How to power my circuit.

[Daniel Pitts]

Hello,

So far, all of my circuits have had the power supplied by an Arduino
(compatible) board. I'm going to try to change the setup here a little,
and design a circuit that has just an ATTiny in its place, but now I
have to think about power supply.

The circuit I'm envisioning is going to have three Texas Instruments
"TLC5916" constant-current sink's, powering one column of a multiplexed
8x8 RGB led matrix. The circuit will also have an ATTiny85 (or ATTiny84,
depending on a few things), and a 74HC238 (to select the row on the LED
matrix).

Now, if I'm reading the specs right, I think I can power the whole thing
on 3 AA batteries (4.5 volts). But if I wanted to have more robust power
handling, I'm not sure what I want to do. I'd like to try to keep the
circuit cheap, and potentially have it powered by a wall socket or USB.
I also don't have the time/money to etch my own PCB, so something that I
can plug into my protoboard is preferable.

Is it just me, or is deciding on a power supply a relatively difficult
problem, compared to other aspects of working on digital circuitry?

Thanks in advance for any suggestions!

--
Daniel.

 

[krw@att.bizzzzzzzzzzzz]

On Tue, 07 Aug 2012 17:39:32 -0700, Daniel Pitts
<newsgroup.nospam@virtualinfinity.net> wrote:

Hello,

So far, all of my circuits have had the power supplied by an Arduino
(compatible) board. I'm going to try to change the setup here a little,
and design a circuit that has just an ATTiny in its place, but now I
have to think about power supply.

The circuit I'm envisioning is going to have three Texas Instruments
"TLC5916" constant-current sink's, powering one column of a multiplexed
8x8 RGB led matrix. The circuit will also have an ATTiny85 (or ATTiny84,
depending on a few things), and a 74HC238 (to select the row on the LED
matrix).

Now, if I'm reading the specs right, I think I can power the whole thing
on 3 AA batteries (4.5 volts). But if I wanted to have more robust power
handling, I'm not sure what I want to do. I'd like to try to keep the
circuit cheap, and potentially have it powered by a wall socket or USB.
I also don't have the time/money to etch my own PCB, so something that I
can plug into my protoboard is preferable.

Is it just me, or is deciding on a power supply a relatively difficult
problem, compared to other aspects of working on digital circuitry?

Thanks in advance for any suggestions!

Just use a 5V wall wart. You can either get a mating connector for your board
or just cut the end off and solder the ends to your circuit. The ones with
switching regulators are probably the best bet because you'll be sure they're
regulated. These are generally specified as 120/240V. Linear or T/R (junk)
will be 120V only (in the US).

 

[Rich Webb]

On Tue, 07 Aug 2012 17:39:32 -0700, Daniel Pitts
<newsgroup.nospam@virtualinfinity.net> wrote:

Is it just me, or is deciding on a power supply a relatively difficult
problem, compared to other aspects of working on digital circuitry?

s/deciding on a power supply/analog design/ -- Yes, although it can be a
lot of fun.

If you don't need portability then krw's suggestion of a regulated wall
wart is spot on and by far the simplest. Do ensure that the output is
actually regulated, though; some are not more than a step-down
transformer with a diode bridge.

--
Rich Webb Norfolk, VA

 

[Les Cargill]

Rich Webb wrote:

On Tue, 07 Aug 2012 17:39:32 -0700, Daniel Pitts
newsgroup.nospam@virtualinfinity.net> wrote:

Is it just me, or is deciding on a power supply a relatively difficult
problem, compared to other aspects of working on digital circuitry?

s/deciding on a power supply/analog design/ -- Yes, although it can be a
lot of fun.

If you don't need portability then krw's suggestion of a regulated wall
wart is spot on and by far the simplest. Do ensure that the output is
actually regulated, though; some are not more than a step-down
transformer with a diode bridge.

Most tend to be switchers these days because switchers can be lighter,
which matters in containerized shipping.

Heck, there's an external "laptop" power supply with my wife's new
cheap desktop.

