high voltage boost DC/DC converters (5V->250V)

[Michael]

Hi there - at work I'm probably going to have to design a high voltage
DC/DC converter sometime in the near future. It looks like it'll be
something along the lines of 5V input and 250V output. Current will be
under 10ma. Probably more like 1-2ma. The more regulated the better,
but I will probably just regulate it down a bit with a FET to clean it
up.

My understanding is that once you start to get to something like an
output voltage 10x or more of the input voltage the standard boost
circuit (cap + diode + inductor) no longer really cuts it. Instead,
I've read that step-up transformers are a better way to go. However,
when I look through Digi-Key, all I see are step down transformers for
stepping down wall power, as well as various specialty transformers.

Any suggestions? Thanks!

-Michael

 

[[email protected]]

Hi there - at work I'm probably going to have to design a high voltage
DC/DC converter sometime in the near future. It looks like it'll be
something along the lines of 5V input and 250V output. Current will be
under 10ma. Probably more like 1-2ma. The more regulated the better,
but I will probably just regulate it down a bit with a FET to clean it
up.

Why would you say that? What's your switching frequency?

I hacked something like this
http://www.cpemma.co.uk/555pwm.html

with a boost converter in place of the motor, and got up to 100V out:
enough to light up a 110V night light with about 10V (eight NiMH AAs
in series).

The inductor I used was pretty beefy, though: from a throw-away
blender. Thick gauge wire.

- another Michael

 

[Nico Coesel]

Hi there - at work I'm probably going to have to design a high voltage
DC/DC converter sometime in the near future. It looks like it'll be
something along the lines of 5V input and 250V output. Current will be
under 10ma. Probably more like 1-2ma. The more regulated the better,
but I will probably just regulate it down a bit with a FET to clean it
up.

My understanding is that once you start to get to something like an
output voltage 10x or more of the input voltage the standard boost
circuit (cap + diode + inductor) no longer really cuts it. Instead,
I've read that step-up transformers are a better way to go. However,
when I look through Digi-Key, all I see are step down transformers for
stepping down wall power, as well as various specialty transformers.

Any suggestions? Thanks!

Wurth (www.we-online.com) has solutions for these sort of 'problems'.
They have standard transformers which can be connected in several ways
to make all kind of ratios. Getting some samples and support from
these guys is never a problem.

 

[Jan Panteltje]

Hi there - at work I'm probably going to have to design a high voltage
DC/DC converter sometime in the near future. It looks like it'll be
something along the lines of 5V input and 250V output. Current will be
under 10ma. Probably more like 1-2ma. The more regulated the better,
but I will probably just regulate it down a bit with a FET to clean it
up.

This always worked for me:
+5V
|
|
|
e
b-------------------------------------------------------------------------------------------
c any PNP |
| |
| |
---------------------------------- B Vout
| | _---------A | A---------------a D k---------------------------------------------------------- + 250V
| R2 2k7 |( | diode 1kV | | B +5
| | |(300turns | | R1 +5 | |
| |-------- |( | | | | | R 2k7
| | 1/2turn) |(------------ | | c c |
| | -- |( 5turns C 10uF 300V |--- b b ----|
| | | |------------- | | e e |
| | | c | | | |______| |
|+ |- ---- b any NPN C1 22nF poly | | | Z zener 3.3V
C3 10uF C2 10uF e | | R 10k R |
|- |+ | | | | | |
/// /// /// /// /// /// /// ///
small potcore 2x any NPN
calculate R1
for Vout
Waveform is sine! select C1 for frequency (few kHz).
connect A with A,m and also B with B.
Note C2 is reverse biased.
C1 must be poly (ceramics may not work right).
C3 and C2 best be tantalum, the transistors Si.
Typos reserved.
Copyright Jan Panteltje 2007
All right reserved.
Usenet patent 0x754
Released under the hardware GPL.
<disclaimer>

The only thing is that you will have to wind 300 turns or so thin wire on a coil former.
Impregnate it too.

 

[Winfield Hill]

Hi there - at work I'm probably going to have to design a high voltage
DC/DC converter sometime in the near future. It looks like it'll be
something along the lines of 5V input and 250V output. Current will be
under 10ma. Probably more like 1-2ma. The more regulated the better,
but I will probably just regulate it down a bit with a FET to clean it
up.

