12V->60V @3-4A SMPS design

[Mook Johnson]

Have you considered push pull. No fancy high side gate drivers to deal
with. Full(two quadrant)flux swing on the core.

 

[Paul Mathews]

Doesn't current mode pretty much get rid of those problems?

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
[email protected] Info for manufacturers:http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com- Hide quoted text -

- Show quoted text -

Current mode feedback works as expected to equalize ONLY when the
feedback is linear. Various components in the feedback chain have
compliance limits. In short, there are many conditions in which your
feedback signal is limiting, and you can't count on it to drive the
power stage back into dc balance. One partial remedy is to use a
completely separate power supply to provide bias for the controller
chip itself. This approach eliminates most, but not all, of the
situations in which the flux-walking problem can occur. Some designers
include an air gap in the transformer to make it more tolerant. --Paul
Mathews

 

[Henry Kiefer]

Hmm.. nasty, nasty peak currents, >100A.

Hi Spehro -

Do you received my private email?

Regards -
Henry

 

[Fred Bloggs]

To make a battery charger circuit that will run from a 12VDC
regulated supply and produce (up to) 60V at 3-4A, what SMPS topology
would be recommended? I've gotten the suggestion of a flyback, but
240W seems a bit on the high side for that.

Input current should max out around 25A (80% efficiency) so I suppose
MOSFETs will be the way to go. Forward converter?

Anyone feel like putting together a real design? We need to make a
dozen or two of them, so it just has to work right the first time,
not necessarily be particularly cheap.

They're NiMH, I want control over the algorithm- thermistor as well
as voltage depression (and discharge) and prefereably the ability to
change battery technology with approximately the same voltage
(currently 33 cells and 16 cells per pack). I also have to deal with
possible charger-charger interaction. Lead acid is definitely not in
the cards, but one of the Li technologies might be, if it can be
proven safe.

Approximately 20 to 60V at 0.1 (?) to 4A, programmable by a micro.
Since it's charging batteries, current control is the ultimate goal.
A trickle charge capability is desirable.


No, within reason (the chargers can't interact to the point where
performance is affected). There is some commoning of one side of the
output lines, so a synchronized setup would be preferable so that
current and temperature could be measured without interaction. That
may be just a detail-- if it's designed in from the start.

The absolute simplest power control scheme would be hysteretic current
control of the the boost topology, which can be paralleled in constant
current mode. This brings up that 20-60V output at 0.1-4A output
current. You need to translate this to constant current mode
programmable from IL->IU with what compliance, and float charge constant
voltage programmable from VL->VU over what current range. Usually a
table or graph required here. That is the easy part. The hard part is
tying in all the charging algorithms and fault protection as a function
of battery chemistry. TI has that BQXXXX series of multi-chemistry
battery charging control IC's. You're going to need something like that
as either the finished IC or embedded/emulated in a PIC controller or
something. The finished circuit will be simple enough, but the design
effort will be non-trivial.

 

[Spehro Pefhany]

The absolute simplest power control scheme would be hysteretic current
control of the the boost topology, which can be paralleled in constant
current mode. This brings up that 20-60V output at 0.1-4A output
current. You need to translate this to constant current mode
programmable from IL->IU with what compliance, and float charge constant
voltage programmable from VL->VU over what current range.
Yes...

Usually a
table or graph required here.
That is the easy part. The hard part is
tying in all the charging algorithms and fault protection as a function
of battery chemistry. TI has that BQXXXX series of multi-chemistry
battery charging control IC's. You're going to need something like that
as either the finished IC or embedded/emulated in a PIC controller or
something.

I'm really not worried about the algorithm.. it's less than a page of
C to implement the most complex NiMH algorithm used by chip makers,
including recovery from over-discharge, which the present micro-based
"commercial" units don't properly implement. The battery manufacturers
also publish (somewhat incomplete) information which is useful for
charge termination parameters for a given cell.

Iout = f(Vcell(t), t, T(t), ncells)

The finished circuit will be simple enough, but the design
effort will be non-trivial.

Agreed. But not a crazy amount of work.


Best regards,
Spehro Pefhany

 

[Spehro Pefhany]

Hi Spehro -

Do you received my private email?

