Switching transistor question

[default]

The short idea:

What I want to do is switch ~1/2 amp with no more than .2 volts drop
(would be happier with .15) . . . all this on a 3 volt supply,
burning no more than about 150 milliwatts, and with no moving parts.

Any other ways to skin this animal?



The application:
I'm building a time-lapse camera using an el-cheapo 1.3 megapixel
electronic camera.

The camera is happy with about 2.2 volts at ~500 ma when fully turned
on - providing the source impedance is low and is designed for a pair
of AAA cells.

Plan A was to power the whole thing with a pair of AA batteries. To
that end, I got it working beautifully, but had to use a relay to
switch the camera on. The relay only sucks down 40 milliamps and is
only on for a short period so it is a viable way to do it.

Ideally I would want a semiconductor switching the camera on and off
if that's possible, but it isn't looking so easy with only 3 volts or
less to work with.

The sequence is to apply power to the camera. Wait 5 seconds or so
for it to initialize its processor, toggle the shutter low for a brief
period then leave the camera on another 9 seconds so it has time to
store the picture in its flash memory, then the camera has power
removed and the processor goes to sleep for ten minutes and the cycle
repeats.

So far the only semiconductor that will switch the ~ 500 ma the camera
needs when on has been a Darlington pair - but the CE voltage is too
high for the camera to power up reliably with only a 3 volt supply.

To bias a small NPN transistor on and into saturation with 500 ma in
the collector would be great but I haven't found a transistor with
enough gain at that current.

Plan B is to use a supply of 4.5 volts - but that is essentially
wasting one whole cell just to satisfy the voltage drop across the
Darlington.

Is there some way of doing this without the penalty of going to an
extra battery cell or driving it with more than 40 milliamps?

 

[mpm]

FET.
An IRF7331 specs at 0.045 Ohms @ Vgs= 2.5V. Good for 5.6A

There is a Zetek part which is even more appropriate, but I don't have
the part number handy.
We have a similar circuit that uses the above with no problem.

 

[John Popelish]

The short idea:

What I want to do is switch ~1/2 amp with no more than .2 volts drop
(would be happier with .15) . . . all this on a 3 volt supply,
burning no more than about 150 milliwatts, and with no moving parts.

Any other ways to skin this animal?

The application:
I'm building a time-lapse camera using an el-cheapo 1.3 megapixel
electronic camera.

The camera is happy with about 2.2 volts at ~500 ma when fully turned
on - providing the source impedance is low and is designed for a pair
of AAA cells.

Take a look at this high current, high gain, low saturation
voltage transistor:
http://www.zetex.com/3.0/pdf/ZTX1047A.pdf

It will drop about .05 volts with 500 mA collector current
and 2 mA base current.

 

[Eeyore]

To bias a small NPN transistor on and into saturation with 500 ma in
the collector would be great but I haven't found a transistor with
enough gain at that current.

Use 2 transistors. It doesn't have to be a darlington connnction. Just use the first
one to create enough base current for the second, so your 'control circuit' need only
supply a milliamp or so.

Graham

 

[default]

Use 2 transistors. It doesn't have to be a darlington connnction. Just use the first
one to create enough base current for the second, so your 'control circuit' need only
supply a milliamp or so.

Graham

That's what I've been trying to do. With a 2N3904 the gain is too
low, when pulling significant collector current - takes ~25 ohm
resistor to turn it on, or more current than my relay uses (125 ohms).
With two transistors, I'm not overloading the controller, but the
efficiency is still poor due to the heavy base current in the
switching transistor.

There are a few mosfets that may work, not too many in a TO92 package.

The best solution may be to build a one transistor blocking oscillator
to step up the control voltage enough to use an inexpensive common
power mosfet.

Think high efficiency - want the battery to last weeks or use solar
cells.

 

[John Popelish]

That's what I've been trying to do. With a 2N3904 the gain is too
low, when pulling significant collector current - takes ~25 ohm
resistor to turn it on, or more current than my relay uses (125 ohms).
With two transistors, I'm not overloading the controller, but the
efficiency is still poor due to the heavy base current in the
switching transistor.