--
Les Cargill

 

[Daniel Pitts]

On 8/8/12 5:27 AM, Rich Webb wrote:

On Tue, 07 Aug 2012 17:39:32 -0700, Daniel Pitts
newsgroup.nospam@virtualinfinity.net> wrote:

Is it just me, or is deciding on a power supply a relatively difficult
problem, compared to other aspects of working on digital circuitry?

s/deciding on a power supply/analog design/ -- Yes, although it can be a
lot of fun.

If you don't need portability then krw's suggestion of a regulated wall
wart is spot on and by far the simplest. Do ensure that the output is
actually regulated, though; some are not more than a step-down
transformer with a diode bridge.

Okay, that all sounds good for now, and that's the approach I'll take.
Although, how do I verify if it is actually regulated or not?

In any case, I would like to know more about how to work on this kind of
thing in the future. For instance, I need to replace my sprinkler
system panel, which uses 24V for the valve controls, but I'll probably
end up using 5V or 3.3V ICs for most of the rest of the circuit.

Also, some idea's I had for projects require higher voltages. I'd be
interested in creating a Geiger Counter, for instance. I'd also like to
play around with EL wire for a project. Both of those require much
higher voltages.

Basically, I'm self-taught here, and would like some guidance on how
best to learn when/why to use various methods of power supply, and how
to design/build those.

Thanks again for all the advice,
Daniel.

 

On Tue, 07 Aug 2012 17:39:32 -0700, Daniel Pitts
<newsgroup.nospam@virtualinfinity.net> wrote:

Hello,

So far, all of my circuits have had the power supplied by an Arduino
(compatible) board. I'm going to try to change the setup here a little,
and design a circuit that has just an ATTiny in its place, but now I
have to think about power supply.

The circuit I'm envisioning is going to have three Texas Instruments
"TLC5916" constant-current sink's, powering one column of a multiplexed
8x8 RGB led matrix. The circuit will also have an ATTiny85 (or ATTiny84,
depending on a few things), and a 74HC238 (to select the row on the LED
matrix).

Now, if I'm reading the specs right, I think I can power the whole thing
on 3 AA batteries (4.5 volts). But if I wanted to have more robust power
handling, I'm not sure what I want to do. I'd like to try to keep the
circuit cheap, and potentially have it powered by a wall socket or USB.
I also don't have the time/money to etch my own PCB, so something that I
can plug into my protoboard is preferable.

Is it just me, or is deciding on a power supply a relatively difficult
problem, compared to other aspects of working on digital circuitry?

Thanks in advance for any suggestions!

I bought a relatively cheap switching supply from Marlin P Jones. It
gets a lot of use with the pic circuits I play with. It has switch
selectable 3, 4.5, 6, 7.5, 9, 12 volts out at one amp.

My pic breadboards have on board "whatever to 5 volt" three terminal
regulators, along with 3 AA batteries, and a RS232 programming
connector along with a solderless breadboard.

The on board 5 volt reg is only a 78L05 so it can't put out more than
100 milliamps to power the pic, then I use higher voltage and power
supplies for the heavy loads.

 

[Rich Webb]

On Wed, 08 Aug 2012 09:09:29 -0700, Daniel Pitts
<newsgroup.nospam@virtualinfinity.net> wrote:

On 8/8/12 5:27 AM, Rich Webb wrote:
On Tue, 07 Aug 2012 17:39:32 -0700, Daniel Pitts
newsgroup.nospam@virtualinfinity.net> wrote:

Is it just me, or is deciding on a power supply a relatively difficult
problem, compared to other aspects of working on digital circuitry?

s/deciding on a power supply/analog design/ -- Yes, although it can be a
lot of fun.

If you don't need portability then krw's suggestion of a regulated wall
wart is spot on and by far the simplest. Do ensure that the output is
actually regulated, though; some are not more than a step-down
transformer with a diode bridge.


Okay, that all sounds good for now, and that's the approach I'll take.
Although, how do I verify if it is actually regulated or not?

Generally (one expects that there may be exceptions) all you would need
to do is to check the unloaded terminal voltage. If it's (pulls random
wall-wart out of drawer) labeled "output 6 VDC 100 mA" and the no-load
voltage is 8 V then it's not really regulated. At its rated load it
should drop the extra 2 V internally (i.e., its output impedance is
about 20 ohms) so whatever it was intended to be plugged into would get
close to 6 V.