My understanding is that once you start to get to something like an
output voltage 10x or more of the input voltage the standard boost
circuit (cap + diode + inductor) no longer really cuts it. Instead,
I've read that step-up transformers are a better way to go. However,
when I look through Digi-Key, all I see are step down transformers for
stepping down wall power, as well as various specialty transformers.

Small smps transformers are very easy to wind.

Here's what I designed for a similar need about
12 years ago. I used an LT1172 switcher IC in
a miniDIP package.

My technician wound the transformer bobbin.

I used an RM8 core with 3B7 material, gapped
(on one side) for A160 (A_L = 160nH/N^2).
The primary was 10 turns of #24 magnet wire
(for L = 25uH), then a layer of tape, and
the secondary was 100 turns of #30 wire.
So you can see it was easy to wind.

 

[[email protected]]

This always worked for me:
+5V
|
|
|
e
b--------------------------------------------------------------------------------------------
c any PNP |
| |
| |
---------------------------------- B Vout
| | _---------A | A---------------a D k---------------------------------------------------------- + 250V
| R2 2k7 |( | diode 1kV | | B +5
| | |(300turns | | R1 +5 | |
| |-------- |( | | | | | R 2k7
| | 1/2turn) |(------------ | | c c |
| | -- |( 5turns C 10uF 300V |--- b b ----|
| | | |------------- | | e e |
| | | c | | | |______| |
|+ |- ---- b any NPN C1 22nF poly | | | Z zener 3.3V
C3 10uF C2 10uF e | | R 10k R |
|- |+ | | | | | |
/// /// /// /// /// /// /// ///
small potcore 2x any NPN
calculate R1
for Vout
Waveform is sine! select C1 for frequency (few kHz).
connect A with A,m and also B with B.
Note C2 is reverse biased.
C1 must be poly (ceramics may not work right).
C3 and C2 best be tantalum, the transistors Si.
Typos reserved.
Copyright Jan Panteltje 2007
All right reserved.
Usenet patent 0x754
Released under the hardware GPL.
<disclaimer>

The only thing is that you will have to wind 300 turns or so thin wire on a coil former.
Impregnate it too.

You might want to look at the articles in the "Circuits" section at
http://www.highvoltageinfo.com/articles.php for a few ways you can
build your power supply.

 

[Joerg]

Hi there - at work I'm probably going to have to design a high voltage
DC/DC converter sometime in the near future. It looks like it'll be
something along the lines of 5V input and 250V output. Current will be
under 10ma. Probably more like 1-2ma. The more regulated the better,
but I will probably just regulate it down a bit with a FET to clean it
up.

My understanding is that once you start to get to something like an
output voltage 10x or more of the input voltage the standard boost
circuit (cap + diode + inductor) no longer really cuts it. Instead,
I've read that step-up transformers are a better way to go. However,
when I look through Digi-Key, all I see are step down transformers for
stepping down wall power, as well as various specialty transformers.

Any suggestions? Thanks!

I went well past 10:1 on bucks. But when the factor nears 20 it does get
old. You need a ferrite transformer like those used in very small
switcher wall warts. Except that you'd operate this transformer in
reverse. The problem is the usual: Very difficult to obtain at
reasonable prices when outside China. Sure, you could get the for a few
dimes from China but only if you buy a whole pallet of them. Also, keep
in mind that the flyback versions are air-gapped.

 

[Joerg]

Small smps transformers are very easy to wind.

Here's what I designed for a similar need about
12 years ago. I used an LT1172 switcher IC in
a miniDIP package.

My technician wound the transformer bobbin.

^^^^^^^^^^^^^^^^^^^


Aha! I always have to put lots of aloe creme on my index fingers after
doing too much of this winding. Else they'll split open in cold weather.

I used an RM8 core with 3B7 material, gapped
(on one side) for A160 (A_L = 160nH/N^2).
The primary was 10 turns of #24 magnet wire
(for L = 25uH), then a layer of tape, and
the secondary was 100 turns of #30 wire.
So you can see it was easy to wind.

