Designing HV transformers.

J

Joseph Casey

Guest
Greetings.
In the past I succesfully built HV transformers for a 2KV voltage
sourse. I drive the primary with a 20KHz sinwave. I used to use UU cores
like those in flyback transformers. But now I am having great
difficulty getting those type of cores - a MAJOR stockist of Ferroxcube
etc tells me they are very hard to get! So my question is: will cores
such as etd54 do the job? Why are single loop UU type cores always used
in flyback transformers?
Regards
Joseph Casey.
 
Joseph Casey wrote...
In the past I succesfully built HV transformers for a 2KV voltage
sourse. I drive the primary with a 20KHz sinwave. I used to use UU
cores like those in flyback transformers. But now I am having great
difficulty getting those type of cores - a MAJOR stockist of Ferroxcube
etc tells me they are very hard to get! So my question is: will cores
such as etd54 do the job? Why are single loop UU type cores always used
in flyback transformers?
Click to expand...
I suppose the large winding area is attractive for high-voltage
designs. But this is less of an issue at higher frequencies.
For example, I used a U core with a bar piece for my 10kV 300kHz
resonant transformer.


--
Thanks,
- Win
 
Joseph,

You didn't mention your location or the power level needed.

If you need 6 to 40 watts at up to 3 KV, you might look into a
piezotransformer. They have high step up ratios and do not suffer from
shunt capacitance problems like conventional wire based transformers
do.

To get a few watts, use a piezoinverter from a CCFL inverter which are
available here in the US. If you need more power (up to 30 watts, or
slightly more), you can use the newer radial piezotransformer
available in Europe.

Regulation is easily obtained as well, you didn't mention whether you
needed regulation.

They operate from 50 kHz to around 200 kHz and produce AC, with the
majority at the lower end of that frequency range.

Regards,

M




On Tue, 24 May 2005 16:33:13 +0100, Joseph Casey <joecasey@indigo.ie>
wrote:

Greetings.
In the past I succesfully built HV transformers for a 2KV voltage
sourse. I drive the primary with a 20KHz sinwave. I used to use UU cores
like those in flyback transformers. But now I am having great
difficulty getting those type of cores - a MAJOR stockist of Ferroxcube
etc tells me they are very hard to get! So my question is: will cores
such as etd54 do the job? Why are single loop UU type cores always used
in flyback transformers?
Regards
Joseph Casey.
Click to expand...
 
Hello Joseph,

I have used regular cores and mostly toroids for transformers that had
to comply with IEC601, plus they needed to be defibrillator proof. In
essence that means they must withstand a 5KV blow.

The main thing is an absense of any sharp edges, wire with certified
insulation properties and very detailed and well guarded production
procedures.

For UU cores you may have to contact Asian manufacturers such as
Dongying Xinji. That's the region where most TVs are built these days
and the ones with CRTs need such cores for their flyback transformer.

Regards, Joerg

http://www.analogconsultants.com
 
The main thing is an absense of any sharp edges,
Click to expand...
Why is this important? I do notice that the TV ones are rounded inside
next to the winding but have corners outside.
 
Mebart wrote:
Joseph,

You didn't mention your location or the power level needed.
Click to expand...
Ireland. The power is about 10 watts intermittent use.

To get a few watts, use a piezoinverter from a CCFL inverter which are
available here in the US. If you need more power (up to 30 watts, or
slightly more), you can use the newer radial piezotransformer
available in Europe.

Regulation is easily obtained as well, you didn't mention whether you
needed regulation.
Click to expand...
It is regulated. I don't want to change the design so I'll stick with
wound transformers but the piezo transformers would be fun to experiment
with.

Thanks
JC
 
Hello Joseph,

The main thing is an absense of any sharp edges,

Why is this important? I do notice that the TV ones are rounded inside
next to the winding but have corners outside.
Click to expand...
Mostly to avoid corona or arcing. Far enough away from the winding it
isn't so important. Another concern is if the wire is to be wound
directly onto the bare or taped core. Any burrs can damage the insulation.

Regards, Joerg

http://www.analogconsultants.com
 
Hi Joel,

You asked the million dollar question! If you want a few units, you
can get them....but the supply isn't reliable in the US for commercial
purposes. Places that want to design them are selling you design time,
and won't even produce it for you unless you want to buy them by the
ton:>:

I've followed the development of these ever since learning about them
in the 1999, thanks to an article in EDN Magazine. Lots of outfits
claim to make them, but they turn out to be BS claims-they don't
advertise their standard products and there is no real content when
you look for info on their websites. Been there, did that, plenty!

