Running a voltage regulator below recommended input voltage.

I'm planning a project which I would like the flexibility to power it from 4x AA Alkaline batteries (6V), 4x AA NiMH (4.8V) or a 12V input. Maximum current would be 100ma.

I would like to use this LDO regulator: http://www.digikey.ca/product-detail/en/NCP5501DT50RKG/NCP5501DT50RKGOSCT-ND/2121569
(from the datasheet, dropout voltage at 100ma is 75mV, therefore minimum input voltage is 5.075V)

This would work great for the 6V or 12V supply, but with 4.8V, it would be running below it's recommended input voltage. Is this Ok? Would the output just be 4.8V - 0.075V?

I know I can just run off 4x NiMH batteries without a regulator, but I want the flexibility to pop in either type of battery with minimal hassle.

Is this OK?

It wouldn't exactly be passing through the low-voltage.
It wouldn't necessarily be destructive, it would just suck.
Put in a switch to go via reg or straight from batt-pack.

Is this Ok?

The datasheet should have some info regarding that. Those guys typically have a minimum input voltage of 1.5 - 2v - the point where they start to turn on. From there to their rated voltage, they simply pass on the input voltage, with minimum voltage drops.

Would the output just be 4.8V - 0.075V?

You will have to try for sure.

I've run older style TO-220 v.regs off batteries, and as I recall, the v.reg continued to run ok for
1 or 2V below 5V, until the microcontroller stopped working, although obviously the v.reg was
no longer was regulating. I've not tried this for any smt v.regs, like on Arduino boards.

My choice, however, would not be to start out with only 4.8V for Vin, as you're already into
the abnormal range, but rather use 5 or 6 NiMH cells.

Yes, it should work. However, I would want to be sure that the Arduino and associated circuitry will work well on 4V, so that it will work even when the NiMh cells are partially discharged. The Arduino should be OK on 4V if you don't mind the ADC reference voltage changing, or you use a different reference from AVcc. Will the other circuitry you are using work on 4V? If not, then use 5 or 6 cells instead of 4.

You could also consider designing the system to run on 3.3V.

Your datasheet suggests that your regulator turns on at 2v Vin. So anything higher than that is passed onto the load / arduino.

Thanks for the input guys.

My choice, however, would not be to start out with only 4.8V for Vin, as you're already into
the abnormal range, but rather use 5 or 6 NiMH cells.

That would work too. Might have to go that route.

You could also consider designing the system to run on 3.3V.

I thought of that as well, it would also consume less power too. I am planning to us a RGB 16x2 LCD on this project, and non that I've found run on 3.3V.

On a side note, can you arbitrarily operate a board at 5V or 3.3V (assuming peripheral components can run at either voltage)? Or are there certain fuses that need to be set? Is a level shifter required when uploading a sketch to a 3.3V board from a 5V programmer?

To run 5v lcd at 3.3v, you can drive a pwm pulse train into a charge pump (two diodes + caps) to create a negative voltage. You can then use that voltage to create the contrast voltage.

John_S:

You could also consider designing the system to run on 3.3V.

I thought of that as well, it would also consume less power too. I am planning to us a RGB 16x2 LCD on this project, and non that I've found run on 3.3V.

I think you will find that your LCD contrast changes when the supply voltage does. In which case, you need to maintain a stable 5V.

John_S:
On a side note, can you arbitrarily operate a board at 5V or 3.3V (assuming peripheral components can run at either voltage)? Or are there certain fuses that need to be set? Is a level shifter required when uploading a sketch to a 3.3V board from a 5V programmer?

If you run a 5V board at 3.3V, there are at least a couple of issues:

  1. You will probably need to change the fuses to reduce the brownout detection threshold.

  2. Assuming the board uses a 16MHz crystal or resonator, you will be slightly overclocking it, because the maximum operating frequency at 3.3V is (if my calculation is correct) 14.14MHz.

you will be slightly overclocking it, because the maximum operating frequency at 3.3V is (if my calculation is correct) 14.14MHz.