What else is handy... a "7.5 V 1000 mA" that measures 9.5 V unloaded and
a "12 V 400 mA" that's a spanking 18.5 V. There's also a "3.6 V 1 A"
that's reading 3.4 V. The last one is noticeably smaller and lighter
than the traditional transformer + bridge style. It's probably a tad low
since it isn't seeing its recommended minimum load (prob 100 mA or so).

A random 7 to 9 V wall wart paired with a simple 7805 linear regulator
on your breadboard might be a better start. You'll start to become
acquainted with filter caps and (possibly) heat-sinks but also another
step towards a DIY project.

In any case, I would like to know more about how to work on this kind of
thing in the future. For instance, I need to replace my sprinkler
system panel, which uses 24V for the valve controls, but I'll probably
end up using 5V or 3.3V ICs for most of the rest of the circuit.

Also, some idea's I had for projects require higher voltages. I'd be
interested in creating a Geiger Counter, for instance. I'd also like to
play around with EL wire for a project. Both of those require much
higher voltages.

Basically, I'm self-taught here, and would like some guidance on how
best to learn when/why to use various methods of power supply, and how
to design/build those.

Wouldn't hurt to pick up a copy of "The Art of Electronics" and its
companion workbook but you could start with some of the manufacturer
application notes at, say, National and TI (now joined at the hip) or
even the Wikipedia entry on voltage regulators.

--
Rich Webb Norfolk, VA

 

[fungus]

On Wednesday, August 8, 2012 2:39:32 AM UTC+2, Daniel Pitts wrote:

Now, if I'm reading the specs right, I think I can power the whole thing
on 3 AA batteries (4.5 volts).

The problem with batteries is...they're
not 1.5V!

They start out at about 1.56V when new,
rapidly drop to about 1.3, gradually drop
to about 1.1, then rapidly drop to zero.

Your 3xAA batteries will start out at 4.6V
then drop down to between 3.3-3.9V for
the rest of their lifespan. You have to
design for that.

Is it just me, or is deciding on a power
supply a relatively difficult problem,
compared to other aspects of working
on digital circuitry?

No, it's difficult.

An easy way to get 5V is with a wall-wart
or USB cable (as you mentioned).

If you want batteries it's a bit harder.
The ATtiny chips work on a wide range of
voltage so they should be OK with 3xAA.
The brightness of the LEDs can vary quite
a lot though.

You can design all sorts of circuits to
get around this but it's far, far easier
to go to ebay.com and type "5V DC boost".

That will get you a little PCB that gives
you a constant 5V DC from batteries. I've
been using them a lot lately and they're
worth every penny for projects like this.

Some of them even have USB sockets so you
can have the option of either battery or
USB power with the same cable.

 

[krw@att.bizzzzzzzzzzzz]

On Wed, 8 Aug 2012 21:37:14 -0700 (PDT), fungus <tooby@artlum.com> wrote:

On Wednesday, August 8, 2012 2:39:32 AM UTC+2, Daniel Pitts wrote:

Now, if I'm reading the specs right, I think I can power the whole thing
on 3 AA batteries (4.5 volts).

The problem with batteries is...they're
not 1.5V!

They start out at about 1.56V when new,
rapidly drop to about 1.3, gradually drop
to about 1.1, then rapidly drop to zero.

Your 3xAA batteries will start out at 4.6V
then drop down to between 3.3-3.9V for
the rest of their lifespan. You have to
design for that.


Is it just me, or is deciding on a power
supply a relatively difficult problem,
compared to other aspects of working
on digital circuitry?


No, it's difficult.

An easy way to get 5V is with a wall-wart
or USB cable (as you mentioned).

If you want batteries it's a bit harder.
The ATtiny chips work on a wide range of
voltage so they should be OK with 3xAA.
The brightness of the LEDs can vary quite
a lot though.

Several families of CMOS will work on wide voltage ranges, as well. I like to
use these, when possible, because having one P/N simplifies BOMs. Brighness
variation can be reduced significantly by driving LEDs with a current source,
at the cost of some complexity. I've had to do whenever driving LEDs directly
from a battery.

You can design all sorts of circuits to
get around this but it's far, far easier
to go to ebay.com and type "5V DC boost".