I've even looked in Asia via the web. The Asian suppliers don't even
want to talk to you unless you want to order 200,000 units per year
from them. In general, Asia is a waste of time-take it from someone
who has tried. One Asian vendor tried to hit tried to sell me samples
shipped from .tw, they wanted 12 dollars each unit and $200 for
shipping 3 units. Needless to say, I did not reply:>:

Regarding the roll your own effort.... They are not terribly hard to
make although you will have to attach wires and build a holder to
contain your piezoelement. The supports have to be placed at 1/4 and
3/4 of the physical length or else the movement of the piezoelement
will cause it to fracture. You can't solder wires to them either,
connections must be low mass. If you can deposit silver (very thin,
can't have much mass) on the piezo, anyone in the semiconductor biz
can attach wires for you.

If you want characterized units, you can do it yourself with an HP
network analyzer. I have a Spice model for one that actually works!
The internal parameters are very important with respect to the driver.

All the CCFL inverters on the market today use constant current
regulation, if you want to regulate it with respect to voltage output,
you gotta roll your own driver. This is not impossible, but forget a
single chip solution:>:

I have megabytes and megabytes of info I've collected on them over the
years, there is alot of good ifo on the web too. Keep in mind, some
call them 'piezotransformers' others call them "piezoelectric
transformers" or "piezo transformer" etc. Google turns up lots of
hits....but getting some of the units themselves will be a challenge.

If you want to play with higher powered button or circular devices,
they are available from a European supplier. The circular ones put out
5 times more power than the conventional rectangular Rosen type
devices.

You can buy a development kit from TI that comes with driver chips
(constant current type) and piezotransformers on a PCB. You can also
buy Panasonic units (without a source of specs) from digi-key. Not
cheap, I might add and they are selling them until the stock runs out.

I found a single US supplier that actually has the units in stock for
immediate sale, but they don't even know what a Spice Model
is......again, these are devices without complete specs.

I am working on a photomultiplier supply using one of these units,
hope it will be efficient.

Hey, what are you trying to do with them, and what do you have for
test equipment?? A network analyzer I hope?? Having a network analyzer
will allow you to characterize your own unit, you won't be reliant on
a manufacturer for specs.

In closing, I can say that these devices are appearing in so called
'personal air purifiers'. They generate high voltage, and the sellers
claim breathing ozone is healthy for you:>: The FDA is confiscating
them as fast as they enter the country, but some are on the market
(surplus, ebay and other non-reputable sellers have them). Search for
piezo+air+purifier on ebay, you will have many hits. I haven't looked
inside any of these yet, but they might have a constant voltage driver
although it's probably not regulated well.

So, tell me more...........

Want 20 or 40 mb worth of web info on them?? Ive got a big collection
of files in my personal library. Say the word and I'll ship them via
email.

M




On Tue, 24 May 2005 10:47:50 -0700, "Joel Kolstad"
<JKolstad71HatesSpam@yahoo.com> wrote:

Mebart,

With a piezo trasnformer, is there any way to "roll your own" like there is
with conventional transformers? I'd love to experiment with piezo
transformers in some switching power supplies, but it's unclear to me how you
can do that in small quantities. Places like
http://www.fuji-piezo.com/piezo.htm seem to want to do a custom design?

Thanks,
---Joel Kolstad
Click to expand...
 
Mebart wrote...
You didn't mention your location or the power level needed.

If you need 6 to 40 watts at up to 3 KV, you might look into
a piezotransformer. They have high step up ratios and do not
suffer from shunt capacitance problems like conventional wire
based transformers do.

To get a few watts, use a piezoinverter from a CCFL inverter
which are available here in the US. If you need more power (up
to 30 watts, or slightly more), you can use the newer radial
piezotransformer available in Europe.
Click to expand...
How high in voltage rating are they available? 5kV, 10kV?
What are some suppliers of these new European radial
piezotransformers?

Regulation is easily obtained as well, you didn't mention
whether you needed regulation.
Click to expand...
How is that done, servo the primary supply voltage?

They operate from 50 kHz to around 200 kHz and produce AC,
with the majority at the lower end of that frequency range.
Click to expand...
Especially at higher voltages I suppose?


--
Thanks,
- Win
 
Hi Mebart,

Thanks for the detailed reply; it's most informative.

It sounds like this could be one of those items that would work well via
"eBay distribution" if we could find a supply who's sell, say, 100 at a
time.

You can buy a development kit from TI that comes with driver chips
(constant current type) and piezotransformers on a PCB. You can also
buy Panasonic units (without a source of specs) from digi-key.
Click to expand...
Yeah, I found the Panasonic units at DigiKey and was amazed to see that
Panasonic then doesn't actually know anything about them on their web site.
Sheesh!

I found a single US supplier that actually has the units in stock for
immediate sale, but they don't even know what a Spice Model
is......again, these are devices without complete specs.
Click to expand...
Specs would be nice, but I (think) characterizing them empirically would be
OK.

