Maker Pro
Maker Pro

Design idea for low voltage SMPS

S

Steve

Jan 1, 1970
0
Well, looking for ideas would be more accurate as I've currently run out of
them!

I need to design a small, single cell, battery powered inverter to produce a
very approximate (within say 10%) constant 2A into a resistive load. The
load may vary such that the cct needs to deliver from about 1.5 to 2.5v. The
inverter final output could happily be AC as the load is resistive.

The particular difficulties I see are that:-
1. The supply is a single rechargeable cell of only a nominal 1.2v which
rules out many of the usual IC SMPS control devices.
2. The unit should need no user power switch and only consume a maximum
of perhaps a few 100uA when the load is disconnected.

Given the low current accuracy requirement, I've been thinking along the
lines of a self oscillating discrete design with a periodic trigger kick that
works when the load is connected. I'm hoping for some suggestions on
obtaining the constant current characteristic. I know I could just produce a
classic DC output design with I feedback, but was hoping to avoid the extra
complexity (cost) and losses (battery life) involved in the rectification.

I've been perusing my old discrete design handbooks for ideas and have drawn
a blank, rather like my brain at present.

I'm currently enquiring whether it would be possible to make the supply a 2
cell battery, as I already have a design that could be modified to be
suitable.
If that avenue fails, have any of you come across anything like this, or have
possible circuit configurations to consider?
 
J

Joerg

Jan 1, 1970
0
Hello Steve,
1. The supply is a single rechargeable cell of only a nominal 1.2v which
rules out many of the usual IC SMPS control devices.
That is right. Your idea to use a blocking oscillator approach is a good
one. Can't use FETs either since there isn't much to turn on the gates
hard enough so it will most likely have to be bipolar transistors

LTC has some PWM controllers that are claimed to work down to 1V. Check
their web site to see if any can be used with external devices large
enough to handle your current requirement.
2. The unit should need no user power switch and only consume a maximum
of perhaps a few 100uA when the load is disconnected.
There is the challenge. First you'll have to detect a load which can be
done simply via a load sensing resistor and a transistor. That can
trigger the converter. Afterwards this resistor needs to be bridged in
order not to waste much power.

Then you have to detect when the load current drops below a given limit
and that's not so easy. You don't want to drop 700mV somewhere on a load
sense resistor so it will have to be done in the millivolt range. Here
you'd have to look at devices made for hearing aid applications. Don't
expect a large variety since hearing aids are nearly all done with ASICs
these days.

Also, whatever you use, if it is for mass production you need to make
sure there are enough of other markets for the parts you design in. Else
they may be obsoleted and then you would be back to square one. You may
be better off with a discrete design, possibly using transistor arrays
if board space is a concern. Also, you may have to chase for some
transistors with enough power and the highest possible beta at that power.
I've been perusing my old discrete design handbooks for ideas and have drawn
a blank, rather like my brain at present.
This is the kind of stuff that books and app notes won't go very deep
into. IOW, a real engineering challenge. I love those.
I'm currently enquiring whether it would be possible to make the supply a 2
cell battery, as I already have a design that could be modified to be
suitable.
Two cells opens the door to a whole fleet of PWM chips. But first I'd
try to figure out whether it can be done with one cell.

Regards, Joerg
 
I

Ian Stirling

Jan 1, 1970
0
Steve said:
Well, looking for ideas would be more accurate as I've currently run out of
them!

I need to design a small, single cell, battery powered inverter to produce a
very approximate (within say 10%) constant 2A into a resistive load. The
load may vary such that the cct needs to deliver from about 1.5 to 2.5v. The
inverter final output could happily be AC as the load is resistive.

The particular difficulties I see are that:-
1. The supply is a single rechargeable cell of only a nominal 1.2v which
rules out many of the usual IC SMPS control devices.

http://www.maxim-ic.com/

There are many small, nearly free devices to boost from 1V - 3/5V.
Howver, though many are "1A", when you read the specs closely,
they won't do that at low input voltages.
An external power FET or saturated bioplar switch would do the job.

If the volume isn't overly important, then a simple switch running
at 25-100Khz would probably be the easy way.
 
