High voltage switching power supply

[Hassan]

Hello;
I would like to design two switching power supplies:
1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA

Does anybody know any useful source of information in the internet so
that I can learn the design concepts of high voltage output switching
power supplies?
Also, I would like to know if there is any specific IC designed for
this purpose.

Thank you in advance for your help.

 

[colin]

Hello;
I would like to design two switching power supplies:
1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA

Does anybody know any useful source of information in the internet so
that I can learn the design concepts of high voltage output switching
power supplies?
Also, I would like to know if there is any specific IC designed for
this purpose.

Thank you in advance for your help.

as its only a few tens of watts i sugest u use one of those smps flyback ics
i think some of them even have a built in power transistor wich wld make it
very easy, or a use a high voltage mosfet, and perhaps a voltage doubler to
reduce the turns on the transformer. cant remember wich ones have off hand
but im using a uc3843 wich drives a bs131 mosfet to produce 200v + from 13v
with a small fluback inductor made out of a toroid. it even has voltage
reference and al the pwm stuff u need, very small foortprint in fact the
whole thing isnt much bigger than a postage stamp.

actualy i found it put it too close to sensitive inputs so i modified it so
the output drive a series resonant LC instead of the mosfet, and managed to
sync it so it resonated. worked quite well although it seemed a bit of a
bodge to get it to sync but the reduction in interference was amasing.

the ic datasheets usualy have lots of design stuff and further aplication
notes wich probably give u all the info u cld want.

i was hoping to find an IC that was capable of driving a series LC
arangement directly and was designed to sync well with the resonant
frequency.

oh and be carefull with the hi volts ...

Colin =^.^=

 

[Scrim]

Hello;
I would like to design two switching power supplies:
1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA

Does anybody know any useful source of information in the internet so
that I can learn the design concepts of high voltage output switching
power supplies?
Also, I would like to know if there is any specific IC designed for
this purpose.

Thank you in advance for your help.

I can't tell you how to do it, but I can warn you to watch out for the
large _intrawinding_ capacitance from all those turns on the secondary. If
you use a flyback scheme you can keep the winding ratio lower if you let the
primary voltage climb high during the flyback period, which means using a
high voltage switching transistor, even with a 40v supply. I think that the
reason the old unsophisticated circuits with extra windings to drive the
transistor base(s) are so effective is because they work resonantly, which
solves the problem caused by the secondary intrawinding capacitance. See AN1
and AN4,fig4 here for the kind of design I mean:
http://www.zetex.com/3.0/3-8-1a.asp

I'm definitely no expert on this, but have experienced some of the problem
when trying to make a ~3000V 6.5mA constant current supply for a HeNe laser.
Good luck, and please let me know if you find any good information on doing
this.

Scrim

 

[Robert Baer]

I can't tell you how to do it, but I can warn you to watch out for the
large _intrawinding_ capacitance from all those turns on the secondary. If
you use a flyback scheme you can keep the winding ratio lower if you let the
primary voltage climb high during the flyback period, which means using a
high voltage switching transistor, even with a 40v supply. I think that the
reason the old unsophisticated circuits with extra windings to drive the
transistor base(s) are so effective is because they work resonantly, which
solves the problem caused by the secondary intrawinding capacitance. See AN1
and AN4,fig4 here for the kind of design I mean:
http://www.zetex.com/3.0/3-8-1a.asp

I'm definitely no expert on this, but have experienced some of the problem
when trying to make a ~3000V 6.5mA constant current supply for a HeNe laser.
Good luck, and please let me know if you find any good information on doing
this.

Scrim

Excellent reference! AN7 also seems to be useful.
Too bad that page 2 of AN7 is screwed up...

 

[Scrim]

Excellent reference! AN7 also seems to be useful.
Too bad that page 2 of AN7 is screwed up...

Looks ok to me. What's wrong with it?

Scrim

 

[Genome]

| Hello;
| I would like to design two switching power supplies:
| 1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
| 2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA
|
| Does anybody know any useful source of information in the internet so
| that I can learn the design concepts of high voltage output switching
| power supplies?
| Also, I would like to know if there is any specific IC designed for
| this purpose.
|
| Thank you in advance for your help.

Look at the UC3827-1 current fed push-pull buck regulator IC from
Unitrode, now part of TI.

DNA

 

[HARRY DELLAMANO]

| Hello;
| I would like to design two switching power supplies:
| 1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
| 2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA
|
| Does anybody know any useful source of information in the internet so
| that I can learn the design concepts of high voltage output switching
| power supplies?
| Also, I would like to know if there is any specific IC designed for
| this purpose.
|
| Thank you in advance for your help.

