Any success at bit rates > 38k?

RFM12 Wireless module discussion.

Any success at bit rates > 38k?

Postby jkingdon » Sun Aug 10, 2008 10:59 pm

Hi,

I'm having a frustrating time trying to get these modules working at higher bit rates. I was wondering what success others are having at bit rates in the 80k range (I'm not trying to get into the analog only > 115k range). I had a prototype apparently working at about 80 kbits, but it turned out to be wildly sensitive to the physical configuration around the clock out pin. If I had a 4" piece of wire connected to clock out it was ok, but it received spurious packets if not. I expected to be able to turn clock out off as I'm not using it, but that makes matters worse!

Anyone had any similar experience?

Regards,
James.
jkingdon
 
Posts: 13
Joined: Sun Aug 10, 2008 10:51 pm

Re: Any success at bit rates > 38k?

Postby seaton » Wed Aug 13, 2008 1:53 am

I remember reading an errata notice, something along the lines of noise problems which was down to the CLK pin, a work around was to disable the CLK output. unfortunately I can't remember where, But I will have a hunt through my documents.

Are you just using just SPI? I don't know if you read my blog entry on the higher bitrates, but may help. http://blog.everythingrobotics.com/2008/06/23/rfm12b-bit-rates-and-throughput/

Stephen....
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Re: Any success at bit rates > 38k?

Postby seaton » Wed Aug 13, 2008 2:18 am

Here it is, which may be the cause of your problem

The clock output (Pin8) may generate high clock spurs in the RF output spectrum.
It is not recommended to use the clock output (Pin8). It can be disabled by setting the “dc” bit in the Power Management Command to (1). The next product revision is intended to solve the problem.


Have your tried disabling and tying it high?


Stephen....
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Re: Any success at bit rates > 38k?

Postby jkingdon » Wed Aug 13, 2008 4:54 pm

Hi Stephen,

Many thanks for the suggestion, I hadn't seen the errata. I had tried disabling the clock output, but bizarrely it seemed to make things worse, not better. I'll repeat the test when I get a chance. Currently I'm moving towards living with the bad data and adding extra redundancy to the command packets so that I don't get false triggers, but this hurts the latency of course.

To answer your previous question, for receive I'm using FFIT and nFFS to grab the data, so I think this is fairly optimal. On the tx, I haven't been able to get the RGIT/SDO line working, so I'm using a messy mix of reading the status register and then putting the data by nFFS. Definitely not optimal, but maybe not an issue with 2mbit SPI.

It looks to be working up to 80kbps now (but I need to do a more thorough test to see how reliable that is), but not at 115. Unfortunately, to work at 80kbps I've had to go back to a 2 byte synchron and I'll probably need a 2 byte 'fire' command, which is going to push the latency to very much the top end of what I wanted.

It may become a none issue as my zigbee modules arrived yesterday and if they work well I may just swap out the RF12s and keep them for less demanding projects.

Regards,
James.
jkingdon
 
Posts: 13
Joined: Sun Aug 10, 2008 10:51 pm

Re: Any success at bit rates > 38k?

Postby seaton » Thu Aug 14, 2008 1:50 am

out of interest, what distances are you getting at the higher speeds?

Regards,

Stephen...
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Re: Any success at bit rates > 38k?

Postby jkingdon » Thu Aug 14, 2008 1:57 pm

seaton wrote:out of interest, what distances are you getting at the higher speeds?

Regards,

Stephen...


Hi Stephen,

I just did a very quick test, with uc clock off, and using C603 for 86kbps, range still looks good. With the standard 16 byte +crc packet test I had consistent coverage from one corner of the house to the other (upstairs bedroom to basement, so through 2 floors), and out into the garden which is about 130 ft from the tx location. I didn't have time to go further to find the limits. They sure are amazing little units for the money. It wasn't so long ago I paid at least thirty times more for one way only links that maxed at 9600 bps.

Regards,
James.
jkingdon
 
Posts: 13
Joined: Sun Aug 10, 2008 10:51 pm

Re: Any success at bit rates > 38k?

Postby jkingdon » Thu Aug 14, 2008 11:28 pm

Just managed to squeeze in time for a little more testing. With 1 byte preamble, 2 byte synchron and a 2 byte trigger command latency is right around 800us at 86 kbps. Test shots are solid at 1/250s, about 1/8th frame black bar at 1/320. Better than I was expecting as I was targeting 600us. It will be interesting to see if the ZigBees can do better or not. I think it will depend on how much of the clever stuff I can turn off - there's way too much smarts going on in there :)

With the 2 byte synchron and 2 byte trigger there were no misses, and no false triggers, but it was admittedly a very quick test session over only 8 ft or so.

James.
jkingdon
 
Posts: 13
Joined: Sun Aug 10, 2008 10:51 pm

Re: Any success at bit rates > 38k?

Postby seaton » Fri Aug 15, 2008 4:37 am

jkingdon wrote:
seaton wrote:out of interest, what distances are you getting at the higher speeds?

