FCC rules for frequency hopping

[perky]

I'm a little confused by the FCC rules for frequency hopping. You must use at least 25 channels if your transmit bandwidth is 250kHz or over, and 50 channels if less. There ia a max dwell time specified of 400ms for both, so each channel cannot tranmit for longer than this, and this gives a 20 second cycle time for 50 channels and 10 second cycle time for 25 channels.

So that in effect gives a duty cycle of 2% for each channel for 50 channels, and 4% for 25 channels. Presumably if you only transmitted for 40ms per channel the cycle times would be 2s for 50 channels and 1s for 25 channels, it would still meet the spec when averaged over 20s and 10s respectively.

There doesn't seem to be any restriction on whether you actually have to transmit something at any given time, so what if you elected simply not to transmit on 42 of those 50 channels and only use 8? How is that any worse than using all 50 channels (assuming you don't break the duty cycle rule)? I don't think there's any rule about the pseudorandom sequence you choose either. So I'm confused why it is specified this way and not by duty cycle.

[joelucid]

so what if you elected simply not to transmit on 42 of those 50 channels and only use 8?

I think you have to use all frequencies because of this sentence in the regulation:

"Each frequency must be used equally on the average by each trans- mitter."

Joe

[raggedyanne]

Frequency hopping schemes are often referred to as frequency hopping spread spectrum (FHSS) To
be considered as a compliant for a frequency hopping system the device or system must meet the
following requirements:
1.  The channel carrier frequencies must be separated by a minimum of 25 kHz, or the 20 dB
bandwidth of the hopping channel, whichever is the greater.
2.  The  channel  hopping  frequencies  which  are  selected  at  the  system,  hopping  rate  must  be pseudo-random  in  nature.  On  average,  each  channel  hopping  frequency  must  be  used
equally.
3.  The receiver bandwidth should match that of the transmitter and will hop in synchronization.
4.  If the 20 dB bandwidth is less than 250 kHz, the system shall use at least 50 channels. The
average  dwell  time  on  a  particular  channel  shall  not  exceed  400  mS  within  a  20  second
period. If the 20 dB bandwidth is 250 kHz or greater, then the system shall use at least 25
channels. The average dwell time shall not exceed 400 mS within a 10 second period.
5.  The maximum allowed 20 dB bandwidth of the hopping channel is 500 kHz.
6.  For  systems  employing  at  least  50  channels,  the  maximum  peak  conducted  output  power
output  is  +30  dBm  (1  W).  For  systems  employing  less than  50  channels  but  at  least  25 channels, the maximum output power is +24 dBm (0.25W).
7.  If the antenna used has a directional gain in excess of 6 dBi then the conducted output power
shall be reduced by the amount in dB that the directional gain of the antenna exceeds 6 dBi.
8.  In any 100 kHz bandwidth outside the frequency band of operation the power shall be at least
20 dB below that in the 100 kHz bandwidth within the band that contains the highest level of
the desired power.
9.  Radiated harmonic and spurious emissions which fall within the restricted bands, as defined
in  FCC  Part  15.205  must  comply  with  the  radiated  emission  limits  specified  in  FCC  Part
15.209.

[john4444]

Hi Perky,

ISTM the intent is to spread all the users out across the band.
If you choose not to transmit on a particular channel but follow the sudo-random  channel hopping,
I don't see that there would be an issue.
However, if your intent is to reserve a couple of channels for your own use, or only transmit on particular channels,
ISTM that would be against the intent.

John

[perky]

I thought my interpretation had to be wrong, thanks all for pointing out the flaw ;-). This does lead to a strange situation where it's perfectly OK to transmit data on all 50 channels, but not transmitting on some of them would would actually be a better situation isn't, but I suppose if they allowed it you could get more uneven spread of spectrum usage.

Problem I'm facing is I've designed a frequency hopping system for 868MHz using 8 channels and 25mW with < 1% duty cycle for a battery powered system. This means transmitting for, say, a short packet every 6 seconds. That would take 48 seconds for a fully cycle. I would like to extend that to 915MHz, but it appears I may have to use at least 50 channels (or 25, but then range might be compromised due to > 250kHz bandwidth criteria) which would extend that to 5 minutes for a full cycle. The number of channels isn't a problem in itself, it's how to get the receivers to lock to the sequence quickly and efficiently as they are also battery powered and holding them in receive mode for a full cycle or more is not particularly good, either for battery life or end user experience.

