App notes of high voltage SMPS?

[jonpaul]

Far: "The thought of an EE leaving a company  .. starting up a competitor.. because they saw a similar co's schems, is for the birds."

Au contraire:

My  firm hired a tech in 1980s to assisit in design and fab  HV, high power ballasts and ignitors

So, worker X copied all the schematics and drawings, BOM, wiring diag, when I was away from the lab, X approached a big customer and quit.

The stolen knowledge, plans and design from X were then  built in his firm  for years.

We started legal actions but useless.

My 1983 design is STILL being built by many firms.

Just my memories!

Jon

[coppercone2]

that stuff sounds like its easily reverse engineered, especially for HV stuff, there are not that many parts makers even if you don't know the part #

why are the prints so important in that case?

[jonpaul]

Employee /tech/thief, stole customers contact info, design, plans, mechanical drawings (~ 100), all manufacturing info.

A 12 kW arc lamp  is not a BOM/SCH/place DK/Mouser device, , most parts are custom made.

Jon

[coppercone2]

I guess if the market is small enough then the reverse engineering savings can make it profitable

but in a bigger market I think every part would just be reverse engineered by a team. I guess it depends on company wealth. This is for iMAX movies? I do enjoy those, but there is not alot of theaters. I know with small production high tech, finding buyers and convincing people that the tech is useful/important is hard

i think the amusing thing though is that someone wants to invest capital into stolen crap, its probobly easier to make a new product with a legitimate team if you have the money to go into production, rather then to try to fight someone over cutthroat market, and deal with thieves. Like its really obsessive.

I imagine that it had something to do with pride/ego/etc then absolute profit (someone decided to start a war). Someone really had a fancy for xenon lamps lol, they MUST get rich off the xenon lamp market, not some other untapped market that would make more money.

And with a product like that, you can probobly cut prices to cut the other guy out for long enough for them to starve no? Usually entry is not that cheap, have them do zero sales for a year. If you have the same customers just lower price to shut down the enemy sales force till someone decides to stop paying them for doing nothing. Hopefully you can develop a new better product in time to sink the giant submarine force of outdated unsold copycat lamps... so long you have a war chest to live off of.

[Faringdon]

Au contraire:

Thanks, Granted Jon Paul, there are circuits whose schematic should be kept secret, but not standard power supply schems, which are all over the web anyway, and have demo boards freely sold.
From what i see, there is essentially no difference in an LLC for 15kv, 2kw output, and an LLC for 2kw, 200V output.

...other than the obvious different turns ratio....and the need for special HV isolation material and clearance.
Also the fact that you are forced to do an "Integrated transformer" at high voltage, because you simply cant get tight coupling.
...This in turn means, that you cant do a tight coupled LLC transformer with a pllel LMag.....(to reduce RMS current in the primary coil).
..Also, there could be noise issues if you are dividing down a high voltage output down to a 2.5V reference, say.
Also, i reckon it might be good to have it on light load at startup, and just have a pre-charge SMPS to take it up to 15kv.....then again, a long soft start at high frequency might do the trick.
Also, high Z nodes can get induced up to high voltage by the high E field.....so "quenching" resistors may be needed at certain points, to prevent nodes getting induced up in voltage.
Also, sharp pointed metal bits shoudl be kept away from circuitry, as they will attract a high E field to them...eg , the corner of a metal tab TO220.
Also, theres the obvious spacing and clearancing, creepaging.

Other than that, i'd say a high voltage LLC (15kv say) has little essential differences from a 200v output one. Would you agree? (obvious differences accepted, eg, the output didoes and caps  will be much higher voltage rated etc etc).

[jonpaul]

another zero information response from " Farrington"  100% wrong

keep simulating, don't fool with HV power supply

j

[T3sl4co1l]

Heh, I wonder at what point a higher-Q resonant system (more like an induction heater than a conventional LLC supply) becomes necessary.  Probably fairly high voltages, and low to modest powers.  In other words, enough to preclude multipliers of much ratio, but not so much that impedances are reasonable (more conventional topologies become applicable at, say, power distribution levels).

Suppose I should do some design exercises to sketch that out some day.

Also whether pulsed vs. CW matters much, which is to say Tesla coil sorts of things, but with a rectifier hanging off the top.

Tim

[Simon]

well getting to a high voltage is a matter of a transformer, it's what you want to do with it after that. Not sure why the 400kV is needed in the first place so the answer can be anything between, a pain to do or don't be silly.

[Kean]

"400kV" step up modules are readily available on ebay...

Thanks to "The revolution in high power SMPS" it should just be a matter of get a sufficiently large number of those to meet the power requirements.

 

[Simon]

yes, that would totally work.

[pcprogrammer]

"400kV" step up modules are readily available on ebay...

