Well, that’s great news. Thanks for sharing.

The XMIT_START_ADJUSTMENT is a delay value I determined experimentally. You add it to the delay value when transmitting. At 4800 baud I wouldn’t think this value would matter at all: take your pick!

Mikal

After spend the whole day testing it’s reliability with 19.6608 MHz. It’s working excellent!!! Thank you very much.

BTW, What is #define XMIT_START_ADJUSTMENT ? Do I need to use 4 or 5 with 19.6608 & 20 MHz?

Thanks for the interesting question. I originally used a spreadsheet to calculate the delay values for the different clock speeds. But when these turned out to be slightly inaccurate, I turned to a logic analyzer for tuning. For Arduino applications, this is an excellent (and cheap!) one: http://www.usbee.com/sx.html.

However, 4800 baud is a relatively slow speed, and therefore fairly tolerant of minor inaccuracies. I’d be willing to bet that, given the values for 4800 baud at 8 and 16 Mhz, you could probably calculate the 20MHz and 19.6608MHz values with sufficient accuracy. In fact, I’ll do it for you:

20MHz:
{4800, 290, 598, 598, 590}

19.6608Mhz:
{4800, 285, 587, 587, 580}

Try replacing these values in the table and see if you get the results you want!

Mikal

If you use calculation, please tech me.
If you use Logic Analyzer, please help me for 4800 baud rate.
[I don't have the expensive equipment ]

Can you tell me what flavor of Arduino you are using and at what baud rate? Is it reception that is corrupted or transmission? My expectation is that on a 16MHz processor you should see perfect transmission up to 115.2K baud and reception up to 31.250K baud. At 38.4K and 57.6K, you’d expect about 1% corruption. On an 8MHz processor these numbers are lower.

All older versions are available back to rev. 2. Just change your link, say, from NewSoftSerial8.zip to NewSoftSerial7.zip.

Please do advise if you see better performance with older versions.

Thanks

Mikal