I've seen a reference in a document somewhere to being able to use one, or a small set, of frequencies for synchronization only and not for data, but haven't been able to find any official FCC guidelines for this approach. Ideally I'd like say 8 channels only for syncing (just in case the one I chose is blocked).

Mark.

[joelucid]

The number of channels isn't a problem in itself, it's how to get the receivers to lock to the sequence quickly and efficiently as they are also battery powered and holding them in receive mode for a full cycle or more is not particularly good, either for battery life or end user experience.

I've just gone through this exercise myself. I think instead of having the server collect clients once they fall out of sync it's much better to have the clients send bursts of packets to find the correct channel of the server.

Sending is more costly than receiving so why is this better? First it doesn't require the server to detect when clients fall out of sync. Then the server can continue to be reached during synchronization. And finally clients can select a sync strategy that matches its power supply limitations - so a coin cell node might send a packet every ten seconds and take 10 minutes to join, but at least it can join.

I used to think control channels might work. But I've since read that low power Bluetooth does not qualify as frequency hopping system because it only uses 3 sync channels.

Joe

[perky]

Unfortunately my system is entirely battery operated, even repeaters. This means the main controller server and the repeaters are asleep for the vast majority of the time. This locking thing is actually quite a pain with battery systems :-(
Mark.

[joelucid]

This locking thing is actually quite a pain with battery systems :-(
Mark.

Yeah, I suppose you could switch the server/repeater into some listen mode variant during locking and work with long preambles. But this scenario does seem like a tough fit for freq hopping. Have you considered LoRa on only a few channels? Say 3 channels to make locking easy and use 500khz LoRa which is inherently FCC compliant.

Joe

[perky]

I haven't looked at LoRa in detail. I'm a bit concerned about the power needed and the time for transmission. I realize that the RF power could be much lower to get the same range as normal FSK, so what are the tradeoffs? Can you get the same distance as narrow band FSK for the same battery power?
Mark.

[joelucid]

I have yet to try LoRa but I think because of the coding gain you'd get better range at the same power level. If you use 500khz fdev at spreading factor 6 for example you'd get 37500 baud and a rx sensitivity of -116 dBm - hard to beat with FSK.

Since the signal is so broad band it already meets the FCC regulations without hopping. You can add hopping on a few channels to avoid getting jammed.

You pay for it all with higher radio costs, but if both sides are on battery it might be worth it to avoid the sync problem.

Joe

[perky]

Thanks Joe, this may end up being the solution. I clearly need to get up to speed with these LoRa things ;-)
Mark.

[raggedyanne]

You do not have to transmit on every channel. Each radio must evenly switch between each channel an allocated period of time, so if ch_1 operates in a 20s period so must every other channel.

So regardless if awake or asleep and not receiving or transmitting the channels are being equally allocated.

If in doubt contact the technical department of the governing body, it is free, clear & concise.

[perky]

You do not have to transmit on every channel. Each radio must evenly switch between each channel an allocated period of time, so if ch_1 operates in a 20s period so must every other channel.

So regardless if awake or asleep and not receiving or transmitting the channels are being equally allocated.

If in doubt contact the technical department of the governing body, it is free, clear & concise.

Can you point us to links to the regulations that say this, because if what you say is true then we could simply choose not to transmit on 49 out of the 50 channels and just use one of them, and it would then be a simple duty cycle restriction with up to 1 watt of power on that channel :-O

[raggedyanne]

AS in my rtc/gps example where a transmitter is changing channels but only transmitting on a single channel that relates to a current minute from 60 transmitters. Which was clarified in an email to a technical administrator of the ACMA regulatory body here in Australia. You need to contact your government regulatory body if in doubt, this is why we pay taxes.

Use this as an example and i am sure you will be pleasantly surprised by what they say, always have the information emailed to you, perky

If it is not written does not mean it is not allowed. If you can interpret the information to suit your needs, then this is what the laws mean until otherwise amended. Laws are not supposed to be ambiguous.