Frequency synthesized RX problem

[charles0198]

I am experiencing two very strange things about my new receiver that I have constructed. The basic design is from F. THOBOIS’s frequency synthesized RX21/23 on 72MHz with rotary switch selectable frequency, as well as PPM shift select (this means that the RX can work with both futaba and JR TX). PIC16F84 is used to program the MC145170 PLL and after programming the PIC goes to sleep mode. The voltage regulator is MAX883.

The first problem is that when I use Futaba or JR TX on any channel there is a very slight jitter (clicking) inside the servo when the sticks are stationary. What could be the reason for this?

The second problem is that when I use JR TX on any channel there is a slight cross talk of adjacent channels. I am using inverter before the basic F. THOBOIS’s design before feeding the 4015 decoder. What could be the reason for this problem?

[Mr.RC-CAM]

The first problem is that when I use Futaba or JR TX on any channel there is a very slight jitter (clicking) inside the servo when the sticks are stationary. What could be the reason for this?Use your o-scope to measure the relative jitter. What do you measure?


The second problem is that when I use JR TX on any channel there is a slight cross talk of adjacent channels.It sounds like you are thinking this might be related to the shift. However, it could easily be related to the channel count. If that is the case, then your decoder's input filtering needs some work. Your o-scope is the tool of choice here too.

RC-CAM

[vintage1]

Depending on how you did the synthesiing bit, there might be some instablitiy in the LO that is causing jitter.

OTOH most servos do jitter a little in my experience.

[charles0198]

Mr.RC-CAM

On the oscilloscope the jitter apperas as slight random PWM. The second problem is only with JR, not with Futaba.

vintage1

The PLL is comparing 5KHz... means R = 10.245MHz/2049 and N = 61.xxxMHz/y where y is the step ratio.

[vintage1]

Well I assume that somewhere you have a varicap that actually sets the LO before you divide-and-compare.

Check that out for ripple voltage on it. I was called upon to design a beast like this once for AM/FM radio.

In practice we had a pretty slow responding system,but it used ti 'hunt' gently around the correct frequency at a very low frequency - - once every few seconds.

With a narrow band system, maybe its hunting a bit faster - like 50hz or so.

That might give very sligh variations in pulse width, since the transmitted frames are not exactly 'square' but have 'sloping edges'

I'd say the JR problem is either in the transmitter, or it has very 'sloping edges' and you need to attend to how you square off the pulses before decoding.

[Mr.RC-CAM]

On the oscilloscope the jitter apperas as slight random PWM. That is what jitter is. But, what do you measure? If it is 1uS then that is acceptable. If it is 10uS then that is horrible. What do you measure? The numbers are what matter.


The second problem is only with JR, not with Futaba.I only offer this as another possible cause of the problem (besides the shift). What is the encoded channel count of the JR versus the encoded channel count of the Futaba? If they are different, then you might have other issues besides the obvious. With some choices of PPM filtering, you can run into issues that starts to act up with certain channel counts. Vastly different framerates can also be another source of frustration. So, if you are getting nowhere attacking this as a shift problem, then keep in mind there are other gotcha's out there.

RC-CAM

[jperch]

I am interesting in seeing the design you mention. Can you point me to it?

Thanks,
Joe

[charles0198]

I have made the receiver based on F. THOBOIS design
http://home.nordnet.fr/~fthobois/rx21.htm
http://home.nordnet.fr/~fthobois/RX23.htm
The receiver has some what improved and is working fine when the Tx is close but by moving the Tx away from the receiver causes servos to jitter slightly. This problem is related more to the antenna position. When the antenna is placed parallel to the RX the servo jitters but when the antenna is placed below or hocked up the servos jitter very slightly. These results are with Futaba Tx. The problem with JR Tx is still there!

The modifications from F. THOBOIS design is the PIC16F84 instead of motorola micocontroller and MAX883 instead of LM2931 or direct battery.

[Acetronics]

:) Hi, Charles,

Have a look to this project:
http://jerome.viriot.free.fr/cours/cours.htm
I'm sure you will recognize something .... many explanations are also available ...
but in french !!!

Alain ;)

[charles0198]

Does anyone have idea about the loop filter design for MC145170?

Acetronics
The RX link you gave uses second loop capacitor of 10nF whereas F. THOBOIS design uses 1nF! Why is the difference and what is the effect of this capacitor?

[Acetronics]

:eek:
Hi, Charles
This design is an improvement of F.Thobois' receiver ...and it has been designed together.

All I can do for you here is send you the MC Datasheet ... but this :mad: :mad: :mad: :mad: kind of file limitation begins to make me furious ... Have an address to post it ???

I Try to touch HF circuits only if it's impossible to avoid it !!!

Alain
:rolleyes:

[charles0198]

I am using the following circuit before feeding to 4015 decoder. (basically an inverter circuit)

The problem is that the adjacent channels have cross talk... means while moving the aileron the throttle servos moves a bit (these 2 channels are adjacent in a JR frame), when elevator is moved the aileron servo moves a bit, when rudder is moved the elevator servo moves a bit. The throttle has no effect on any other channel as its the first in the frame. What could be the reason and how to solve?