There are a few mosfets that may work, not too many in a TO92 package.

The best solution may be to build a one transistor blocking oscillator
to step up the control voltage enough to use an inexpensive common
power mosfet.

Think high efficiency - want the battery to last weeks or use solar
cells.

Or buy a transistor made for that application, like the
ZTX1047A.

 

[default]

Or buy a transistor made for that application, like the
ZTX1047A.

Of course. Thank you. I did peruse that family of parts after you told
me about them. Sent for a Mouser catalog too they have several
suitable mosfets.

I have a good selection of parts, so my first thought is to try to use
what I already have - I'm funding this project. But if I build
several it may be better to buy something . Mouser has some Supertex
mosfets that should work - $.89 each in a TO92 package.

 

[[email protected]]

The short idea:

What I want to do is switch ~1/2 amp with no more than .2 volts drop
(would be happier with .15) . . . all this on a 3 volt supply,
burning no more than about 150 milliwatts, and with no moving parts.

Any other ways to skin this animal?

The application:
I'm building a time-lapse camera using an el-cheapo 1.3 megapixel
electronic camera.

The camera is happy with about 2.2 volts at ~500 ma when fully turned
on - providing the source impedance is low and is designed for a pair
of AAA cells.

Plan A was to power the whole thing with a pair of AA batteries. To
that end, I got it working beautifully, but had to use a relay to
switch the camera on. The relay only sucks down 40 milliamps and is
only on for a short period so it is a viable way to do it.

Ideally I would want a semiconductor switching the camera on and off
if that's possible, but it isn't looking so easy with only 3 volts or
less to work with.

The sequence is to apply power to the camera. Wait 5 seconds or so
for it to initialize its processor, toggle the shutter low for a brief
period then leave the camera on another 9 seconds so it has time to
store the picture in its flash memory, then the camera has power
removed and the processor goes to sleep for ten minutes and the cycle
repeats.

So far the only semiconductor that will switch the ~ 500 ma the camera
needs when on has been a Darlington pair - but the CE voltage is too
high for the camera to power up reliably with only a 3 volt supply.

To bias a small NPN transistor on and into saturation with 500 ma in
the collector would be great but I haven't found a transistor with
enough gain at that current.

Plan B is to use a supply of 4.5 volts - but that is essentially
wasting one whole cell just to satisfy the voltage drop across the
Darlington.

Is there some way of doing this without the penalty of going to an
extra battery cell or driving it with more than 40 milliamps?

I'd go for a FET.

You may want to drive the gate with a slowly rising signal rather than
just whacking it. There is always a possibility that the device being
fed by the fet will latch up IF it is normally turned on with a switch
on the device itself. In a sense, you are turning on the device in a
manner the designer didn't expect.

Typically a chip road test will catch such problems, but not every
manufacturer tests their chips in such a manner.

 

[default]

On Sun, 01 Jul 2007 05:45:14 -0000, [email protected] wrote:

Snip

I'd go for a FET.

You may want to drive the gate with a slowly rising signal rather than
just whacking it. There is always a possibility that the device being
fed by the fet will latch up IF it is normally turned on with a switch
on the device itself. In a sense, you are turning on the device in a
manner the designer didn't expect.

Typically a chip road test will catch such problems, but not every
manufacturer tests their chips in such a manner.

Yes, a mosfet seems tailor made. Most of the ones I have handy
require about 4.5 volts to turn on - which was what was stopping me
from using them. Way lower drop than any bipolar so it is the obvious
winner. But the ones I found were $2.50 each / $9 shipping. I paid
$9 for the cameras I'm using. One of my criteria is to keep the cost
low because there's a good chance I won't recover all the cameras I
deploy.

The Subconscious Engineering Department must have been working on this
problem overnight, because this morning I have a slick elegant
solution that doesn't require buying any parts. The picaxe 08M has a
PWM output function programmed into it, actually has two. I can use
that to toggle a voltage doubler with a few small caps and diodes and
switch the FET on. Eureka!

I'm using only two, out of five, I/O pins. So there's still hope for
using a servo to pan the camera and passive IR to trigger action.
(something I may want to add once I get one or two doing what I want).