Regards,

Stephen...


Hi Stephen,

I just did a very quick test, with uc clock off, and using C603 for 86kbps, range still looks good. With the standard 16 byte +crc packet test I had consistent coverage from one corner of the house to the other (upstairs bedroom to basement, so through 2 floors), and out into the garden which is about 130 ft from the tx location. I didn't have time to go further to find the limits. They sure are amazing little units for the money. It wasn't so long ago I paid at least thirty times more for one way only links that maxed at 9600 bps.

Regards,
James.


Great stuff, I agree on the value for money as I know what you mean. For the higher bit rates have you tried increasing the Rx bandwidth and the Tx deviation?

Stephen...
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Re: Any success at bit rates > 38k?

Postby seaton » Fri Aug 15, 2008 4:44 am

jkingdon wrote:Just managed to squeeze in time for a little more testing. With 1 byte preamble, 2 byte synchron and a 2 byte trigger command latency is right around 800us at 86 kbps. Test shots are solid at 1/250s, about 1/8th frame black bar at 1/320. Better than I was expecting as I was targeting 600us. It will be interesting to see if the ZigBees can do better or not. I think it will depend on how much of the clever stuff I can turn off - there's way too much smarts going on in there :)

With the 2 byte synchron and 2 byte trigger there were no misses, and no false triggers, but it was admittedly a very quick test session over only 8 ft or so.

James.


You can also drop down to a single byte synchron if required, but this is fixed in the hardware so will be prone to false triggers if others are operating close by.

Yes I guess with the Zigbee modules it how much of the mesh and network ability you can disable etc, but they are runniing in the 2.4GHZ so you will easily get 250kbps on them. Also you may want to try the nordic RF modules that have the shockburst mode that allow you to get a very high bandwidth out in small packets, sparkfun have them, but you could get the chips elsewhere, I have a couple at home but yet to try them. I know these will definitely be capable of the higher bit rates, but the price is nowhere near that of the the RFM12 modules.

Stephen...
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Re: Any success at bit rates > 38k?

Postby jkingdon » Fri Aug 15, 2008 1:41 pm

seaton wrote:
For the higher bit rates have you tried increasing the Rx bandwidth and the Tx deviation?

Stephen...


Yes, when I was having problems with the noise on the receiver I sent an email to Hope support. They didn't exactly answer my question, but one of the things they suggested was increasing the bandwidth and deviation, so it's now running with 400kHz bandwidth and 240kHz deviation - their suggested values.

Out of interest, where did you find the errata? I had a search around their web site but couldn't see any.

Thanks for the pointer to the Nordic chips. I thought the Microchip zigbees modules were amazing, but the Nordic ones look to be even better. I haven't tracked down any prices yet. Microchip's modules are $10 for single units, which is still very affordable for my project. Now I'm past the 'frustrated & depressed' stage with the RF12s I'm beginning to wonder if I need to spend the time getting the zigbees working. It's difficult to see that they are going to give me anything much extra for the effort (in the specific case of a flash trigger), and I already have 10 RF12s ready to use.

Regards,
James.
jkingdon
 
Posts: 13
Joined: Sun Aug 10, 2008 10:51 pm

Re: Any success at bit rates > 38k?

Postby seaton » Sat Aug 16, 2008 3:16 am

jkingdon wrote:Yes, when I was having problems with the noise on the receiver I sent an email to Hope support. They didn't exactly answer my question, but one of the things they suggested was increasing the bandwidth and deviation, so it's now running with 400kHz bandwidth and 240kHz deviation - their suggested values.

Out of interest, where did you find the errata? I had a search around their web site but couldn't see any.


LOL not from Hope, their support, much like their documentation is lacking to say the least, I had found out (through much google translations) that HOPE use Integrations chips (http://www.integration.com), obviously licensed the technology from them. Integration have much better documentation :) They have even answered a couple of questions for me (but I didn't tell them I was using hopes modules if you know what I mean ;) )

Thanks for the pointer to the Nordic chips. I thought the Microchip zigbees modules were amazing, but the Nordic ones look to be even better. I haven't tracked down any prices yet. Microchip's modules are $10 for single units, which is still very affordable for my project. Now I'm past the 'frustrated & depressed' stage with the RF12s I'm beginning to wonder if I need to spend the time getting the zigbees working. It's difficult to see that they are going to give me anything much extra for the effort (in the specific case of a flash trigger), and I already have 10 RF12s ready to use.


