Hi all,
Recently I'm working on the TL16C752C UART chip and I have noticed some problem with it.
1. Data Receiving Error
Error occurs quite often when receiving data. I noticed the "Short STOP Bit Errata", and did the following tests:
- Use 8-N-1 setting for TL16C752C to receive data sent from USB UART with 8-N-2 setting, the data receiving is all right without error.
- Use 8-N-2 setting for TL16C752C to receive data sent from USB UART with 8-N-2 setting, still a lot of error.
- Use 8-E-1 or 8-O-1 to receive data in 8-E-1 or 8-O-1 format, respectively, still a lot of error.
- Decrease the divisor by 1, there is no error.
BTW, I have made measurements and confirmed that the stop bit sent by my USB UART is exactly 104us at 9600bps.
So, I think the workaround 1 (to use 2 stop bits) in the errata only takes effect when the TL16C752C is configured as 8-N-1 and the data sent to it is actually 8-N-2.
And decreasing the divisor works, because in this case the TL16C752C samples the RX data in higher frequency, which makes the stop bit appear longer. However this is not a good way, as I have to modify the driver and write in the code something like "if the UART is a TL16C752C, make sure to decrease the divisor by 1 before writing to DLL/DLH"...
I guess it's pretty sure that it is a bug with TL16C752C. Is there any new version coming to correct this bug?
2. "One Baud Time" when 485EN = 1
The main reason I use TL16C752C is the automatic RS-485 flow control feature. However, when I set 485EN bit, all the outgoing characters on TX pin appear to have 1 extra bit.
I noticed that according to the datasheet, the DTR# will be asserted one baud time before transmitting data on TX pin. But when TX FIFO is enabled and there is some data in it, I don't think there should be the 1 baud time interval between characters.
I'm sure TX FIFO is enabled when I'm sending data, and AFR = 0x14 (only RCVEN and 485EN is set, all other bits including DLY[2:0] are cleared.
Below is the captured waveform. Channel 1 (yellow) is DTR#, channel 2 (blue) is TX. The TX data is in 8-N-1 format, but because of the "1 baud time", it looks like 8-N-2. And DTR# goes low for a short period after the first character, but stay high until all the characters are transmitted.
The correct 8-N-1 waveform should be like this (captured when 485EN = 0):
Any suggestions regarding this problem? I'm not 100% sure that I have completely understood the behavior of DTR# in RS-485 mode.
3. Where is DLY3
I guess this is a minor mistake in the datasheet. It says, After both are empty (after the last stop bit of the last data byte), the DTRx output stays high for a programmable delay of 0 to 15 bit times, as set by DLY[3:0]. However, there is no DLY3 bit in the AFR, only DLY[2:0] so the programmable delay can only be 0 to 7 bit times.
Anyone can confirm this? Or is DLY3 hidden somewhere that I did not find?
Thanks in advance for any comments.
Best regards,
Albert
