Searching \ for '[EE] Tips for designing a high current DC-DC conve' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/power.htm?key=current
Search entire site for: 'Tips for designing a high current DC-DC conve'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Tips for designing a high current DC-DC conve'
2008\04\16@192550 by Matthew Miller

flavicon
face
Hello,

I am not as familiar as I want to be on this topic and I'm looking for
pointers. ;) My goal is to modify an electric cart to be charged by solar
panels mounted on the cart's roof. The cart charging circuitry requires a 48
volt input and the maximum output of the solar panels (in series) is 34V and
7A.

What I need then is a converter that can handle about a max of 5A output at
48 volts. In my research, I'm drawn to a boost type converter since I need a
constant and higher voltage output than the input voltage. A boost converter
is also attractive because of its simplicity. Designing a flyback type
converter would be more difficult for me, but I want something that works
well!

Can anyone give me pro/con about a boost topology converter for this
application?

Thanks!  Matthew

2008\04\16@193534 by Xiaofan Chen

face picon face
On Thu, Apr 17, 2008 at 7:25 AM, Matthew Miller <spam_OUTnamiller2TakeThisOuTspamnaxs.net> wrote:
> I am not as familiar as I want to be on this topic and I'm looking for
> pointers. ;) My goal is to modify an electric cart to be charged by solar
> panels mounted on the cart's roof. The cart charging circuitry requires a 48
> volt input and the maximum output of the solar panels (in series) is 34V and
> 7A.
>
> What I need then is a converter that can handle about a max of 5A output at
> 48 volts.

This is not achievable since the efficiency will need to be over 100%.

Max input power: 34V x 7A = 238W
Max output power: 48V x 5A = 240W.

Xiaofan

2008\04\16@195240 by Richard Prosser

picon face
Matthew,
I can only really think of the pros.
1. Simple design - simple drive requirements.
2. High conversion efficiency. (we get over 96% when converting AC
mains to 420V DC at about 70kHz and >1kw and this includes losses in
the rectifier diodes)
3 Ability to perform a simple optomisation to maximise power output.
(i.e adjust the pwm ratio slightly to maximise charge current. Wobble
the ratio around slightly to ensure you are operating at the peak
point for the existing conditions.)
4. Can be controlled by a simple micro - eg PIC, etc.
5. You don't need an additional diode to prevent backfeed under low
light conditions, the boost diode will perform this function.

cons.
1 Maybe slightly more expensive than the simplest switched capacitor
or self oscillating push-pull transformer  system. But a switched cap
doubler will be really struggling at 5Amps and a self oscillating
push-pull oscillator will not have the same amount of control
available. (unless you make it more complicated & push the cost up
anyway).

2. You don't get isolation between the input and output side, but you
don't need it anyway.

I'm sure others will add additional thoughts.

I guess this is a hobby thing - not a product for sale?  If it's a
saleable product, then you may need to look at RFI, etc issues which
will complicate things.. If you get above 60V then electrical safety
comes into it also but it looks like you'll be well below that.

RP


On 17/04/2008, Matthew Miller <.....namiller2KILLspamspam@spam@naxs.net> wrote:
{Quote hidden}

> --

2008\04\16@202230 by Matthew Miller

flavicon
face
Hi Richard,

On Thu, Apr 17, 2008 at 11:52:31AM +1200, Richard Prosser wrote:

> 5. You don't need an additional diode to prevent backfeed under low
> light conditions, the boost diode will perform this function.

Oh, I hadn't thought of this! Thanks for the idea.

> 2. You don't get isolation between the input and output side, but you
> don't need it anyway.

I don't think isolation should be necessary here...

> I guess this is a hobby thing - not a product for sale?  If it's a
> saleable product, then you may need to look at RFI, etc issues which
> will complicate things.. If you get above 60V then electrical safety
> comes into it also but it looks like you'll be well below that.