Regards -
Henry

Yes, Henry, thanks. I'm going to look at your German patent over
breakfast. ;-)


German Patent office-- Fusball & Technik?
http://www.patent-und-markenamt.de/infos/fussball/wm2006/index.html


Best regards,
Spehro Pefhany

 

[John Larkin]

I'm really not worried about the algorithm.. it's less than a page of
C to implement the most complex NiMH algorithm used by chip makers,
including recovery from over-discharge, which the present micro-based
"commercial" units don't properly implement. The battery manufacturers
also publish (somewhat incomplete) information which is useful for
charge termination parameters for a given cell.

Iout = f(Vcell(t), t, T(t), ncells)


Agreed. But not a crazy amount of work.


Best regards,
Spehro Pefhany

Go with the fashion, "digital power". Compute PWM of the boost
converter drive(s) based on measurements.

John

 

[Harry Dellamano]

To make a battery charger circuit that will run from a 12VDC regulated
supply and produce (up to) 60V at 3-4A, what SMPS topology would be
recommended? I've gotten the suggestion of a flyback, but 240W seems
a bit on the high side for that.

Input current should max out around 25A (80% efficiency) so I suppose
MOSFETs will be the way to go. Forward converter?

Anyone feel like putting together a real design? We need to make a
dozen or two of them, so it just has to work right the first time, not
necessarily be particularly cheap.

Best regards,
Spehro Pefhany

If you can't get OTS then the best topology is the Boost, similar to a
flyback but no custom transformer, the push pull and forward converters have
discontinuous input current and custom transformers. Any resonate topology
is not viable at low voltages. Your 25A input current is better handled with
two interleaved boost channels 180D out of phase. The UCC2820 will handle
all the drive functions at about 200KHz PWM.
http://focus.ti.com/lit/ds/symlink/ucc28220.pdf
If the input inductors are still two big, go to three or four channels at
120D each, the inductors would be OTS. This would cut your output storage
caps in half. It is always a good idea to use more silicon and spread the
heat. At 90% efficiency, that is 27 watts of power loss. This can be
increased to about 93% with synchronous rectifiers for the high side diodes.

YMMV
Harry

 

[Spehro Pefhany]

Go with the fashion, "digital power". Compute PWM of the boost
converter drive(s) based on measurements.

John

So long as a glitch doesn't cause smoke, I'd be happy to do so.

Given the voltage range, I'm not sure about the boost converter
topology though. I can't go much under 12V without running into
problems at the low end with a simple boost converter.
Best regards,
Spehro Pefhany

 

[Spehro Pefhany]

If you can't get OTS then the best topology is the Boost, similar to a
flyback but no custom transformer, the push pull and forward converters have
discontinuous input current and custom transformers. Any resonate topology
is not viable at low voltages. Your 25A input current is better handled with
two interleaved boost channels 180D out of phase. The UCC2820 will handle
all the drive functions at about 200KHz PWM.
http://focus.ti.com/lit/ds/symlink/ucc28220.pdf
If the input inductors are still two big, go to three or four channels at
120D each, the inductors would be OTS. This would cut your output storage
caps in half. It is always a good idea to use more silicon and spread the
heat. At 90% efficiency, that is 27 watts of power loss. This can be
increased to about 93% with synchronous rectifiers for the high side diodes.

YMMV
Harry

Thanks very much, Harry, we'll take a close look at doing just that.


Best regards,
Spehro Pefhany

 

[Henry Kiefer]

Yes, Henry, thanks. I'm going to look at your German patent over
breakfast. ;-)


German Patent office-- Fusball & Technik?
http://www.patent-und-markenamt.de/infos/fussball/wm2006/index.html

I don't know how you reached there

You can either use:
- http://depatisnet.dpma.de/DepatisNet/depatisnet?window=1
or better there:
- http://v3.espacenet.com/textdoc?DB=EPODOC&IDX=DE19616621&F=0

Maybe you get a session timeout. Start over.

