too, but that style of blue-skies R&D has gone out of fashion. It paid off hugely in the long term, but accountants like to do their discounted cash flows on more predictable profits.
Mine were written where the busy executive could walk to my desk or tell me to walk to his office if he felt curious - back then they were all male.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.