[Mr.RC-CAM]

PIC16F84 is used to program the MC145170 PLL and after programming the PIC goes to sleep mode. Sleep is good. But, is the WDT and all interrupts disabled? Are you sure?


The problem is that the adjacent channels have cross talk...
Is the interaction seen on the raw PPM output of the FM demodulator?

RC-CAM

[charles0198]

Yes, the WDT and all interrupts are disabled!
__CONFIG _CP_OFF & _WDT_OFF & _PWRTE_ON & _RC_OSC
...
main call Init_PORTs
load_freq movlw 0x08 ; clear all interrupt flag bits
...

When an oscilloscope is connected to observe raw PPM output of the FM demodulator, there is a 50Hz "hum" coupled and is observable on the PPM signal (slow forward and backward shift of PPM frame), so its very difficult to check that! Maybe the VCO is too sensitive...

[Mr.RC-CAM]

so its very difficult to check that!
You really need to observe the PPM train in its raw form and determine if the problem can be seen there. Otherwise, you will be just throwing rocks at the problem.

As for your 50 Hz hum, perhaps you are suffering from a ground loop between the scope and the Rx. If the Rx is being powered from a bench supply, then pull it off it and power the Rx from a real battery pack.

RC-CAM

[charles0198]

The raw PPM frame seems to be fine and the the PPM pulse width and duty cycle remains same of subsequent channels! The problem is somewhere in the inverter or the coupling of the two transistors.

Another difference in the JR and Futaba frame is that the first pulse is not a channel in case of Futaba but for JR the first pulse is throttle.

[Mr.RC-CAM]

Then you should consider my comments in post #6. Your PPM filtering is suspect (C19, C20). Confirm your existing values as well as try other R-C values.

RC-CAM

[charles0198]

I am fed up... the circuit is very simple but the problem is #&^#((!__... :(

[lazy-b]

Charles: Why not try using a regular Inverter, i think this will give you a sharper edges or more square pulses.....or you might put some sort of Positive-Feed to the Transistor circuit (Hysteresis) ......better just install a Shimitt Trigger Inverter.....or use an 8 pin OP-AMP to make an Inverter......1st make it work, then just fine a way to simplify the circuit later.

Your should able to catch those cross-talk using an oscilloscope, make sure your are using a Hi-Impedance probe, to avoid loading the test circuit......Mr.RC-CAM is right, its got to be some where in the filtering.

Its look like your Syntherize work perfectly, its only a matter of decoding those raw ppm pulses, or you might just write a program to decode those Raw ppm pulses.

I guess its now much easier for you to make a SCANNER, just need to find a valid pulse of between 1 to 2 msec, to validate a existence of PPM signal.

[charles0198]

As I have mentioned before the RX is working fine with Futaba TX, but with JR TX there is a cross talk between adjacent channels (only the PPM pulse ahead of the affected pulse has the slight mixing effect but not the reverse).

The problem could be in:
1) inverter or 4015 decoder
2) JR Tx
3) data slicer of MC3362

I have checked and confirmed the first 2 with oscilloscope and they are working fine. I have compared the inverted and original demodulated JR frame and the inversion is fine. The 4015 is also working good as it is working with Futaba. I have also checked the JR Tx frame before and after connecting the encoder to RF deck and its fine.

The problem that I have observed on the oscilloscope is that the negative pulse width in the PPM frame slightly changes by moving the pulse position of the channel ahead. As the inverter is just inverting the frame pulses, the 4015 is triggered by the rising edge of the pulse (infact the falling edge in original demodulated frame), so the change in the pulse width slightly affects the PWM signal to the servo. But the time between the rising edges of pulses of the original demodulated frame is fixed and is not effected, so by somehow if the 4015 is triggered from the rising edge in original frame then it will work.

Maybe the data slicer comparator of MC3362 is giving the problem! I am using exactly the same component values from F. Thobois design.

[Mr.RC-CAM]

Maybe the data slicer comparator of MC3362 is giving the problem! If that is true, then you will see the problem with the 0-scope at the data slicer output. If the PPM looks good there, then keep moving downstream.

[charles0198]

I have discovered a very strange thing about the Futaba and JR systems... might be I may be wrong!

Regarding the JR unwanted channle mixing, I adjusted the 455KHz quadrature tank (coil) of MC3362 (T4 in case of F.Thobois design) and the variable capacitor across the 10.245MHz crystal (that actually adjusts the crystal frequency slightly), at some position with JR TX the system is working fine (no mixing and no slight jitter). I thought this is due to the off center frequency of the TX but same results with another JR TX.

At the above mentioned position of 455KHz tank and variable capacitor the Futaba TX doesn't work. By readjusting the 455KHz tank and variable cap. according to the F.Thobois tuning procedure the RX is working with Futaba TX (with some very slight jitter) but with JR TX again the same problem.

Any thoughts, solution?