PINNED - Plug & Play PGDT for Radio Control

[Burgerman]

This picture shows more external receivers. These are vertically polarised, and another horizontal one but at 90 degrees to the main RX. This means all 3 planes are covered.

And in 4 separate antennas. With 4 separate receivers, in 4 separate places. An inch or two is all that's needed.

[Burgerman]

Another of the rear one.

This setup will give a pretty secure safe signal path, as the RC system uses the strongest signal from any of the four receivers.

Why is it needed?
Because 2.4ghz doesn't go through anything well, doesn't go around corners, and suffers from signal degradation from reflections. Its very like light. If you can hit an antenna with a torch, at every position and angle as you turn the chair around you are pretty good to go!

Woodies "cheap" transmitter, will work, and the AR500 will work too. If you don't expect perfection. But I wouldn't trust it 100 percent, especially in and around my house, inside my van, around areas where there is a lot of 2.4ghz noise. Like wifi, phones, Bluetooth, microwaves, baby alarms, video cameras etc.

[rustyjames]

Woody, that's fine, I'm not in any hurry, thanks.

[woodygb]

I'm really surprised that you get reception problems with the std DSM system.

We ...Robot builders ... bury the receiver inside the bot ..sitting inches away from unshielded motors ..and get very little in the way of reception problems...admittedly the range we are driving at is only 30 feet or so ...but the environment is possibly the worst that you come across.

[rustyjames]

Burgerman, thanks for the enlightenment on that and the photos. I have a lot to learn when it comes to RC.

[Burgerman]

Well I am flying. I have a logger http://www.spektrumrc.com/Products/Defa ... ID=SPM9540

If I fly over my street, on DSM or DSM2 in fact, I get masses of frame losses, holds etc. And in fact control losses. There spots may be small but matter. There are thousands of wireless routers and other things throwing out tons of 2.4 noise. Go up to just 50 feet and you can see many hundreds of houses... Range in these circumstances is highly unpredictable.

In and around my house, with a DSM2 receiver and no satellite I get fail safes in many rooms and in certain spots. These may only be 2 square inches. Then it works again. My heli chassis is carbon fibre, has a heavy battery. If I am outdoors in a field, and go just 100 meters away, carrying it, leave TX on car, and slowly rotate it in my hands the light "goes out" when the frame blocks the signal. 2.4 does not see through my heli. So you need more recievers on the other side... The bottom, etc.


DSMX helps massively. with the mass of noise now in built up areas. old original DSM picks 1 channel and sits on it till you get shot down...
DSM2 picks 2 channels. But the same thing can happen. Especially when it picks 2 right next to each other...
DSMX is newer is a HUGE improvement and uses 2 channels, and both of these hop all over the band like a machine gun. So every channel would need to be blocked to lose a connection.

So DSMX = BETTER! Where is a 2.4ghz noisy area like houses, streets, colleges, or any built up area.
But that only helps where the signals can get through.

Where there are materials in the way you have another issue. Hence treating the model or powerchair to a bunch of receivers located all over, so that if you rotate it, you can ALWAYS see one or more RX.

Because it may not matter at 30 feet, but at 60 feet you have 1/4 the signal. At 120 feet you have 1/16th of the signal. And I just drove my chair around the end of my street. That's about 350 feet... I would hate it to be shot down as I turned or by interference from everyones wifi...

[Burgerman]

http://www.youtube.com/watch?v=tXFfJUGUH_U

Horizon make these things...

Range isn't an issue. (well it is for me but that's a different purpose, not required for a powerchair. I boosted mine to well over 4 miles for a drone)
Problems are 3.
that the RX cant receive much if an antenna is at 90 degrees to your transmitter (polarisation) So polarise all of them differently so you cover all 3d.
that the RX cant read a signal if its "end on" but the rest of them will not be.
that the RX cant see your transmitter through metal, batteries, circuit boards, very well or at all at a distance.

A slow 6mph chair travels 8.8 feet into your car, traffic, or through a restaurant in 1 second.
After signal is lost that's how long it takes before the chair starts to slow. That takes another 5 feet at best, likely more.

[Burgerman]

http://www.rcmodelreviews.com/dx8dsmxreview.shtml

new DSMX
Hops on all channels, 2 at once.

From about 9 mins.



http://www.youtube.com/watch?v=Vy7Gg6Tj57Q

DSM2 and its issues. Uses 2 frequencies.

DSM uses just 1 frequency...

[woodygb]

B.M.
You seem to be getting over technical... at least for the use that I envisaged this R/C interface for ...which is ...remote parking and fetching of the chair ( without passenger) within a restricted range of ..say 30 feet max.

I believe that bog standard DSM2 should be just fine for doing this ...Yes/No ?

Your way is of course best ..but seems a little OTT if the range is sensibly and responsibly restricted by the user.