 

bigclive did a test of these:

[PartialDischarge]

I think I may have the record for owning the most dangerous HV source in this forum, a Haefely 100kVac with enough current to be very spooky  .
With the rectifiers I can go to 140kVdc

[Faringdon]

another zero information response from " Farrington"  100% wrong

Thanks..
Though the pinnacle of HV SMPS is for sure, that you need to use LLC.
1...because HV txformer will have high leakage, and thats no problem for LLC
2...LLC with FWB output has absolutely no overvoltage ringing on the output diodes

And...the LLC,  like all  "current slosher" converters, has the problem of "severe reverse recovery" in the primary diodes (the ones in the FET package).

https://www.eevblog.com/forum/renewable-energy/avoiding-reverse-recovery-problems-in-llc-converters/

As the above thread tells it...This is totally resolved by adding series diodes in series with the FETs...and sic diodes in parallel with this diode_FET twosome.
Infineon have a different way...they say they have software (in their app note) which detects when severe reverse recovery is about to happen, and then they just avoid switching on the FETs.

On no forum, has any engineer, ever refuted the "series diode" method for LLC.

If any engineer does refute it, then that engineer also is refuting the usefulness of the  infineon software described above.

If  an LLC does not use "infineon software method", or  "series diode method", then it is just waiting for the conditions which will bring severe reverse recovery to it....then it will be destroyed, or seriously weakened...causing eventual destruction...probably it dying during normal operation and people thinking it just died of "natural causes", eg  Fake components or ESD from assembly handling etc etc.

[T3sl4co1l]

another zero information response from " Farrington"  100% wrong

Thanks..
Though the pinnacle of HV SMPS is for sure, that you need to use LLC.
1...because HV txformer will have high leakage, and thats no problem for LLC
2...LLC with FWB output has absolutely no overvoltage ringing on the output diodes

Well, FWB would be rather wasteful, more to the point you can use full wave doubler, or higher order types, just fine.

And...the LLC,  like all  "current slosher" converters, has the problem of "severe reverse recovery" in the primary diodes (the ones in the FET package).

https://www.eevblog.com/forum/renewable-energy/avoiding-reverse-recovery-problems-in-llc-converters/

As the above thread tells it...This is totally resolved by adding series diodes in series with the FETs...and sic diodes in parallel with this diode_FET twosome.
Infineon have a different way...they say they have software (in their app note) which detects when severe reverse recovery is about to happen, and then they just avoid switching on the FETs.

On no forum, has any engineer, ever refuted the "series diode" method for LLC.

If any engineer does refute it, then that engineer also is refuting the usefulness of the  infineon software described above.

If  an LLC does not use "infineon software method", or  "series diode method", then it is just waiting for the conditions which will bring severe reverse recovery to it....then it will be destroyed, or seriously weakened...causing eventual destruction...probably it dying during normal operation and people thinking it just died of "natural causes", eg  Fake components or ESD from assembly handling etc etc.

Such features are standard in dedicated controllers e.g. UCC256403.  First by just not operating in that regime by design, and second by briefly halting operation, or faulting out entirely as the case may be.

It's also not "eventual[ly] destructive" when designed properly.  Switching loss may increase, but by how much, is known at design time, and can be controlled by design and detected with additional circuitry if not provided by the controller.

Tim

[Faringdon]

Such features are standard in dedicated controllers e.g. UCC256403.

Thanks, i looked at the UCC256403 datasheet, though find that unfortunately it cannot assure avoidance of  severe reverse recovery incidence in the LLC..

The following App Note
https://www.infineon.com/dgdl/Infineon-ApplicationNote_EVAL_600W_12V_LLC_C7_with_600V_C7_XMC-ApplicationNotes-v01_00-EN.pdf?fileId=5546d46253f6505701544cc1d15c20d7

...shows what is needed to absolutley always avoid severe reverse recovery  in LLC.
...Page 16 of the above confirms that unless these measures are taken...we are vulnerable to severe reverse recovery in LLC.
One single incidence of  severe reverse recovery in LLC can kill its FETs.........the current spike is that large, specially from a 390V bus.

Page 56 and 57 of the above App note show what measures are needed...and as can be seen, these measures are not available in UCC256403.

UCC256403 datasheet
https://www.ti.com/lit/ds/symlink/ucc256403.pdf?ts=1665905071034&ref_url=https%253A%252F%252Fwww.google.com%252F

The controller shown in the above App Note, employs all protections as available in the venerable ICE2HS01G controller...but even admits that those measures are not enough to avoid severe reverse recovery in all cases.....so it adds the necessary measures by way of the XMC microcontroller, and its ADC's, and the software it runs.

The alternative, is AYK, to use the "series diode method".

[T3sl4co1l]

That... says nothing.
They're using the same qualitative language I am.
They clearly state that it "can" cause "stress".
There is no proof that it will lead to rapid wear or destruction.
In particular, the pages you have pointed to, provide no analysis, illustrating a given device, in a given circuit and layout, will suffer destruction.  Nor that they will be free of such.

