Cheaps disposable joysticks from China

[fabrasil]

Folks,

i just create that thread for share info about all cheaps and disposable joystick from China.
In my country (Brazil) a lot of companies are selling they power wheelchairs with this ones and some times without any kind of support after sale including for authorized technicians. They only provide the new joystick, however, many times our customers are unable to buy new ones - poor people.

I know that those who have a power wheelchair must be able to maintain it, just like when you buy a motor vehicle but i've been trying to help these people, a way to give back to the world everything I've already conquered and even for life.


Cheers,
Fabio

[fabrasil]

See that two joys and mcu's from China

A single board of chipset that joy chinese

Anoter one

both exrernals:

Externally they are identical

Cheers

[Burgerman]

The issue with joysticks is not the hardware but the software. PG drives for e.g have decades worth of experience and updates to make them drive well, and be safe, as well as user programable to set them up correctly for each user/motor/powerchair and options.

So the real questions are:
How much power (amps?)
Are they USER programmable? So that different impedance motors from a different type of chair for e.g could be used?
And after all that are they any good?

And what are you trying to do with these? For e.g, do you have the ability to read the code, from the chip you show, if you plan to replace it?

[fabrasil]

Dear BM

The issue with joysticks is not the hardware but the software. PG drives for e.g have decades worth of experience and updates to make them drive well, and be safe, as well as user programable to set them up correctly for each user/motor/powerchair and options.

I agree with you about software knowledge.

So the real questions are:
How much power (amps?)

About i could see (personal experience) - about 15 joy_driver installed and opened and examine the electronic circuit its 80% real 35Amps and 20% 45Amps with very short peak power - i damage two units due that.

Are they USER programmable? So that different impedance motors from a different type of chair for e.g could be used?
And after all that are they any good?

Brands of MCU as Micon (FDA Aprlved in 2018), Norlha are OEM programmable, and i didnt see parameters like impedance. Are very limited - about 25 options.

And what are you trying to do with these? For e.g, do you have the ability to read the code, from the chip you show, if you plan to replace it?

The pcb that has the mcu scraped have no idea what they are using but they are very similar, change some details.
Using 8 g065n07 mosfets for both sides of motors (promissed deal well with brushed 250w/24v), they could be fake ones - 2 mosfets for each channel.

I just want to discovery more things about then and how to change parameters of then.


Cheers

[Burgerman]

If they were correctly implemented and the software properly written then it should not be possible to damage them that way.

>>>>>
About i could see (personal experience) - about 15 joy_driver installed and opened and examine the electronic circuit its 80% real 35Amps and 20% 45Amps with very short peak power - i damage two units due that.

They should use current sensing, temperature sensing, and roll back the power until in a safe zone. They do not sound either powerful enough to use properly or reliable / safe enough. But they are cheap.This is why. No impedance parameter? Then probably they dont have a positive feedback loop in the software or dont measure current at all. So they will be terrible to try and use on any powerchair. They wont drive properly.

[snaker]

My old chair uses this joystick. The controller is called newVsi (a PG brand). It claims 50A rated but it's weaker than vr2 50A (I guess it's only 35A). I tried to connect it to PC through woodygb's cable and the PG software did recognize it. Its parameters are less than vr2 but include all essential ones.

[Burgerman]

I think the original poster has a chinese cheap clone/copy? As in, PG had nothing to do with it.

[ex-Gooserider]

It seems possible that they might be a copy / clone of the PG product given that the Chinese are good at copying everything else....

IIRC it is pretty easy and cheap to make the cable that talks to VSI and the programming software is accessible - so I'd think it worth trying to see what happens if you try using the PG software to talk to it....

ex-Gooserider

[fabrasil]

They should use current sensing, temperature sensing, and roll back the power until in a safe zone. They do not sound either powerful enough to use properly or reliable / safe enough. But they are cheap.This is why. No impedance parameter? Then probably they dont have a positive feedback loop in the software or dont measure current at all. So they will be terrible to try and use on any powerchair. They wont drive properly.

Hi there!

Yes, they do not had any kind of feedback except single voltage for both motor and brakes. Tested it here. Reliable? Maybe not but for some people is the single choice (at least @ Brazil).

I am still studyng it but mains mainly problem is time.