[rustyjames]

Woody, a range of 30' would work for me

[Burgerman]

>believe that bog standard DSM2 should be just fine for doing this ...Yes/No ?

Quite probably. 99.9 percent of the time. Unless you plan on driving it from room to room remotely as I do. Range in itself isn't the problem. You will get about 1km on flat ground line of site. But it will be intermittent across a house with dead spots.

But a better RX doesn't cost much compared to a new kitchen or car door! And all new Spektrum gear is now DSMX and DSM2 is no longer available. So if buying new you will have this anyway.

So the real extra cost for having better signal at all angles etc, and peace of mind, is only slightly extra for a receiver with a couple of satellite receivers.

[Burgerman]

Neither good nor bad, but takes some getting used to. Is this...

Stock mobility controllers have "round" joysticks. As in no corners. So you cant get 100 left and 100 forwards at the same time.

RC transmitters allow this. It feels very strange... Both better and worse. Faster and less predictable...

My transmitter allows a software emulated "round" joystick. Usually used with cyclic pitch control with helicopters. It feels better with this set up.

[Burgerman]

Here is a quick test.

Sat in a BM2...
Driving an empty old (but un-used) BM1 from 10+ years ago...

This is set to max speed (6mph) and driven slowly by RC using Woodys new RC interface. All programming of the controller is at 100 percent accelerations, decelerations etc. It follows the sticks exactly.

This is to show how easy and accurate it is. And I can turn it OFF by transmitter. And also ON by transmitter...

I have been doing this with RC powerchairs for years, but this is with Woodys better digital plug and play RC interface.

http://www.wheelchairdriver.com/gopro/rc.mp4 As usual the damned dog gets in the way a dozen times.

[SMA Dad]

Success!!! I'll buy one...
Congratulations Woody!...and thanks for the cool video Burgerman.

It looks like the resolution is good. I know John would say something if he could detect the digi-pot steps or processing lag.

Still seeing if I can use the existing digipot to output the 6v centre and 1.2v-1.3v swing... the spec sheet suggests that I may not be able to use it.

Woody, were you able to make the MCP42X1 work and did you use a 6 v center with a 1.2 v swing?
If so, I'm going to put the op amps, speed controllers, and r/c filter components back on the shelf and head in this direction with my nunchuck project.

Again....Congrats!!! and put me on the waiting list.

[falco peregrinus]

Well done, Woody! Burgerman, you did better steering that thing by remote control than my wife does steering hers by joystick!
Falco

[woodygb]

Woody, were you able to make the MCP42X1 work and did you use a 6 v center with a 1.2 v swing?

Yes...I did finally manage ...at least according to my cheap multi meter ... a 6v centre and 1.2v swing .

I also played around with interfacing a Nunchuck using some code from the web ...this was successful... but I shelved it after a couple of days as I couldn't see a way to detect a signal loss ( the Arduino retains the last known good signal when I disconnect it ) .

I did throw the problem at Lenny and he suggested that I try using time and seeing if the input has changed as a way of implementing a failsafe .

I will revisit the Nunchuck some time soon.

[woodygb]

B.M.

Would you still like the compact version with the tweaked software .. ( minimal deadband ) ?..if so I'll post it to you today.

[Burgerman]

Please. And it will allow me to test. I also have an old DX5e here, acquired cheap for comparative range tests against my boosted transmitter. so can check and see if its neutrals are consistently accurate or repeatable enough.

They can be a bit of an issue when connecting to such things as the quad copter controller for some people due to the somewhat flexy plastic stick units etc. So best test?

____

For what its worth the cheapie 5E had the exact same range at an open airfield as my stock £1k JR transmitter. Both had the same 100mW RF power. That's DSM2 and DSMX for the JR radio. At about 700 meters with a good RX.

In built up areas, like down my street, through walls etc with both transmitters on my bed, DSM2 has about half the range of DSMX due to RF noise. So 80 meters and 160 meters. Lots of Wi-Fi... And the DSM2 was much less predictable on different days.

Once the booster was added to the JR, from 100mW to 3200mW RF (a 32x power increase) output range increases everywhere to about 5 times greater. But signal dropouts through walls etc practically vanishes at any sensible range.

Why do multiple RX positions and high power matter? Because I want to have a belt and braces approach so I feel safe driving it inside the halls at mobility shows or in other public noisy places. Peace of mind.

[woodygb]

Any test data is much appreciated...

I've tried to anticipate all events ..but real world testing is much much better than me bench testing and watching data on the computer screen.

Have you got a AR500 receiver with the DX5E ...?

[Burgerman]

I have a 400 or 500 - I forget which. Has 1 short and 1 longer antenna. It has exactly the same range as a satellite, or an expensive 10 or 12 channel RX. But no diversity. It is DSMX but backwards compatible. It will work line of sight reliably, up to about 60 meters. Or in a house through a few walls especially if located high up away from chairs metalwork so it cannot be shaded. Like in a backrest.

