Author Topic: Send WAY Fewer Bytes  (Read 3512 times)

ChemE

  • Sr. Member
  • ****
  • Posts: 419
  • Country: us
Re: Send WAY Fewer Bytes
« Reply #15 on: December 16, 2016, 05:24:23 PM »
I've tested out moving the node further from the gateway and using powerlevel(0) and 300 kbps this is what I get:

# of preamble bytes || max range (ft)
0          ||          about 6'
1          ||          about 20'
2          ||          about 30' but RSSI is -90db b/c power level is so low

Adding a third sync byte to extend range does not appear to benefit my setup.  It may even work out that I would be better off increasing the power level a little and dropping back down to one preamble byte though I would need an oscilloscope to be able to collect the precise power consumption during broadcast to confirm or refute this theory.

EDIT: Going to powerlevel(31) and down to 1 preamble byte works perfectly fine for ranges of 100'.  It is too cold to move the node outside so I'm not going to work out how far I need to go to need the second preamble byte :)
« Last Edit: December 16, 2016, 05:30:13 PM by ChemE »

joelucid

  • Hero Member
  • *****
  • Posts: 867
Re: Send WAY Fewer Bytes
« Reply #16 on: December 17, 2016, 02:41:09 AM »
Quote
# of preamble bytes || max range (ft)
0          ||          about 6'
1          ||          about 20'
2          ||          about 30' but RSSI is -90db b/c power level is so low

That makes sense: the bit synchronizer needs 12 bit preamble to sync according to the datasheet. Interestingly I just ran a range test with 300kbit this week to optimize radio parameters and found that I could improve range by going up to 6 bytes preamble. I do not have a good explanation for this experimental observation though.

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: Send WAY Fewer Bytes
« Reply #17 on: December 17, 2016, 08:43:19 AM »
That's a very interesting result!  You're getting much longer range at relatively little incremental cost.

How much improvement is there in going from 2 bytes of pre-amble to 6-bytes?


perky

  • Hero Member
  • *****
  • Posts: 873
  • Country: gb
Re: Send WAY Fewer Bytes
« Reply #18 on: December 17, 2016, 09:35:55 AM »
I think this might be evidence that the bit synchronizer is possibly PLL based, the more transitions the more accurate the final sampling windows are. That may result in a lower BER for a given S/N.
Mark.

joelucid

  • Hero Member
  • *****
  • Posts: 867
Re: Send WAY Fewer Bytes
« Reply #19 on: December 18, 2016, 06:57:26 AM »
Quote
That's a very interesting result!  You're getting much longer range at relatively little incremental cost.

How much improvement is there in going from 2 bytes of pre-amble to 6-bytes?

I'm not sure how much. I originally measured this result a couple of months back when I first measured out the best modulation index for given bitrates. I only retested this week again because I still don't fully understand this result. So I measured that it still exists, not the magnitude.

I didn't see this with lower bitrates than 300kbit btw.

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: Send WAY Fewer Bytes
« Reply #20 on: December 18, 2016, 10:18:33 AM »
I think what's likely going on is that multi-path reflections and/or noise/interference can throw off the bit synchronizer, creating a false negative.  So, if you send multiple pre-amble bytes, it gives the receiver more chances to catch one that's "well formed," which then it uses to set the bit synchronizer and launch into further demodulation.