I mostly go non-gapped. Push pull or AC-coupled half-bridge. And don't
mis-count. One-Mississippi, two-Mississippi, ...

 

[Tam/WB2TT]

Hi there - at work I'm probably going to have to design a high voltage
DC/DC converter sometime in the near future. It looks like it'll be
something along the lines of 5V input and 250V output. Current will be
under 10ma. Probably more like 1-2ma. The more regulated the better,
but I will probably just regulate it down a bit with a FET to clean it
up.

My understanding is that once you start to get to something like an
output voltage 10x or more of the input voltage the standard boost
circuit (cap + diode + inductor) no longer really cuts it. Instead,
I've read that step-up transformers are a better way to go. However,
when I look through Digi-Key, all I see are step down transformers for
stepping down wall power, as well as various specialty transformers.

Any suggestions? Thanks!

-Michael

If you really want to do it on the cheap, get a 60 Hz 120:6VCT transformer.
Apply push pull drive to the 6V windings at about 100 Hz, and take the HV
off the 120 winding with a full wave doubler circuit. You will want to take
the drive off a Dff to insure 50% duty cycle, otherwise there will be a DC
component in the transformer. There should be a reasonable looking square
wave on the 120 winding.

Tam

 

[John Larkin]

Why would you say that? What's your switching frequency?

I hacked something like this
http://www.cpemma.co.uk/555pwm.html

with a boost converter in place of the motor, and got up to 100V out:
enough to light up a 110V night light with about 10V (eight NiMH AAs
in series).

The inductor I used was pretty beefy, though: from a throw-away
blender. Thick gauge wire.

You're boosting about 10:1. He needs 50:1. High boost ratios are
tough. The energy in the inductor is 0.5*L*I^2. When the fet turns off
and things fly up, the winding, fet, and diode capacitance have to be
charged, and that energy is 0.5*C*V^2. As the boost voltage goes up,
the energy lost in those capacitances goes up as V^2. At typical
numbers like 50 or 100:1, there's just nothing left for the load. The
"fix" is less L, which requires a beefier, higher-voltage fet, which
adds drain capacitance, and the whole thing bites you in the butt.

Using a transformer, or an autotransformer, helps a lot. Mouser and
Digikey sell lots of dual-winding inductors, which can be configured
as a center-tapped autotransformer: +5 on one end, fet drain in the
middle, flyback from the other end.

It could be that a spiffy bipolar, like one of the super-saturating
Zetex parts maybe, would have a better ratio of saturation current to
output capacitance than a fet.

John



John

 

[Robert Latest]

Here's what I designed for a similar need about
12 years ago. I used an LT1172 switcher IC in
a miniDIP package.

Along the lines of page 10 in the LT1172 datasheet, top image?

robert

 

[Mike Harrison]

I went well past 10:1 on bucks. But when the factor nears 20 it does get
old. You need a ferrite transformer like those used in very small
switcher wall warts. Except that you'd operate this transformer in
reverse. The problem is the usual: Very difficult to obtain at
reasonable prices when outside China. Sure, you could get the for a few
dimes from China but only if you buy a whole pallet of them. Also, keep
in mind that the flyback versions are air-gapped.

I think a few people do standard magnetics for Power Integrations' Topswitch family - these may be
suitable.
Also look at transformers for photo flash/strobe charging - you may be able to find these off the
shelf

 

[Winfield]

Along the lines of page 10 in the LT1172 datasheet, top image?

Close, that's a 20-30V to 5V step-down, whereas mine
was a 5V to 220V step-up. BTW, I see they specify a
1:3 transformer, whereas wouldn't turning it around,
making 3:1, be more appropriate?

I used a 1:10 ratio, so the 220V output corresponded
to a 22+5 = 27V flyback swing, plus a little leakage
inductance spike. Looking at my old RIS-177 drawing,
I don't see a clamp or snubber. Hmm, seems to have
worked well all these years, but I do know better.
Note, they suggest a 25V zener plus a fast diode.

 

[Robert Latest]

Close, that's a 20-30V to 5V step-down, whereas mine
was a 5V to 220V step-up.