It's not for sale. I'm teaching a class this summer on alt. energy and
sustainability for high school kids and part of the class is to modify the
golf cart. The kids are putting everything together, I've just got to design
the electronics. :)

Thanks for your input! Matthew

2008\04\16@202821 by Matthew Miller

flavicon
face
On Thu, Apr 17, 2008 at 07:35:12AM +0800, Xiaofan Chen wrote:
> On Thu, Apr 17, 2008 at 7:25 AM, Matthew Miller <namiller2spamKILLspamnaxs.net> wrote:
> > I am not as familiar as I want to be on this topic and I'm looking for
> > pointers. ;) My goal is to modify an electric cart to be charged by solar
> > panels mounted on the cart's roof. The cart charging circuitry requires a 48
> > volt input and the maximum output of the solar panels (in series) is 34V and
> > 7A.
> >
> > What I need then is a converter that can handle about a max of 5A output at
> > 48 volts.
>
> This is not achievable since the efficiency will need to be over 100%.
>
> Max input power: 34V x 7A = 238W
> Max output power: 48V x 5A = 240W.

Hi Xiaofan,

You're right. I was just using round about numbers. Why type 4.958A when I
can type 5A? Come on! :) Any way, having a bit of head room can sometimes
prevent problems.

Is the boost topology good for this kind of application?

Thanks. Matthew

2008\04\16@204828 by David VanHorn

picon face
>  Is the boost topology good for this kind of application?


Certainly!  In fact, you can probably find something off the shelf for
this, since a lot of telco stuff is nominally 48V.

2008\04\16@205818 by Jinx

face picon face
> This is not achievable since the efficiency will need to be over 100%.
>
> Max input power: 34V x 7A = 238W
> Max output power: 48V x 5A = 240W

Although the sun is shining a lot more of the time than the cart will be
in use (probably)

I'm no expert with SMPS so can't help with an actual circuit for your
numbers but many of the ICs I looked at (MC34063, LM25xx and
others) for a project had app notes for external switching elements
(bipolars and FETs) for high-current outputs. A basic circuit for the
LM2576 can deliver 3A without additional parts, so it's not a huge
leap to 5A

2008\04\16@210656 by Apptech

face
flavicon
face
{Quote hidden}

If this is a one off then buy versus build needs to be
considered.
That's possibly "buy and modify" but you may still find that
it's cheaper and easier to do that.

Assuming build from scratch -

This is in the capabilities of a single switch simple boost
converter, although commercial solutions at this level may
often use some form of push pull topology.

You say the charging circuit needs 48V. Is that a 48V
battery or a lower voltage battery with a charger that needs
a higher voltage than the battery. That is an important
question as it can severely affect the solution. If you want
something that works with an existing 'black box' then
optimum efficiency may not be the aim. if you want best
energy use then more details of the whole system would be
useful.

eg
xxx solar panel with N cells.
Vmax = 34V (68 cells?)(unusual)(measured?)
I max = 7A (measured)(short circuit?)(name plate?)...

Assuming that the 34V is Voc in full sun then Voptimum
(Vmpp) will be about 34 * (0.75 to 0.8) = 25 to 27V

Assuming a 48V battery = 4 x 12V the required charge voltage
will be more like 56 V.

So step up will be say 25 to 48 ~= 2:1

A boost converter with the inductor "standing" on the 25V
rail will need to produce another 25V or so.
Efficiencies approaching 95% should be possible with care.

On the input side power in (first cut) = say 25*7 = 175
Watts
Question: What resistance in the primary side will cause 1%
losses?

1% = 1.75W say 2W
For 2W at 7A R = P/I^2 = 2/49 ~= 0.04 ohm.

So per 1% primary loss you can have 40 milliohm for wiring,
inductor, MOSFET (hot) and more.

Secondary (boost mode losses) involve ~= 3.%A the wiring and
7A pulses in the inductor, diode and 50V reservoir capacitor

Assuming Schottky diode and say 0.5V drop you get 7A x 0.5V
x 50% = 1.75W say 2W losses or another 1%.

If you are aiming at 95% overall and allow ~= 50% in input
and out put sides you are allowed a total input equivalent
resistance of about 2.5(%) x 40 milliohm = 100 milliohm.