I am taking advantage of the fact that the internal processor for the
camera is turning the camera on when batteries are inserted - That way
I can over-ride the 40+ second automatic power down the camera has and
only keep it on for 15 seconds.

As for the idea of ramping up the power to the processor inside the
camera (that what you were saying?) that seems like a bad idea. For
a time - the supply would be sitting on the cusp of good/bad power.
Anyhow it appears as if the camera designers already thought of that -
they monitor battery charge and if the power comes on with a low
battery the camera doesn't come on but just alerts you that the
battery is low. There appears to be about a half second or so when
this is going on during power up - screen takes awhile to initialize
and you can't snap a picture until it is done.

Another thing that makes me skeptical - I used a few surplus remote
controls to provide a signal for break beam infrared detectors -
cheaper and easier than building the signal source - the ones I was
using definitely didn't like coming up slowly - they would latch up.
(I had some idea to provide power during momentary outages via a
"super cap" which, when it charged, kept the supply from coming on
quickly and that caused a latch up)

 

[Fred Bloggs]

...- the ones I was
using definitely didn't like coming up slowly - they would latch up.

That is true but it was not "latch-up." Almost every imaginable power
reset architecture malfunctions with slow voltage turn-on. But these
maximums are on the order of 15ms...
The 500mA load is about the magnitude where you may have to start
thinking about exceeding the VGS threshold by a safe margin, otherwise
the RGS,ON may be much higher than you think. I wouldn't use a 4.5V
threshold in a doubler app powered off 3V.

 

[[email protected]]

On Sun, 01 Jul 2007 05:45:14 -0000, [email protected] wrote:

Snip






Yes, a mosfet seems tailor made. Most of the ones I have handy
require about 4.5 volts to turn on - which was what was stopping me
from using them. Way lower drop than any bipolar so it is the obvious
winner. But the ones I found were $2.50 each / $9 shipping. I paid
$9 for the cameras I'm using. One of my criteria is to keep the cost
low because there's a good chance I won't recover all the cameras I
deploy.

The Subconscious Engineering Department must have been working on this
problem overnight, because this morning I have a slick elegant
solution that doesn't require buying any parts. The picaxe 08M has a
PWM output function programmed into it, actually has two. I can use
that to toggle a voltage doubler with a few small caps and diodes and
switch the FET on. Eureka!

I'm using only two, out of five, I/O pins. So there's still hope for
using a servo to pan the camera and passive IR to trigger action.
(something I may want to add once I get one or two doing what I want).

I am taking advantage of the fact that the internal processor for the
camera is turning the camera on when batteries are inserted - That way
I can over-ride the 40+ second automatic power down the camera has and
only keep it on for 15 seconds.

As for the idea of ramping up the power to the processor inside the
camera (that what you were saying?) that seems like a bad idea. For
a time - the supply would be sitting on the cusp of good/bad power.
Anyhow it appears as if the camera designers already thought of that -
they monitor battery charge and if the power comes on with a low
battery the camera doesn't come on but just alerts you that the
battery is low. There appears to be about a half second or so when
this is going on during power up - screen takes awhile to initialize
and you can't snap a picture until it is done.

Another thing that makes me skeptical - I used a few surplus remote
controls to provide a signal for break beam infrared detectors -
cheaper and easier than building the signal source - the ones I was
using definitely didn't like coming up slowly - they would latch up.
(I had some idea to provide power during momentary outages via a
"super cap" which, when it charged, kept the supply from coming on
quickly and that caused a latch up)

Latch-up is more likely from a fast rising voltage. I'm thinking of a
few hundred microseconds, not miliseconds. Yes, part of the road test
for chips is a slow turn on, i.e. a few hundred miliseconds. You
really should be able to apply DC at any voltage and no have the chip
latch or behave strangely. The problem with your design is that it is
probably full of cheap ass Chinese parts, so the parts themselves may
not be very robust.

The power fet turn on is a real life problem. Some designers don't
trust the shutdown pins n chips, and like to waste money by using
series P-fets because it is something they control. ]In all fairness,
perhaps not every chip in the system has a shutdown pin, so they
decide to just power off entire sections of circuitry with the pass
fet.