Cheers

[fabrasil]

My old chair uses this joystick. The controller is called newVsi (a PG brand). It claims 50A rated but it's weaker than vr2 50A (I guess it's only 35A). I tried to connect it to PC through woodygb's cable and the PG software did recognize it. Its parameters are less than vr2 but include all essential ones.

Ji there

Did you have pics of this joy? Its old?An real PG?

Cheers

[Burgerman]

If it wasnt real PG drives the software wouldnt recognise it or work. Unless they also copied that. Which is possible but unlikely.

[snaker]

I donated that chair to an elder. The chair is a Chinese brand Wisking, the joystick (controller) is newVsi. You can google the world "newVsi".
https://www.google.com/search?q=%22newV ... 6&dpr=1.33

[fabrasil]

I donated that chair to an elder. The chair is a Chinese brand Wisking, the joystick (controller) is newVsi. You can google the world "newVsi".
https://www.google.com/search?q=%22newV ... 6&dpr=1.33

Yep. You are right.

Look that:
http://scooterland.com.au/wp-content/up ... manual.pdf
https://www.monsterscooterparts.com/pg- ... 61-01.html
even https://www.pridemobility.com/pdf/owner ... er_boi.pdf

[martin007]

IIRC it is pretty easy and cheap to make the cable that talks to VSI and the programming software is accessible - so I'd think it worth trying to see what happens if you try using the PG software to talk to it....

Can you do that with just a cable and a computer?
And program the joystick?

[Burgerman]

With EVERY modern sysem. Some also need a dongle to decode. And if its fake chinesium it likely wont work,

[martin007]

With EVERY modern sysem. Some also need a dongle to decode.

Which don't require the dongle?
I'll try to stay away from the chinese.

[Burgerman]

All themodern ones do need. And of those R-Net is the best system, and far better to program. And the only one we can get OEM access.

[ex-Gooserider]

With EVERY modern sysem. Some also need a dongle to decode.

Which don't require the dongle?
I'll try to stay away from the chinese.

The older and 'obsolete' Pilot + controllers from P&G don't need a dongle and can be programed with a fairly simple and inexpensively made cable (directions elsewhere on the forum, use the search) They are no longer manufactured, but can be purchased used on eBay. They aren't as flexible / capable as R-Net, which replaced them, but are quite usable on a basic chair

Some of the other P&G series controllers will also work with the Pilot+ software and cable, but I'm not sure offhand which ones.

R-Net is nicer but needs a quite expensive, and somewhat difficult to get Dongle in order to program it.

As mentioned previously, the Chinese products are a bit of an unknown. Some CLAIM to be possible P&G knockoffs, but it is unknown if the P&G software / cable would work with them. Given the low cost of making a Pilot cable I'd be inclined to give it a try as a cheap experiment.

ex-Gooserider

[Burgerman]

R-Net is nicer but needs a quite expensive, and somewhat difficult to get Dongle in order to program it.

A DEALER level programmer/dongle is available easily from every chair manufacturer and is right there on the order sheet. (this may not be the case in the future) Just tick the box or order one later from your chair supplier. This CAN be used if needed for all programming needs.

As for cost, they are expensive. 2 to 4 hundred. But I for e.g have 3 complete R-Net sets of equipment to use on my chairs or as spares, as well as 2 complete R-Net powerchairs in use already. I will have no other restricted system in the future.

Thats about 20 to 30 grands worth of powerchairs at retail, and doesent include the 3 compete spare sets of R-Net gear for future use on other chairs or as spares. In order to properly use these on different chairs, and in order to make any of my powerchairs drive PROPERLY these programmers are totally essential.

So its a tiny investment compared to the overall price of your kit to have an ESSENTIAL programmer. Its currently the best control system by far, and it is the ONLY one that allows us to get proper programming capability. So its a small price to pay. Plus things may not stay this way forever. They may make PGDT r-net stuff as innacessible as all the other manufacturers already have. Destroying future programming capability and system flexibility on other chairs/motors in the future. Tying us to their "experts" like pride, dynamics new linx system, or worse, permobil/myra does. So do yourselves a huge favour. Stock up on R-Net chairs, components, and get a PROGRAMMER while its still possible!!!

[slomobile]

Regardless how good these newVSI controllers are, "The newVSI is supplied with industry-standard cable options, with either Beau or Anderson connections to the batteries and motors."