Hey, what are you trying to do with them, and what do you have for
test equipment?? A network analyzer I hope?? Having a network analyzer
will allow you to characterize your own unit, you won't be reliant on
a manufacturer for specs.
Click to expand...
I have a pet project of a CCFL driver, but it's somewhat higher power than
most (~25W) and making the driver board more efficient and smaller is always
desirable. Constant current regulation (adjustable for the sake of dimming)
is what I'm using. I do have access to a network analyzer.

Want 20 or 40 mb worth of web info on them?? Ive got a big collection
of files in my personal library. Say the word and I'll ship them via
email.
Click to expand...
Sure! Although I'm not certain I actually have any e-mail accounts that'll
accept a file that big. I don't suppose there's an FTP or HTTP site you
could place a ZIP archive of them all?

Thanks,
---Joel
 
On 24 May 2005 14:57:43 -0700, Winfield Hill
<hill_a@t_rowland-dotties-harvard-dot.s-edu> wrote:

Mebart wrote...

You didn't mention your location or the power level needed.

If you need 6 to 40 watts at up to 3 KV, you might look into
a piezotransformer. They have high step up ratios and do not
suffer from shunt capacitance problems like conventional wire
based transformers do.

To get a few watts, use a piezoinverter from a CCFL inverter
which are available here in the US. If you need more power (up
to 30 watts, or slightly more), you can use the newer radial
piezotransformer available in Europe.

How high in voltage rating are they available? 5kV, 10kV?
What are some suppliers of these new European radial
piezotransformers?
Click to expand...

Currently, the industry believes that if if ain't for a ccfl inverter,
then your application has no merit. This is not an exaggeration. Most
have ratings of 6 to 8 kV peak to peak. There is actually very little
that limits the output however, except that the devices aren't built
with larger spacings to allow higher outputs.


Regulation is easily obtained as well, you didn't mention
whether you needed regulation.

How is that done, servo the primary supply voltage?
Click to expand...
Sample the output and feed it to a conventional error amp feedback
loop circuit.


They operate from 50 kHz to around 200 kHz and produce AC,
with the majority at the lower end of that frequency range.

Especially at higher voltages I suppose?
Click to expand...
No, actually not at all. The higher operating frequency devices are
generally physically smaller and therefore have a higher frequency of
operation and put out less power because they are smaller. Again, this
isn't necessarily the way it could be....except that it is the mindset
of the designer. Again, if it ain't for a ccfl inverter, it doesn't
count:>:

There are some first class technical references on the web. Some of
these go into great depth regarding the physical dimensions and the
particular material used to make them and many questions are answered
regarding what makes 'em tick.

Hope this helps. Feel free to ask other questions or I'm happy to send
a care package in the form of web documents I've collected over the
recent history of these devices.

M
 
On Tue, 24 May 2005 16:03:17 -0700, "Joel Kolstad"
<JKolstad71HatesSpam@yahoo.com> wrote:

Yeah, I found the Panasonic units at DigiKey and was amazed to see that
Panasonic then doesn't actually know anything about them on their web site.
Sheesh!
Click to expand...
This is fairly typical of the entire industry. No one has specs
listed, and the one single vendor who does has some serious errors in
his listing! Complete characterization necessary for Spice simulations
is non-existant. I am NOT kidding about this!


Sure! Although I'm not certain I actually have any e-mail accounts that'll
accept a file that big. I don't suppose there's an FTP or HTTP site you
could place a ZIP archive of them all?
Click to expand...
I'm not sure-I know yahoo gives 20 mb of free web space if you start a
mailing list. I have 10 mb of personal web space, but that won't even
start to cover all the materail I have.

How often do you check your mail??? I can forward a mb or two each day
till you holler STOP. Is your email address in the header ok?

Regards,

M
 
Joseph Casey wrote:

Greetings.
In the past I succesfully built HV transformers for a 2KV voltage
sourse. I drive the primary with a 20KHz sinwave. I used to use UU cores
like those in flyback transformers. But now I am having great
difficulty getting those type of cores - a MAJOR stockist of Ferroxcube
etc tells me they are very hard to get! So my question is: will cores
such as etd54 do the job? Why are single loop UU type cores always used
in flyback transformers?
Click to expand...
Cores are likely to be available according to market demand.

I find the ETD and EE series to be best supported overall for general
purpose.

Doesn't matter if it's flyback or forward.

Download Epcos's Ferrite Magnetic Designer utility. It's great.

Graham
 
Hi Mebart,

<Mebart> wrote in message news:q0o791h66191g0e85sa2q3l94pbadnit02@4ax.com...
I'm not sure-I know yahoo gives 20 mb of free web space if you start a
mailing list. I have 10 mb of personal web space, but that won't even
start to cover all the materail I have.
Click to expand...
OK.