Latch-up testing, generally only done on the initial design, is done
with the power supply applied. That is, the chip is powered and you
push/pull the pins other than supply and ground. Thus most QA
procedures don't test powering up the chip via a pass fet.

 

[neon]

why don't you get a 2n2907 pnp to switch 500ma in the reverse direction then the sat is in the mv. you must use force beta to saturation. as you know once conducting current flow bothway.

 

[default]

That is true but it was not "latch-up." Almost every imaginable power
reset architecture malfunctions with slow voltage turn-on. But these
maximums are on the order of 15ms...

The 500mA load is about the magnitude where you may have to start
thinking about exceeding the VGS threshold by a safe margin, otherwise
the RGS,ON may be much higher than you think. I wouldn't use a 4.5V
threshold in a doubler app powered off 3V.

Tell me more (I'm opinionated, obstinate, stupid (at times)
bullheaded, intractable, etc.) but this is something I haven't
encountered.

And would 6 volts for what is supposed to be a 4.5 (max) turn on good
enough? (for a 30 amp mosfet pulling less than an amp?)

 

[default]

Latch-up is more likely from a fast rising voltage. I'm thinking of a
few hundred microseconds, not miliseconds. Yes, part of the road test
for chips is a slow turn on, i.e. a few hundred miliseconds. You
really should be able to apply DC at any voltage and no have the chip
latch or behave strangely. The problem with your design is that it is
probably full of cheap ass Chinese parts, so the parts themselves may
not be very robust.

Micro seconds? I'm thinking PWM output toggling a voltage doubler
with a filter cap and bleeder resistor. I worry that it will be too
slow to turn on.

Yes it is full of cheap ass Chinese parts - I wouldn't be trying this
with expensive parts.

The power fet turn on is a real life problem. Some designers don't
trust the shutdown pins n chips, and like to waste money by using
series P-fets because it is something they control. ]In all fairness,
perhaps not every chip in the system has a shutdown pin, so they
decide to just power off entire sections of circuitry with the pass
fet.

My picaxe, if I understand it correctly must be totem pole outputs.
My camera trigger is a .1 uf ceramic with no bleeder (just trying to
eliminate some reverse current so went with a cap instead of
cap/bleeder or open collector). come to think of it, you make it high
it sources and make it low it sinks.

I suspect you are talking about some more proprietary knowledge from
working or designing these cheap Chinese camera things - I'm just
tinkering and trying to build something to satisfy my curiosity,
learn, and just maybe wind up with something with market value. I do
know its been fun so far.

I do want to build the best device I can within financial limits.

Latch-up testing, generally only done on the initial design, is done
with the power supply applied. That is, the chip is powered and you
push/pull the pins other than supply and ground. Thus most QA
procedures don't test powering up the chip via a pass fet.

Now you seem to be arguing against slow power up.

The camera comes up 100% of the time when you slide the batteries in
and close the contacts on the batteries - one of those pull out and
hinges away battery compartment with the series connection for the
batteries on the sliding portion.

This thing is probably relatively immune to contact bounce or garbage
power. My mosfet is bound to be better than sliding metal contacts,
wouldn't you agree?

Right now I can barely lift my arms from kayaking - the camera and
breadboard were taped to the dash on my truck to watch the boat ramp -
maybe tomorrow I'll fool with driving mosfets with the pwm output.
Now it is time for homebrew beer and "zoning out."

 

[LVMarc]

Take a look at this high current, high gain, low saturation voltage
transistor:
http://www.zetex.com/3.0/pdf/ZTX1047A.pdf

It will drop about .05 volts with 500 mA collector current and 2 mA base
current.

Amazing perfroamcne! JP thanks for the link and insight inot this part!
I did not know they beat vsat down sooo low!

Marc Popek

 

[neon]

to switch 3 v with mv ofsrt use a transistor into the reverse mode the beta is horrible but the switch is in the mv as opposed to .2 or higher, force beta of ten you get your MV saturation . try 2n2907 for positive clamps.