IIRC it is pretty easy and cheap to make the cable that talks to VSI and the programming software is accessible

That makes these a good choice as a "universal" backup/emergency controller. Equip your chair with either of these connector sets, and you have a place to plug in one of these joystick/power module combo units. Whether your stock JSM, PM, or their wiring fails, this thing can get you back home. Right?

You can google the world "newVsi".

Yep. You are right. Look that:
http://scooterland.com.au/wp-content/up ... manual.pdf
https://www.monsterscooterparts.com/pg- ... 61-01.html
even https://www.pridemobility.com/pdf/owner ... er_boi.pdf

Beau part S9G-5409CCT (PG part 63.20.012) female chair side.

newVSI Beau

Anderson SB50(Battery) PowerPole45 Motor and Brake symmetrical chair and controller side

newVSI Anderson

Can anyone find a source for Beau part S9G-5409CCT (PG part 63.20.012) female and its male counterpart ?
Beau appears to be part of Molex now, but I cannot find that connector in the Molex catalog. Just this mouser page https://www.mouser.com/catalog/catalogUSD/645/2570.pdf

I have an old P&G PM with the Beau male connector which I liberated and a Jet 3 chair with the female socket on the chair. This allows me to build new control systems however I want (Fanatec CSL elite pedals, DE Sabertooth 2x60 w/Kangaroo for hands free driving) and plug it into the chair while testing, but plug the stock VSI controller back in to have a useful, movable chair when a testing session is over.

[slomobile]

FWIW the female chair side Beau connector I have also bears markings
INDUSTRIKABLAGE AB
ART. 307122-00-0
Ser. 1123

The controller they came from is Penny + Giles Drives Technology LTD
Christchurch Dorset O.K.
Type No. D49323/4
Serial No. 97067266

If anyone has info on what that came from or anything about it, let me know. thanks

[Burgerman]

They are about 40A. Takes around 100 or so to turn my chair in the hallway. So it might just about be able to get you home in on the straight and narrow. POSSIBLY. If its configured correctly. They are used on very small low power slow geared high impedance portable type chairs.

Also this https://www.dimensionengineering.com/pr ... rtooth2x60 isnt adequate it will only supply half the current needed. And it uses just 4 mosfets instead of 16 in a powerchair unit or 32 in the roboteq for heat/impedance reasons. It will go into orbit.

[slomobile]

I've used a DE Sabertooth 2x25 on a wheelchair to push my Chevy C2500 pickup around during calendar parking when it wouldn't start. They work just fine. Definitely wont go into orbit, they soft limit current to protect themselves if over temp or over current, and return to full power automatically as soon as they cool. The 2x60 is fan cooled as well as heatsinks on both sides of the board. Even the Rnet 120 can only do 100A steady state. The 120A rating is only 10 second boost Sabertooth can also do 120A boost for a few seconds.

They might not live as long as R-net 120 but for under $200 brand new, easily remote controllable, very flexible, usable with gaming controls, and response programmable via UART with free downloadable software where you just draw the response curve you want, they fit into many situations. Probably not something you want to rely on daily if you are quadraplegic and cannot replace it by pulling a plug and insering another. But that is the functionality that having a common connector set allows, and lots of folks in chairs do have the ability to change a plug.

[Burgerman]

But 60A isnt adequate. It takes almost double that per motor to turn in place on my hallway carpet...
And what control script or software do you use, and does it include essential load compensation etc?

Note motor current turning in place in my hallway...
M1 and M2 current.

102A and 113A doing nothing but turning on a thick carpet. How do you think 60A max is going to work out in more extreme situations like on a ramp or grass? If I move the stick slightly until 60A per motor is seen, the chair just sits there doing nothing! It would move forwards and back. Certainly wouldnt turn.

[martin007]

The older and 'obsolete' Pilot + controllers from P&G don't need a dongle and can be programed with a fairly simple and inexpensively made cable (directions elsewhere on the forum, use the search) They are no longer manufactured, but can be purchased used on eBay. They aren't as flexible / capable as R-Net, which replaced them, but are quite usable on a basic chair

Some of the other P&G series controllers will also work with the Pilot+ software and cable, but I'm not sure offhand which ones.

R-Net is nicer but needs a quite expensive, and somewhat difficult to get Dongle in order to program it.

