Radio Link Performance

It has been fine here for the past three days.
The room where the emon base is located is our study and it has been occupied a lot lately with working from home during COVID restrictions.

It’s hard to say. I don’t think so. I have been using the internet for the past 4-6 hours and the longest drop out has been approximately 2 minutes. There have been 20min drop outs earlier in the day when the internet was not being used.

The big antenna is pointed to the ISP’s tower approximately 4 km away. The small antenna is one end of a point to point radio link that connects the internet service to the house. The two radios on the shed are connected via a single cat 5 cable. At the house there is another small antenna that is connected to the WAN port on the router (Cambium Networks cnPilot r195 Series Wi-Fi Router).

There is insulation in the wall - glass fibre I think. No foil.
Just to be sure I took the emon base outside this morning for 30 min or so. There was no significant improvement in the radio signal and it still experienced a 5 minute dropout.

Does your emonTx have a RFM69CW radio (there’s a ‘Learn’ page to identify it if you’re not sure, most emonTx V3.4s do have it, but I have an early one with a RFM12B) and how is it powered - via 5 V USB or the a.c. adapter?

Does the shed’s mains power originate in the house?

Do you have anything else that might be working on the 434 MHz band?

The next thing I’d try would be to kill the power to the internet radios for a few hours.
(or even a full day, if that’s possible)
If, after doing that your problem goes away, then you’d know for sure where the source of your
interference is.

Another possibility is a 3rd order product. (also referred to as intermodulation or intermod)
i.e. the second harmonic of one signal is mixing with the fundamental of another signal with the
result that a signal at 433 MHz is produced.

Difficult to track down, and given your location, I’d say quite unlikely.
But, weirder thangs have happened. :wink:

The emonTx units that I have (there are two) are only a couple of years old. When I got them I remember checking that they had the newer radios. One is in the power shed and the second is in the house. They are both powered by the ac adapter. In watching them for the past couple of days their transmission times are currently being shown as 4 seconds apart.
With respect to the origin of the power, it is in fact quite the opposite. The shed houses the batteries and inverters for our off grid power supply and the house and other buildings are supplied from the shed.
I am not aware of anything else working in the 434Mhz band. Having said that, there is an old Efergy energy monitor in third building that I think operates on 434Mhz where the transmitter is still connected. I should disconnect it as an unwanted dark load. Our closest neighbor is 1km away so interference from anything they are operating is unlikely.

Unfortunately I am going away for 4 weeks today so this will have to wait till I come home.

When you get back…

If you have a 5 V USB d.c. supply, you could increase the power on the Shed emonTx. The later versions of rfmTxLib allow you to set the r.f. power. The default is 25 (the same power as the RFM12B), but you can increase it by 6 dB, but at a risk: if you transmit without the antenna connected, there’s a real risk that you will destroy the RFM69CW. You’ll need to edit and reload the sketch to change the power, it’s not available via the on-line configuration commands.

If you try increasing the r.f. power with only the a.c. adapter supplying the emonTx, it’s liable to crash each time it tries to transmit. (The power supply inside the emonTx is designed to not draw the current needed by the radio from the adapter, to avoid too big a dent in the waveform it’s trying to measure.)

Did you mean the house is fed from the shed? Either way, there’s an a.c. mains cable between the two? What I was thinking, could you have (say) a Pi Zero in the shed on the end of an Ethernet powerline adapter, with the other end in the house? Those aren’t my favourite devices, but they are another possibility.

I might leave increasing the transmitter power until I have checked some different locations for the emon base - but thank you for the suggestion. I got rid of the small gum tree in the direct line of sight. When I get back I should have some extended data to look for more patterns in.

Yes I did. Post edited.

There is about 150m of cable between the house and the power shed. The power shed feeds the first building we lived in and the new house is fed from the first building. I did try an Ethernet powerline adapter previously but could not get it working reliably either. The link was not stable and hence the direct radio link which does appear to work well.

I meant to add, you could also replace the antenna you’re using with a Yagi. That would give more power in the direction it is pointing, at the expense of everywhere else (which you’re not interested in at the transmitter).

e.g.
https://au.element14.com/lprs/yagi-434a/antenna-yagi-7-element-434mhz/dp/2096215

(about 9 dB more on-axis for this one)

I did wonder about the efficacy of that approach and was hoping that the RF experts out there would recommend this approach if it was worthwhile. Again, something to think about when I come back. I might have an antenna real estate problem though - the yagi is much bigger than the simple monopole.

