Two Notecard types and 4 antenna types tested!

Following on from this exchange, I was interested to find out how the WBEX Notecard (based on the Quectel B91-EX modem, LTE cat-1) faired compared to the NBGL-500 (based on the Quectel BG95 M3 modem, LTE-M) version in Australia. I set up an experiment testing the effect of modem (2 types) and antenna (4 types + diversity) on signal strength. The results were surprising (at least to me) so I thought I’d share them.

I ran the experiment with 2 x WBEX and 2 x NBGL-500 Notecards each syncing with Notehub at the same time, repeating every 15-min with the average calculated after 3-6 time periods per antenna type. Signal strength and quality were read from the Notehub session data. The antenna types I tested were:

  1. Chip antenna (Ingnion TRIO mXTEND, NN03-310). Nominal 3.8db peak gain, 1.1dB gain@700MHz. Antenna impedance matching as per Ignion specifications and PCB layout checked by Ignion. Impedance optimised for Band 28 (700MHz), the most important band in my rural setting.

  2. Whip antenna with magnetic base (Yetnorson) . Nominal 7db peak gain.

  3. Log periodic directional antenna (Yetnorson). Nominal 14dB peak gain.

  4. Patch antenna stick-on. (Taoglas FXUB63). Nominal 5dB peak gain.

For the latter 3 antenna types the experiment was run with and without the diversity antenna on one of the WBEX notecards with the chip antenna connected to the diversity connector. The results are summarised for the RSRP parameter rather than RSSI, as the former is the received power in the bands of interest and the latter, total received power (incl noise). I know signal quality is important too but RSRQ did not vary much within each antenna group so results are not presented. Error bars in the graphs are standard deviations (n=2).

The conclusions I drew from the results are:

  1. The WBEX Notecards performed MUCH worse than the NBGL-500, irrespective of antenna type! For the chip antenna the difference was 4.3db, for the whip 5.9db and the patch antenna 4.4 db. For the log periodic antenna the difference was a staggering 15.1db (or ~5x the signal strength!) between Notecard types.

  2. The diversity antenna option of the WBEX made virtually NO difference to signal strength irrespective of antenna type!

  3. By far the best performing antenna was the log periodic antenna (no surprise) but the magnitude of the difference was an eye-opener. Interestingly, the WBEX with a log-periodic antenna was only ~3db better compared to the whip and chip antennas, but with the NBGL-500, the signal strength was massively improved as noted above.

  4. The patch antenna performed quite well and was on par with the whip antenna and better than the chip antenna. It should be noted that the patch antennas were >10cm away from the PCBs. If it is mounted close (<15mm) to the PCB copper pour (e.g. mounted inside an enclosure, the efficiency will be destroyed!). See this research paper.

  5. The chip antenna was ~5db worse than the whip and 6db worse than the patch antennas (i.e. ~2x lower signal strength!).

I hope anyone else struggling to make sense of Notecard types and antenna types finds this useful. Bear in mind horses for courses – for an indoor, or high bandwidth use-case, the WBEX may have an advantage, but for my application (outdoor and small bandwidth), the NBGL-500 is the go. If I’ve missed or misconstrued anything I’d be happy to hear?

@rberkelm Thank you very much for sharing this information – this is very comprehensive! A few questions:

  • You mentioned rural area, could you please describe the overall terrain/structures and distances or possible interferences between your test bed and the towers?
  • Did you also run a WBEX test with patch + patch vs. WBEX patch + chip?

Very much looking for the analysis of your results by Blues:

  • Is it a tradeoff between “diversity” (meaning automatic alternate frequencies) and signal strength?
  • Are the other factors to consider, other than the ground plane discussed by @rberkelm ?

My use case, and I suspect also for @rberkelm, is getting data in/out reliably after something has gone wrong with the mobile carrier – sometimes for many hours – timely or speed not a priority.

Hi @Francois - thank you. My test bed was outside and next to my steel shed. There are hills, valleys and a lot of trees between me and the nearest tower 2.9km away (as the crow flies). For reference my flagship android phone reports an RSRP of between -105 to -110db most of the time, so definitely in the marginal range.
I did some extra tests for you using the patch + patch diversity antenna on a WBEX and patch antenna by itself on another WBEX. Repeated 6 times at 15min intervals. Did the same for the whip antenna, just for completeness. Results are:
extra diversity tests2

So, no real difference for the patch antenna but a reasonable difference for the whip antenna, but with higher variance.
I’m not sure exactly how Quectel implement their diversity antenna. One way is by selection or switching where both signals are presented and the processor simply picks the best. Another method is some smart combining of the signals, but from what I understand, that is more power hungry. Based on what I see from my data, I suspect the former. So, if you are in a reasonably non-reflecting area, you may as well present the best antenna in the first place? But if that is your use-case, The NBGL-500, LTE-M option seems to be far superior in terms of signal strength.
That said, I’m just a babe-in-the-woods when it comes to antennas so, like you, I’d also very much like to hear from those more familiar with these matters.

PS. I said in my first post that the WBEX Notecard was based on the Quectel B91-EX modem. It is of course an EG91-EX… apologies.

Hi @rberkelm – thank you for running those tests and your description of the environment.
Your Quectel implementation question is right on:
Option A (selection of the best signal): useful when the edge of coverage present changing signal patterns. Perhaps in a mobile scenario? Environmental changes? In my case my devices do not move, few towers, two carriers. Not much changes except weather and tower traffic.
Option B (combining signals): would that really reduce RSRP (that’s above my skills…) or just add performance to data traffic? In my case, power is not an issue as my devices are permanently powered 5vDC.
Unfortunately whip is not an option in my scenario, patch only. There are more expensive (and larger path antennas), so it raises the question – instead of buying two patch antennas, would it make more sense to purchase a larger higher gain model?
Let’s hope Blues or someone else in this forum can shed some light on all this!

This is very interesting research @rberkelm! One key piece of data that is missing though is whether you’re connecting to the same tower and same tower antenna with each test. Can you try capturing a trace log (documented here) before the Notecard establishes its initial connection so we can look at the band, channel, carrier mcc/mnc/TAC/cellid, and access technology (GSM vs LTE vs 3G) used for each test case.

Thanks!

Hi Rob
No I didn’t control for which tower the Notecards were connecting to. If I had done so, the effect would’ve been to reduce variation which in turn would’ve made the result even more significant (statistically anyway). The conclusions don’t change, but granted, you would get more context.

instead of buying two patch antennas, would it make more sense to purchase a larger higher gain model?

You mean a higher gain omni-directional antenna? I think as a passive antenna, there is only so much “gain” that can be had from omni directional. Unless you are thinking an antenna amplifier - you said you are permanently powered - do you have mains? Otherwise a directional antenna is about the only thing that is going to give you a substantial improvement.