J

Joerg

Jan 1, 1970
0
Hi Ian,
There are many small, nearly free devices to boost from 1V - 3/5V.
Howver, though many are "1A", when you read the specs closely,
they won't do that at low input voltages.
An external power FET or saturated bioplar switch would do the job.
Most of them will indeed not deliver full current at low input voltages.
But there may be a chance to use several in parallel and combine the
outputs with Schottky diodes. It may not be as easy as it sounds to make
sure that such ganging will not render them unstable. Then also,
Schottkys will cause a hefty penalty in overall efficiency

If this is for a mass product my main concern would be long term
availability of the parts.

Regards, Joerg
 
W

Winfield Hill

Jan 1, 1970
0
Steve wrote...
I need to design a small, single cell, battery powered inverter to
produce a very approximate (within say 10%) constant 2A into a
resistive load. The load may vary such that the cct needs to
deliver from about 1.5 to 2.5v. The inverter final output could
happily be AC as the load is resistive.

The particular difficulties I see are that:-
1. The supply is a single rechargeable cell of only a nominal 1.2v
which rules out many of the usual IC SMPS control devices.
2. The unit should need no user power switch and only consume a
maximum of perhaps a few 100uA when the load is disconnected.

You can easily do this if you design a two-stage converter, as we
discussed here on s.e.d. a few months back, IIRC. The first stage
is a low-power low-voltage converter giving you 4 to 5V to start
or even run the second converter. The second stage operates off
the 5V to run a control circuit and drive the large MOSFET, etc.,
and is able to inexpensively and efficiently take the 0.9 to 1.2V
source, etc., and give you your desired output. For low outputs,
like 1.5 to 2.5V, you could add a second transformer winding that
makes a 5V output to take over the task of the first stage. The
advantage of this approach is that you can use low-cost highly-
efficient low-voltage SMPS control ICs that are normally used to
create 1.5 to 2.5V from 5V power, except with a 1.2V power input.

The thread used a self-oscillating circuit for the first stage.
 
P

Pooh Bear

Jan 1, 1970
0
Joerg said:
Hello Steve,

That is right. Your idea to use a blocking oscillator approach is a good
one. Can't use FETs either since there isn't much to turn on the gates
hard enough so it will most likely have to be bipolar transistors

How about a low power step up of the 1.2V for the control circuitry ?


Graham
 
J

Joerg

Jan 1, 1970
0
Hi Graham,
How about a low power step up of the 1.2V for the control circuitry ?
That's an option but it isn't that low in power. The gate drive for a
large FET at reasonable SMPS frequencies requires quite a bit. But it
might be an idea that works.

Regards, Joerg
 
M

Mook Johnson

Jan 1, 1970
0
Another thing to consider. What happens to your 1.2 volts when you pull 2 -
4 amps from it? (battery current with 1.8A @ 2.5V output)

You'll be working with more like 1 volt or lower depending on the battery
size.

I've seen someboost designs that bootstrap the power for the IC fron the
boosted output. might be worth a look.
 
S

Steve

Jan 1, 1970
0
You can easily do this if you design a two-stage converter, as we
discussed here on s.e.d. a few months back, IIRC.

Can you remember what sort of subject line it was Win, I'd like to browse
back?
I'll take a look at the two stage approach, thanks.

and On Tue, 02 Nov 2004 21:28:39 GMT, Joerg
That is right. Your idea to use a blocking oscillator approach is a good
one. Can't use FETs either since there isn't much to turn on the gates
hard enough so it will most likely have to be bipolar transistors

LTC has some PWM controllers that are claimed to work down to 1V. Check
their web site to see if any can be used with external devices large
enough to handle your current requirement.

Thanks, I'll take a look at those. Actually I prefer a discrete design for
this for reasons of cost and parts availability.
There is the challenge. First you'll have to detect a load which can be
done simply via a load sensing resistor and a transistor. That can
trigger the converter. Afterwards this resistor needs to be bridged in
order not to waste much power.