If the FET is half of this that's 50 milliohm. A FET with a
real world 50 milliohm Rdson will need a spec sheet one of
under 25 milliohm say 0.02 ohm as the spec sheets almost
always quote pulsed values at very low duty cycles.

FET rating needs to be say 20A, 50V (abs min, more better),
<= 20 mohm Rdson.
The IRFP064 is a good start. TO247 package.
$6 Digikey
Claimed 0.008 Rdson
55V 110A
A bit more voltage would be advised!

IXFH110N10P is a bit better and a bit dearer
http://ixdev.ixys.com/DataSheet/99212.pdf


search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=IRFP064NPBF-ND
http://www.irf.com/product-info/datasheets/data/irfp064npbf.pdf


I'll stop there.
MUCH more can be said but lets see if we are going in the
right direction for you.

This and friends could be useful
Note that Schottky diodes really really don't like over
voltage - rate conservatively.

       http://www.vishay.com/docs/88941/v30100s.pdf



       Russell McMahon

2008\04\16@211150 by Apptech

face
flavicon
face
>>  Is the boost topology good for this kind of application?

> Certainly!  In fact, you can probably find something off
> the shelf for
> this, since a lot of telco stuff is nominally 48V.

Indeed. Although the PV panel's variable input voltage will
reduce your options.

if you want maximum power point tracking for peak energy use
you will need further control.


       Russell

2008\04\16@215756 by gardenyu

picon face
part 1 1666 bytes content-type:text/plain; charset="gb2312" (decoded quoted-printable)

I can't believe a high school will require such a converter for a simple class.
Just by some commercial one, take the label off and tell them you made yourself.
 Not joking this time, I think some IC manufacturere should have some chip for boost converter, no transformer is necessary here. you don't really need to program anything sometime.



> Date: Wed, 16 Apr 2008 20:08:12 -0400> From: .....namiller2KILLspamspam.....naxs.net> To: EraseMEpiclistspam_OUTspamTakeThisOuTmit.edu> Subject: Re: [EE] Tips for designing a high current DC-DC converter> > Hi Richard,> > On Thu, Apr 17, 2008 at 11:52:31AM +1200, Richard Prosser wrote:> > > 5. You don't need an additional diode to prevent backfeed under low> > light conditions, the boost diode will perform this function.> > Oh, I hadn't thought of this! Thanks for the idea.> > > 2. You don't get isolation between the input and output side, but you> > don't need it anyway.> > I don't think isolation should be necessary here...> > > I guess this is a hobby thing - not a product for sale? If it's a> > saleable product, then you may need to look at RFI, etc issues which> > will complicate things.. If you get above 60V then electrical safety> > comes into it also but it looks like you'll be well below that.> > It's not for sale. I'm teaching a class this summer on alt. energy and> sustainability for high school kids and part of
 the class is to modify the> golf cart. The kids are putting everything together, I've just got to design> the electronics. :)> > Thanks for your input! Matthew>

2008\04\16@230754 by Apptech

face
flavicon
face
> It's not for sale. I'm teaching a class this summer on
> alt. energy and
> sustainability for high school kids and part of the class
> is to modify the
> golf cart. The kids are putting everything together, I've
> just got to design
> the electronics. :)

That both tightens and loosens the spec.

Efficiency and minimum cost are not as important as
bullet-proofness.

Many major IC manufacturers provide ICs that are well suited
to this task and have application notes that cover it well.
Linear Technology are a good place to start. It's not an
overly hard task to do "well enough" and if you rate the
components conservatively it should work quite well. With a
single winding inductor you don't get the leakage inductance
woes that beset some fancier moderate power designs.

If you wanted absolute minimum spec a (blush) 555 set at ~=
50% duty cycle driving a FET with a simple gate driver and
suitable inductor and output diode would do a "good enough"
[tm] job. But, I'd recommend a dedicated IC driving a FET
similar to the one I mentioned.

