Searching \ for 'Constant current switcher?' in subject line. ()
Help us get a faster server
FAQ page: www.piclist.com/techref/power/priswitch.htm?key=switcher
Search entire site for: 'Constant current switcher?'.

Truncated match.
'Constant current switcher?'
2007\03\10@001652 by

I have studied and have a pretty good idea about how switching power
supplies work. However, what I have seen is designed to create a regulated
voltage output.

What I need is the ability to have a regulated current output. In this case
about 800ma. The source voltage is 170DC (half bridge from 120VAC) and the
expected voltage across the load is between 2 and 5 volts (or preferably up
to 10 volts).

Note: I fully understand the concerns of operating directly off the AC line
without any isolation. This circuit will be fully isolated from anything
else, and in a sealed, earth-grounded chassis, so I am not two worried about
that.

Important parameters:

Relatively efficient - can't deal with too much heat.

Needs to work from -10F to 150F

Needs to work from input = 140VDC to 200VDC with some ripple

Needs to be pretty small (maybe 1" x 1" x 1" ? ?)

Accuracy is not all the important.

Temperature coefficient is not that important, output could vary be 10% over
temperature.

Supply reject is not that important, again 10% over range.

Dynamic resistance not that important. Current could vary as much as 10%
over a load voltage of  2V to 5V.

Thanks,

Bob Ammerman

>
>
> What I need is the ability to have a regulated current output. In this
> case
> about 800ma. The source voltage is 170DC (half bridge from 120VAC) and the
> expected voltage across the load is between 2 and 5 volts (or preferably
> up
> to 10 volts).
>

Sounds like you need a buck switcher to regulate the voltage developed
across a current sense resistor.
10 volts @ 800 mA is 8 watts so your physical size requirement can probably be met.

The usual way to get regulated current is to use a traditional voltage regulated
output design,
but have the feedback voltage be derived from a sense resistor in the load path,
instead of the output. You then use an op-amp to provide the gain needed to
get the current sense signal up to reference level of the traditional voltage
feedback network.

e.g. put a 0.1 ohm sense resistor into the ground leg of your load.
You get 80 mV signal at required current. For 1.25 V reference typical for converters,
needed gain is 15.625. (1.25V/80mv). Your op-amp should be able to have ground as
an output (rail to rail) but as long as it can get below the reference, and can
accept grounds as input, you should be ok.

You'll have as much compliance voltage as your supply can provide (170V in this case)
so you may want a 'soft start' circuit and some sort of clamping to prevent the
full 170V from being output. (e.g. Max on time/duty cycle limit,
zener to output to provide overriding feedback once voltage gets to
some 'limit' level. e.g. 15 V.

I would suggest that if this is a "few off" application that you look at using
switcher wall warts used for cell phone or lap top chargers, and just hack the
feedback loop as described above (op-amp).

Robert

Robert Ammerman wrote:

{Quote hidden}

On 3/9/07, Robert Ammerman <rammermanverizon.net> wrote:
> I have studied and have a pretty good idea about how switching power
> supplies work. However, what I have seen is designed to create a regulated
> voltage output.
>
> What I need is the ability to have a regulated current output. In this case
> about 800ma. The source voltage is 170DC (half bridge from 120VAC) and the
> expected voltage across the load is between 2 and 5 volts (or preferably up
> to 10 volts).

5V and 10V at 800mA is not the same.  Keeping dissipation small on
the current regulator is the key. Definitely I'll vote for an 170VDC
to 6VDC converter followed by a constant current generator. Power
dissipation will be maximum (6V-2V)*0.8=3.2W and minimum
(6V-5V)*0.8A=0.8W. More power at 10V load voltage swing.

At 1 inch cubic dimension is quite difficult to evacuate the heat, so
probably will require external heatsink.
You seem to be assuming that the current regulator has to be linear. That's
not the case at all - straightforward to have a switching current regulator,
in which case you don't have a problem with heat. Given 80% efficiency
(which isn't hard to achieve) your maximum power dissipation in the
regulator is only 2W when the output's 10V and 800mA - much simpler than a 2
stage process.

On 3/10/07, Vasile Surducan <piclist9gmail.com> wrote:
>
> 5V and 10V at 800mA is not the same.  Keeping dissipation small on
> the current regulator is the key. Definitely I'll vote for an 170VDC
> to 6VDC converter followed by a constant current generator. Power
> dissipation will be maximum (6V-2V)*0.8=3.2W and minimum
> (6V-5V)*0.8A=0.8W. More power at 10V load voltage swing.
>
I'm agree. However depends how clean must be the output. If you need
less than say 5mV ripple, then I have doubts the single stage will be
ok (at least I've never solved such small ripple in one single stage).

On 3/10/07, Chris McSweeny <cpmcsweenygmail.com> wrote:
{Quote hidden}

> -
On Fri, 2007-03-09 at 23:51 -0500, Robert Ammerman wrote:
> I have studied and have a pretty good idea about how switching power
> supplies work. However, what I have seen is designed to create a regulated
> voltage output.
>
> What I need is the ability to have a regulated current output. In this case
> about 800ma. The source voltage is 170DC (half bridge from 120VAC) and the
> expected voltage across the load is between 2 and 5 volts (or preferably up
> to 10 volts).

It's pretty much the exact same thing as the "voltage output" switchers
you've seen, the only difference being instead of the feedback being
related to the voltage of the output (often through a voltage divider)
you instead use the voltage across a current sense resistor as the
feedback.

Everything else will be very similar. Only places you have the be extra
careful of are ensuring your circuit behaves "OK" during extremes, i.e.
short circuit on the output, or open circuit on the output.

TTYL

>
>
> Everything else will be very similar. Only places you have the be extra
> careful of are ensuring your circuit behaves "OK" during extremes, i.e.
> short circuit on the output, or open circuit on the output.

Agreed, an open in the feedback would be most amusing.

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...