Then you have to detect when the load current drops below a given limit
and that's not so easy. You don't want to drop 700mV somewhere on a load
sense resistor so it will have to be done in the millivolt range. Here
you'd have to look at devices made for hearing aid applications. Don't
expect a large variety since hearing aids are nearly all done with ASICs
these days.

Also, whatever you use, if it is for mass production you need to make
sure there are enough of other markets for the parts you design in. Else
they may be obsoleted and then you would be back to square one.

You've got that right! I've been bitten by that before with many components,
such as MCUs and specialist display controllers. The problem is often one of
availability rather than obsolescence.

I've been toying with a self oscillating design with a FET and an
autotransformer, with an overwinding for gate drive. Since I needed no
output components imparting a load, it does indeed stop oscillating when the
load is removed. I can re-trigger it with a simple periodic oscillator for
low consumption when unloaded. I haven't yet dreamt up a satisfactory
constant current characteristic though.

It IS an interesting exercise though, and does prove I can use a low loss FET
rather than a saturated NPN. I keep remembering the constant current
requirement only needing to hold to 10% and wondering if I can achieve it
within some magnetics without high gain active feedback. The requirement is
for a low cost design probably to be manufactured in the far east.

Mook:
Yes it's a valid concern regarding supply volts drop. The cell is a 8A/hr
unit capable of delivering large currents. I estimate it needs to operate
down to around 1.1v or just under, under max load.

Meanwhile, thanks everyone for the responses so far.
 
J

Joerg

Jan 1, 1970
0
Hi Steve,
Thanks, I'll take a look at those. Actually I prefer a discrete design for
this for reasons of cost and parts availability.
That's the way to go, especially when it is manufactured abroad. For an
even better deal it pays to obtain a preferred parts list with pricing
from your prospective manufacturer.
I've been toying with a self oscillating design with a FET and an
autotransformer, with an overwinding for gate drive. Since I needed no
output components imparting a load, it does indeed stop oscillating when the
load is removed. I can re-trigger it with a simple periodic oscillator for
low consumption when unloaded. I haven't yet dreamt up a satisfactory
constant current characteristic though.

It IS an interesting exercise though, and does prove I can use a low loss FET
rather than a saturated NPN. I keep remembering the constant current
requirement only needing to hold to 10% and wondering if I can achieve it
within some magnetics without high gain active feedback. The requirement is
for a low cost design probably to be manufactured in the far east.
You can achieve proper limits and range via the properties of the
magnetics but this is playing with fire. The materials tolerance needs
to be very tight. Also, you'd have to make absolutely sure that the
contract manufacturer won't change horses some day because they could
get a better deal from another ferrite manufacturer. How often did you
hear from a purchaser "oh, but the rep assured us that this is a 100%
second source"? There would have to be a very tight ECO control on the
transformer, or the whole unit for that matter.

It would be better if you could provide some kind of feedback to
regulate the output current. If you can post a schematic we could shoot
off ideas.

The two-step approach that Winfield and Graham suggested can get you out
of this bind but it would mean a cost increase. So maybe the first thing
would be to see if your existing idea could be furnished with feedback
at a lower cost.
Mook:
Yes it's a valid concern regarding supply volts drop. The cell is a 8A/hr
unit capable of delivering large currents. I estimate it needs to operate
down to around 1.1v or just under, under max load.
That would already be close to a 10% drop, even more reason for a
feedback strategy.

Regards, Joerg
 
S

Steve

Jan 1, 1970
0
The two-step approach that Winfield and Graham suggested can get you out
of this bind but it would mean a cost increase. So maybe the first thing
would be to see if your existing idea could be furnished with feedback
at a lower cost.
I've just finished prototyping a two stage design along the lines Win &
Graham suggested and while somewhat unfinished it satisfies the main criteria
apart from standing current. I think I can improve that, but it's still a
bit more complex than I had intended. However having talked to the customer
again this afternoon I have alarm bells ringing in my head. Something to do
with their apparent continued inability to find someone to sign agreement on
my specification and extent of supply that they've had for over a week.
Sometimes you just get a feeling... So despite their pressing urgency I've
pleasantly let them know that work is stalled - at close to a first draft!
;-)

So thanks again folks for your suggestions - they gave me a kick start,
 
Top