Well, of course, modulo that. I need some 1kV/10uA, preferably from a
single AA (for an insulation tester; I'm tired of lugging around the
mains-powered one), but I'm so swamped with other things that I'll probably
never get around to it.

robert

 

[Winfield]

Well, of course, modulo that. I need some 1kV/10uA, preferably
from a single AA (for an insulation tester; I'm tired of lugging
around the mains-powered one), but I'm so swamped with other
things that I'll probably never get around to it.

1000:1 is a bit of a challenge for the transformer,
all kinds of painful issues, like excess winding
capacitance, etc., may come up, especially at low
power. Not to say it can't be done, or hasn't been
done, but I'd be inclined to first step up from 1V
to 5 or 10V, then go the rest of the way with a
flyback transformer.

 

[Michael]

Small smps transformers are very easy to wind.

Here's what I designed for a similar need about
12 years ago. I used an LT1172 switcher IC in
a miniDIP package.

My technician wound the transformer bobbin.

I used an RM8 core with 3B7 material, gapped
(on one side) for A160 (A_L = 160nH/N^2).
The primary was 10 turns of #24 magnet wire
(for L = 25uH), then a layer of tape, and
the secondary was 100 turns of #30 wire.
So you can see it was easy to wind.

Hi Winfield - so is the standard path to wind your own for prototypes,
then have a custom transformer wound when you hit production? I had
been hoping to find a readily available transformer (as my quantities
will never be high enough to warrant a full custom run of
transformers, but will be high enough to make me scared at the thought
of hand winding all of the buggers). Thanks!

-Michael

 

[Winfield Hill]

Hi Winfield - so is the standard path to wind your own for
prototypes, then have a custom transformer wound when you
hit production? I had been hoping to find a readily available
transformer (as my quantities will never be high enough to
warrant a full custom run of transformers, but will be high
enough to make me scared at the thought of hand winding all
of the buggers). Thanks!

There are hordes of custom transformer shops out
there that'll make quantities as low as 5 pieces,
at affordable prices, hoping you'll come back for
more later. Who can say, your product may take off.

 

[Tam/WB2TT]

Hi Winfield - so is the standard path to wind your own for prototypes,
then have a custom transformer wound when you hit production? I had
been hoping to find a readily available transformer (as my quantities
will never be high enough to warrant a full custom run of
transformers, but will be high enough to make me scared at the thought
of hand winding all of the buggers). Thanks!

-Michael

Electronic photo flash units commonly work off 3 or 6 volts, and deliver
several hundred V DC. You should be able to find a schematic for such, or
even fix a junked flash unit; maybe from a disposable single use camera.

Tam

 

[Joerg]

I think a few people do standard magnetics for Power Integrations' Topswitch family - these may be
suitable.

A bit too pricey for my taste, and single-sourced. I'd rather roll my own.

Also look at transformers for photo flash/strobe charging - you may be able to find these off the
shelf

And those for CFL backlights since those have HV windings.

 

[Joerg]

You're boosting about 10:1. He needs 50:1. High boost ratios are
tough. The energy in the inductor is 0.5*L*I^2. When the fet turns off
and things fly up, the winding, fet, and diode capacitance have to be
charged, and that energy is 0.5*C*V^2. As the boost voltage goes up,
the energy lost in those capacitances goes up as V^2. At typical
numbers like 50 or 100:1, there's just nothing left for the load. The
"fix" is less L, which requires a beefier, higher-voltage fet, which
adds drain capacitance, and the whole thing bites you in the butt.

Using a transformer, or an autotransformer, helps a lot. Mouser and
Digikey sell lots of dual-winding inductors, which can be configured
as a center-tapped autotransformer: +5 on one end, fet drain in the
middle, flyback from the other end.

Whenever I looked I found them to be rather expensive, even at 1k qties.
Then when my clients went into production and we sourced it all in China
the cost of a custom made dropped to 10% or so. Why is that? Somewhere
in between there must be somebody trying to make oodles of money. So I
never buy more than a handful.

It could be that a spiffy bipolar, like one of the super-saturating
Zetex parts maybe, would have a better ratio of saturation current to
output capacitance than a fet.

Zetex makes really nice transistors. Unfortunately mostly
single-sourced, but good.