If a bit less efficiency is OK an eg IRFP250 in TO247
package would do well enough for about $1.50 from Digikey.

Inductor can be a commercial one with the right specs or a
self wound air cored one (RFI, what RFI?) of about 100 uH
(or see app note for IC you are using) and fat enough wire
to get resistance down to a loss level that is acceptable.



       Russell



2008\04\17@010514 by Xiaofan Chen

face picon face
On 4/17/08, Matthew Miller <namiller2spamspam_OUTnaxs.net> wrote:
> > This is not achievable since the efficiency will need to be over 100%.
> >
> > Max input power: 34V x 7A = 238W
> > Max output power: 48V x 5A = 240W.
>
> You're right. I was just using round about numbers. Why type 4.958A when I
> can type 5A? Come on! :)

You can not expect 100% efficiency, not even 99%. Without
some serious design efforts, it is not so easy to achieve 90%
efficiency in this case.

> Any way, having a bit of head room can sometimes
> prevent problems.
>
> Is the boost topology good for this kind of application?
>

Yes I will think so. For this one-off non-commercial
design, leave plenty of headroom for all the power
stage components (MOSFET, inductor and capacitors)
and assume only 70% efficiency.

A dedicated MOSFET driver and a PWM controller
are recommended. Microchip has some good
MOSFET drivers (from the Telcom acquisition).
I have not used any of Microchip's PWM controller.

Xiaofan

2008\04\17@011006 by Xiaofan Chen

face picon face
On 4/17/08, Matthew Miller <@spam@namiller2KILLspamspamnaxs.net> wrote:
> > I guess this is a hobby thing - not a product for sale?  If it's a
> > saleable product, then you may need to look at RFI, etc issues which
> > will complicate things.. If you get above 60V then electrical safety
> > comes into it also but it looks like you'll be well below that.
>
> It's not for sale. I'm teaching a class this summer on alt. energy and
> sustainability for high school kids and part of the class is to modify the
> golf cart. The kids are putting everything together, I've just got to design
> the electronics. :)
>

Then you need to be very careful about safety. For example,
be sure to have some load. Sometime if the loop contol is not good,
the output can be very high voltage for boost converter if no loads
are connected.

I would not recommend to do this with high school kids.

Xiaofan

2008\04\17@035016 by Apptech

face
flavicon
face
Look at this page

       http://others.servebeer.com/misc/mc34063

And especially

       http://others.servebeer.com/misc/mc34063/boost_converter_mc34063.pdf

The latter is a copy of a page from Jinx's website of a
simple boost converter using a cheap and available mc34063
and a single FET.

As shown, with suiotable component changes the circuit would
do your task.
See the data sheets and app notes on the first page above
for details.

The 0R1 drops (AFAIR) 0.3V at target current so say 0.3/10A
= 0.03 ohm in your application = suitable piece of wire.

200 uH would probably change.
Core certainly would.
Micrometals website has a calculator program that allows
sizing a core but a self wound aircore one would do to
start.

You'd need a power supply below panel voltage for this. The
MC34063 is good to 40V AFAIR BUT the FET gate won't be.
A 12V supply with a zener or whatever is fine.

Use suitable Schottky for output (eg one I mentioned earlier
would do).

Joe's waveforms are much nicer than you usually get in
practice :-).

I can show you a cheap gate driver circuit if you want
better low drive - on drive is probably very fine.

At say 10 uS cycle (MC34063 is slow) and say 5 uS charge
time L =~~ V.t/I = 25.5/7 ~= 20 uH.

20 uH at 7A is not too hard to do on a largish ferrite
bobbin.
The FET shown there will work with more losses than the
other one I mentioned.
At 7A and 0.1 ohm Rdson you get 5 Watts x 50% duty cycle x
1/2 for waveform = say under 2 Watts.
The TO247 package needs a minor heatsink at that power.

As shown the cct runs in the 20W range. Note that it has
magical efficincy. Switching noise in meters can do funny
things and confuse you badly. Ask me how I know :-(.