Look at the UC3827-1 current fed push-pull buck regulator IC from
Unitrode, now part of TI.

DNA

Ok DNA, i'll bite, why would you use a UC3827-1 on a 30W single output
application?
Regards
Harry

 

[Winfield Hill]

Scrim wrote...

I can't tell you how to do it, but I can warn you to watch out for the
large _intrawinding_ capacitance from all those turns on the secondary.

Sadly I haven't had the time today to pull out material from my own
high-voltage transformer design files, or copies of web-info stored
on my computer, but here's one reference, Linear Technology's AN39,
http://www.linear-tech.com/pub/document.html?pub_type=app&document=43


Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[colin]

Scrim wrote...

Sadly I haven't had the time today to pull out material from my own
high-voltage transformer design files, or copies of web-info stored
on my computer, but here's one reference, Linear Technology's AN39,
http://www.linear-tech.com/pub/document.html?pub_type=app&document=43


Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

yes interwinding capacitance can be a killer, the capacitance on the
secondary goes up with the number of turns and when its reflected back to
the primary it increase by the square of the turns ratio. so its
proportional to turns ^3

one way to get round this is to use resonant technique, long time ago i made
a cdi gnition unit for a motorbike, this used same technique to keep the
eficiency high as posible before mosfets were available, it used a simple
cmos hex shmit inverter, and free ran as a relaxation oscilator untill the
output gets going into resonance then the colector voltage faling below a
few volts trigered the next cycle of turning the transistor on, thus the
transistor always turned on at zero voltage so no switching loss, a largish
capacitor could then be used acros the colector to eliminate turn off losses
too and aded to the resonance, but too high and you need a larger
inductor/transformer becuase some of the stored energy goes into charging up
the capcitance rather than into the output. it was quite simple just used 3
of the 6 hex inverters and a single transistor driver.

i keep looking for a smps pwm control ic that uses the same technique but
most just seem to ramp the frequency and hope it hits the resonant point.


Colin =^.^=

 

[Robert Baer]

Looks ok to me. What's wrong with it?

Scrim

Look at the screwed up legend for the graph on the right, as well as
the "chopped off" legend below it.

 

[Scrim]

Look at the screwed up legend for the graph on the right, as well as
the "chopped off" legend below it.

OK, I see it.

 

[Scrim]

Scrim wrote...

Sadly I haven't had the time today to pull out material from my own
high-voltage transformer design files, or copies of web-info stored
on my computer, but here's one reference, Linear Technology's AN39,
http://www.linear-tech.com/pub/document.html?pub_type=app&document=43


Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

Nice Link Win. That describes exactly the problem I had trying to build an
I.C. based HeNe laser psu (3000V, 6.5mA constant current). But by the time
you get to 3000V the problem is very bad. I've seen the split bobbin
approach used in some HT circuits, like gas oven ignition systems, which
dump a mains charged capacitor into the primary to produce a spark out of
the secondary.
I pulled a HeNe supply 'brick' apart once though (it was a switching supply
set in a solid brick of encapsulant, which I softened and removed by
heating, but not very successfully!). This ran straight off of the mains
(240V) and the transformer was conventionally wound (i.e. no slit bobbin)
and had a surprisingly low turns ratio. Having spent foolishly large amounts
of time trying to wind low capacitance transformers I gave up on using
switching supply I.C.'s for HT supplies. My conclusion was that the best
approach would be to use a simple low voltage driven old fashioned
transistor resonant step up system with some op amps etc to controll it.

This was very annoying because what I really wanted to do was have the psu
run straight off the mains, perhaps a copy of the system used in the HeNe
brick, but I never found out how that worked.

If you find time to dig up any more info or can point to some please do.

Scrim

 

[Scrim]

Look at the screwed up legend for the graph on the right, as well as
the "chopped off" legend below it.

By the way, don't confuse 'INTERWINDING' and 'INTRAWINDING' capacitance. The
former refers to that between primary and secondary, whilst the later refers
to capacitance between the windings of the secondary and is also refered to
as distributed capacitance.

Scrim

 

[Genome]

|
| | >
| > | > | Hello;
| > | I would like to design two switching power supplies:
| > | 1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
| > | 2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA
| > |
| > | Does anybody know any useful source of information in the internet
so
| > | that I can learn the design concepts of high voltage output
switching
| > | power supplies?
| > | Also, I would like to know if there is any specific IC designed
for
| > | this purpose.
| > |
| > | Thank you in advance for your help.
| >
| > Look at the UC3827-1 current fed push-pull buck regulator IC from
| > Unitrode, now part of TI.
| >
| > DNA
|
| Ok DNA, i'll bite, why would you use a UC3827-1 on a 30W single output
| application?
| Regards
| Harry
|

Err..., it's the only method I know for generating high voltages and I
think its cute.