Boost regulators will cost power, twice (higher voltage for the ballast
resistor to dump and the inefficiency of the boost regulator). It's better to
use a current source, if "constant" brightness is needed.

That will get you a little PCB that gives
you a constant 5V DC from batteries. I've
been using them a lot lately and they're
worth every penny for projects like this.

Some of them even have USB sockets so you
can have the option of either battery or
USB power with the same cable.

 

[fungus]

On Thursday, August 9, 2012 3:19:09 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:

Boost regulators will cost power, twice
(higher voltage for the ballast resistor
to dump and the inefficiency of the boost
regulator).

I've measured them at about 90% efficient
on 3x 1.5V batteries.

It's better to use a current source,
if "constant" brightness is needed.

I don't think a textbook current driver
will really be more efficient than one
of those little boards. Plus you more
of them if there's multiple LEDs.

Having a known, regulated voltage to
work with makes life sooo much easier.

 

[krw@att.bizzzzzzzzzzzz]

On Thu, 9 Aug 2012 09:31:11 -0700 (PDT), fungus <tooby@artlum.com> wrote:

On Thursday, August 9, 2012 3:19:09 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:

Boost regulators will cost power, twice
(higher voltage for the ballast resistor
to dump and the inefficiency of the boost
regulator).

I've measured them at about 90% efficient
on 3x 1.5V batteries.

I can believe that but even at 90% efficiency, you're throwing away >10% of
your batteries' capacity *and* (Vboost-Vbatt)*Iled.

It's better to use a current source,
if "constant" brightness is needed.


I don't think a textbook current driver
will really be more efficient than one
of those little boards. Plus you more
of them if there's multiple LEDs.

It certainly is. You're both boosting to a higher voltage *AND* dissipating
more in the ballast resistor because of it. To light an LED, current is the
only thing that matters. You're throwing away everything above battery
voltage, plus the 11% of the total for the boost regulator. The current
source is nothing more than a voltage variable resistor so it dissipates
*less* power than a resistor at high voltage (assuming the current is set to
light the LED at the minimum voltage).

Having a known, regulated voltage to
work with makes life sooo much easier.

Different discussion. Wall warts are even simpler.

 

[Daniel Pitts]

On 8/9/12 9:31 AM, fungus wrote:

On Thursday, August 9, 2012 3:19:09 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:

Boost regulators will cost power, twice
(higher voltage for the ballast resistor
to dump and the inefficiency of the boost
regulator).

I've measured them at about 90% efficient
on 3x 1.5V batteries.

It's better to use a current source,
if "constant" brightness is needed.


I don't think a textbook current driver
will really be more efficient than one
of those little boards. Plus you more
of them if there's multiple LEDs.
Actually, I'm planning on using the TLC5916 from TI, which is a

constant-current sink 8bit shift register, for exactly this purpose.
That way brightness will be consistent.

 

[fungus]

On Thursday, August 9, 2012 9:11:55 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:

I don't think a textbook current driver
will really be more efficient than one
of those little boards.

It certainly is. You're both boosting to a higher voltage *AND* dissipating
more in the ballast resistor because of it.

Yes...but current sources also need a higher
voltage to allow the circuit to do its thing.

eg. In the context of this project you'll
need 4xAA for it to work.

Even 4xAA is marginal for a 3.6V LED. With
a cheapo current source (eg. twin-BJT) you'll
run the risk of not being able to sustain
the LED current until the battery is fully
drained, making efficiency moot.

ie. In most practical contexts the extra
efficiency probably cancels out.

 

[fungus]

On Friday, August 10, 2012 2:01:47 AM UTC+2, Daniel Pitts wrote:

Actually, I'm planning on using the TLC5916 from TI, which is a
constant-current sink 8bit shift register, for exactly this purpose.
That way brightness will be consistent.

Looks like the ideal chip for this...
you get a shift register *and* current
source all in one.

 

[fungus]

On Friday, August 10, 2012 2:01:47 AM UTC+2, Daniel Pitts wrote:

Actually, I'm planning on using the TLC5916 from TI, which is a

constant-current sink 8bit shift register, for exactly this purpose.

That way brightness will be consistent.

