Up-Link for ER9X [Magi(C)] a vision...

[HC1969]

Thanks Mike!

Which is used in the (positive or negative)?

[MikeB]

I'm not sure, but as long as the short (high pulse in your diagram) is always the same width, it doesn't matter. The time from rising edge to rising edge is the same as falling edge to falling edge.

Mike.

[HC1969]

I'm not sure, but as long as the short (high pulse in your diagram) is always the same width, it doesn't matter. The time from rising edge to rising edge is the same as falling edge to falling edge.

Mike.

Ok, thanks!

[HC1969]

I started development with an 8-channel independent switch .
1 servo channel through 8 switching outputs (256 stage).

[HC1969]

Completed the test / measurement experimental FW , which I measured the achievable resolution. I found that in the current HW up to 5 -bit ( 5 -channel ) control can be safely constructed .
Those who are interested in the technical details :
The servo PWM signal 1 - 2ms pulse . 1Mips the MCU 's (now I use the internal 4MHz RC oscillator) .
The 16 -bit Timer1 meter that measures 1MHz signal source .
Thus, the measured value between 1000 and 2000 ( 1ms , 2ms) .
The difference between the two in 1000 ( measurement bandwidth ) .
8 channel must share this bandwidth ( 1000/256 = 3.906 ) to 256
Since only integer values ​​can be set so that the three - band limits needed to be chosen.
Due to the need of measuring CPU time taken to process a minimum of 20 measurements per digit value .
This went on under the unstable connections /interfere with each other . Of course, redesigning the hardware ( the MCU is driven by 20MHz ), this could be improved , but I do not want to.
If this back in the 20's incremental count ( 1000/20 ) = 50 incremet / positions. It follows that max. 5 bits ( 2 ^ 5 = 32) can be realized. So it will be 31 increment ( 31 ms) between the positions . It has been a stable, secure each digit .
The eight hardware outputs can be five and three other independent switch outputs accordingly.
Now, this output can be connected to 1-1 plus three independent Mellis , what if we make parallel , for example, then . the output load can be applied to ( theoretically ) 1A (it's the original 500mA ) .
So it would be five independent outputs, which 3pcs 1A ,and 2pcs 0.5A .
or
the remaining 3 pieces could be used for any other purpose output .
Now the philosophy of what to do with them ...

[HC1969]

Dear MikeB!
I am confident that this topic was also read.
I made the switch to channel 5. I do this for free.
However, I can not control this current mixes.
Binary timings can not be setup!
I need a mixing process by which the PWM signal can be set directly (adding).
currently:
bit0 = 30 uS
Bit1 = 60 uS
Bit 2 = 120 us
bit3 = 240 uS
bit4 = 480 uS

An example:
Let:
bit0 = THR switch
Bit1 = RUD switch
Bit2 = ELE switch
bit3 = AIL switch
bit4 = GEAR switch

If everything is turned off (00000) PWM = 1000 uS.
If turned is THR + ELE switch, then (00101) PWM = 1000 + 30 + 120 = 1150 uS
If turned is all switch, then (11111) PWM = 1000 + 30 +60 +120 + 240 +480 = 1930 uS
I tried to channel offsets, but can not be solved. The mixes adding he did not work as it should for me.
You may have an idea for it?

[KAL]

Why using existing switches? I won't miss this switches for several things like aileron-rudder-mix, dual-rate and many more ...
I would however add 5 additional switches connected with a resistor array to simulate / replace one of the pots.

At the moment I have no exact idea of the configuration but I thought of 2 sets of 5 resistors (100R, 200R, 400R, 800R, 1K6)
like in the attached drawing (sorry for the bad quality, no better program than paint at the moment).
The values of the resistors are only to show the principle ...

This would simulate a 6k2 linear poti with 32 steps

[Kilrah]

30us = 5.86%

So start with -100% MAX, then add each of those as your bits
bit0=6% MAX Switch(xxx)
bit1=11% MAX Switch(yyy)

etc

Adjust limits to get 1000 to 1930, then set subtrim to be in the middle of the limits. Should work...

[jhsa]

er9x has HALF, not MAX..

João

[HC1969]

Thank you, but this is not the problem.
I realized that the radio mix menu Weight value and real PWM signal is not linear!
I measured carefully in context.

The measurements were performed with a FrSky D8R receiver output.
I used the following mixes (use CH5):

5 switch, digital weight and HALF, and -100 subtrim:

Several variations performed the measurements and is always calculated the PWM timing at 1%.
Base timing (digital 0000): 992 uS

I found a difference of one (himself) on the linearity of switches:
Calculation: (measured timing-base timing)/weight

Code: Select all

Weight        Measured timing     calculated 1% (error)
3%                          1016 uS                   8 uS/%
6%                          1048 uS                   9.333 uS/%
12%                        1112 uS                   10 uS/%
24%                        1232 uS                   10 uS/%
48%                        1480 uS                   10.1666 uS/%

Unfortunately this is still not to. This non-linearity is amended more depending on how much switch how it is turned on!
I measured some combination of:

Code: Select all

Weight        Measured timing     calculated 1% (error)
3%+6%                    1080 uS                  9.77 uS/%
3%+12%                  1144 uS                  10.13 uS/%
3%+24%                  1264 uS                  10.074 uS/%
3%+6%+12%              1200 uS                   9.904 uS/%
3%+48%                  1512 uS                  10.196 uS/%
3%+6%+12%+24%+48%    1940 uS               10.193 uS/%

The question is whether these errors (nonlinearity), it just depends on the ER9X to or affect the FrSky system as well?

