High voltage buck and forward converter .... - or one stage direct

[Klaus Kragelund]

Hi

I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.

The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.

Two questions:

1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.

2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)

Thanks

Klaus

 

[HarryD]

Hi

I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.

The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.

Two questions:

1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.

2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)

Thanks

Klaus

I would consider a non isolated Buck Converter, bucking down to 150V.
Follow this with a non-isolated Boost Converter, boosting up to 75V.
Follow this 75V to 150V input bus with an isolated forward converter that
can easily handle the 2X input voltage range. The first two stages would
normally pass or convert as required and operate at a max duty of 5:1. The
two intermediate bulk caps need to handle only 150V max.
Going directly from 16V-600V in a forward converted is very difficult! It
would take a 800V MOSFET and duty as low as 1.3%.

Cheers,
Harry

 

[Joerg]

Hi

I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.

The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.

Two questions:

1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.

Sorry, no idea there since I never use "fat" converter chips. Their
production lifetime is often low and they are hardly ever 2nd sourced.
Also, IMHO having power components and control circuitry on the same
chip isn't so great.

2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)

Done that recently but the other way around, 10-15V in and a few hundred
volts out. Half-bridges and similar architectures do not like to be
regulated but they can be. So, yesterday it was ready and I had to fire
it up. Only about 5W but the principle of operation is always similar.
For fun and to test stability I turned the HV converter down to close to
zero. The output was still nicely regulated but of course the duty cycle
looked like s..t. Similar to an idling Harley-Davidson with really
fouled-up spark plugs. Even the trusty old Tek 2465 had trouble
triggering onto the fuzz. Without the metal enclosure it would probably
have been an EMI nightmare.

Regulating a bridge converter can be compared to some really old rotary
engine aircraft from almost a century ago. The cylinders spun around and
the "crankshaft" was bolted to the fuselage. Since a throttle was not
available and the engine would always deliver full power the only way to
land them (and survive) was to reduce power by cutting the ignition to
several cyclinder. Rat-tat-poof-rat-tat-BANG-poof ...

As Harry mentioned, with such wide input ranges you need FETs that can
push a lot of current around and at the same time can stomach a kilovolt
or so. Spend some time on the snubber design to make sure the spikes
never ever exceed the breakdown limits.

 

[Mike Monett]

[...]

Regulating a bridge converter can be compared to some really old
rotary engine aircraft from almost a century ago. The cylinders
spun around and the "crankshaft" was bolted to the fuselage. Since
a throttle was not available and the engine would always deliver
full power the only way to land them (and survive) was to reduce
power by cutting the ignition to several cyclinder.
Rat-tat-poof-rat-tat-BANG-poof.

I thought they cut the ignition to all cylinders. Not so!

Here's some more info for those aspiring to be fighter pilots:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
8 October 2004, 03:07 PM

By Tripehound

Rotary throttle

Not a throttle like in your car, but a way of increasing or
decreasing the amount of fuel mixture fed to the engine, in three
separate steps.

There was an engine air intake pipe that used something like a ball
valve to regulate the amount of air going to the engine. This was
generally operated via a linkage from a handle in a "throttle
quadrant" located on the left side of the cockpit. The amount of
fuel going to this intake pipe was regulated by a "fine fuel" valve
from the pressurized fuel tank, or from a gravity tank, to a jet
inside the ball valve, where the incoming fuel was atomized. The air
ball valve acted like a Venturi tube, supplying suction to help move
the fuel. This "fine fuel" valve was operated by a separate handle
located in or near the "throttle quadrant". The Germans seemed to
favor locating the "fine fuel" control on the control column and
operating the valve by a twist knob via a Bowden(?) cable. Castor
oil was carried in a separate tank and pumped from this tank thru a
sight flow regulator in the cockpit to the fuel mixture feed line as
it entered the crankcase.

Both air and fuel had to be increased or decreased in steps to avoid
leaning or flooding, either one resulting in a dead engine. As the
aircraft climbed and the air became less dense, the fuel had to be
cut back to avoid flooding and a dead engine. As the aircraft
descended the fuel had to be increased to avoid leaning out and a
dead engine. Oil regulation didn't seem to be quite as demanding of
immediate attention as the air/fuel mix, but still necessary.

