Frsky compatibility

The value on FSCTRL0 is 8 bits value from 0 to 255...we can test every 10 th value 25 times ..till we bind..The value no need to be exact...theres is a range.I can bind succesfully with my module from
170 to 245....less or over these values not binding.

Hello again.
FYI the 26MHz xtal in my old V8 hack module runs at 25999524 Hz.
Multiply by 92 to get a ballpark 2.4G frequency. This module works with my AVR solution, although i only have the V8 protocol working atm and won't attempt the 2way D8 until the Fall this year.
I have a module without PA from an ebay vendor called Rayax. This did not work until i detuned the xtal by replacing the xtal loading capacitors. Prior to the mod' it's xtal was running at 26000155.2Hz.
I also have a Skyartec module with PA. This worked out of the box with it's xtal running at 26000073.6Hz. I calculated that to pull it's frequency down to be compatible with the FrSky module, the FSCTRL0 reg would be -31. On testing with a FrSky Rx this seemed to be in the correct zone as the limits of reception were FSCTRL0 values of +17 and -51.

I hope this helps, but clearly we need to measure more FrSky xtals to work out what the nominal frequency is. All my measurements were taken by looking at the Fosc /192 signal which i recall is presented on the GDO0 pin.

Regards, Rick.

Midelic, the value in FSCTRL0 is signed (2's compliment ..... invert and add 1).
170 and 245 are both negative so your xtal is running high compared to FrSky.

Can this instability be explained by relatively small frequency deviation - ~32kHz for FrSky-2 and ~29kHz for FrSky-1?

Skyartec uses ~51kHz which should be more resilient. I don't have the FrSky modules myself and can't try it, but could you set DEVIATION register for FrSky to 0x50 (corresponds to ~51kHz)?

It'll either make the signal unrecognizable at all or make it more resilient.

On the other hand, your quoted crystal differences translate to frequency deviation instability of ~0.8Hz (it's around 25ppm). It is hard to believe that the detector is so sensitive that it fails to decode the signal.

It is still puzzling why only CC2500 is so capricious, why all other chips works with similar (or worse) crystals pretty reliably.

Admittedly, they use much wider frequency channels. My beloved nRF24L01+ uses deviation either 160kHz (for 250kbit/sec and 1Mbit/sec) or 320kHz (for 2Mbit/sec).

Frequency range of CC2500 is 2400 – 2483.5 MHz

The base freq on frsky is 2404 Mhz...and my module adjusted freq is lower than that,... (15khz lower till to the end of the my base channel)..There are 4 registers for tuning cc2500 base freq.
FREQ2,FREQ1,FREQ0,and FSCTRL0....first 3 make a 24 bits word and you can do coarse(rough) tuning freq ..and FSCTRL0(freq offset) is for fine tuning..+-210Khz.
It is possible tuning using these registers and a frsky rx as a reference(no need measuring the Xtal or detune it) for a tx module (or a frsky tx as reference for a rx ...on binding on channel 0).
A software program to search for the channel zero ends using frsky gear as reference..and found the center of the base channel.
Btw I build a Frsky Rx based on these findings and it seems to work fine.

@victzh
I think the problem of frsky compared with others is narrow channel bandwidth used..frsky have 135.417 kHz BW.

PhracturedBlue wrote: with the 2-way, I wonder if we can somehow get close enough to bind, then use the telemetry response and use freqest to determine the adjustment.

Has anyone looked at the SPI on the Rx side.
I wonder if the FrSky Rx's use the FREQEST reg to modify the FSCTRL0 reg in order to track frequency drift from the external transmitter (The TX module). The CC2500 datasheet mentions this.
I have 3 FrSky TX modules and 3 RX's atm. I'd like to measure the xtals in all of them, but the cpus have to be held in reset to do this. Hmmm.

Wish i could help you Guys, but i'm really busy with other Hobby stuff.

Unfortunately I don;t have a spi sample from rx binding ,...only from hopping data.
which module are you testing,... the one with PA?
I think you at one of the end of frsky base channel ...for this reason you lose connection easily.I think your base channel is out of FSCTRL0 range(+-210Khz) ...in this case should tweak also FREQ0 register....

victzh wrote: Can this instability be explained by relatively small frequency deviation - ~32kHz for FrSky-2 and ~29kHz for FrSky-1?

I think you're correct ..... what irony that it comes down to the DEVIATION register.
Just doing some numbers on paper as i don't have RF Studio to help. Not that RF Studio is of any help when the xtal frequency is below 26Mhz !.

For the V8 (1way) protocol.
Base frequency = 2400999786 Hz at 26Mhz nominal frequency.
Channel Bandwidth = 135416 Hz
Channel Spacing = 299926 Hz
GFSK deviation = +/- 28564 Hz

As My V8 RF deck was running at 25999524 ish.
This puts the Base Carrier Frequency at 2400955829 Hz ... down by about 44Khz but greater than the deviation.

... Not sure about this but i guess if your 26Mhz xtal is out by about 310 Hz from whatever the FrSky norm might be then you could be in trouble.

I think i might increase the V8 deviation value to that of FrSky 2way (D8 series) to give greater resilience. (Must look more closely at the two way register initialise sequence).

This is just a hunch, but i'm wondering whether PhracturedBlue's problems were originally down to having the PA and LNA enabled at the same time. Maybe there was some sort of regenerative effect going on.

Hope this helps someone.
Regards, Rick.

If is out by 310Khz,,FSCTRL0 is doing nothing,,,should tweak FREQ0 reg.

midelic wrote: If is out by 310Khz,,FSCTRL0 is doing nothing,,,should tweak FREQ0 reg.

I meant 310 Hz at 26 MHz oscillator frequency .... actually i think its more like double that for the local oscillator to see no deviation of the transmitted carrier.

p.s. midelic have you got that fbl100 code working ?.

Rick,,so you are ...I thought is other Rick.
I'm still waiting for the other nrf module..I have only one...In the mean time I was busy with frsky gear,,trying to make it work...
I was using the same cheap cc2500 pcb as pb but different lot and the base frequency was shifted .I ordered also one cc2500 module with PA to see if I can make it work.I tried only 2 way protocol.
Rick by chance do you have spi sample from frsky rx binding data?

SRST strobe?????
or SRES?
I see you mod. FREQ0 register.

After SRES I had 100ms delay (original reset routine)...and my cc2500 module have still base freq shifted...maybe it was just my luck with these modules.