The only way to solve the task when each position (32) is in agreement as to calibrate!
I need some immediate timing options in the menu mix menu (PWM timing setup)!

[MikeB]

I'm not sure I agree.
100% on er9x/openTx is 512 uS.
-100% should be 988 uS pulse
+100% should be 2012 uS pulse
total range 1024 uS
3% of 1024uS is 30.72uS
So:
3% should be 988+30.72 = 1018.7
6% should be 988+61.44 = 1049.4
12% should be 988+122.88 = 1110.9
24% should be 988+245.76 = 1233.7
48% should be 988+491.52 = 1479.5

These match quite closely with your measured values.

On the limits page I display the actual calculated pulse width being sent for a channel. Move the highlight down to channel 5 and the number on the top line is the actual pulse being sent in uS.

Mike.

[HC1969]

Thank you Mike!
It is conceivable that the FrSky transmitter module are slightly distorted by the timing?

[MikeB]

Yes, we send the pulse as PPM (with some slight jitter), the Tx module has to measure this, send it to the receiver that then has to generate a pulse out. While all have crystals, each could be running at a slightly different timing. Then there is the accuracy of your 'scope to consider as well.

You could remove the module from the Tx and measure the actual pulse on the PPM signal, maybe used channel 1 to make it easier to trigger the 'scope.

Mike.

[HC1969]

It works! Thank you!!!
The values ​​given in the calibration table and re-counted and mid-range set to every work!
Smaller, more refinement, you need to (to be eliminated in the switching).

[HC1969]

READY!
Please see:
[BBvideo 640,480]http://youtu.be/E6jbXgE_I14[/BBvideo]

[KAL]

thumbs up.jpg (1.64 KiB) Viewed 16134 times

[LTMNO]

READY!

When are we selling them?

[GTiDon]

I may have missed the point here a little.
Why use a digital uplink? Why not use 6 different pwm levels on one channel? Is it so u can have any combination?

Will it be theoretically possible to have a telemetry uplink to replace data radios for flight controllers?

Sent from my C6603 using Tapatalk

[MikeB]

It is actually 32 different PWM levels on a single channel, so you CAN have any combination. The result is 5 separate digital outputs at the receiver, each of which is independantly controlled.

Mike.

[HC1969]

READY!

When are we selling them?

The original hardware is now available on eBay. If the order should indicate that firmware it asks for "D" (RCS-1D).
See: http://cgi.ebay.com/ws/eBayISAPI.dll?Vi ... 0719332329

[HC1969]

In short, what this is:
Be over 1pc servo channel 5pcs ,independently of each other (!) Feed on and off electrical things.
The outputs are controlled to 8 pieces 0.5A Output from 2-2pcs switch 2 comes out. If they are connected in parallel with the output, then 1A Capacity controlled. The remaining 2pcs (3+2 = 5) 0.5A Output remains of the original in size.
The outputs of these types of (digital) controls are nowhere (HK than not).
The outputs are fully independent from each other 32pcs strict control servo position adjustability should the transmitter side. This ER9X and OpenTX a very easy mix.
Only in such a setting, you must mix:

Inside is a mix :

Then they just need to be adding!
The proper operation of that channel page (-100) must be moved (in this example, Channel 5):

A sample of my five channels individually did a switch, but it can be anything else, of course!
Anyone familiar with the digital weighting (2^x), it will be familiar with the weight values​​!
00000
00001
00010
00011
00100
00101
00110
00111
01000
etc. ....

[HC1969]

Opened her own a topic that is in all the info together.
All in one topic for Multi switch

[GTiDon]

Very nice thanks for explaining again. This would free up a couple of channels on my FPV plane so I can actually use my pan tilt now.

Sent from my C6603 using Tapatalk

[GTiDon]

Will programming this be possible? Currently I use a virtual switch time off 1s time on 0.1s to run Led strobes controlled by the radio. Ersky9x.
Any ideas how to program this at a specific pwm frequency?

Sent from my C6603 using Tapatalk

[Kilrah]

Use that switch to control the mixer line with the weight that corresponds to the output the strobes are on.

[HC1969]

Project ready. Home page (use google translator):
http://rc-miskolc.emiter.hu/rc-miskolc/ ... Itemid=204

[HC1969]

RCS-1D digita switch (5pcs) via Taranis:
[BBvideo 640,350]http://youtu.be/QfElb10gsIo[/BBvideo]

[jhsa]

Would it be possible to adapt this to run on a arduino pro mini?

Thanks

João

[HC1969]

It is possible, if you can adjust the precise PWM signals (steps).