The coup button, or "blip" switch, allowed quick changes in engine
speed by cutting ignition to all cylinders. On landing, for example,
the engine was adjusted to take-off power then the coup button used
to cut the engine for brief intervals to decrease air speed for
landing. If the button was held too long the unburned fuel/oil mix
would flood the engine to where it wouldn't restart when the button
was let go, and/or worst case the expelled unburned fuel mix would
catch fire in the cowling.

Wooden airplanes and iron men!

Gregory -

Have a look at this thread from a few years back.

http://www.theaerodrome.com/forum/aircraft/2302-throttle-
down.html?highlight=throttle

(It's three pages - too long to post but quite interesting.)

Tripehound's explanation is very good. I would only add that some
makes and models of rotaries had different methods of control in
addition to the mixture adjustments and coup button. The 160hp
Gnome, for example, could be set to fire all 9, 7, 5, or just 3
cylinders on each revolution. The cylinders that were not firing
still cycled, but had no spark. They filled with fuel/oil and then
expelled it, unburnt, into the cowl. Messy and primitive, but it
worked. Mostly.

There are photos in Mike Vine's Return to Rhinebeck of ORA's repro
rotary-powered Caudron G3 being overhauled. The mechanic has the
nacelle up on sawhorses with the cowl and engine removed - you would
not believe the GUNK on the firewall. It's as thick as cake
frosting.

Last edited by EricGoedkoop; 9 October 2004 at 07:36 AM.

http://www.theaerodrome.com/forum/aircraft/17121-rotary-engine-
throttle.html
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Regards,

Mike Monett

 

[Joerg]

[...]

Regulating a bridge converter can be compared to some really old
rotary engine aircraft from almost a century ago. The cylinders
spun around and the "crankshaft" was bolted to the fuselage. Since
a throttle was not available and the engine would always deliver
full power the only way to land them (and survive) was to reduce
power by cutting the ignition to several cyclinder.
Rat-tat-poof-rat-tat-BANG-poof.

I thought they cut the ignition to all cylinders. Not so!

It depended on the model and brand of aircraft engine.

Here's some more info for those aspiring to be fighter pilots:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
8 October 2004, 03:07 PM

By Tripehound

Rotary throttle

Not a throttle like in your car, but a way of increasing or
decreasing the amount of fuel mixture fed to the engine, in three
separate steps.

There was an engine air intake pipe that used something like a ball
valve to regulate the amount of air going to the engine. This was
generally operated via a linkage from a handle in a "throttle
quadrant" located on the left side of the cockpit. The amount of
fuel going to this intake pipe was regulated by a "fine fuel" valve
from the pressurized fuel tank, or from a gravity tank, to a jet
inside the ball valve, where the incoming fuel was atomized. The air
ball valve acted like a Venturi tube, supplying suction to help move
the fuel. This "fine fuel" valve was operated by a separate handle
located in or near the "throttle quadrant". The Germans seemed to
favor locating the "fine fuel" control on the control column and
operating the valve by a twist knob via a Bowden(?) cable. Castor
oil was carried in a separate tank and pumped from this tank thru a
sight flow regulator in the cockpit to the fuel mixture feed line as
it entered the crankcase.

Both air and fuel had to be increased or decreased in steps to avoid
leaning or flooding, either one resulting in a dead engine. As the
aircraft climbed and the air became less dense, the fuel had to be
cut back to avoid flooding and a dead engine. As the aircraft
descended the fuel had to be increased to avoid leaning out and a
dead engine. Oil regulation didn't seem to be quite as demanding of
immediate attention as the air/fuel mix, but still necessary.

The coup button, or "blip" switch, allowed quick changes in engine
speed by cutting ignition to all cylinders. On landing, for example,
the engine was adjusted to take-off power then the coup button used
to cut the engine for brief intervals to decrease air speed for
landing. If the button was held too long the unburned fuel/oil mix
would flood the engine to where it wouldn't restart when the button
was let go, and/or worst case the expelled unburned fuel mix would
catch fire in the cowling.

Wooden airplanes and iron men!

Gregory -

Have a look at this thread from a few years back.

When I did parachuting in Belgium they still used the word "coupe" to
tell the pilot to cut power. Of course, him being the bush pilot type he
often didn't and then you found out that air does have some mass ...
jump ... *THWACK*

http://www.theaerodrome.com/forum/aircraft/2302-throttle-
down.html?highlight=throttle

(It's three pages - too long to post but quite interesting.)