As mentioned previously, the Chinese products are a bit of an unknown. Some CLAIM to be possible P&G knockoffs, but it is unknown if the P&G software / cable would work with them. Given the low cost of making a Pilot cable I'd be inclined to give it a try as a cheap experiment.

ex-Gooserider

I understand.

[slomobile]

And what control script or software do you use, and does it include essential load compensation etc?

Software:
https://www.dimensionengineering.com/so ... ribe/html/
https://www.dimensionengineering.com/info/arduino
https://www.dimensionengineering.com/info/dotnet
It may do IR compensation, I don't know, read the software manual and see. You can tune the expo just like in RC. I greatly prefer its layout to Rnet software. I don't require IR compensation because I will use encoders with the Kangaroo add on which I think will afford a superior level of control to IR comp. I still only have the 2x25 and it is actually sufficient for most things. No kangaroo or 2x60 yet, but planned soon as I can afford. Encoders currently are read by the Arduino.

It is plenty drivable without either encoders or IR compensation. It just means when you get stuck on a threshold, you give it a little more stick to get over, then reduce on the other side. Turning in place on the 2x25 works fine, it just takes a lot of stick movement to get started, then need to reduce quickly. Often easier to just let go when pointing the desired direction to stop on a dime. Feels sluggish starting out because it takes a lot of power and it lets you feel it. That is a good thing if you need to manage your consumption to preserve range. A rolling turn is much easier on the batteries if you can manage it, and this gives you the feedback to understand it without an amp meter. And when you let go of the stick, it actually stops instead of carrying on a couple seconds. It does not have all the nanny safety stuff built in. You want protection from yourself, you'll have to implement it yourself, including the braking circuit, but thats just 1 switch mosfet and a couple lines of Arduino code. And you can set the brake timeout to any value you like, or monitor the encoders to only apply parking brake when the wheels are actually stopped. Avoiding being tossed on fast downhills which is a hazard with Rnet. Its accessible and open enough that a person can do just that. It isn't for everyone. But I bet that you specifically BM would like it if you tried it. If you absolutely have to have IR compensation it can be implemented by reading current in software.

In addition to using the DEScribe software, I write my own Arduino routines which I suppose are equivalent to Roboteq scripts. Except they are well understood by a large Arduino DIY community instead of being an obscure language understood by... Lenny? Anyone else?

MFR product claims:
"Peak currents of 120A per channel are achievable for a few seconds." Just like Rnet. 102A and 113A are less than 120A.
"Sabertooth 2X60 uses 0.8 milliohm MOSFETs in its bridge."
"Overcurrent and thermal protection means you'll never have to worry about killing the driver with accidental stalls or by hooking up too big a motor."

[Burgerman]

If you say so. All of your description sounds exactly liike my worst nightmare. It would never be properly usable without load compensation. Or with non linear joystick turn rates.

Sounds like a weak version of the roboteq (150A) 32 mosfets, with its own software (loads of settings for everything you mentioned built in) that took many extra years of code by lenny and testing to make actually usable with any good level of control comparable to a commertial system like r-net. MUCH harder than expected. And he is still doing it. That 114A was just a turn in place on a flat floor. I hit its 120A limit all the time in actual useage. And its around 30% Worse (so runs out of torque and stalls) with faster 8mph motors which is why I use 6mph ones.

So you think will work OK but your standards of what you consider as control and usability must be a long way from mine! Watch robotwars. Almost every bot can be driven. But mostly they are only partly under control.

[martin007]

And what control script or software do you use, and does it include essential load compensation etc?

Software:
https://www.dimensionengineering.com/so ... ribe/html/
https://www.dimensionengineering.com/info/arduino
https://www.dimensionengineering.com/info/dotnet
It may do IR compensation, I don't know, read the software manual and see. You can tune the expo just like in RC. I greatly prefer its layout to Rnet software. I don't require IR compensation because I will use encoders with the Kangaroo add on which I think will afford a superior level of control to IR comp. I still only have the 2x25 and it is actually sufficient for most things. No kangaroo or 2x60 yet, but planned soon as I can afford. Encoders currently are read by the Arduino.

