Hello,
So my brother asked me a question a few days ago. What is the difference between TTL and CMOS? Well I was sorta able to answer, although my knowledge of those is only theory and not actual application specific. So what are the pros and cons between the two? Like, why would one choose CMOS over TTL, and vise versa?
Difference is in the type of transistors used, MOSFETs vs BJTs.
Used to be that CMOS was low power, TTL was high speed.
Used to be that CMOS was more susceptible to electrostatic damage, TTL less so.
CMOS less power hungry than TTL, with CMOS power needs based on switching speeds, with TTL power usage more steady state.
Nowadays, CMOS used for a lot more stuff, using one vs the other more a matter of what functionality you want.
By functionality, you mean if you want high speed or low power consumption, chow how much faster is TTL over CMOS?
funkyguy4000:
By functionality, you mean if you want high speed or low power consumption, chow how much faster is TTL over CMOS?
Anymore, you will not know unless you read the datasheet for the specific part. It’s no longer true that CMOS is always slower than TTL because there are so many newer packages that were never made into TTL and that run at really high speeds. Read the datasheets for the parts in question. Propogation delays are usually on the top half of the first page, you won’t have to go far into it.
There are also some voltage differences and CMOS can be run at higher voltages. In fact, some packages run faster at higher voltages. Of course if you are running your system at 10V-15V to take advantage it better be all TTL or have level converters in it. Not all CMOS chips can be run at these voltages, check the datasheet.
More information:
More information still:
http://www.kpsec.freeuk.com/components/74series.htm
http://www.kpsec.freeuk.com/components/cmos.htm
http://www.ti.com/lit/ds/symlink/sn74ls595.pdf http://www.ti.com/lit/ds/symlink/sn74hc595.pdf
An example - look at the tables for Tpd from SRCK to Qa-h. LS595, max is 18/25nS for High & low outputs HC595, max is 40nS.
With 16 MHz clock of 62.5nS the slower part works. With 32MHZ clock of 31.25nS, the slower part would not.
So choice of part depends on situation.
Interesting, so is TTL being zoned out?
so is TTL being zoned out?
No, it is not only about speed but drive capacity. In general you can get more current out of TTL, there are TTL devices that do not have the equivalent CMOS parts so functionality is an issue as well.
The fastest TTL will still go faster than the fastest CMOS. It is just that the slowest CMOS is not as slow as it used to be.
Interesting What are some of the other more common factors? Just like a list that I could research on my own.
funkyguy4000: Interesting What are some of the other more common factors? Just like a list that I could research on my own.
Several factors help one decide which IC family type were best for any one specific application, power consumption per function, input noise immunity, Vcc operating voltage range, etc.
TTL logic using npn/pnp transistors were 'king of the hill' for many decades for most applications, but gradually CMOS logic using mosfet transistors improved where they are the dominate tech used these days.
Lefty
Here's a good chart showing how the different logic families can play with each other.
Oh that is a great read!
The chart of the logic voltage thresholds is quite intriguing. Especially the LowV one. Is that what the devices that take in 3.3v logic are or is that something totally different and unrelated?
For 3.3V devices, the levels are typically a factor of the supply voltage.
For ATMega328s:
With Vcc = 3.3V:
Low Input must be <= 0.3 x Vcc, = 0.99V
High input must be >= 0.6 x Vcc = 1.98V
1V to 1.97V: input is not defined, may be seen as high or low.
With Vcc = 3V:
Low output will not higher than 0.6V with 10mA load
High output wil not be less than 2.3V with 10mA load
Numbers are from Section 29 of the '328 datasheet.
Okay so its pretty much unrelated. Thank you so much everybody!
If you remember anything or have anything more, feel free to throw it in here. Otherwise, thank you
Of course CMOS scales to much faster clock rates with smaller processes and lower voltages - a 1.8V CMOS logic family would be interesting.
I have an old Sci Am special on microelectronics, published some time in the 80's - they forecast that I2L (integrated injection logic) would take over from MOSFET technology. Never happened!
The other feature of TTL that ought to be mentioned is that the static power dissipation of gates is many orders of magnitude higher than CMOS (factor of millions I think) - this means TTL cannot be used for VLSI at all since the quiescent power dissipation would be measured in kW and MW. A chip with 50 TTL gates on it is feasible, with a million gates: totally impossible.
Aren't FTDI chips TTL? Such as the one and the arduino Duemilanove?
funkyguy4000: Aren't FTDI chips TTL? Such as the one and the arduino Duemilanove?
You mean this part?
http://www.ftdichip.com/Support/Documents/DataSheets/ICs/DS_FT232R.pdf
Though it interfaces CMOS to TTL, I can't see it being TTL. Maybe BiCMOS? The datasheet does not what technology the chip is made from.
funkyguy4000: Aren't FTDI chips TTL? Such as the one and the arduino Duemilanove?
I think you would be hard pressed to find any mass produced IC developed in the last 10 years (perhaps even longer) that is bipolar (as pure TTL is based on) tech, rather then the various CMOS derived tech. Once they got the speed and output drive to be competitive with bipolar TTL, the other overwhelming features of CMOS made it the mainstream tech to use in IC development. I think newer bipolar IC designs are probably only used in few low volume specialized niche applications, it at all?
cmiyc: I would suggest starting here: https://www.google.com/search?q=difference+between+TTL+and+CMOS
Couldn't resist the thread necromancy to say that's almost exactly how I found this page, 7+ years later :-) (excepting I included the term 'Arduino'). Thank you all (belatedly) for your insights.