Tripehound's explanation is very good. I would only add that some
makes and models of rotaries had different methods of control in
addition to the mixture adjustments and coup button. The 160hp
Gnome, for example, could be set to fire all 9, 7, 5, or just 3
cylinders on each revolution. The cylinders that were not firing
still cycled, but had no spark. They filled with fuel/oil and then
expelled it, unburnt, into the cowl. Messy and primitive, but it
worked. Mostly.

Yes, and here is a nice link with a picture of the Gnome engine:
http://www.nationalmuseum.af.mil/factsheets/factsheet.asp?id=830

What he probably meant with "Mostly" is that when the blipping wasn't
done right you could land with a trail of fire and black smoke. The more
"sophisticated" engines would blip alternatively on various cylinders,
maybe to reduce the chance of the big bang followed by a fire.

[...]

 

[legg]

Hi

I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.

The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.

Two questions:

1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.

2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)

Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.

No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.

The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.

'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?

RL

 

[Klaus Kragelund]

Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.

No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.

The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.

'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?

RL- Skjul tekst i anførselstegn -

- Vis tekst i anførselstegn -

Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.

If we assume the input ac is rectified to a DC link voltage and then
bucked down by a converter to 24V (primary conv), then normally the
duty-cycle would be minimum 8% and that should be no problem.

The half-bridge converter runs off this 24V voltage and bucks down to
5V (secondary conv). In principle I could adjust the 24V voltage up to
increase the ratio of the secondary conv and ease the losses of the
primary conv.

With respect to the primary buck, I found the Viper100a which seems to
be able to be used in a buck configuration. Alternatively it could be
done by standard PWM ic, high-side driver and a N-channel MOSFET -
which could be all second sourced components.

Thanks

Klaus

 

[Joerg]

Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.

100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be.

[...]

 

[legg]

Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.

If we assume the input ac is rectified to a DC link voltage and then
bucked down by a converter to 24V (primary conv), then normally the
duty-cycle would be minimum 8% and that should be no problem.

The half-bridge converter runs off this 24V voltage and bucks down to
5V (secondary conv). In principle I could adjust the 24V voltage up to
increase the ratio of the secondary conv and ease the losses of the
primary conv.

With respect to the primary buck, I found the Viper100a which seems to
be able to be used in a buck configuration. Alternatively it could be
done by standard PWM ic, high-side driver and a N-channel MOSFET -
which could be all second sourced components.

Thanks

Klaus

Why does the input range go down to 16V?

RL

 

[Klaus Kragelund]

Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.

100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be.

[...]

Actually the 100ms is because the control guys cannot shut down their
stuff fast enough, so I am force to live with it.

I did actually think about the snip-off FET, but then that would be
have to be rated for the current at 16V and voltage rated for 700V, so
I could equally well just use a big FET in the buck converter. Another
idea was to use the fact that the input is ac, so could just turn on
the big FET for each half-cycle when the voltage passes 24V - to
charge the DC link. But I think the losses will be very high.

Thanks

Klaus

 

[Klaus Kragelund]

Why does the input range go down to 16V?

RL- Skjul tekst i anførselstegn -

- Vis tekst i anførselstegn -

Well, because we have both 16VDC and a 230Vac input and we use the
same terminals (16V is from a battery)

Thanks

Klaus

 

[Joerg]

On Thu, 20 Nov 2008 06:58:51 -0800 (PST), Klaus Kragelund
Hi
I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.
The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.
Two questions:
1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.
2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)
Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.
No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.
The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.
'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?
RL- Skjul tekst i anførselstegn -
- Vis tekst i anførselstegn -
Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.

100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be.

[...]

Actually the 100ms is because the control guys cannot shut down their
stuff fast enough, so I am force to live with it.

I did actually think about the snip-off FET, but then that would be
have to be rated for the current at 16V and voltage rated for 700V, so
I could equally well just use a big FET in the buck converter. Another
idea was to use the fact that the input is ac, so could just turn on
the big FET for each half-cycle when the voltage passes 24V - to
charge the DC link. But I think the losses will be very high.

The losses don't have to be high especially since it's probably 50Hz or
60Hz AC. The main issue in the AC case are the capacitors. Having to
have all those at the highest expected voltage rating is going put a
real dent into the budget because anything past 350V or maybe 400V will
be large, expensive and not liked by the purchasing folks. I guess in
the end you'll just have to price all this out.

 

[Joerg]

Hey, I understood that!