I ran into the ISP representative this morning. He told me that the big radio is operating at 2.4 GHz and the little one at 5 GHz. A scheduled upgrade will see the big radio move to 5 GHz in the near future.

but it does not look like it had any significant effect.


I might set up a camera to see if there is anything significant happening during the drop outs when I come back.

Sorry coming in at a different angle here, but something I was interested in doing a while back was finding someone capable of doing impedance matching tests on the rfm69 modules. It’s been discussed before that they’re probably not a 50 Ohm output on their own.

I didn’t mention that because, at that size, it was unlikely to make an difference. Covered in wet foliage and/or full of sap and fully grown, it could have been a different story.

Indeed, although you don’t necessarily have to have 7 elements - that’s the first commercially available one I found to illustrate that they were available. The smallest you’re likely to get will be 3 elements (back to front: reflector, driven element, director, so roughly ¼ m in length overall), but that won’t be quite so directional and thus the on-axis gain will be less.

@danbates
From the data sheet:
2.4. Module Consumption
The tables below give the electrical specifications of the transceiver under the following conditions: Supply voltage VBAT1=
VBAT2=VDD=3.3 V, temperature = 25 °C, Pout = +13dBm, 2-level FSK modulation without pre-filtering, FDA = 5 kHz, Bit
Rate = 4.8 kb/s and terminated in a matched 50 Ohm impedance, unless otherwise specified.”

also from section 2.4.4. Transmitter, Table 7, Transmitter Specification

“RF output power in 50 ohms”

and

3.3.6. Power Amplifiers
Three power amplifier blocks are embedded in the RFM69CW. The first one, herein referred to as PA0, can generate up to +13 dBm into a 50 Ohm load.”

These statements seem to indicate that the expected matching impedance is 50 Ω. I don’t think anyone with experience in r.f. matters expects the load to be exactly 50 + j0 Ω, but that is the design objective.

I tried shutting the power off to the internet radios today for 2 hours while there was a dropout - unfortunately no change.
I then tried taking the emon base to the emonTx and this did not help either.

So a friend of mine with an antenna analyser had a look at the feed and the antenna and helped identify that the stub antenna had a faulty connection in it. We installed another smaller stub antenna that we had handy and I will assess the performance over the next two weeks. Current receive strength is about -73dB.

Which, although less than you had before, should be adequate.
Good to hear you found the problem.

Was the problem visible? If so, what are we looking for? Because this user Issue with emonPi Antenna - #5 by glyn.hudson might have the same problem.

Interesting, were there any visual clues on the antenna? Was using the antenna analyser the only way to determine the fault?

Could we offer to send you a new antenna to test to see if this issue is inherent?

There were no visual clues. In fact the first time we put the antenna itself on the analyser it looked fine. It was only when we started pulling it about a bit that the fault was evident and then it could be induced easily.
At one stage we tried an antenna similar to this one but sized for 434MHz which worked well.


Seeing how easy it was to make, I might try making one to replace the little stub antenna that is there now.

2 Likes

Give it a go. It’s well worth the effort. You’ll find it performs better than the rubber duck.

The one pictured looks like it’s cut for ADSB. (1090 MHz)

There are variations on that theme. A VHF one I have in my loft has radial ‘spokes’ that are horizontal, rather than pointing down.

@Bill.Thomson - Do you have the dimensions and angles for the various configurations?

For operation at 433.92 MHz, the driven element should be cut to 164 mm.

With antennas, one thing to keep in mind is velocity factor. i.e. RF travels slower in a conductor
than it does in free space. For a simple wire element, the VF is ~0.95, for coaxial cable, it varies
quite a bit. Typically, it ranges from 0.66 to 0.88.

Here’s a link to a 1/4 Wave Ground Plane Antenna Calculator

https://m0ukd.com/calculators/quarter-wave-ground-plane-antenna-calculator/

From the page at the link above:

A quarter wave monopole mounted against a perfect ground will have an impedance of around 36Ω but by bending the radials down at an angle of 45°, we increase this to around 50Ω whilst at the same time lowering the radiation angle more towards the horizon. (42° is the theoretical perfect angle for 50Ω feed, but who’s measuring!)

I was actually repeating something @Bill.Thomson said to me a while ago, but he hadn’t repeated it in this thread, so I thought to do that. I’ve never seen anyone here test these modules for their matched load.