DNA

 

[colin]

Scrim wrote...

Sadly I haven't had the time today to pull out material from my own
high-voltage transformer design files, or copies of web-info stored
on my computer, but here's one reference, Linear Technology's AN39,
http://www.linear-tech.com/pub/document.html?pub_type=app&document=43


Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

the lt1070 mentioned in that app note looks realy good for making the
minimalist parts count converter. and so many circuit aplications if you
look at the spec sheet for it, its a goldmine. i wldnt of thought of many of
them. they have seem to have every thinkable combination of pos/negative
input/output, buck/boost, forward/flyback ...

however the internal switch is only rated at 70 volts so i wldnt recomend
the 40:1 turns ratio you would need for the 3kv op if if u just use the
internal transistor. ive never trtied it tho ...

uc3843 still the best bet in my opnion with a 1kv mosfet u cld almost get
away with an no step up transformer for 1kv and just use a tripler for 3kv.
a turns ratio of 5:1 for the 1kv suply gives a switch voltage of 200v and
say 7:1 and 430v for the 3kv.

although on reading the lt1070 again it does show you can drive a power
transistor using comon base/gate type arangement.

i just further reduced the emi from my hv inverter, i used 2 of those 70v
quad schotky arays to form a voltage multiplier (8 diodes to give X4),
takes up hardly any board space, so now my resonant inverter only has to
generate 70v max and makes the toriod a lot smaller. cant see any cross
interference at all now, even on the circuit right next to it, although god
knows why i didnt just put it further away in the first place.


Colin =^.^=

 

[colin]

Nice Link Win. That describes exactly the problem I had trying to build an
I.C. based HeNe laser psu (3000V, 6.5mA constant current). But by the time
you get to 3000V the problem is very bad. I've seen the split bobbin
approach used in some HT circuits, like gas oven ignition systems, which
dump a mains charged capacitor into the primary to produce a spark out of
the secondary.
I pulled a HeNe supply 'brick' apart once though (it was a switching supply
set in a solid brick of encapsulant, which I softened and removed by
heating, but not very successfully!). This ran straight off of the mains
(240V) and the transformer was conventionally wound (i.e. no slit bobbin)
and had a surprisingly low turns ratio. Having spent foolishly large amounts
of time trying to wind low capacitance transformers I gave up on using
switching supply I.C.'s for HT supplies. My conclusion was that the best
approach would be to use a simple low voltage driven old fashioned
transistor resonant step up system with some op amps etc to controll it.

This was very annoying because what I really wanted to do was have the psu
run straight off the mains, perhaps a copy of the system used in the HeNe
brick, but I never found out how that worked.

If you find time to dig up any more info or can point to some please do.

Scrim

i dont think its realy that dificult as long as you dont try to go from say
12v to the hv al in one step with a transformer, if so your making things
very dificult for yourself.

whatever input voltage you have 12v + or more ....

1) use hi voltage switch in flyback mode to generate hi voltage spike at
turn off acros trsnformer (small 400v, even upto 1000v mosfet are easily
available)
2) use low number of turns as posible on pri and secondary (wind it neatly
space the turns if necesary and use several layers of tape between each
layer) use of oversized core gives you a very low primary turns.
3) tune them for resonance if you feel like it
4) use a doubler or tripler etc ..

you can get away with transformers with very low turns ratios ie 5:1 and
hence avoid winding cap. if you take apart a tv line output transformer wich
generates 25kv youl find it has surprisingly few turns on the hv winding.

in fact look at the tea1541 app note gives you diagrams etc for tv eht
supply direct from mains.

i havnt thought about how you would make it a constant curent output but
constant power output would be realy easy, just make the switch turn off at
a preset current.

Colin =^.^=

 

[HARRY DELLAMANO]

|
| | >
| > | > | Hello;
| > | I would like to design two switching power supplies:
| > | 1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
| > | 2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA
| > |
| > | Does anybody know any useful source of information in the internet
so
| > | that I can learn the design concepts of high voltage output
switching
| > | power supplies?
| > | Also, I would like to know if there is any specific IC designed
for
| > | this purpose.
| > |
| > | Thank you in advance for your help.
| >
| > Look at the UC3827-1 current fed push-pull buck regulator IC from
| > Unitrode, now part of TI.
| >
| > DNA
|
| Ok DNA, i'll bite, why would you use a UC3827-1 on a 30W single output
| application?
| Regards
| Harry
|

Err..., it's the only method I know for generating high voltages and I
think its cute.