Nor have you provided proof that the add-diodes arrangement will definitely survive.  I'm sure I could construct one poorly enough, or operate it badly enough, to still cause destruction.  No, a more nuanced argument must be made.  Example: the application of the two topologies are substantially similar (thus their operating conditions are directly comparable), and the operating conditions are wider for one case, but at such-and-such expense to efficiency and BOM cost.

Can you see how that works?  Please, take the time to develop your critical thinking skills.  It will be of great value to you in both professional and social environments.  The best time to do this was, well, way back in school, but the second best time is right now!

Tim

[Faringdon]

Nor have you provided proof that the add-diodes arrangement will definitely survive.

Thanks,  AYK, the added series diode makes  sure that the reverse recovery current spikes through the FET  are insignificantly small.
AYK, its these reverse recovery spikes that are the problem for LLC.

Quoting from the infineon App note (pg 16)  linked in #39....

The other challenging operating condition is capacitive region operation in the boost mode below fo, which
in a properly designed LLC converter should be avoided at all times, yet in a few conditions may be
unavoidable, albeit briefly.

...the countenance that it may be "unavoidable"..
AYK, they woudlnt have written that software algorithm if they though they  would be OK without it.
The fact that Infineon have written the software as described shows that they need to mitigate the situation of reverse recovery.
...Other than the diode, and that software, there are not really any other ways to  be sure that a general spec LLC will be able to avoid severe reverse recovery in those special moments....rare(?) that they may be.

If LLC can be OK without that software...then we have to ask...why did they bother writing it? and  why are they   controlling that LLC with that XMC micro, instead of their ICE2HS01G LLC controller?

Have infineon written software that is worthless, and doesnt do what they say it does?

[T3sl4co1l]

...the countenance that it may be "unavoidable"..

"Albeit briefly"

So come on, show your simulation that shows TJ rising to destruction in that brief moment.  Where is it?

AYK, they woudlnt have written that software algorithm if they though they  would be OK without it.

As you know, people wouldn't have ported Doom to the Ti-83 calculator if they thought they would be OK without it.

I accept this as incontrovertible fact and invite you to join my religion, It Runs Doom. https://www.reddit.com/r/itrunsdoom/

Some day, we will have power supplies running Doom!

If LLC can be OK without that software...then we have to ask...why did they bother writing it? and  why are they   controlling that LLC with that XMC micro, instead of their ICE2HS01G LLC controller?

Have infineon written software that is worthless, and doesnt do what they say it does?

Why did I bother writing this on a decidedly inferior platform?
https://github.com/T3sl4co1l/Reverb
Surely there is no practical application for such a function.  Or if there were, one could go out and buy a commercial one for a couple bucks!  Who would waste their time demonstrating their skills, or the capability of a platform, with such a worthless function?

In other words, it's quite likely they built it for any number of goals, including a power supply with incrementally improved behavior, or a demonstration of the real-time capabilities of their DSP, or...

Tim

[Faringdon]

Thanks,
AYK, there are a large number of articles written about severe reverse recovery incidences in LLC.....ill link them in here when i dig them out....its  a while since ive read them...but the general concensus was that LLC has a severe reverse recovery problem (in "that"  zone of operation that causes it).
AYK, none of  the shipped products may ever suffer this...or just maybe, all of them could.
Which is why the series diode, or the infineon software , seems like a good "get out clause"

Put an LLC with series diode, alongside an LLC without series diode, and give them repeated short cct output tests, for days and days......i know which one my money's on to last the longest.

[Terry Bites]

You an aspiring bond villain I see.
If you need a high energy 400kV pulse, that can be arranged.
Build a proper Marx generator.

[LinuxHata]

Well, I'm also interested in these, but my questions are more structured and logical, I hope.

There are a lot of forward/flyback SMPS power supplies for relatively low voltages available, say 3.3-120V range.

What if I took such power supply, say, 12 volts 50 amps (600W)
and want to convert it to high voltage one?, say, 300 volts, 2 amps.
As I understand, I'll need to re-wind the transformer, replace all diodes and caps and most likely, optocoupler, and adjust output feedback divider in a proper way, to receive the required output voltage.

Will this be enough, or I'll need to adjust something at the primary side, too?
And what if I need to have 1200 volts at output? I guess, transformer shape should be changed and 2ndary made in sections?

[Jwillis]

The Tesla coils original concept was to demonstrate wireless electricity. There are some applications today that apply that concept but as a practical application for a high power grid the draw backs out way the benefits. 
The idea of a SMPS having the requirements of upward of 400kV and 5000W would have to have some practical need for it. Another issue is what would the energy source be. Power in equals power out minus losses. Not to mention such a device has little advantage benefit over a simple transformer at those voltages when you take into account complexity, cost, and cost to maintain. Also the filtering required would exceed the copper content of an equivalent transformer. Just imagine the EMI coming off such a device. 
400kV at 5000W is a mere 12.5mA, which is fine for a deadly taser or mass shock therapy, other than that theirs really no practical use for it.