5
\$\begingroup\$

It's a noob question, but every circuit I make uses 5V: 74LS uses 5V, ATMEGA328 uses 5V and so on.

But in reality, everybody's designing circuits and shove a voltage regulator with 9V or 12V power source. Wouldn't a company who makes 6V batteries sell by billions?

\$\endgroup\$
5
  • \$\begingroup\$ Every circuit I make uses 3.3 volts since it can be linearly regulated from a single lithium ion cell. Alternatively, I can use 2 alkaline cells, or a lithium manganese single cell if it doesn't draw much current. \$\endgroup\$
    – HL-SDK
    Jul 30, 2014 at 17:58
  • \$\begingroup\$ 6V = 4 alkaline cells, and you can get 6V batteries. I'm told the Aston Martin Lagonda used specially made 6V TTL logic for some crazy reason, but I can't find a cite for that factoid. \$\endgroup\$
    – pjc50
    Jul 30, 2014 at 18:55
  • \$\begingroup\$ 6v lantern batteries (basically 4 super-D cells in a metal can) were not uncommon in the era when 5v TTL became a norm, but that likely has little to nothing to do with the choice. \$\endgroup\$ Jul 30, 2014 at 18:55
  • 5
    \$\begingroup\$ Very little transistor-based logic was running from battery power until long after 5V was settled on as a supply voltage. Computers used to be large, stationary objects. \$\endgroup\$
    – Matt B.
    Jul 30, 2014 at 20:32
  • \$\begingroup\$ An Alkaline cell's 1.5 nomimal voltage is under 0 load. Under realistic load, you generally get 1.2-1.3 volts. So 4 of them give 6V nominal, but 4.8-5.2 volts practical, which actually works pretty well for 5v (TTL-level compatible) CMOS circuits, which are often not very picky about the exact volatage in the first place. So in practice 5v is not that inconvienent for batteries. \$\endgroup\$ Jul 31, 2014 at 18:57

5 Answers 5

12
\$\begingroup\$

The simple reason is that back in the Dark Ages (like the 1950's and 1960's) nobody in their right mind would try to power logic with batteries. The first logic families, RTL, DTL, and the big winner, TTL, were bipolar, not CMOS. Currents were in the vicinity of 2 - 20 mA / gate, depending on interconnection. So any reasonable sized logic circuit would pull amps of current, and systems pulling 100's of amps were not unheard of. Not only that, the only widely available rechargeable battery chemistries were lead-acid and NiCad. Neither of these was going to provide 10's to hundreds of amp-hours of capacity in anything portable.

With the widespread acceptance of TTL, the successor CMOS families (74HC, HCT, ACT, etc ad nauseum) were designed to be compatible with TTL, which meant 5 volt operation (although you'll notice that 74HC works over 3 - 6 volts). The big exception was the CD4000 series, which had (and have) a much wider operating range. Then, as fab processes got smaller and the devices faster, lower-voltage families started to dominate until, as KGregory points out, 5V is really not much used commercially, at least for entirely new designs.

As for the origins of 5V/TTL, that was a set of design tradeoffs. If you find a schematic of a TTL gate, you'll notice that it needs at least 3 diode drops internally, plus various resistor drops. What that doesn't tell you is the choice of reference currents (1.6 mA for a low input) which was in part determined by the current levels required to produce acceptable switching speeds. These current levels in turn set limits on the internal resistor values, and the voltages needed to feed them. You also need to factor in the state of semiconductor fab capability - the first TTL circuits were at the edge of what could be reliably produced. Imagine - 20 to 100 gates on a chip! That's (gulp) hundreds of transistors, with the masks all laid out by hand. All of this, including power dissipation limits, resulted in the standard TTL supply voltage spec of 4.75 to 5.25 volts. As it turned out, this was an adequately wide margin for practical systems, and the speed (10 - 20 MHz) was adequate for a wide range of applications. So TTL became king. Even then, if you wanted faster speed there were other families available, like 74S and ECL, but those puppies were even bigger power hogs than TTL. Go look up the construction techniques for the first Cray computers.

\$\endgroup\$
1
3
\$\begingroup\$

Criteria number 1 : Base-emittor junctions of a transistor that are reverse polarised cannot have much more than 6 volts. They start 'leaking' and accumulate time based damage.

Criteria number 2 : fist logic chips were bipolar and had quite a bit a static current consumption resulting in heat. higher voltage means more heat...

criteria number 3 : early chip technology used for digital suffered from scaling problems. they needed quite a bit of distance to 'hold' a standoff voltage. making chips impractical and expensive ( the cost of a chip is defined in square millimeters of surface... )

Throw that in a heap and you end up with something that works between 3 and 5 volts. At 3 volts the transistors did not switch 'fast' enough to get nice clean pulses so they settled at 5 volts. All criteria met

Now, for early MOS technology they ran into another problem. They only had NMOS transistors. There were no P-MOS ( they hadn't figured out the process of implantation yet, they were depositing doped regions through crystal growth in an oven and then etching it. ) So they stacked nmos transistors to make totem-pole systems. problem is that you now need an additional voltage to switch top and bottom. So they could have used ground , 5 volts and 10 volts ( to turn on the top transistor you raise its gate 5 volts above its source which sits at 5 volts. Problem is that this was not compatible with bipolar logic. so they flipped the stuff around. they used -5 volts and used that as the 'ground' level. to create a compatible output all they needed was a mos from the 5 volts to the output pin. turn on the top mos ang you get 5 volts out. turn it off and you get 0 volts out. the internal logic used -5v as a logic 0 and 0 volts as a logic 1. Early cpu's in NMOS technology actually have a -5 volts pin.