How often do you check your mail???
Click to expand...
Yahoo! messenger runs whenever this PC is on (more or less work hours) and it
alerts me whenever I have new Yahoo! mail.

I can forward a mb or two each day
till you holler STOP. Is your email address in the header ok?
Click to expand...
Sure, that'd be great. Try sending... say... 5MB chunks, I'm pretty sure
Yahoo! mail will swallow that.

The e-mail address in the header is correct after removing the "HATESSPAM"
part. :)

Thanks,
---Joel
 
Winfield Hill wrote:

Joseph Casey wrote...

In the past I succesfully built HV transformers for a 2KV voltage
sourse. I drive the primary with a 20KHz sinwave. I used to use UU
cores like those in flyback transformers. But now I am having great
difficulty getting those type of cores - a MAJOR stockist of Ferroxcube
etc tells me they are very hard to get! So my question is: will cores
such as etd54 do the job? Why are single loop UU type cores always used
in flyback transformers?

I suppose the large winding area is attractive for high-voltage
designs. But this is less of an issue at higher frequencies.
For example, I used a U core with a bar piece for my 10kV 300kHz
resonant transformer.
Click to expand...
Large winding area was traditionally good to accomadate the necessary safety
margins.

Use triple insulated wire and your problems go away.

Graham
 
"Joel Kolstad" <JKolstad71HatesSpam@yahoo.com> a écrit dans le message de
news:1197ottpdrpmjcf@corp.supernews.com...
Hi Mebart,

Mebart> wrote in message
news:q0o791h66191g0e85sa2q3l94pbadnit02@4ax.com...
I'm not sure-I know yahoo gives 20 mb of free web space if you start a
mailing list. I have 10 mb of personal web space, but that won't even
start to cover all the materail I have.

OK.

How often do you check your mail???

Yahoo! messenger runs whenever this PC is on (more or less work hours) and
it
alerts me whenever I have new Yahoo! mail.

I can forward a mb or two each day
till you holler STOP. Is your email address in the header ok?

Sure, that'd be great. Try sending... say... 5MB chunks, I'm pretty sure
Yahoo! mail will swallow that.

The e-mail address in the header is correct after removing the "HATESSPAM"
part. :)
Click to expand...

Make a zip of all files and just use www.yousendit.com

You upload your file (up to 1GB), then the other guy receives an email with
a link to download it.


--
Thanks,
Fred.
 
Mebart wrote:

Joseph,

You didn't mention your location or the power level needed.

If you need 6 to 40 watts at up to 3 KV, you might look into a
piezotransformer. They have high step up ratios and do not suffer from
shunt capacitance problems like conventional wire based transformers
do.

To get a few watts, use a piezoinverter from a CCFL inverter which are
available here in the US. If you need more power (up to 30 watts, or
slightly more), you can use the newer radial piezotransformer
available in Europe.

Regulation is easily obtained as well, you didn't mention whether you
needed regulation.

They operate from 50 kHz to around 200 kHz and produce AC, with the
majority at the lower end of that frequency range.

Regards,

M




On Tue, 24 May 2005 16:33:13 +0100, Joseph Casey <joecasey@indigo.ie
wrote:


Greetings.
In the past I succesfully built HV transformers for a 2KV voltage
sourse. I drive the primary with a 20KHz sinwave. I used to use UU cores
like those in flyback transformers. But now I am having great
difficulty getting those type of cores - a MAJOR stockist of Ferroxcube
etc tells me they are very hard to get! So my question is: will cores
such as etd54 do the job? Why are single loop UU type cores always used
in flyback transformers?
Regards
Joseph Casey.


Are they reliable to 180C?
Click to expand...
Are small ones available - similar to the FerroxCube size P14/8?
 
"Pooh Bear" <rabbitsfriendsandrelations@hotmail.com> wrote in message
news:4293E37C.FA04AA25@hotmail.com...

Use triple insulated wire and your problems go away.
Click to expand...
.... or are replaced by unwanted resonances and capacitive losses. Distance
is Good!
 
But this is less of an issue at higher frequencies.
For example, I used a U core with a bar piece for my 10kV 300kHz
resonant transformer.


Thanks Win.
Click to expand...
Why is space less of an issue at higher frequencies? I presume you mean
less of an issue for insulation/arcing but I can't see why.
 
Joseph Casey wrote...
But this is less of an issue at higher frequencies.
For example, I used a U core with a bar piece for my
10kV 300kHz resonant transformer.

Thanks Win.
Why is space less of an issue at higher frequencies? I presume
you mean less of an issue for insulation/arcing but I can't see why.
Click to expand...
No, many fewer turns are required at higher frequencies, and
in most instances the core area can be smaller as well.


--
Thanks,
- Win
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top