Let me know how you go with the Nordic Chips as far as pricing, with the nordics it looks like you will need some external circuitry, then there may be layout issues, or use a pre build one like the ones from Sparkfun, probably the only thing you will get going zigbee is the higher bandwidth (plus you get some nice things like RSSI values and 802.15.4), don't know about the range though. The higher bandwidth would be nice to get higher sync sppeds, but my camera maxes out at 1/250 so really these modules pretty well cover my needs atm, and really I think it's hard to beat the RFM12 for cost as it's fully integrated solution, no need for any laying out of circuits or additional components, unfortunately you have to go through a pretty steep learning curve and no real documentation or examples, but once you get over it then they are fantastic little modules, one of the biggest issues I found when starting is where is the problem? on the Tx or Rx side of things? ( a real chicken or the egg scenario)

Actually to tell the truth, my triggr has been on the back seat lately as I'm in the process of developing a wireless sensor mesh using these modules, these and an AVR (ATMEGA168) are a lot cheaper than the WSN mote that are available, the only problem is these are not 802.15.4 compliant on the mac layer, so currently rolling my own.
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Re: Any success at bit rates > 38k?

Postby jkingdon » Mon Aug 18, 2008 3:21 am

I found a supplier for Nordic modules, at only $3 each... in quantities of 50000. Oops :)

I also got the Microchip Zigbee modules going, but the lowest latency I could get out of them was 1ms, even after disabling as much of the high level functions as I could and sending a 0 length payload (used the sequence id for data). Looks like the RF12s win out for this application. The only doubt is that the Microchip demo software has an option for an undocumented turbo mode at 625 kbps which might make a difference if it worked. Sadly, it didn't work when I tried it. At 250kbps the modules were solid throughout the house (longer distances will have to wait), so it looks like they'll be great for projects with larger packets which can benefit from the throughput.

James.
jkingdon
 
Posts: 13
Joined: Sun Aug 10, 2008 10:51 pm

Re: Any success at bit rates > 38k?

Postby jkingdon » Tue Aug 19, 2008 1:20 am

Ok, sorry to be swamping your message board, but I thought it worth updating from my last note on the Microchip Zigbee units. I left a question with Microchip support and they got back to me in a matter of hours. With the extra information I was able to get the 625 kbps mode working and also disable the CSMA/CA function (listen before tx to avoid collisions). This got the latency down to 840us, which is better but still not as good as the RF12. I think I'll save the Zigbees for something with bigger packets for them to get their teeth into. At 625 kbps I still had range throughout the house, but at maximum distance there were signs of dropped packets depending on module orientation (they have a major dip along the central axis). Still not bad range for the speed and power, and given that the house is flooded with wifi and dect phone signals. I think these are going to be really useful modules for the right project.

James.
jkingdon
 
Posts: 13
Joined: Sun Aug 10, 2008 10:51 pm

Re: Any success at bit rates > 38k?

Postby seaton » Fri Aug 22, 2008 4:36 am

jkingdon wrote:I found a supplier for Nordic modules, at only $3 each... in quantities of 50000. Oops :)


Dang!!! Price is right just not the QTY bugger

I also got the Microchip Zigbee modules going, but the lowest latency I could get out of them was 1ms, even after disabling as much of the high level functions as I could and sending a 0 length payload (used the sequence id for data). Looks like the RF12s win out for this application. The only doubt is that the Microchip demo software has an option for an undocumented turbo mode at 625 kbps which might make a difference if it worked. Sadly, it didn't work when I tried it. At 250kbps the modules were solid throughout the house (longer distances will have to wait), so it looks like they'll be great for projects with larger packets which can benefit from the throughput.


Well done, I was looking at ordering a couple of these a few months back, but thought better of it as I already don't have enough time as it is. Are these 2.4GHz?
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Re: Any success at bit rates > 38k?

Postby seaton » Fri Aug 22, 2008 4:50 am

jkingdon wrote:Ok, sorry to be swamping your message board, but I thought it worth updating from my last note on the Microchip Zigbee units. I left a question with Microchip support and they got back to me in a matter of hours. With the extra information I was able to get the 625 kbps mode working and also disable the CSMA/CA function (listen before tx to avoid collisions). This got the latency down to 840us, which is better but still not as good as the RF12. I think I'll save the Zigbees for something with bigger packets for them to get their teeth into. At 625 kbps I still had range throughout the house, but at maximum distance there were signs of dropped packets depending on module orientation (they have a major dip along the central axis). Still not bad range for the speed and power, and given that the house is flooded with wifi and dect phone signals. I think these are going to be really useful modules for the right project.


Thanks for the update, I think that any of these zigbee/802.15.4 type modules are all going to have similar problems probably due all the behind the scenes things they do that we have no control over. I had a look at the 802.15.4 spec the other day for a bit of light reading (as one does) and there is alot going on there. I guess the RFM12 is pretty well raw data being send and received so a lot less latency involved and not much happening behind the scenes.

What type of camera are you using? As it sounds like your max sync is 1/250 anyway. Do you know what PW sync to? Or what frequency they run at?

Stephen...
seaton
Site Admin
 
Posts: 134
Joined: Mon Jan 23, 2006 8:48 am
Location: Bunbury, Western Australia

Next

Return to RFM12

Who is online

Users browsing this forum: No registered users and 1 guest

cron