From espacenet there is a auto-translation link. Translation looks good
for being automatic.
I let translated it there:
http://www.worldlingo.com/wl/epo/ep...iption&OPS=ops.espacenet.com&LOCALE=en_EP&T=1

Hope that helps.

cu -
Henry

 

[john jardine]

To make a battery charger circuit that will run from a 12VDC regulated
supply and produce (up to) 60V at 3-4A, what SMPS topology would be
recommended? I've gotten the suggestion of a flyback, but 240W seems
a bit on the high side for that.

Input current should max out around 25A (80% efficiency) so I suppose
MOSFETs will be the way to go. Forward converter?

Anyone feel like putting together a real design? We need to make a
dozen or two of them, so it just has to work right the first time, not
necessarily be particularly cheap.

Best regards,
Spehro Pefhany

http://www.speff.com

I did a 12V car battery, to 50V at 3Amps boost, for a stepper motor supply.
Uses an ETD44 with 10 turns of 6mm tri-rated. Single BUKxxx (can't read it)
with 2"x3" heatsink. PWM gate drive and seperate current sense resistor and
comparitor to kill the gate on any overcurrent. Took about an hour to design
and build.
Sadly, this was before I learned that these things are far too complicated
to be designed so casually and I lost confidence.
Still running after 6 years though.

 

[Clifford Heath]

I did a 12V car battery, to 50V at 3Amps boost

I need to do something casual like that sometime soon, to drive a
servo'd rcar rooftop antenna rotator... so if you could casually
answer a couple of questions I'd appreciate it.

Did you do a boost, flyback or some other topology?

On what is tri-rated wire actually rated?

Clifford Heath.

 

[john jardine]

I need to do something casual like that sometime soon, to drive a
servo'd rcar rooftop antenna rotator... so if you could casually
answer a couple of questions I'd appreciate it.

Did you do a boost, flyback or some other topology?

On what is tri-rated wire actually rated?

Clifford Heath.

[Sounds interesting project!]
'Tis a standard 'Boost' ie, Supply > Inductor > Fet > deck.
Best I remember it, is that "Tri-rated" is multistand wire with insulation
rated for a working temperature range up to 105 degC and has some approval
details printed along it's length. (as cable, nothing special).

The other week and suspicious of the size of some "10mm^2 tri rate", I miked
up the cores, counted 'em and did the sums. Turns out it was 7mm^2. Looks
like the world Copper shortage is having an effect.
Got to keep stuff casual as SMPS topics seem to intrude on a lot of
jealously protected turf and delicate egos. Hence It'd be tantamount to
iconoclasm to dare mention how straightforward my recent design was for a
customer's 1400W boost

 

[Clifford Heath]

'Tis a standard 'Boost' ie, Supply > Inductor > Fet > deck.
Best I remember it, is that "Tri-rated" is multistand wire with insulation

Thanks John. I'm not familiar with working at these power levels.
Though I think I understand most of the theory, I'm sure in practise
that learning layout skills will be interesting.

[Sounds interesting project!]

<http://www.melbfox.com.au/>. We go transmitter-hunting once a month.
A number of teams have power-rotated antennas with all kinds of DSP
hooked up to the RSSI, but when you get an intermittent transmitter
they often miss the pulse simply because they're pointing the wrong
way. Hand-rotated antennas work well, but need to be smaller in size.

My notion is to build a rotator that can hunt and circle near a manual
(or maybe even automatically detected) likely direction, potentially
synchronizing with a periodically-pulsing transmitter. Computer
control means it can be made not to wind up the cable (no more than
one turn each way), so no slip-ring coupling is needed, and compass
information from the GPS means it'll keep on target as the car turns
corners. I have a nice 18:1 worm-drive gearbox with a 1" output shaft,
and a powerful DC servomotor from an old HP 7978 tape drive - about
2Kg, driven by a linear amp from +-45V supplies! I'll use PWM not the
linear, but just need more than 12V.

When you get close, you jump out and run with a hand-held Yagi+sniffer.
One of the local guys builds this nice bit of kit:
<http://www.users.bigpond.net.au/vk3yng/foxhunt/2m_sniffer/manual.htm>.
This is the device that requires the dual-gate GaAs-FET I was asking
about recently, for its combination of sensitivity, low noise performance,
and huge dynamic range.

Clifford Heath.