Well, because we have both 16VDC and a 230Vac input and we use the
same terminals (16V is from a battery)

Other idea: Why not have two switchers? One that runs off the 230VAC
rail and another that runs off the battery rail. Assuming the 230VAC is
desired to take precedence you could then run the 16V switcher on idle
and it only takes over when the 230VAC rail fails to deliver. You could
even completely disable the 230VAC switcher in case that dreaded spike
comes along and let the 16V take over for those 100msec.

 

[JosephKK]

Hi
I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.
The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.
Two questions:
1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.
2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)
Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.

No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.

The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.

'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?

RL- Skjul tekst i anførselstegn -

- Vis tekst i anførselstegn -

Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.

100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be.

[...]

Actually the 100ms is because the control guys cannot shut down their
stuff fast enough, so I am force to live with it.

I did actually think about the snip-off FET, but then that would be
have to be rated for the current at 16V and voltage rated for 700V, so
I could equally well just use a big FET in the buck converter. Another
idea was to use the fact that the input is ac, so could just turn on
the big FET for each half-cycle when the voltage passes 24V - to
charge the DC link. But I think the losses will be very high.

Thanks

Klaus

I was thinking along those lines. How bout a zener and power
transistor (emitter follower) skimming device in front of your
switcher.



--------+-------------+
.-. |
| | |
| | |
'-' |
| |/
+---------o-|
| |>
| |
| +----------------
|
|
|
o
z
A
|
------------o------------------------------
(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

 

[Joerg]

Klaus Kragelund wrote:
On Thu, 20 Nov 2008 06:58:51 -0800 (PST), Klaus Kragelund
Hi
I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.
The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.
Two questions:
1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.
2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)
Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.
No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.
The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.
'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?
RL- Skjul tekst i anførselstegn -
- Vis tekst i anførselstegn -
Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.
100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be.

[...]

Actually the 100ms is because the control guys cannot shut down their
stuff fast enough, so I am force to live with it.

I did actually think about the snip-off FET, but then that would be
have to be rated for the current at 16V and voltage rated for 700V, so
I could equally well just use a big FET in the buck converter. Another
idea was to use the fact that the input is ac, so could just turn on
the big FET for each half-cycle when the voltage passes 24V - to
charge the DC link. But I think the losses will be very high.

Thanks

Klaus

I was thinking along those lines. How bout a zener and power
transistor (emitter follower) skimming device in front of your
switcher.



--------+-------------+
.-. |
| | |
| | |
'-' |
| |/
+---------o-|
| |>
| |
| +----------------
|
|
|
o
z
A
|
------------o------------------------------
(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

Better to really switch off. Else, if an overvoltage condition persists
and it's not a big old fat transistor on a heat sink ... phssst ... *POOF*

 

[Klaus Kragelund]

KlausKragelund wrote:
On Thu, 20 Nov 2008 06:58:51 -0800 (PST),KlausKragelund
Hi
I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.
The supply needs to be double isolated and my first thought was to use
a HV buck converter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.
Two questions:
1. Do you know an integrated IC than can buck 25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.
2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)
Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.
No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.
The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.
'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?
RL- Skjul tekst i anførselstegn -
- Vis tekst i anførselstegn -
Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.
100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be.
[...]

Actually the 100ms is because the control guys cannot shut down their
stuff fast enough, so I am force to live with it.

I did actually think about the snip-off FET, but then that would be
have to be rated for the current at 16V and voltage rated for 700V, so
I could equally well just use a big FET in the buck converter. Another
idea was to use the fact that the input is ac, so could just turn on
the big FET for each half-cycle when the voltage passes 24V - to
charge the DC link. But I think the losses will be very high.

The losses don't have to be high especially since it's probably 50Hz or
60Hz AC. The main issue in the AC case are the capacitors. Having to
have all those at the highest expected voltage rating is going put a
real dent into the budget because anything past 350V or maybe 400V will
be large, expensive and not liked by the purchasing folks. I guess in
the end you'll just have to price all this out.

--

I will try to do a simulation of that. Regarding the voltage rating of
the capacitors, I think I can use ones rated for the 230Vac and simply
regard the overvoltage situation as a transient (as it indeed is).
Even better simply only turn the FET on if the voltage on the ac cycle
is say 24V, so this way the caps can be rated only for something like
35VDC and then much less volume.

