I just wanted to give everyone who is waiting for a 3DR 433MHz radio a heads up about a shipping delay. A few days ago I discovered an interference problem with the 3DR 433MHz radios that can cause the range to be much less than expected. We're still investigating the problem, but meanwhile I've suggested to Sam that they delay shipping the 433MHz variant of the radio until we've solved it.
Our current theory is that the 480MHz clock in the USB host controller on some PCs is the source of the noise, and that 480MHz is close enough to the 433 MHz radio band to cause the problem. This theory is supported by testing with the USB host controller forced into USB 1.1 12MBit/s mode ("full speed") instead of the more common "high speed" 480MBit/s mode.
The above graph shows a particularly extreme version of the problem on one of my laptops. This graph was produced with a modified firmware that uses the Si1000 as a simple scanner, looking at the background signal level at frequencies from 420MHz up to 950MHz. The graph shows the background noise levels in "433-scan.txt" (the red dots) for a USB version of the 3DR 433 radio. The green dots are for the FTDI version of the radio connected to the same laptop. Those noise levels are far too high, and would severely limit range when connected to this laptop.
We have found some workarounds - in particular running the USB host controller in 'companion' mode, where it forces the controller to run all ports in "full speed" 12MBit/s mode. That helps, but we are not very happy with the solution, and the noise is still higher than we would like.
So we are still investigating the best solution and meanwhile I've suggested to 3DRobotics that they hold off shipping the 433MHz radios until we have properly solved the problem.
The 900MHz radios are unaffected by this and we've got great range on those with lots of different ground station laptops.
I'll keep you all updated via this forum as we make progress on the noise problem. Meanwhile all I can do is apologise for the delay. We thought it was better to ensure the problem is properly fixed rather than having some people disappointed in the range.
This weekend we have some time to look into the 433MHz radios to see what we can find. Some early testing from last weekend with Tridge indicated the problem was radiated. Disconnecting the antenna from the unit reduced the noise levels completely. This means that the power supply regulation is OK and that the noise path is not going internally through the board (+5v rails, USB lines on the pcb etc). It does however point to radiated coupling from the cable connected to the PC.
Some of the things we will be looking at include:
Noise profiles on all lines (+5, D+, D-,GND) coming from the PC with a FET probe.
Ferrite beads/chokes. --- the opto fiber cable above looks cool!..
Changing the antenna to be ground independent (1/2 wave sleeve dipole) to reduce coupling of noise from the board.
Shielding of the unit.
Will post an update on what we find....
Thanks for the update Seppo, I'm looking forward to your test results!
we are too looking foward for solution......
Had a chance to see what was going on.
The USB data lines are definitely noisy through the 433MHz band.
The level of noise on the usb lines is in the order of ~-65-> -70dBm, especially during data transfer across the bus. Background noise levels when probing (with a fet probe) on the usb cable shield with the USB radio unplugged is around -95dBm.
Using a FET probe without a ground clip connection, essentially couples any field between what you're touching with the tip and the ground of the probe. It is not an absolute measurement, but a pretty good indicator of relative levels of noise when used like this.
Power output of both modules is +9.8dBm and +9.2dBm which is as expected.
A sensitivity test was done with connecting the modules together via double shielded coaxial cable and a lot of attenuation (incl a step attenuator to increase in 1dB steps). One module was placed into a shielded, diecast aluminium box and taped shut with aluminium tape to reduce the effects of radiated coupling. The module in the box was powered by a battery with no other cables coming out of it.
With the default settings, an attenuation of 112dB started to increase the error correction count slowly - this is not super accurate, but gives a pretty good idea of where the limit is.
The radio settings were default, with an air rate of 64kbps and 10 hops.
Taking out the transmitter power of approx +10dBm from the path loss of 112dB, gives a sensitivity of around -102dBm @64k which seems reasonable.
Looking at the non-usb version running off the FTDI cable, the noise levels are much lower, and probing the individual serial lines going into the module, noise levels are around -95 to -100dBm.
Tridge mentioned that he might have a chance to look at the 433MHz module running off an FTDI cable for a range test shortly.