           Russell







2008\04\17@042203 by Alan B. Pearce

face picon face
>It's not for sale. I'm teaching a class this summer on alt. energy and
>sustainability for high school kids and part of the class is to modify
>the golf cart. The kids are putting everything together, I've just got
>to design the electronics. :)

Gee, why didn't they projects like that when I was around ...;)

A boost converter is what you need, I would suggest you investigate the
Linear technology website. The LT1071/LT1171 look like they may do what you
want, although you may need a couple running in parallel to get the current
level you want. The do also have an app note on using a multiphase converter
for very high power, which may be worth investigating.

http://www.linear.com/pc/categoryProducts.jsp?navId=H0,C1,C1003,C1042,C1031&sort=1&direction=desc&action=SORT


2008\04\17@052825 by Apptech

face
flavicon
face
> A boost converter is what you need, I would suggest you
> investigate the
> Linear technology website. The LT1071/LT1171 look like
> they may do what you
> want, although you may need a couple running in parallel
> to get the current
> level you want. The do also have an app note on using a
> multiphase converter
> for very high power, which may be worth investigating.
>
> http://www.linear.com/pc/categoryProducts.jsp?navId=H0,C1,C1003,C1042,C1031&sort=1&direction=desc&action=SORT

The LT1071 is rated at only 2.5A
The LT1070 is rated at 5A.

Datasheet for either.

   http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1042,C1031,C1061,P1266,D2406

Some good application circuits in there.

Power rating is about 100W at 25 Vin.
Unusually high Vinmax of 60V makes it potentially suited to
this job.

Note that the LT1070 is an ollllllllllllllldie but a goodie
BUT the oldness means it is slow (as is the MC34063).
100 kHz.
Main result is a physically larger inductor is needed than
if the more recent gee whizz chips are used.



2008\04\17@073815 by Alan B. Pearce

face picon face
>The LT1071 is rated at only 2.5A

Which is why I suggested he might need multiples in parallel.

>The LT1070 is rated at 5A.

Didn't pick that one on my very quick scan of the page, but it certainly
looks like it might do as a single device.

>Note that the LT1070 is an ollllllllllllllldie but a goodie
>BUT the oldness means it is slow (as is the MC34063).
>100 kHz.
>Main result is a physically larger inductor is needed than
>if the more recent gee whizz chips are used.

However another advantage of Linear Technology is that they will provide
free samples, no questions asked, just like Microchip. IIRC you can get 2
samples of a device, max 3 devices in one shipment per month. So it is
possible to hedge ones bets by filling the shipment with similar devices,
and have enough on hand if the experimenters manage to blow 1 or 2 up.

And as you say, they do also have some excellent app notes. The other trick
I like about their datasheets is the table of similar application devices on
the last page, so if the device for which you have the datasheet doesn't
quite do what you want, then there is a fair chance the table will point you
at one that will. It can be a bit like some of the interesting time wasting
web sites Russell points us at, following the datasheet trail.

2008\04\17@091552 by Matthew Miller

flavicon
face
Hi David,

On Wed, Apr 16, 2008 at 08:48:27PM -0400, David VanHorn wrote:
> >  Is the boost topology good for this kind of application?
>
> Certainly!  In fact, you can probably find something off the shelf for
> this, since a lot of telco stuff is nominally 48V.

More positive input! I have a textbook that I've been reading, plus lots of
websites, and it's good to know that I'm going in sort-of the right
direction.

Do you have any search terms that I could use to find this off the shelf
equipment?

Thanks.  Matthew

2008\04\17@134200 by James Salisbury

flavicon
face
Hi all,

It is not as simple as you may think to get the maximum power out of a
solar cell. A solar cell is a current, not voltage source, couple it up
to a battery load and you will sit low down the power curve, increase
the impedance of the load slightly and the terminal voltage will rise
but the current will be the same, more power.
http://en.wikipedia.org/wiki/Maximum_power_point_tracker

2008\04\17@184046 by Matthew Miller

flavicon
face
Russell, thanks for your lengthly comments!

