1. Home
  2. Electronics Design
  3. Power generator...

Power generator...

From 7mm dia PCB to kVAs...

I've a requirement for a 'small power' AC generator:

  • sine wave
  • frequency 5kHz to 50kHz
  • voltage output up to 25kV peak The load is roughly 150pF, with small losses (almost a glass capacitor).

Gasp, the 150pF load translates to 1.1A at 50kHz/25kV and that's 27.5kVA of circulating power.

The 1:10 frequency range (might reduce to 10kHz-50kHz so a 1:5 ratio) almost prevents using resonance as this would translate to a 1:100 tuning cap ratio, with an already highish starting value, and a variable inductor at those voltage levels is sure impractical.

At 5/10kHz step up transformer will have a big core section and can't be ferrite: the low saturation induction will call for high secondary turn count and with the required high isolation that will make for high leakage inductance which won't go with the 1A current. More, if low enough leakage inductance were possible, the step up ratio, (say 50 with a 500V primary supply) will make for huge current and losses at the high frequency end.

One thing I'm thinking about is, apart from using nano-crystalline core material (saturation around 1T) in order to cut on turns count, is to size the transformer so that its secondary side leakage inductance resonates slightly below the 50kHz limit, say 40kHz, with the 150p load cap. That won't reduce the primary current, still 50A, but will reduce the necessary primary voltage, hence losses...

What do you think. Any other cute idea?

Oh, it's a one off.

--
Thanks,
Fred.
Reply to
Fred Bartoli
Loading thread data ...

Is the frequency fixed in use, or does it have to be easily variable? I'm thinking you could get away with a variable reluctance, double tuned transformer (aka Telsa coil). That is pretty low, so instead of ferrite, something stripwound or nanocrystalline (as you mentioned) or amorphous (pricey) might do.

Otherwise, a pair of 10kW transmitter tubes will do it directly. ;-)

Tim

-- Deep Friar: a very philosophical monk. Website:

formatting link

Reply to
Tim Williams

On a sunny day (Thu, 11 Feb 2010 22:25:24 +0100) it happened Fred Bartoli wrote in :

kVA? You could use L - C, and resonance no?

amp -- R -- L -- | === | 150 pF ///

Zero power :-) You just need to switch caps and perhaps waste some power in a series R (in series with L) to decrease Q. I did something like that and switched inductors with relays :-) Awfull sparks, special HV caps, wound te inductors myself. But the voltage was lower.

Reply to
Jan Panteltje

Jan Panteltje a écrit :

series with L) to decrease Q.

Can't do that. While decreasing the frequency in a 1:10 ratio, you increase the tuning cap in a 1:100 ratio (up to 15nF!!!) , and circulating current in a 1:10 ratio, so that at 5kHz, you'll have 275kVA reactive power...

Oops!

--
Thanks,
Fred.
Reply to
Fred Bartoli

Tim Williams a écrit :

--
Thanks,
Fred.
Reply to
Fred Bartoli

Tim Williams a écrit :

The frequency need not be adjusted while running.

I've thought about dual coupled tuned circuits, but a 25kV double tuned xfrmer with adjustable inductance in the 7H/70mH range doesn't seem obvious for a onesie at first sight...

--
Thanks,
Fred.
Reply to
Fred Bartoli

On a sunny day (Thu, 11 Feb 2010 23:28:55 +0100) it happened Fred Bartoli wrote in :

series with L) to decrease Q.

That makes no sense no me. You leave the cap at 150 pF, it is the load. For lower frequencies you need a bigger inductor, but even lower you need no inductor at all, as for lower frequencies the 150 pF will not draw a lot of current. All you need then is a 1:n transformer. Just a lot of relays to switch things.

Reply to
Jan Panteltje

Welllll,

25kV is ~NST scale stuff, so you'll need an awful lot of wire, giving an awful lot more capacitance, which will in turn increase the current draw. Also, it's a >1.1A NST. Oops. Well, it's not like NSTs are usually more than 1kVA anyway, and you're dealing with 20, so it's not *that* awful just yet, it's going to be big however you do it.

Now, you need enough wire to get 25kV and 70mH, which is on the high side for something nearly air cored, but at 25kV you've got sheer turns count, you don't really have to worry about inductance. Some of that inductance will also be taken up by your winding capacitance, so it's not a big deal, at the expense of magnetizing current, which could easily become a big deal.

Consider an average TV flyback transformer. If it's 160V input and reaches about 0.3A in 60us, then it's 32mH primary, and if the primary generates a

1kV peak, the secondary generates a 30kV peak with a 30:1 turns ratio, or a 900:1 inductance ratio, so the secondary is 28.8 henries of very small wire on a ferrite core. SWAGging that ferrite core at 0.5uH/T^2 suggests 253 turns primary and 7.6 kiloturns secondary (not horrible, albeit >40AWG!).

Now, a 100:1 inductance ratio isn't that impossible, but it does imply having a fair sized core with enough gap to change effective permeability over the same range. If an ungapped loop has an effective airgap of maybe

0.1 mm, you need to be able to stretch that to 10mm (ignoring fringing). That's not too bad, and seems believable for typical stuff.

So all that said, magnetic design is a big problem. Saturation flux is fairly constant with respect to gap, so you need a core big enough to accommodate the lowest frequency (which will be the ungapped state). 25kV at 5kHz is sqrt(2)*Vrms/(pi*F) = 2.25 V.s/turn, quite a lot. Let's look at some fancy metglas cores,

formatting link
On the last page, the AMCC-1000 (geez, this thing sounds huge) shows 3uH/T^2 at 0.5mm gap, saturating around 1000At (nice high field, at least). That's

3000uVs/T, so you'd need 750 turns on that. Hey, that's not actually too bad. That core is listed as 33mm leg width, 85mm thick, 40 x 105mm window, and 7.1kg though. $$$! As for current capacity, 1A is about 24AWG, or 25 thou diameter, or about 623 mm^2 cross section, so it'll pack into maybe 6mm winding height, not counting insulation. Plenty of room left for primary and a large amount of insulation, should be able to get away with a somewhat smaller core. Meanwhile, the air core inductivity should be around ~20nH/T^2, or 12mH minimum inductance. Maximum inductance is 16.8H. Should be plenty of range?

It's actually looking not implausible. Electronic or other means may still be preferrable though..

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
Reply to
Tim Williams

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.