Rather than calculate it, you can read the datasheet for you.

dhenry:

you will be slightly overclocking it, because the maximum operating frequency at 3.3V is (if my calculation is correct) 14.14MHz.

Rather than calculate it, you can read the datasheet for you.

I calculated it from the data in the datasheet in the absence of a figure for max clock speed @ 3.3V being provided. From figure 29.1 on page 322 to be precise.

Thanks dc42, gives me more to think about. One more question, can I run a (properly fused and clocked) 3.3V board temporarily at 5V when I program it with a 5V programmer?

John_S:
Thanks dc42, gives me more to think about. One more question, can I run a (properly fused and clocked) 3.3V board temporarily at 5V when I program it with a 5V programmer?

Yes, if all the components on the board and connected to the board's output pins and programming pin are rated at 3.3V. If there is a 3.3V regulator still connected to the board, in theory you might damage by feeding 5V into its output. That's unlikely, but it may feed power from its output pin back to its input pin.

Some ISPs (e.g. AVRISP mkII) require the target device to be powered, then they sense that voltage and program it at the same voltage.

If you run a 5V board at 3.3V, there are at least a couple of issues:

  1. You will probably need to change the fuses to reduce the brownout detection threshold.

  2. Assuming the board uses a 16MHz crystal or resonator, you will be slightly overclocking it, because the maximum operating frequency at 3.3V is (if my calculation is correct) 14.14MHz.

If the Brownout is set for 4V when Vcc=5V, won't it automatrically change to 0.8*3.3V = 2.64V for Vcc=3.3V?
The uC has no absolute knowledge of Vcc, or does it?

On one of my boards I have a jumper to select Vcc = 3.3V or 5V, and I've seen no problems whatsoever when
set to Vcc=3.3V and using a 16-Mhz crystal plus Duemilanove and UNO chips burned with the stock bootloaders
[ie for 5V]. However, some people warn against using 16-Mhz xtal at 3.3V [but I'm gonna run it till it mucks
up if ever].

oric_dan(333):
If the Brownout is set for 4V when Vcc=5V, won’t it automatrically change to 0.8*3.3V = 2.64V for Vcc=3.3V?
The uC has no absolute knowledge of Vcc, or does it?

Yes, it does. The brownout detector uses an internal (but not very precise) voltage reference. See table 29.13 in the atmega328p datasheet,

In retrospect, I guess that makes sense.

I’ve been trying to decipher exactly what fuse settings are made in the bootloaders. I found the
following in “Makefile” in the IDE optiboot directory,

atmega328_isp: atmega328
atmega328_isp: TARGET = atmega328
atmega328_isp: MCU_TARGET = atmega328p

512 byte boot, SPIEN

atmega328_isp: HFUSE = DE

Low power xtal (16MHz) 16KCK/14CK+65ms

atmega328_isp: LFUSE = FF

2.7V brownout

atmega328_isp: EFUSE = 05
atmega328_isp: isp

This would explain why I can run the chips at 3.3V without the brownout triggering.

I’ve not yet found whether the watchdog is enabled or not. Takes a lot of searching to nail this stuff
down.

I've not yet found whether the watchdog is enabled or not.

:slight_smile:

If you haven't tried digging through the zillions and zillions of Arduino IDE subdirectories to try and
find a smidgeon of information, then an entire new universe awaits you, LOL.

I rigged up an experiment today.
I did a load test using a 7805 vs an LM2904-5 (an LDO lin reg).

        no load   60?    47?
*7805*
Vin      Vout    Vout    Vout
4.0      3.23    2.62
4.5      3.72    3.07
5.0      4.2     3.58
6.0      4.99    4.56
7.0      4.99    4.99

*2940*
4.0      4.0     3.9     3.9
4.5      4.3     4.42    4.42
5.0      4.95    4.93    4.89
6.0      4.97    4.97    4.97
7.0      4.98    4.98    4.99

The 7805 blows at this low V_in thing and the LDO seems to hold up,
but I'm not convinced that there isn't a "gotcha" in there.
Anyway, such are my findings.