Once they could construct both PMOS and NMOS (what we now call the CMOS process : complementary metal oxide semiconductor : meaning both n and p , although that metal - oxide ... for a long time was not true... it started like that, went away ( we used doped polysilicon as the gate no need for metal.... ) and now is back ) the negative voltage wa sno longer needed.

there were other technologies around like ECL that also required a negative voltage and used 5 volts and -3 volts as their supply rails ( although the logic levels for ecl are like 1 volt and - 1.2 volts or something like that . consumption of power in ECL is a constant , you just throw current from one loop to another) so that way they maintained compatibility with existing power supply systems..

it's all historical and based on practicality for early integrated circuit technology.

A cray computer like the cray 1 for example did not have a 'regulator' as we understand it now. they used a rotary convertor. a motor drove a generator that made a 6 phase output voltage at 400 hertz. they rectified that and ended up with very little ripple due to the 6 phases. so they needed minimal capacitors ( the cray 1 sucked hundreds of amperes on its power rails... being a fully ECL machine )

the 'regulator' just controlled the field coil of the generator to adjust the output of the generator. so they did not use a transistor to regulate the hundreds of ampere. just control the strenght of the spinning magnet an u regulate output voltage of the generator.

there's all kind of trickery like that in these early machines.

\$\endgroup\$
3
  • \$\begingroup\$ 100 amps seems like a lot, but at 5v it's only 500w -- the same as 5 light bulbs. \$\endgroup\$
    – tcrosley
    Jun 5, 2015 at 17:55
  • \$\begingroup\$ Current in TTL is always in mA never in Amps. Hence o/p power is always in milliWats not in Watts. \$\endgroup\$ Jun 5, 2015 at 18:19
  • \$\begingroup\$ I was referring to your statement about the Cray 1 consuming hundreds of amps. Of course it used ECL, not TTL. My comment didn't say anything about TTL. I was just pointing out that 100 amps at 5v is really that much in terms of watts. \$\endgroup\$
    – tcrosley
    Jun 5, 2015 at 18:25
1
\$\begingroup\$

The logic voltage depends largely on the CMOS fabrication processes while the battery voltage depends on battery chemistry. Both are very much independent. Furthermore, not every logic application uses a battery and, in fact, many are powered by AC.

Most importantly, however, is the fact the logic voltage is a major contributor to power consumption in digital logic. Therefore, even if you have to linearly regulate the logic voltage down from some higher voltage, you will save power by reducing the logic voltage. When using a switching regulator (aka DC-DC converter), the power savings will be even greater.

For this reason, we see logic voltages steadily trending downward with newer logic devices. 5V logic, is in fact mostly obsolete these days and is mostly only used by hobbyists today. 3.3V is much more common for board-level logic, while 2.5V and 1.8V logic are growing in popularity. Furthermore, many ICs use even lower logic voltages internally (1.5V and lower is common).

Another issue is the fact that, even though a battery is nominally 9V or 12V (or any voltage it may be), the actual voltage may fluctuate quite a bit. For this reason, it would generally be a good idea to use a voltage regulator to guarantee a more stable and consistent supply voltage in order to ensure more stable and consistent operation. This means that most designers will look for a power source that is guaranteed to be higher than their operating voltage since they intend to regulate it down anyway.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ TTL != CMOS... that came later \$\endgroup\$ Jul 30, 2014 at 18:52
  • \$\begingroup\$ True, but that underscores my point about the evolution towards lower voltages and reduced power consumption. \$\endgroup\$
    – kjgregory
    Jul 30, 2014 at 19:28
  • \$\begingroup\$ I'm not sure how this explains why TTL was designed to use 5V... \$\endgroup\$ Jul 30, 2014 at 19:55
  • \$\begingroup\$ It's driven by the device physics and dependent on figuring out how to reliably make transistors that operate at a given logic voltage. It is also somewhat arbitrary (e.g. 5V is a nice round number as opposed to, say 5.1634V). The availability of batteries that can supply said voltage is very much not a significant factor in deciding what logic voltage to design for. \$\endgroup\$
    – kjgregory
    Jul 30, 2014 at 20:01
0
\$\begingroup\$

Going out on a limb, and making a WAG which doesn't even reach SWAG status, my bet is that batteries were barely a consideration during the bad old TTL days. Digital computers were huge arrays of TTL chips, and batteries would have lasted moments. Calculators and similar handheld products tended toward 9V batteries, which have fairly large capacity, and those were replaced often.

In some ways, today's mobile device industry shows a remarkable convergence of better logic families and better battery tech. To really properly appreciate modern battery technology, it is required that you owned a TI-SR50 calculator at some point in your life.

\$\endgroup\$
0
\$\begingroup\$

As already said, old 5V ttl systems drain lots of power so battery powering them seem unreasonable. But there's a reason for 5V voltage regulator. Very simple voltage regulator consist of zener diode reference with a transistor as a voltage follower. 5.6 V zener is most temperature stable. After 0.6 V drop in voltage follower, there's 5 V stable power.

Note, even if 5V battery exist its voltage range would not suit to TTL requirements. TTL normally accept 4.75 to 5.25 V range, which is (plus-minus) 5% tolerance. Best suited four NiCd cells battery provides voltage range is 4.4 (completely drained) to 5.6 V (fully charged) so you need an IC at least 13% supply voltage tolerance.

Most suitable for battery powered application is CMOS technology which is historically accept very wide voltage range, say 3 to 15 V (see CD4000 series chips). So it accept 4.5 V, 9 V or 12 V without any transformations.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.