When the input is DC, then the FET should be on always, so the FET
would just turn on at any voltage below 35VDC to protect the caps and
to function in either ac and DC case (the ac and DC are never present
at the same time and use the same terminals)


Thanks

Klaus

 

[Klaus Kragelund]

Hey, I understood that!



Other idea: Why not have two switchers? One that runs off the 230VAC
rail and another that runs off the battery rail. Assuming the 230VAC is
desired to take precedence you could then run the 16V switcher on idle
and it only takes over when the 230VAC rail fails to deliver. You could
even completely disable the 230VAC switcher in case that dreaded spike
comes along and let the 16V take over for those 100msec.

- Skjul tekst i anførselstegn -

- Vis tekst i anførselstegn -

Infact as I mention in another post just now written the ac and DC use
the same terminals to avoid production mistakes

Regards

Klaus

 

[Klaus Kragelund]

KlausKragelundwrote:

KlausKragelund wrote:
On 26 Nov., 18:06, Joerg <[email protected]>
wrote:
KlausKragelund wrote:
On Thu, 20 Nov 2008 06:58:51 -0800 (PST),KlausKragelund
Hi
I need to design a wide input range converter (16VDC to 600VDC) with
two 5V outputs with a total power of 25W.
The supply needs to be double isolated and my first thought was to use
a HVbuckconverter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.
Two questions:
1. Do you know an integrated IC than canbuck25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.
2. What about converting directly from the input to 5V by using the
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will be quite
different)
Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.
No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.
The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.
'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?
RL- Skjul tekst i anførselstegn -
- Vis tekst i anførselstegn -
Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.
100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be.
[...]
Actually the 100ms is because the control guys cannot shut down their
stuff fast enough, so I am force to live with it.
I did actually think about the snip-off FET, but then that would be
have to be rated for the current at 16V and voltage rated for 700V, so
I could equally well just use a big FET in thebuckconverter. Another
idea was to use the fact that the input is ac, so could just turn on
the big FET for each half-cycle when the voltage passes 24V - to
charge the DC link. But I think the losses will be very high.
The losses don't have to be high especially since it's probably 50Hz or
60Hz AC. The main issue in the AC case are the capacitors. Having to
have all those at the highest expected voltage rating is going put a
real dent into the budget because anything past 350V or maybe 400V will
be large, expensive and not liked by the purchasing folks. I guess in
the end you'll just have to price all this out.
--

I will try to do a simulation of that. Regarding the voltage rating of
the capacitors, I think I can use ones rated for the 230Vac and simply
regard the overvoltage situation as a transient (as it indeed is).
Even better simply only turn the FET on if the voltage on the ac cycle
is say 24V, so this way the caps can be rated only for something like
35VDC and then much less volume.

When the input is DC, then the FET should be on always, so the FET
would just turn on at any voltage below 35VDC to protect the caps and
to function in either ac and DC case (the ac and DC are never present
at the same time and use the same terminals)

Yes, that could work. Especially since you mentioned that both voltage
options come in via the same rail. Of course, in the AC case that will
really mess up the power factor but it might not be of much concern for
a 25W supply.

Yep, normally the power factor is only looked upon when the power from
the product is above 50W (75W in some special cases).

Moreover, to reduce the peak current, the caps should be the minimum
value possible to allow for more "on-time" during the ac cycle.
Minimum cap size is the specified hold-time of the SMPS

Thanks

Klaus

 

[Klaus Kragelund]

KlausKragelundwrote:

KlausKragelundwrote:
On 27 Nov., 21:32, Joerg <[email protected]>
wrote:
KlausKragelund wrote:
On 26 Nov., 18:06, Joerg <[email protected]>
wrote:
KlausKragelund wrote:
On Thu, 20 Nov 2008 06:58:51 -0800 (PST),KlausKragelund
Hi
I need to design a wide input range converter (16VDC to 600VDC)with
two 5V outputs with a total power of 25W.
The supply needs to be double isolated and my first thought wasto use
a HVbuckconverter to step the voltage to 16V and then a forward
converter (half-bridge) to step down to 5V and provide safety
isolation.
Two questions:
1. Do you know an integrated IC than canbuck25W without being forced
to resort to a UC28xx and level shifting/current sensing (should
preferable be current mode controller). Something like the Onsemi
NCP1010, but just higher currents.
2. What about converting directly from the input to 5V by usingthe
forward transformer. It will be quite low duty-cycle, so does anyone
have experience with this? (also the transformer design will bequite
different)
Suggest you reconsider your source's input configuration, even if it
means using two sets of fixed alternative input hardware, or
switched/folded conversion topologies.
No single source would seem to have the arbitrary 40:1 voltage
compliance you've specified, without exhibiting loading
characteristics that would normally be sensibly exploited to reduce
circuit complexity, even if it meant burning a continuous minimum
load.
The high end you've specified implies multipliers of 1.6 for creepage,
clearance and insulation distances and thicknesses, when compared to
240VAC circuitry. Safe limiting and fusing over the range also
approaches the impractical.
'Low duty cycle' could imply high pulsed power, for a 25W continuous
rating. Intermittent operational requirements, on the other hand,
could either nix simpler circuitry or allow liberties to be taken with
thermal ratings. What do you actually mean?
RL- Skjul tekst i anførselstegn -
- Vis tekst i anførselstegn -
Well, the 600VDC comes from an ac overvoltage event on the input
(normally this voltage is 230Vac/310VDC). This overvoltage event is
only present for 100ms maximum, so this will not stress the components
over the lifetime, but anyway, 100ms is an eternity in transient
thermal response time terms.
100msec? Interesting, that sounds just like an aircraft spec. Anyhow, if
100msec is abs max and it is guaranteed that those surges won't appear
machine-gun style you could snip those off with a simple pass transistor
that has enough thermal mass for it. One that is always closed except
when the voltage exceeds 350V or so, whatever a safe spec might be..
[...]
Actually the 100ms is because the control guys cannot shut down their
stuff fast enough, so I am force to live with it.
I did actually think about the snip-off FET, but then that would be
have to be rated for the current at 16V and voltage rated for 700V,so
I could equally well just use a big FET in thebuckconverter. Another
idea was to use the fact that the input is ac, so could just turn on
the big FET for each half-cycle when the voltage passes 24V - to
charge the DC link. But I think the losses will be very high.
The losses don't have to be high especially since it's probably 50Hzor
60Hz AC. The main issue in the AC case are the capacitors. Having to
have all those at the highest expected voltage rating is going put a
real dent into the budget because anything past 350V or maybe 400V will
be large, expensive and not liked by the purchasing folks. I guess in
the end you'll just have to price all this out.
--
I will try to do a simulation of that. Regarding the voltage rating of
the capacitors, I think I can use ones rated for the 230Vac and simply
regard the overvoltage situation as a transient (as it indeed is).
Even better simply only turn the FET on if the voltage on the ac cycle
is say 24V, so this way the caps can be rated only for something like
35VDC and then much less volume.
When the input is DC, then the FET should be on always, so the FET
would just turn on at any voltage below 35VDC to protect the caps and
to function in either ac and DC case (the ac and DC are never present
at the same time and use the same terminals)
Yes, that could work. Especially since you mentioned that both voltage
options come in via the same rail. Of course, in the AC case that will
really mess up the power factor but it might not be of much concern for
a 25W supply.

Yep, normally the power factor is only looked upon when the power from
the product is above 50W (75W in some special cases).

Yeah, but you know how the guys in Brussels operate. Pretty soon they
might institute PFC requirements for bicycle lamps ;-)

Yep, on a simelar subject it seems they will entirely remove the quasi-
peak detector and use only peak detector for emission measurements -
so all dirty tricks like spreadspectrum modulation techniques is no
use in the future

That could possibly be done by slowly charging a larger electrolytic via
a resistor and providing a diode across that resistor. So it won't mess
with the PFC but can still tide you over the specified number of cycles.

Very nice idea indeed

Regards

Klaus

 

[legg]

Well, because we have both 16VDC and a 230Vac input and we use the
same terminals (16V is from a battery)

Perhaps you'd have a more realistic product if you split the design
function between two converters in the same style of package - perhaps
the only internal difference would be the front end electronics. This
is a fairly common method of dealing with extended voltage ranges and
it makes sense.

Although it would be nice to claim such a wide compliance, it will
mean multiple input circuits that rely on the integrity of
housekeeping monitors or inrush limiters to connect them suitably.
This a complication and a cost that is probably not currently
justified by commercial demand, even at this relatively low power
level.

Customers damanding the performance, without also being prepared to
shoulder the financial impact, are best avoided. Designers claiming to
have beaten both should feel free to market the results, without
disclosure on usenet and without any advice from me.

RL