It's such an odd voltage. Most DC is in increments of 1.5V (single battery cell voltage). So how did 5V become the defacto standard for logic chips etc.?
Three words: TTL
Transistors are not standardized for 5V, but ICs are or were. Due to electron tunneling on the microscopic IC traces on the silicon slice the IC voltage is dropping and the 5V IC will join Vacuum Tubes in oblivion. USB is 5V so it will be with us for a good while yet until USB gets reinvented. Who or why came up with 5V???
You have 5 fingers - I think it is THAT simple. A nice number in a suitable range - safely higher than a few silicon junction voltage drops but not too high to keep power down.
The world was not designed around the 1.5v alkaline battery. Besides, an alkaline battery is only 1.5v when fresh and unloaded - under load and/or partially depleted, it's less.... If you want 5v out, use 4x1.5v and a "nice" LDO, or 5x1.5v and a cheapie regulator... Or one of the inexpensive DC-DC converters now available for less than the cost of a cup of coffee (search DC-DC step-up or DC-DC step-down on ebay - don't search for buck or boost, because all the listings include both words in the title, even when it is one or the other - just like cheap arduino nano/micro/pro-mini clones include all three names and often 328p and 32u4 in the title of the listing).
5v as the standard logic levels for IC's dates back to the days of old TTL logic, decades in the past, when people couldn't dream of the sorts of electronics we have now.
5v is a nice voltage nowadays too; it's high enough that you can get 3.3v from it with a simple, cheap regulator (eg, a 1117-series with 1.2-1.5v dropout), and use it to charge a LiPo/LiIon battery without a switching regulator and it's higher expense and bulky inductor - and in turn the 3.7v nominal from that can go through a nice LDO (with 0.4v dropout or less) to drive a 3.3v device. 5v is also high enough that you can get 10 watts or so with an inexpensive cable and acceptable voltage drop across the cable, which turns out to be enough for a lot of applications - at 3.3v, you lose about 2.25 times as much of the power (as a fraction - 1.5x the current, meaning same cable gives 1.5x the voltage drop, but with the total voltage being 2/3rds as much
Beyond that, you can run a "5v" arduino pro mini or nano, for example, with 4.5v (the '328p is in spec at 16MHz down to IIRC 3.8, and in practice they usually work at room temp down at 3.3) 
As an aside, for like $3 on ebay you can get a board that you stick a 18650 battery onto that outputs 5v and 3.3v and charges from a microusb cable... (search 18650 battery shield)
Why 120 / 240 VAC instead of an even 100 / 200?
DrAzzy:
As an aside, for like $3 on ebay you can get a board that you stick a 18650 battery onto that outputs 5v and 3.3v and charges from a microusb cable... (search 18650 battery shield)
That's a good idea... Must look into that.
SamR:
It's such an odd voltage. Most DC is in increments of 1.5V (single battery cell voltage). So how did 5V become the defacto standard for logic chips etc.?
Discrete logic powered by batteries? No, that was never the expected thing, early logic families are
very power hungry, bulky and non portable - mains power was assumed(*). 5 is a nice round number,
I suspect its as simple as that.
(*) These chips were developed for mainframe computers, for instance.
MorganS:
Three words: TTL
DTL was there first I think.
MarkT:
DTL was there first I think.
Probably more like RTL [Really Terrible Logic]. ;D
ReverseEMF:
Probably more like RTL [Really Terrible Logic]. ;D
That aparently used 3.6V: NTE9903 RTL 3-Input NOR Gate Medium Power VCC=+3.6V 8-Pin Can, Grieder Elektronik Bauteile AG
So how did 5V become the defacto standard for logic chips etc.?
Well when the 7400 ttl chips came out you have to remember where the electronic industry was. The 20% resistor was the normal resistor used ( and most were 20% to 10%, as the manufacturers would sel the tighter % tol to sell as 10% and 5%), analog voltage meter were used, maybe 3% to 5% accuracy. There was little chance of making a chip run at 3 volts. To get enough over head I would think 5 V was the lowest that they could get a chip to work that they could produce in any numbers. A good 5V power supply would be anywhere between 4.7V and 5.2V and it would drift. A lot of stuff was ran on +/- 12V. Things have changed a lot since then.
1steve:
Well when the 7400 ttl chips came out you have to remember where the electronic industry was.analog voltage meter were used, maybe 3% to 5% accuracy
In some less enlightend countries perhaps, here in the UK we had access to the AVO 8 from 1951 onwards;
The instrument has an accuracy of ±1% of FSD on DC current ranges, ±2% of FSD on DC voltage ranges, ±2.25% of FSD on all AC ranges and ±5% of reading (at centre scale only) on resistance ranges.[11] Its maximum current draw of 50 μA at full-scale deflection (corresponding to 20,000 ohms per volt) is sufficient in most cases to reduce voltage measurement error due to circuit loading by the meter to an acceptable level.
Had mine around 25 years, still works well.
Also, old valve (tube) heaters ran on 6.3V (why 6.3? :o ) - it was relatively simple to build 5V supplies with the transformers. (BT,DT)
In the good old days, just after they came out with TTL, they offered Hi Noise Immunity.
This was 12 volt logic with similar construction as TTL.
This made speeds slow but was popular in high noise environments.
The Intel 8080 had a 12V clock signal, I seem to remember.
srnet:
In some less enlightend countries perhaps, here in the UK we had access to the from 1951 onwards;The instrument has an accuracy of ±1% of FSD on DC current ranges, ±2% of FSD on DC voltage ranges, ±2.25% of FSD on all AC ranges and ±5% of reading (at centre scale only) on resistance ranges.[11] Its maximum current draw of 50 μA at full-scale deflection (corresponding to 20,000 ohms per volt) is sufficient in most cases to reduce voltage measurement error due to circuit loading by the meter to an acceptable level.
Had mine around 25 years, still works well.
How many people remember why there was a mirror strip on the label with scales behind the pointer.
Do you remember the cost of one? What would that be in today money? I can remember a Electronic calculator being chained to a deck in early 60's in the engineer Dept, all it did was +-/*.
I feel old.
You're amongst geriatric friends Larry.
I build my Heathkit valve voltmeter more than 50years ago.
The high impedance of a valve voltmeter was a big improvement over the one my dad was using.
Still working, and sitting on my shelf, with a high voltage probe.
Leo..