Why not a voltage selector on the Arduino board?

giantsfan3:
And pretty much the only disadvantage to having such a voltage selector, as far as I can see, is that beginners may mistakenly leave it on 3.3V and use a 5V shield.

Worse would be using a 3.3V sheild with the voltage kept at 5V. It tends to let the 'magic smoke' out.

retrolefty:
Many people have run their chips at 16Mhz with 3.3vdc Vcc without problems, but that is indeed working outside the datasheet specifications for the device, effectively 'overclocking' when running at that speed and Vcc voltage

Despite the datasheet spec, I've never heard of a case where this has actually not worked, so I expect the spec is uber-conservative. Overclocking successfully is usually a matter of heat dissipation, and older designs like the AVRs have been through enough revisions that they are probably pretty efficient at that, compared to the earlier versions.

The fact that products like the Seeeduino have been available for some time that do this also makes me think that 16MHz at 3v3 is probably quite an acceptable design in practice, except perhaps for critical applications. I'm also regularly running ATtiny85s at 16MHz quite happily 3v3, which is similarly out of the datasheet spec. I haven't observed any issues at all so far.

I'm also regularly running ATtiny85s at 16MHz quite happily @16MHZ

That's a classic if(true) statement! :wink:

Looking at the datasheet graph, I see that is indeed the case (although 12 Mhz might be an option).

Yeah, this same topic has come up half a dozen times in the past week or so. If you interpolate
the datasheet carefully, you'll see the 3.3V point corresponds to 13.3 Mhz, so a little closer to
16 Mhz. However, a lot of guys still say this isn't 16 Mhz.

Yep. But as far as the original topic of this thread is concerned:
Given that people haven't reported problems with 16 Mhz @ 3.3V and more importantly, given that a LOT of parts nowadays are spec'ed to run at 3.3V or below, and usually NOT at 5V, I think it would be more useful than not, if the Arduino devs placed this kind of voltage selection option on the board.

Common parts that run at 3.3V level:
Sd cards, Xbee (pretty much all wireless modules to my knowledge), Accelerometer, Many, many other sensors, etc.

Also, I've seen quite a lot of questions on this forum that could have been solved/simplified by just running the system at the 3.3V level necessary for the interfacing.

Moreover, running the circuit at 3.3V would naturally pave the way to being powered by a Li-ion battery (plus maybe a regulator in between)

retrolefty:

I'm also regularly running ATtiny85s at 16MHz quite happily @16MHZ

That's a classic if(true) statement! :wink:

"The first rule of Tautology Club is: The first rule of Tautology Club."

Yep. But as far as the original topic of this thread is concerned:

My home brew boards all have jumpers to select Vdd = 5V or 3.3V. I've been running 328's
at 3.3V and 16 Mhz fine for year now, but all the guys say wait till next year, ya never know.
[just do it, it'll work fine - famous last words, I hope].

oric_dan(333):
My home brew boards all have jumpers to select Vdd = 5V or 3.3V. I've been running 328's
at 3.3V and 16 Mhz fine for year now

That's your choice, but anecdotal tales of running parts outside their spec proves nothing. Going around telling people it's OK to do so isn't doing anybody a service.

are there any possible difficulties or disadvantages to this?

Cost; board space; reliability; etc.

Why Not a 3V3 option? Indeed.
The cost might be in the high $.50 or so.. burdened. Change that 150mA regulator out for a 500mA one and add a 3 pin jumper for Vcc selection. But This Thread really poses a good question, Why Not a 3V3 Option?. From the inception of the Arduino project in 2005 - 6? to now there's been a major paradigm shift in all digital electronics due to "The Need for Speed". Cmos is now king for many reasons and along with it some issues, one being the gate capacitor. Lowering the supply voltage does two things, one is to lower the overall circuit impedance which raises the drive level available to charge the gate region the second is a result of the first. In lowering the voltage the necessary silicon active area becomes smaller and thus capacitance's both of design and stray, etc. become proportionally smaller allowing speed to go up again until the original factors become an issue again. The new 3V3 is 1V5 - 8. So for a 'New' Arduino I should think that provisions to serve both voltages is well in order. If the spirit of a really basic and reasonably indestructible "Trainer" that is well within the budget of nearly anyone who can use the Internet is to be preserved. The Secondary thought that the "Trainer" is surprisingly sophisticated only adds to it's appeal. Some will grow beyond the "Trainer" and for those there are other devices but the form is so well ingrained in the public eye that if anyone wants to sell a "trainer" it had better be "Arduino Compatible" which is a third argument for a dual voltage board as All of the newer devices are 3V3 Max and many now are lower still.