On Thu, Apr 17, 2008 at 01:06:08PM +1200, Apptech wrote:

{Quote hidden}

There is an existing charging circuit in the cart. I simply want the
switcher to supply the 48V that the plug-in charger would provide.

In another message you mentioned FET driver circuitry. Could you say more
about about that?

Thanks.  Matthew

2008\04\17@211017 by Apptech

face
flavicon
face
> It is not as simple as you may think to get the maximum
> power out of a
> solar cell.

It's as easy as I think :-).
ie not totally easy.

But a good approximation can be obtained by holding the
panel at around 75% - 80% of its open circuit voltage for a
given degree of solar insolation (aka "sunshine").

There are many patents on systems claiming to improve power
point tracking, but the basic principles were established
long enough ago that they are patent free.


       Russell


> http://en.wikipedia.org/wiki/Maximum_power_point_tracker

2008\04\22@082901 by Alan B. Pearce

face picon face
Those wishing to get involved with designing DC-DC converters may like to
peruse this article from a recent edition of EDN.

http://www.edn.com/contents/images/6544742.pdf

2008\04\22@123901 by Vasile Surducan

face picon face
On 4/17/08, Apptech <KILLspamapptechKILLspamspamparadise.net.nz> wrote:
> > It is not as simple as you may think to get the maximum
> > power out of a
> > solar cell.
>
> It's as easy as I think :-).

It's as easy as you believe it is.

It looks like a problem of matching the load impedance with the solar
cell output impedance. Being a real current source it doesn't have
infinite internal resistance, so it also has a finite voltage swing.
Two identical solar cells may have two different internal resistance
for the same load voltage and the same light conditions. Which one is
better, that one  with lower resitance or that one with higher
resistance ?

http://www.maxim-ic.com/appnotes.cfm/an_pk/1029

"This input capacitor also minimizes supply-voltage fluctuations by
lowering the solar cell's output impedance."

Have really doubts about the sentence above.

2008\04\22@201017 by Apptech

face
flavicon
face
>> > It is not as simple as you may think to get the maximum
>> > power out of a
>> > solar cell.

>> It's as easy as I think :-).

> It's as easy as you believe it is.

(ie harder than it may seem).

{Quote hidden}

The basic trick is to maximise a target variable by targeted
search. "Targeting" is where the patent fun is at. The time
honoured technique walks a hopefully single peak hill to
find the summit and then wanders to and fro around that
point to see if the summit has moved. It uses the slope of
the hill either side to determine where the summit is. If
the slope is the same bothe sides of where you are the
summit has moved and you walk uphill until you (hopefully)
find the new location. Target vraiables can include charge
current for a battery load and energy transfer rate for
almost any load.

If the summit is multi peaked it can have problems. In the
case of solar energy the driver TENDS to be single variable
(even if the cells age or differe from their friends). In
the case of other systems eg wave or wind there may be
multiple driving variables. (Incidentally these are the same
problems which "natural selection's" search algorithn must
face but in that case the hill is utterly guaranteed to have
multiple independent peaks and the interaction of drivers is
so complex as to bury the effect of small variables in the
noise leta lone the effect of other variables. Where you
have a large effect or an external selection mechanism (eg
in the latter case a breeder looking for curly eye lashes)
then 'natural selection' works OK.). In the case of solar,
fortunately, such issues largely don't occur.

BUT

The circuit suggested is interesting and useful


> http://www.maxim-ic.com/appnotes.cfm/an_pk/1029

But essentially can be reduced in functional description to:
Converter IC1 is capable of starting at very low voltage and
is used to power IC2 which is capable of a higher energy
level.

> "This input capacitor also minimizes supply-voltage
> fluctuations by
> lowering the solar cell's output impedance."
>
> Have really doubts about the sentence above.

Maybe "Having a decoupling / reservoir capacitor on the
input supply was found to make the circuit work better" :-)


       R


More... (looser matching)
- Last day of these posts
- In 2008 , 2009 only
- Today
- New search...