It is plenty drivable without either encoders or IR compensation. It just means when you get stuck on a threshold, you give it a little more stick to get over, then reduce on the other side. Turning in place on the 2x25 works fine, it just takes a lot of stick movement to get started, then need to reduce quickly. Often easier to just let go when pointing the desired direction to stop on a dime. Feels sluggish starting out because it takes a lot of power and it lets you feel it. That is a good thing if you need to manage your consumption to preserve range. A rolling turn is much easier on the batteries if you can manage it, and this gives you the feedback to understand it without an amp meter. And when you let go of the stick, it actually stops instead of carrying on a couple seconds. It does not have all the nanny safety stuff built in. You want protection from yourself, you'll have to implement it yourself, including the braking circuit, but thats just 1 switch mosfet and a couple lines of Arduino code. And you can set the brake timeout to any value you like, or monitor the encoders to only apply parking brake when the wheels are actually stopped. Avoiding being tossed on fast downhills which is a hazard with Rnet. Its accessible and open enough that a person can do just that. It isn't for everyone. But I bet that you specifically BM would like it if you tried it. If you absolutely have to have IR compensation it can be implemented by reading current in software.

In addition to using the DEScribe software, I write my own Arduino routines which I suppose are equivalent to Roboteq scripts. Except they are well understood by a large Arduino DIY community instead of being an obscure language understood by... Lenny? Anyone else?

MFR product claims:
"Peak currents of 120A per channel are achievable for a few seconds." Just like Rnet. 102A and 113A are less than 120A.
"Sabertooth 2X60 uses 0.8 milliohm MOSFETs in its bridge."
"Overcurrent and thermal protection means you'll never have to worry about killing the driver with accidental stalls or by hooking up too big a motor."

An interesting project.
However I see that it is not cheap.
But this hardware does not replace the joystick.

[martin007]

And what control script or software do you use, and does it include essential load compensation etc?

Software:
https://www.dimensionengineering.com/so ... ribe/html/
https://www.dimensionengineering.com/info/arduino
https://www.dimensionengineering.com/info/dotnet
It may do IR compensation, I don't know, read the software manual and see. You can tune the expo just like in RC. I greatly prefer its layout to Rnet software. I don't require IR compensation because I will use encoders with the Kangaroo add on which I think will afford a superior level of control to IR comp. I still only have the 2x25 and it is actually sufficient for most things. No kangaroo or 2x60 yet, but planned soon as I can afford. Encoders currently are read by the Arduino.

It is plenty drivable without either encoders or IR compensation. It just means when you get stuck on a threshold, you give it a little more stick to get over, then reduce on the other side. Turning in place on the 2x25 works fine, it just takes a lot of stick movement to get started, then need to reduce quickly. Often easier to just let go when pointing the desired direction to stop on a dime. Feels sluggish starting out because it takes a lot of power and it lets you feel it. That is a good thing if you need to manage your consumption to preserve range. A rolling turn is much easier on the batteries if you can manage it, and this gives you the feedback to understand it without an amp meter. And when you let go of the stick, it actually stops instead of carrying on a couple seconds. It does not have all the nanny safety stuff built in. You want protection from yourself, you'll have to implement it yourself, including the braking circuit, but thats just 1 switch mosfet and a couple lines of Arduino code. And you can set the brake timeout to any value you like, or monitor the encoders to only apply parking brake when the wheels are actually stopped. Avoiding being tossed on fast downhills which is a hazard with Rnet. Its accessible and open enough that a person can do just that. It isn't for everyone. But I bet that you specifically BM would like it if you tried it. If you absolutely have to have IR compensation it can be implemented by reading current in software.

In addition to using the DEScribe software, I write my own Arduino routines which I suppose are equivalent to Roboteq scripts. Except they are well understood by a large Arduino DIY community instead of being an obscure language understood by... Lenny? Anyone else?

MFR product claims:
"Peak currents of 120A per channel are achievable for a few seconds." Just like Rnet. 102A and 113A are less than 120A.
"Sabertooth 2X60 uses 0.8 milliohm MOSFETs in its bridge."
"Overcurrent and thermal protection means you'll never have to worry about killing the driver with accidental stalls or by hooking up too big a motor."

An interesting project.
However I see that it is not cheap.
But this hardware does not replace the joystick.

[Burgerman]

It attempts to replace the control module. Fun project but it will never come close to a 45 uk pound 120Ah R-Net with decades of software and firmware development on it. So as I see it ultimately pointless exersize! Other than an interesting learning experiment.