PS: Have you seen the MAX7219 LED driver chip?
It's similar to the TLC5916 but has built in
multiplexing for 8 rows of LEDs.

 

[krw@att.bizzzzzzzzzzzz]

On Fri, 10 Aug 2012 00:28:23 -0700 (PDT), fungus <tooby@artlum.com> wrote:

On Thursday, August 9, 2012 9:11:55 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:

I don't think a textbook current driver
will really be more efficient than one
of those little boards.

It certainly is. You're both boosting to a higher voltage *AND* dissipating
more in the ballast resistor because of it.

Yes...but current sources also need a higher
voltage to allow the circuit to do its thing.

Geez. With the current source the current *does* its thing, without burning
the extra power.

eg. In the context of this project you'll
need 4xAA for it to work.

..6V overhead is all that's needed.

Even 4xAA is marginal for a 3.6V LED.

What LED? We used blues down to 3V and they were plenty bright to be seen
outside during full sunlight (the product is used outside).

With
a cheapo current source (eg. twin-BJT) you'll
run the risk of not being able to sustain
the LED current until the battery is fully
drained, making efficiency moot.

Nonsense. It works. BJT + FET

ie. In most practical contexts the extra
efficiency probably cancels out.

Simply wrong, as shown.

 

[krw@att.bizzzzzzzzzzzz]

On Fri, 10 Aug 2012 00:37:06 -0700 (PDT), fungus <tooby@artlum.com> wrote:

On Friday, August 10, 2012 2:01:47 AM UTC+2, Daniel Pitts wrote:

Actually, I'm planning on using the TLC5916 from TI, which is a
constant-current sink 8bit shift register, for exactly this purpose.
That way brightness will be consistent.

Looks like the ideal chip for this...
you get a shift register *and* current
source all in one.

....and you just said it couldn't be done.

 

[krw@att.bizzzzzzzzzzzz]

On Fri, 10 Aug 2012 03:13:51 -0700 (PDT), fungus <tooby@artlum.com> wrote:

On Friday, August 10, 2012 2:01:47 AM UTC+2, Daniel Pitts wrote:

Actually, I'm planning on using the TLC5916 from TI, which is a

constant-current sink 8bit shift register, for exactly this purpose.

That way brightness will be consistent.

PS: Have you seen the MAX7219 LED driver chip?
It's similar to the TLC5916 but has built in
multiplexing for 8 rows of LEDs.

But it's Maxim. Good luck actually getting them.

 

[JW]

On Fri, 10 Aug 2012 08:56:15 -0400 "krw@att.bizzzzzzzzzzzz"
<krw@att.bizzzzzzzzzzzz> wrote in Message id:
<i61a28dbkjfnu729kccufca1d1df6qhjb9@4ax.com>:

On Fri, 10 Aug 2012 03:13:51 -0700 (PDT), fungus <tooby@artlum.com> wrote:

On Friday, August 10, 2012 2:01:47 AM UTC+2, Daniel Pitts wrote:

Actually, I'm planning on using the TLC5916 from TI, which is a

constant-current sink 8bit shift register, for exactly this purpose.

That way brightness will be consistent.

PS: Have you seen the MAX7219 LED driver chip?
It's similar to the TLC5916 but has built in
multiplexing for 8 rows of LEDs.

But it's Maxim. Good luck actually getting them.

Well, in that particular chips case, not a problem.
http://www.findchips.com/avail?part=MAX7219

 

[Michael A. Terrell]

"krw@att.bizzzzzzzzzzzz" wrote:

On Fri, 10 Aug 2012 03:13:51 -0700 (PDT), fungus <tooby@artlum.com> wrote:

On Friday, August 10, 2012 2:01:47 AM UTC+2, Daniel Pitts wrote:

Actually, I'm planning on using the TLC5916 from TI, which is a

constant-current sink 8bit shift register, for exactly this purpose.

That way brightness will be consistent.

PS: Have you seen the MAX7219 LED driver chip?
It's similar to the TLC5916 but has built in
multiplexing for 8 rows of LEDs.

But it's Maxim. Good luck actually getting them.

There are plenty of them on Ebay, and in several packages.
<http://www.ebay.com/sch/i.html?_trksid=p4340.m570.l1313&_nkw=MAX7219&_sacat=0>