Its this one https://www.spektrumrc.com/Products/Def ... D=SPMAR400 400...

[woodygb]

The AR400 failsafe is what I'd like you to test if you can.

Removes servo output pulses to all channels except the throttle channel
during failsafe.

...I've programmed the Arduino to recognise a signal of 1 or less ( null ) to be a failsafe condition...removing a signal wire from the receiver certainly makes the Arduino failsafe..but haven't had a receiver to test that it also works in actuality.

[Burgerman]

That's easy enough.

I have my GEAR switch set to turn on/off the pilot plus.

Off is easy, flick switch. On needs 2 switch actuations... Why would that be?

[woodygb]

It should really be a spring loaded toggle switch to get the action .... not an ON/OFF switch. ..hence my use of the rudder channel.

The signal sent should be a momentary one as a pull to GND turns the Omni ON and OFF...
Your Gear Retract is causing the Arduino to send a GND signal to turn ON the Omni and maintaining the GND signal ... it needs to return to a 5v positive before once again being used to send a GND OFF signal.... this is happening when your cycling back thru the Gear switches action when it reverts to a Positive 5v...THEN outputs the GND OFF as it returns.

Make sense?

Edit ...I just realised from your description that you seem to have the GND and POSITIVE signal position reversed on the switch ...or if you like the channel needs reversing ... THEN you'd turn it ON with a single action and require the switch to be cycled to turn it OFF.

Not something that needs to be stressed about if your happy to live with it.

[woodygb]

A signal greater than or equal to the programmed neutral ( 1500 ) plus 200 = 1700 ish will send the ( pull to GND/low ) ON/OFF signal ...but the signal must be toggled back to under 1700 ( pull to positive/high ) before the next signal ON/OFF ( pull to GND/low ) can be sent.

I had my Rudder channel reversed so as to have the ON/OFF on a flick to my left ...but that was just my preference.

[Burgerman]

Ah. I get it now.

>>> It should really be a spring loaded toggle switch to get the action .... not an ON/OFF switch. ..hence my use of the rudder channel. The signal sent should be a momentary one as a pull to GND turns the Omni ON and OFF...

Easy to fix that...

Should I add your interface to an updated version of my very old RC powerchair page?

http://www.wheelchairdriver.com/radio-c ... rchair.htm

[Burgerman]

OK latest interface with tightened up deadbands.

Tested first with the olde DX5e and the AR400 RX. Calibrated ok but took a couple of goes before it sat with the lights on your interface off. And they flicker a little. Plugged in, worked as expected, and fail safe does what you expect and it behaves as if there's no receiver connected. So that's good.

Plugged into my AR1000 with 2048 resolution etc. Calibrated dead simple first time. Recalibrated the Omni although it seemed OK anyway and tested. All channels are reversed! Oh well. Recalibrate everything again and it works perfectly. Now, the deadband feels very similar. It may be slightly tighter I think, maybe its in my mind! Hard to tell. Couple of clicks less trim seems to be needed before movement/brakes click in each direction.

And I don't think you will get it better. The deadband I don't like seems to be in the pilot plus system, and most people wouldn't ever notice it.

[woodygb]

I tightened the Arduino R/C interface deadband as much as I possibly could .

When the led lights flicker the MEASURED R/C signal has drifted away from the calibrated neutral ( nominally 1500 ) by + - 3 and the digipot will start outputting the first step of 128 .... so a volts fluctuation of approx. 0.01 when the lights flicker.

NOTE:- I could easily disguise the flicker by having the LED's react to a different signal level ... the digipot would however still output at the previous level...but thought that for this test it was best not to do so.

The resolution is set by the number of steps provided by the Digipot ...using a Transmitter / Receiver combo with a super resolution won't improve the number of available steps.

I'd have to agree that a deadband is most likely built into... and set in stone... inside the controller.

[woodygb]

I did mention the resolution and probable flicker a couple of pages back.

I have 40 steps each way on my trim ..which equals 100 of the signal range 1900 - 1500 - 1100 ..so from 1400 - 1600 ...and the LED's were lighting up after 4 -5 clicks.

The digipot has 128 steps in each direction and your presently losing ...due to the way I'm presently programming my deadband ..the first 6 steps = 0.06v.

I need some sort of deadband ..but it's an easy fix to make the deadband = 1 step ..

The LED's may however blink on/off @ neutral as the signal ...( at least on mine ) ..drifts a little.

[Burgerman]

Yes on my DX5e too. But its no problem. Its not surprising since the sticks are all plastic and creaky! It works OK but its not what you would call well made and its old...

No LED on with the 12x. They are just off all the time. Takes 1 click trim either way to cause a light on or flicker.

Doesn't cause problems anyway, because it gets nowhere near the Omnis deadbands.

[woodygb]

OK...Post back the one you don't want