DNA

We like cute, can you tell us more?
My guess is that you have a Buck followed by a Current Fed Push-Pull that
has multi outputs that are stacked in series for the high voltage. You then
close the loop back to the Buck which is PWM'ed for output voltage control.
Now that said, why is it cute? Well since it is current fed each output
section can be a single winding across two stacked caps on one side and two
stacked diodes on the other. Each section would yield 2*N*Ep/winding. The
P-P going close to 50% duty would have ripple a canceling effect on the
output voltage and require little added filtering.
Cute for power >100W but is it necessary for 30W when a Flyback will do?
Please advise,
Harry

 

[Genome]

|
| | >
| > | > |
| > | | > | >
| > | > | > | > | Hello;
| > | > | I would like to design two switching power supplies:
| > | > | 1- Vin= 40 V DC & Vo= 1000 V DC @ 20mA
| > | > | 2- Vin= 40 V DC & Vo= 3000 V DC @ 10mA
| > | > |
| > | > | Does anybody know any useful source of information in the
internet
| > so
| > | > | that I can learn the design concepts of high voltage output
| > switching
| > | > | power supplies?
| > | > | Also, I would like to know if there is any specific IC
designed
| > for
| > | > | this purpose.
| > | > |
| > | > | Thank you in advance for your help.
| > | >
| > | > Look at the UC3827-1 current fed push-pull buck regulator IC
from
| > | > Unitrode, now part of TI.
| > | >
| > | > DNA
| > |
| > | Ok DNA, i'll bite, why would you use a UC3827-1 on a 30W single
output
| > | application?
| > | Regards
| > | Harry
| > |
| >
| > Err..., it's the only method I know for generating high voltages and
I
| > think its cute.
| >
| > DNA
| >
| We like cute, can you tell us more?
| My guess is that you have a Buck followed by a Current Fed Push-Pull
that
| has multi outputs that are stacked in series for the high voltage. You
then
| close the loop back to the Buck which is PWM'ed for output voltage
control.
| Now that said, why is it cute? Well since it is current fed each
output
| section can be a single winding across two stacked caps on one side
and two
| stacked diodes on the other. Each section would yield 2*N*Ep/winding.
The
| P-P going close to 50% duty would have ripple a canceling effect on
the
| output voltage and require little added filtering.
| Cute for power >100W but is it necessary for 30W when a Flyback will
do?
| Please advise,
| Harry
|

Ow, Leave me alone...... I like the fact you can stack windings which
will track each other. I like the fact that the output dioes are less
stressed when compared to a flyback. I like the fact that the IC 'drops'
in to the OPs requirements. I like the fact that it's a buck so you
don't have to worry about RHPZs and such stuff. It's just nice and I'm
lazy.

DNA

 

[Robert Baer]

i dont think its realy that dificult as long as you dont try to go from say
12v to the hv al in one step with a transformer, if so your making things
very dificult for yourself.

whatever input voltage you have 12v + or more ....

1) use hi voltage switch in flyback mode to generate hi voltage spike at
turn off acros trsnformer (small 400v, even upto 1000v mosfet are easily
available)
2) use low number of turns as posible on pri and secondary (wind it neatly
space the turns if necesary and use several layers of tape between each
layer) use of oversized core gives you a very low primary turns.
3) tune them for resonance if you feel like it
4) use a doubler or tripler etc ..

you can get away with transformers with very low turns ratios ie 5:1 and
hence avoid winding cap. if you take apart a tv line output transformer wich
generates 25kv youl find it has surprisingly few turns on the hv winding.

in fact look at the tea1541 app note gives you diagrams etc for tv eht
supply direct from mains.

i havnt thought about how you would make it a constant curent output but
constant power output would be realy easy, just make the switch turn off at
a preset current.

Colin =^.^=

Please be so kind to give web link to that Tea1541 app note.

 

[colin]

*snip*

Please be so kind to give web link to that Tea1541 app note.

this is the tea1541 app note i was looking at earlier although it doesnt
give eht circuit its just to regulate the supply to the eht supply doh
!although its still pretty usefulll

http://www.semiconductors.philips.com/acrobat/applicationnotes/AN10205_1.pdf

this is the one i was realy meaning but its for a tv deflection ic
TDA4855/58, but u can ignore the deflection parts.

http://www-eu2.semiconductors.com/acrobat/applicationnotes/AN96052_1.pdf

i just looked for uc3842 and eht on google im sure theres more to be found
....


Colin =^.^=