Independent tests prove lack of frequency hopping with XPS

[Kiwi]

Bruce,

I have a question that will likely show just how unknowledgeable I am with some of this.

On my scanner the Y axis starts at -100 and ends at 0. Its calibrated in 20 dBm increments.

How do I relate the wattage of the signal being received from that. Would Zero be 1 watt or is it not relevant.

You see a Spektrum transmitter placed 4 meters from the antenna on my scanner measures about -55 give or take a whisker. Is that 500 Mw or am I totally out of whack with this.

For anyone patiently waiting I have some snippets of information worth sharing I think and they can be in this thread until I get my Wi-Spy running again.

With Spektrum it always seems to look for two channels very well spaced apart. It does not look for a clear channel at either end, for example 1 and 11. But it always seems to look for something with five or six channels apart.

I had four radios running all at once last weekend and the amazing thing was they actually shared channels to achieve that spacing. For example one radio would take 1 and 6. The next would take 6 and 11. Then the third radio would take 4 and 9 while the fourth radio would grab 5 and 11.

Everything worked extremely well and there was nothing at all going on between sets. You just could not tell unless you had the equipment to see all this witchcraft going on.

The reason I have asked Xjet about the dBm levels is at first appearances it appears the XPS system is slightly more powerful. That is if what I am seeing is a true reference to signal strength.

FASST is a real beast to see working as it blasts across the scanner and really only leaves a bunch of dots to let you know where its been. XPS and Spektrum are very easy to monitor compared to the hopping frequencies.

Anyway guys I really do apologize for being slow at getting the simple stuff done. I was set to give it 100% this weekend when the darned scanner took the pip and stopped working. I'm sure I can fix it tomorrow morning when the guys are at work.

Murphies law once again.

[XJet]

	Quote: Originally Posted by Kiwi
	Bruce, I have a question that will likely show just how unknowledgeable I am with some of this. On my scanner the Y axis starts at -100 and ends at 0. Its calibrated in 20 dBm increments. How do I relate the wattage of the signal being received from that. Would Zero be 1 watt or is it not relevant.

0dBm is equal to a power-level of 1mW

That means, if your receiver antenna is receiving/absorbing 1mW of power the SA would show a 0dBm reading (the "m" refers to a milliwatt).

The scale is logarithmic which means that every multiple of 10dB is equal to a change in power by one order of magnitude.

ie: 0dBm = 1mW, -10dBm = 0.1mW, -20dBm = 0.01mW, -30dBm = 0.001mw, etc etc.

Without knowing the exact radiation pattern of the antennas you're using, the path-loss and the magnitude/phase of any reflections it's not possible to equate a dBm reading at the receiver antenna to a power input at the transmitter antenna.

That's because not all the transmitter's power will go into the receiver's antenna -- in fact only a very tiny amount will (which is why the XBeePro receiver has a sensitivity of around -100dBm (or about 0.0000000000001 watts, which is a pretty tiny amount of power). Of course you can see that a receiver capable of responding to such a *tiny* amount of power will have to work *very* hard to avoid being swamped and desensitized by signals that my be millions +60dB, or billions (+90dB) of times stronger.

However, even though you can't measure the actual transmitter power, you can still make relative measurements.

For instance, if you measure a signal reading -60dBm, and another at -80dBm, you know that the second signal is one hundredth the strength of the first one.

If you place two transmitters equally distant from the SA you should get a *reasonable* indication of their *relative* powers too.

If the first transmitter produces a reading of (say) -20dBm and the second a reading of -23dBm then the first one is *twice* as powerful as the second (remember it's a logarithmic scale so twice the power represents only a 3dB change).

	Quote:
	You see a Spektrum transmitter placed 4 meters from the antenna on my scanner measures about -55 give or take a whisker. Is that 500 Mw or am I totally out of whack with this.

It won't translate to an absolute power level at the transmitter end without factoring in a whole bunch of other variables.

	Quote:
	The reason I have asked Xjet about the dBm levels is at first appearances it appears the XPS system is slightly more powerful. That is if what I am seeing is a true reference to signal strength.

Yes, XPS has a 60mW (+18dBm) output power which is quite high. That 60mW becomes an ERP of 100mW (+20dBm) thanks to the use of a dipole antenna on the transmitter module. The dipole effectively provides a degree of power gain by concentrating its signal in directions perpendicular to the antenna itself at the cost of the amount radiated towards the ends of the antenna (which is why you don't point the transmitter antenna at the plane :-)

So if you see the XPS reading -55dBm and another transmitter placed at exactly the same difference produces a reading of -58dBm (3dB *less*) then that second transmitter is probably radiating around half the power (50mW).

[sweetpea]

Very interesting that Spektrum appears to be always 5-6 apart. Not surprised on the same freq issue though. Since all the 2.4 systems claim only partial use on the freq they most likely don't see each other as "interference".

