UART & Data Rate: 8-bit vs. 32-bit

  • Thread starter Thread starter EvLer
  • Start date Start date
  • Tags Tags
    Data Rate
AI Thread Summary
The discussion centers around the tolerance of 8-bit versus 32-bit UARTs (Universal Asynchronous Receiver/Transmitters) regarding data rate differences. Participants clarify that the term "32-bit UART" likely refers to the external parallel interface rather than the actual data frame, which typically consists of 8 bits. It is noted that UART timing is synchronized to the start bit, and if the data rate is inconsistent, it can lead to sampling errors, particularly affecting 32-bit UARTs. The conversation also touches on the concept of data buffering and interrupt timing, suggesting that 32-bit UARTs may have advantages in certain contexts. Ultimately, the question posed appears to be a theoretical one aimed at understanding UART concepts, with some participants expressing confusion over the existence of 32-bit UARTs.
EvLer
Messages
454
Reaction score
0
Let's say you have a 8-bit and 32-bit uart (universal asynchr. receiver/transmitter), which one of them would you say would be more tolerant to the differences in the bata bit rate? in general...
 
Engineering news on Phys.org
I don't know what you mean by an 8-bit UART.. UARTs are serial transceivers.

- Warren
 
By 8-bit I mean that data byte is 8 bits, by 32 --- data "unit" is 32 bits.
I think it's an issue of when to sample, since the clock is much faster than the data line.
 
I don't see how the data width has anything to do with "tolerance" to the data rate.

I suppose you could argue that if the serial data rate slowed down by, say, 1 us per bit, an 8-bit word would only be "slowed down" by 8 us, while a 32-bit word would be "slowed down" by 32 us.

- Warren
 
ok, maybe a bad exam question... it was applied specifically to the transmitter, which did not make sense to me, that's why i asked ... thanks anyway ...
 
I don't know about the current generations, but UARTs used to come with a 1, 2 or 4 word receive data buffer.
Nit picking aside they were referred to as 8, 16 and 32 bit UARTs.
The 32 bit UART was more tolerant of interrupt timing when the stored data was transferred to system memory.

Perhaps the question referred to this.
 
Here is what I got after reading some various stuff on the internet...
UART timing is synchronized only to the start bit of the incoming data, so if the data is a bit too fast or too slow, UART might end up sampling on the data bit transitions, so actually 32 bit UART is less tolerant.
 
But... doesn't every byte have its own start bit, even in a "32-bit" UART?

- Warren
 
chroot said:
But... doesn't every byte have its own start bit, even in a "32-bit" UART?

- Warren

Yeah, 7- 8- (and 9-) bit frames are the only ones I'm aware of.

http://en.wikipedia.org/wiki/RS-232

I'm still confused by the OP's mention of 32-bit UARTs. All that I can think of is that he means the outside parallel interface to the UART -- different bus widths for different memory configurations and uCs. But I think you're correct, that the actual frames (start bit, stop bit, parity, etc.) are confined to basically one byte of data +/- 1 bit.
 
  • #10
Those were my thoughts, berkeman -- that the "32-bit" moniker only refers to its external parallel interface -- and thus that it has no bearing at all on its tolerance to serial timing issues.

- Warren
 
  • #11
ok, sorry if I confused anybody... in lab we did program an 8-bit UART. And this whole deal with 32 bit frame is a theoretical question that was intended to test our conceptual understanding of the UART and the fact that if we sample too soon or too early it has affect on the sampling of the following bits, i.e. one may land in the transition between bits rather than in the middle of the bit.
 
  • #12
So, is it a trick question?

- Warren
 
  • #13
yah... prof's "trick" question :rolleyes:
and seems like 32-bit UARTs don't even exist based on what you guys say...
 
Back
Top