Boob

Why Not a 3V3 option? Indeed.

Well the new Due board solved that, it's 3.3v only, slap on a 5 volt shield and it might ruin your day. :wink:

The original Arduino board used a simple 3 pin jumper chip to select USB or external DC power for the board. But I guess that must have resulted in too many questions for problems for begineers so we now have the auto-voltage selector circuit which to me is a waste of board space that could have been used for some more useful function.

Lefty

All I see is an outstanding reason to make a 5/3V3 board. It certainly isn't the cost and interfacing is not difficult. I figured it out in about 3 hours...

Bob

Well the new Due board solved that, it's 3.3v only, slap on a 5 volt shield and it might ruin your day.

I realize this may be anathema to Atmel advocates, but I just bought a Digilent Chipkit uC32
board [Arduino-UNO form-factor], and it uses an 80-Mhz PIC32 chip that runs at 3.3V, but
all the digital I/O pins are 5V tolerant. My little Pololu Zumo robot wants speed, speed, and
more speed. [back to my cave now].

I realize this may be anathema to Atmel advocates,

Well not to me, as I feel the Due should have been designed to be hardware compatible with all legacy shield boards, but was told (I think) that the 'desired market target price' for the Due would have been to high.

Lefty

5V shield only ruins your day if you power it from 5V.
So power it from 3.3V instead and all is honky-dory.

Gives me an idea for a DUE shield. Bring all pins in, but only 3.3V for power. Any device must be 3.3V device.

2nd shield - Bring 5V in, but provide offset female headers that are all 3.3/5V buffered with TXB0108 chips or equivalent. Fully compatible with 5V shields, but only 3.3V is exchanged with the DUE.

My home brew boards all have jumpers to select Vdd = 5V or 3.3V. I've been running 328's
at 3.3V and 16 Mhz fine for year now

That's your choice, but anecdotal tales of running parts outside their spec proves nothing. Going around telling people it's OK to do so isn't doing anybody a service.

Well, normally I'm a lot more conservative, but in this case, I say full speed ahead and
d*mn the torpedoes. [It works until you're at the bottom of the sea].

5V shield only ruins your day if you power it from 5V.
So power it from 3.3V instead and all is honky-dory.

Now we have the reverse problem to a short time ago. Previously, some vendors were selling
shields, like for XBee and SD cards, that did not have 5V-to-3.3V level-shifters on them, and
people were blowing up the XBees and SD Cards. So, the vendors finally came around to
redesigning the shields better.

So, maybe the shields should have the 5V vs 3.3V option, then the DUE is safe. Can of worms.

CrossRoads:
5V shield only ruins your day if you power it from 5V.
So power it from 3.3V instead and all is honky-dory.

So any active components on any specific legacy 5 volt shield will run at 3.3vd? Also there is still 5vdc on the 5V shield pin even on a Due board, so any old 5 volt shield will still run at 5vdc just due (no pun) to the fact that it's mounted to a Due board.

Gives me an idea for a DUE shield. Bring all pins in, but only 3.3V for power. Any device must be 3.3V device.

Easy to design a shield for Due use only or for Uno use only, trying to get a shield compatible for both is what we are asking about.

2nd shield - Bring 5V in, but provide offset female headers that are all 3.3/5V buffered with TXB0108 chips or equivalent. Fully compatible with 5V shields, but only 3.3V is exchanged with the DUE.

Maybe, not familiar with that TSB0108 chip. Does it support the high output current that a 8 bit AVR output pin can supply?
Lefty

"Maybe, not familiar with that TSB0108 chip. Does it support the high output current that a 8 bit AVR output pin can supply?"
TXB010x would be for logic level translation of control lines between Due IO and 5V devices on the shield.
http://www.ti.com/product/txb0101
Also available with 2,4,6,8 buffered lines:
txb0102
txb0104
txb0106
txb0108