Queestion as topic : does anyone here know of a microcontroller that can operate with supply and I/O between 4 and 7 volts ? If I could have anything I wanted then a SiLabs 8051F320 would be perfect but they can only go to 5.25V. Alternatively almost anything in the TI MSP430 series would do at a pinch but they also only seem to support supply voltages up to 5.5V max.
Of course I could regulate the supply but then I would also need to opto-isolate all the I/O which is not viable.
then old CMOS 1802 is the only one I know of. Oh, and the Motorola MC14500.
It's not difficult, it needn't take up much room, and it makes the system better. Why not do it? Or if that's too much, use resistive dividers or resistors/ zeners to get the voltage down. that helps too, you can add a cap to filter the inputs, EMC and all that.
You don't need to opto-isolate, just level-shift. Many existing ICs can do that. In fact, many PC CPU runs below 2V and level-shift to 3.3V/5V. Many microcontroller runs below 2V for much lower operating power. some has open-collector outputs to aid lvel-shifting.
Maybe I don't understand what you're asking, but why not use a 3V micro and shift it up 4V with a zener in its ground pin(s). I'm not sure how the external I/O you need are ground referenced, Sprow.
I mispoke thanks for pointing that out. Of course any level-shifting method would do, not just opto-isolating the I/O. However I want it all done on the MCU, 'cos otherwise it'll treble the board area (<
1"" square) and also treble the costs. I'm certain this can't be an unusual requirement - is there really no MCU currently available with higher than 5(ish) volts tolerant I/O ?
The old RCA1802 is about it for choices. It runs at up to 10v, so 7 volts would not be a problem. But the old 40 pin .100" spaced pins make for a large chip by today's standards.
But if you choose one of the lower power MCU's and run it at 3v ot 3.3v or
5v with a small voltage regulator, you can keep the size down. You can use some of the CMOS 14Cxxx series IC's as level converters for up to 12-16v max. I do not know how big you want to make your PCB nor how many I/O pins you wanting to use, nor what your going to drive with the I/O pins. One nifty method is to use a small transistors for both In and Out, since they are open collector they could work Ok that way. You use two transistors Out and two In on the same I/O pin, thus you get a non-inverting level shifter.
I take it from your answer that Atmel don't have such a chip in their armoury ? I'm interested to hear from a chip mfr what the reasons behind this might be ? Of course I understand that CPU performance is all optimised for certain voltages and that device densities would suffer. But still... a device with a higher supply and I/O voltage tolerance must surely have many applications ? Anything that runs of batteries would be easier to design, for instance. So how about persuading the chaps back at base for us ? :-)
1) Because higher voltage means higher power, and most of us need to minimize power.
2) It would require a MASSIVE investment in process design and test equipment. There's no way you could ever sell enough to break even.
That's putting it mildly. NRE aside, I would think that geometry restrictions would mean the parts would have to be a lot more expensive than lower-voltage equivalents.
Apparently not, or somebody would still be selling them. ;)
Why is that? Most cells put out 1.1-1.5V. It would seem that lower voltages would be a better answer than higher ones.
I wouldn't hold my breath.
--
Grant Edwards grante Yow! Yow! Is this sexual
at intercourse yet?? Is it,
visi.com huh, is it??
Well Dave, most chip mfr designs are a result of solving THEIR problems, not ours (even though their "Mission Statements" would have you believe otherwise). Or maybe they solve their top buyers (auto mfr) problems, but thats about it (and they kick and fight the whole way in any case).
CPU's that work under wide voltages ranges would make our life much easier, but would cause them big headaches. Likewise for microcontrollers with floating point, or at least a fixed point CPU with a friggin divide instruction (i.e., where is the divide ARM?, oh Thumb2 thanks finally).
But nooooo, they like to make nice sleek simple CPU's that solves all their problem (the origins of the wacky RISC craze), in return, we have to pay the price by wasting our time trying to figure out how to optimize a binary divide algorithm (see related thread in this newsgroup) all because they want to save some die space so they can include a 20th serial inteface that no one needs (but is easy to implement). Or we have to waste our time rewriting/rethinking all our algorithms to not use a divide. Or deal with the headaches of fixed point math and the typically non portable C code that goes along with it(what are we still in the 1970's?).
I'm sure all the reasons why this is so is because it makes the most sense for FORD or others who buy millions of these chips who require the lowest recurring cost and thus is less concerned about NRE (software development). Why if I want floating point do I have to buy a 388 pin 1 mm ball grid array microcontroller to get it (MPC565 with has more serial ports then I can count on two hands)? Because its best for FORD, for embedded products which sell in the thousands or ten's of thousands, most of chips out there don't make much sense and you have a snowballs chance in hell changing it. Just my opinion.
Unfortunately for your design, the entire semiconductor world has been shifting to smaller dies sizes, lower power contraints, and lower voltages. Pretty soon the 1.8v systems will be common place. That is what is driving the market now. As for running off of batteries, many systems run off of one Lithium-Ion cell at 3.7v quite readily. Many other battery powered systems use one 1.2v or two 1.2v cells and use DC-DC converters to generate higher voltages such as for LCD back lighting and such. Thus the old four, six and eight cell battery designs have all been replaced by the one or two cell battery designs. Flash card devices all run at 3.3v or less now. So unfortunately there is simply no market for 10v MCU's anymore. The manufacturers aren't going to make something that doesn't make a profit for them. It costs many millions of dollars for a factory to tool up to make the chips. You can't make obsolete chips if no one wants them. Even if you wanted to buy several million 10v MCU's. none of the manufacturers would be able to accomodate you as they have all been switching to ever smaller die sizes and equipment and no one has old chip manufacturing facilities left to make these kind of large die high power chips anymore.
I do not see much need in a higher supply voltage, but higher I/O voltages would be nice. I think that the largest problem is due to the common use of multifunction pins, i.e. pins that can be programmed both as inputs or outputs, which requires a lot of electronics on the pin. Designing such pins for high I/O voltages would cause a lot of reverse biasing problems.
However, if dedicated input on output pins are used, the outputs could be simply open collector(drain) types, in which the external I/O voltage could be quite high, provided that there are no real or parasitic diodes from the output pin (collector/drain) to the Vdd. If the pin can sink sufficient currents (>20 mA) LEDs and small relays could be driven directly.
On the input side a high current protection diode would be required between each input and Vdd and the input can be used as voltage input, if the input voltage is always between 0 and Vdd or as a current input, with an external series resistor, if the input voltage can swing above Vdd. An extra resistor from input to ground may be needed to move the "low" state threshold sufficiently low.
A CPU with a low internal power consumption can be driven with a series resistor and a shunting zener from any voltage. The zener should keep the Vdd below the maximum allowed Vdd and should be big enough, so that it can also dissipate the total worst case input pin current flowing trough the external series resistors through the input protection diodes to Vdd and through the zener to ground
As such, I do not see a problem if the CPU core Vdd is quite low (1.5-3.3V) but it sure would be nice to have (high voltage tolerant) _current_ inputs and outputs (instead of voltage I/O).
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.