I did have a look at the antennas as well quickly, with the antenna on the end of a cable, measuring the return loss gave around -3dB, which means around 50% of the power is getting reflected at the antenna interface !.. Did some quick experimenting and touching the shield of the feed coax cable with a metal handled scalpel (closest metal thing I had) Length=13.5cm, the return loss improved to around -10dB, which is nice. This means that there is definitely room for improvement with the antenna that gets used. Touching the ground with a piece of metal gave the antenna another half essentially making a dipole.. (the coax is still radiating as well, but it's better than before)..
I havent had a chance to range test things yet, it's late and dark here in Brisbane, but there are a few things to try. The module itself looks quite reasonable and I think with the combination of the FTDI based cable for reduced noise and a nicer antenna range should improve!.. ( fingers crossed :) )
.....and keep fingers crossed......
Thanks for the detailed analysis Seppo!
I did a flight test using two FTDI radios today and had great success. The difference between using a FTDI radio on the ground and using a USB radio is huge.
The setup I used this time was two 'air' radios. On the ground station I had the antenna taped to a fence like this:
as you can see, the radio kept a good link to the limit of what i was comfortable flying such a small plane within line of sight. To get a real idea of range I then dropped the transmit power in the two radios to 2dBm, which is 12dB below the legal maximum for Australia, and 8dB below the legal limit in Europe. I also setup the radios with a DUTY_CYCLE of 10%, which is required in Europe. This was the result:
It lost link briefly at 600m range, but this is with 1/16th of the power allowed in Australia, and 1/6th of the power allowed in Europe. That shows it is capable of very good range (around 3x the range it achieved here).
One thing I noticed is that with the FTDI cables the noise levels in the plane were actually higher than on the ground. Perhaps the lack of a ground plane in the air makes the noise worse? The ground noise was much more consistent, and it had the fence as a ground plane.
The other thing I noticed is that the 10% duty cycle caused a noticable lag in the reporting of mode changes from the ground station. I think the solution in Europe will be to run it with a 20% or 30% duty cycle, and take advantage of the fact that the duty cycle is counted over an hour, so if you know you only fly for half of each hour, you can use a 20% duty cycle.
I think these results show that with the FTDI 'air' versions of the radios that the range is above what we were aiming for (we initially wanted 1 mile when the project started). So I'm going to recommend to Sam that they consider shipping the 433MHz radios just as 'air' radios until the noise issues with the USB version is completely solved.
Excellent report and good news. The FTDI solution is a good one for now and many of our users have FTDI cables lying around (we also sell a nice long one for $18)
What are the main pros and cons between the 900 and 433 and how do they compare to the Xbee? I am considering buy either the 900 or 433 and here there are no restrictions so I could use either.
So firstly (appart from cost) how do they compare to the Xbee for speed and range.
Which out of the 900 and 433 would be the best choice and why?
John: both should outperform the comperable Xbees in range and packet loss, since they're optimized for MAVlink. Speed is not an issue, since that's just something you set. We find that 57k is the best trade-off between throughput and reliable range, on both the Xbees and 3DR radios.
I've never used the 433s, but the 900s are the standard in the US and offer much better range than higher frequencies, such as 2.4 or 5.6.
Chris Im currently using 900 Xbees here. When you say the range is a lot better, what is the distance your getting with the new 3DR900 your using?
As Chris said, both should perform well. The key is which fits into your local radio regulations best.
If you really had no regulatory restrictions (which would be very unusual!) then the best range would be from the 433 radios, as they have a lower 'path loss'. So a 433MHz radio will in theory get about twice the range of a 900MHz radio for the same transmit power and receive sensitivity. That is why HAM radio operators will often prefer lower frequencies.
In practice you will have local regulations that change the choice. Where 900MHz radios are allowed they usually allow for much higher power levels than the 433MHz band. For example, in Australia you can transmit on 915MHz with 40 times as much power as you can use at 433MHz, so the 900MHz radios are clear winners in Australia. The same is true in the US, where you can transmit over a wide frequency band at up to 1W in the 900MHz band.
Another key factor is how much noise you have. We're had a lot more trouble with noise from the 433MHz radios than with the 900MHz radios, particularly because the USB bus in the ground station operates at 480MHz, which is close enough to the radios frequency to cause interference. This noise has a much bigger impact on range than the 'path loss' advantage of 433MHz.
Regardless, we expect you will be happy with either one! Just make sure you choose the one that fits your local radio rules the best.
Thanks for the in depth explanation Andrew. This is a big help.