-136dbm is the best for SX1272. But SX1276 can have a higher sensitivity of -149dbm at 7.8khz of bandwidth (need a TCXO).
@LoRaTracker Backing up my assumptions 
Frequency Range: 902-928MHz;
Noise Figure: 1.2dB at 915MHz
Gain: 24dB at 915MHz
P1: +13dBm
IP3: +25dBm
$155 on eBay: http://www.rfbayinc.com/upload/files/lna/lna-915.pdf
Assumptions are OK, even when they are wrong.
Real world test are what counts.
Have you tested the LNA you linked to with LoRa ?
@LoRaTracker
No, this is just exploring the idea if it is possible and what it would take. Getting feedback on assumptions is part of the process, and I appreciate that you took the effort of sharing your experience.
£400 for extra 6db?..hummm perhaps ask NASA or Jodrell Bank & add a couple of Zeroes to the price?! Tee hee 
Perhaps split front ends and place at focus of dish for TX & RX and just pass I & Q in digital form (separate the SX125x from SX1301?) ? Then u perhaps save the 2 x 3db cable loss - more like 0.5-1db each???
TX/RX from top of high mountains a la stellar observatories? Reduce atmospheric absorption and overall path loss? Trip to Andes or Hawaii anyone?! (And get away from RF noise sources reducing noise floor as called out by LoRaTracker?)…at this rate might even get required TX down below legally allowed limits?! 
@Jeff-UK
For the 33 cm ham radio band in the USA, the max legal power for spread spectrum is 100 Watt.
Atmospheric absorbtion at UHF is minimal, maybe 0.001dB a km, so even from the top of Everest you would only save 0.008dB.
But again, with the receiver typically seeing -105dB of noise, how can you receive signals at -149dBm which is -45dB below noise when the maximum spec is -20dB ?
There are simple tests you can carry out that will show that you cannot get even close to the data sheet stated sensitivity, in the real world that is, due to the impact of noise. What can be achieved in the laboratory is of little consequence.
At first I will start with simplififng the setup.
Use one way communication.
The back way you can do via internet or Short Wave.
This is how HAM radio amateurs are doing this by EME comincation.
You can extend the setup later on, but first you have to proof the simple thinks.
Another option is to try with simple thinks.
- Narrow the band and use TCXO – theoretically it will give you extra 6…9dB, but no idea how those numbers translate in real life
I though the noise floor was bandwidth dependant: -174dbm at 1Hz bandwidth, -144dbm at 1KHz, -114dbm at 1MHz bandwidth… and so on.
So at 7.8kHz, I hope to see less than -105dbm noise.
The -105dBm of noise, seen by the LoRa device, seems to match the SNR figures for the bandwidths we commonly use, 62.5Khz plus.
Yes at lower bandwidths sensitivty increases but you still get nowhere near the sensitivity quoted in the datasheet.
I expect too. The bad real live numbers are comming from the LoRa main advantage.
In practice when you have -120dbm with Lora you receive many disturbances - the main signal + scattered signal + reflected signal + jammed channels - just to name few. Yes the spread spectrum of Lora mitigate them very good. Still not perfect.
That is why I expect better number with highly directional antenna toward the sky
I have only tried with a 10dB gain antenna, pointed at the sky and I detect no differance (in detected noise) between that antenna and a less directional one.
Gonna try this:
2 Likes
Good luck.
The video claims that for the coils to be 28mm ‘outside’ diameter, you should use a former of 21mm but with 1.4mm diameter wire simple sums suggest the former should be 25.6mm diameter.
Do you know what the real world measured performance of that antennas is ?
Cheap antenna here - not as good as the first ones I bought but still OK.
They are advertised as female, but I was posted male versions (!).
They are the sort that have to be modified from 915 to 868 MHz… By adding a bit of single core wire, I got the SWR down from 12 to just under 1.2
N.B. The male version is for the TTN Gateway. The female version can be used on the RAK831.
Not bothered about getting the wrong version - but I still need to buy a female one for my RAK831.
ATM, I don’t want to resolder the “correct” connector on the RAK831, so I’m having to use this adaptor to use a male antenna
Were you able to measure the gain ?
The description is hardly reassuring, was 868Mhz ever used for Wi-Fi routers ?
Yes, description is rubbish! also testing was qualitative only.
On the TTN gateway, I find these modified antenna (must be male) performed slightly better than the ones originally supplied (brief non extensive test).
One thing I did notice is that it seemed better for these antenna to be straight (i.e. not bent) - probably something to do with the antenna being less affected by nearby metal (the PCB?)
When I put them both on a Heltec (use with SMA cable) and the RAK831 (Just a female antenna or use male with adaptor), the RSSI increase was huge.
As always YMMV - just posting the links for people to know where to get them cheaply - to try out with some feedback that they worked for me 
It would be great if you could test them more scientifically using your techniques - only if you have the time of course, but it would be worth putting on your blog perhaps?