Did XPS do the same thing or had you not tried that yet?

[Julez]

	Quote:
	Very interesting that Spektrum appears to be always 5-6 apart. Not surprised on the same freq issue though. Since all the 2.4 systems claim only partial use on the freq they most likely don't see each other as "interference".

Yes, I think with a duty cycle of 10%, sharing one channel with another system bears less risk than having 2 channels not far enough away from each other.

[NormS]

	Quote: Originally Posted by Julez
	Yes, I think with a duty cycle of 10%, sharing one channel with another system bears less risk than having 2 channels not far enough away from each other.

Spektrum uses DSSS not FHSS. Since each transmitter/receiver uses a different spreading function (based on the GUID) they can share the same channel with little to no interference, the same way CDMA cellphones share channels.

[Julez]

We are both correct, and our comments do not exclude each other.
Additionally to the code domain, which you mention, there is the time domain, which I mentioned.
There will have to be 10 Spektrum systems on the same channel, before the probability of the systems stepping on each other reaches 100%.
With only 2 systems on one channel, the probability is only 10%.
In the case, where 2 systems do not only transmit on one channel, but also at exactly the same time, then the code domain will come into play.

Cheers,

Julez

[XJet]

	Quote: Originally Posted by Julez
	In the case, where 2 systems do not only transmit on one channel, but also at exactly the same time, then the code domain will come into play.

Except in the case of XPS where all systems share a common chipping code -- which is also the reason why the XBeePro modules have a CSMA mechanism for avoiding collisions which (it appears, based on other people's test results) JD has disabled, thus significantly increasing the chances of packet collisions between XPS systems sharing the same channel.

[Four Stroker]

Do we know that each Spektrum transmitter has a unique chipping code ? Or is the GUID just a number in the data packet ?

[sweetpea]

Here is a DIY version of channel hopping using the Xbee boards.

http://www.rcgroups.com/forums/showthread.php?t=801036


I thought this was an interesting read. I would not say its ready for GS planes.

[XJet]

The big problem with channel-hopping at the level these guys are playing with is the potentially very long "holds" (periods of no data transfer) that it can introduce.

This leaves the programmer with a dillema...

Do you hop when interference levels are low (ie: *some* communications is still possible) and thus risk introducing unnecessary holds into the link, or do you wait until it's much higher and risk even longer (but less frequent) holds?

Because the switching overhead is so high, it's not a good idea to change frequencies unless it's absolutely necessary but how do you judge "absolutely necessary"?

Leave the hopping too late (once the 2-way link is lost) and the latency involved in finding a clear channel, hopping to it and resyncing with the other end can be so long as to be almost a waste of time.

When you're flying a fast model, or flying close to the ground, even a 1 second lockout is potentially disastrous.

Hop on demand sounds like a great idea but it is one fraught with compromises and potential problems.

That's probably why Spektrum uses redundancy and FASST uses constant hopping.

[sweetpea]

Actually if your read the whole thread they have gone to continuos hopping. Its slower than FASST though.

[XJet]

I notice that the DIY system in question has been built around Jenic modules and not XBeePro ones. I'm not familiar with the Jenic units so can't compare the intrinsic frequency agility they offer in terms of times to perform energy scans and actually change frequencies, however the following would apply to a XBeePro-based module configured in this way:

Continuous hopping implemented over a reliable-transport layer is prone to significant "holds" and can actually cause a whole lot more problems than it solves.

The problem is that the transmitter expects to get an ACK signal back from the transmitter and if it doesn't get one, it resends a minimum of three extra times. So if you blindly hop onto a channel with ahigh noise threshold, you're going to hit latency.

What's more if/when sync is lost between the transmitter and receiver, you've got to re-scan the available channels to find the new transmitting frequency before the link can be restored.

To make things worse -- the noise profile seen by the receiver can be radically different to that seen by the transmitter so establishing what the *new* frequency should be (for resync purposes) becomes very problematic. If the transmitter chooses a frequency that *it* thinks is quiet and starts transmitting there, the receiver may never hear it because "up there" that frequency might be obliterated by high noise levels not visible at ground level.

FASST doesn't have this problem because its receiver always knows where the next hop in the sequence is and the exact amount of time before the transmitter switches to that frequency. Even if sync is lost, all the receiver has to do is choose a nice quiet channel (at its end) and wait for the transmitter to appear on that channel. With a bodged XBeePro FH setup, this could be an unacceptable delay producing a nasty period of "no control" from the pilot's perspective.

And, with XPS/XBeePro you have a whole bunch of extra variables that can skew the effective switching time significantly which further adds to the resync/hop problems.

While it's easy to design and demonstrate a FH system with the XBeePro modules, such a system is far from robust and I certainly would never dare to fly with such a botch-job :-)

It'll work fine -- until you get enough noise or sufficient multi-pathing to provoke a resync which means it'll likely fail just when you need it most.