[jsantos]

I thought that I'd share what I did for my hall effect sensor. After a misstep and a lot of research, I ended up getting my hall effect sensors from eBay for $0.99 a piece. Do a search for "Arduino hall effect sensor" to find them. The sensor is a 3144 sensor mounted on a PCB, complete with LED and resistors. See the attached photos.

1. Photo 1 - On the left is the first sensor I used. This is a Hamlin sensor and IT DOES NOT WORK. In the middle is the sensor that I got from eBay and on the right is the eBay sensor encased in plastic and epoxy.

2. Photo 2 - Shows the sensor mounted on my machine, lighted up.

[DaQue]

What kind of repeatability do you get?

Sent from my Nexus 5 using Tapatalk

[Stevos758]

I thought that I'd share what I did for my hall effect sensor. After a misstep and a lot of research, I ended up getting my hall effect sensors from eBay for $0.99 a piece. Do a search for "Arduino hall effect sensor" to find them. The sensor is a 3144 sensor mounted on a PCB, complete with LED and resistors. See the attached photos.

1. Photo 1 - On the left is the first sensor I used. This is a Hamlin sensor and IT DOES NOT WORK. In the middle is the sensor that I got from eBay and on the right is the eBay sensor encased in plastic and epoxy.

2. Photo 2 - Shows the sensor mounted on my machine, lighted up.

Very nice. I cant quite read it but is it a 681 ohm resistor? That should be close enough to my 550k ohm correct?

I think I will order some. Cant hurt to try I guess.

[jsantos]

What kind of repeatability do you get?

Sent from my Nexus 5 using Tapatalk

The sensors are incredibly accurate and repeatable. I've got an x- and y- fence mounted on my machine and the coordinates of those fences have not varied in the couple of months that I've been using the machine. What I find really amazing is that the DRO readings on those fences have stayed the same, down to the last 4 digits. And I'm using metric!

So when I'm cutting something, I just mount the material flush to the fences and I don't have to look for the x- and y- edges of the material!

[jsantos]

Very nice. I cant quite read it but is it a 681 ohm resistor? That should be close enough to my 550k ohm correct?

I think I will order some. Cant hurt to try I guess.

Impressive eyes! Yes, it's a 680-ohm resistor.

[Stevos758]

The sensors are incredibly accurate and repeatable. I've got an x- and y- fence mounted on my machine and the coordinates of those fences have not varied in the couple of months that I've been using the machine. What I find really amazing is that the DRO readings on those fences have stayed the same, down to the last 4 digits. And I'm using metric!

So when I'm cutting something, I just mount the material flush to the fences and I don't have to look for the x- and y- edges of the material!

Awesome! My hot glued on ones worked ok but these look nicer and will be easier to handle. Did you just epoxy them on?

[jsantos]

Awesome! My hot glued on ones worked ok but these look nicer and will be easier to handle. Did you just epoxy them on?

I routed a cavity in the plastic to mount the sensors in and then poured epoxy on it. I also removed the pins on the PCB and replaced them with wires.

[Falcon69]

look pretty good jsantos!

I've been busy on mine too. Finally figured out the normally closed thing. Now to figure out the kit and make them work so they have a front panel indication for my control panel. Then I can order up the new PCB's!

These new switches are user selectable N/O or N/C!

The blue LED in the video represents the LED inside the optocoupler in the G540 (or other breakout board that has them). Basically, the new switches have a mosfet that triggers the switch input/output. So, Kinda like hooking up a mechanical switch, just with an extra power and ground to light the LED's. Also, there's a bare wire, if it's grounded, it puts the switch into N/O mode. N/C by default.

[nlancaster]

Falcon I really like the idea of NC hal sensors. PM me when you have the new kits ready because I want those for my CRP4848.

[Falcon69]

Working on them.

But I'm trying to decide on the length of cable needed for each switch. Wire is expensive, but I'd like to find a common length to make. I was thinking maybe 16ft on each switch? Or maybe have 4 of them @ 8ft, 4 @ 16ft?

The new kit is pretty complicated. Taking time to layout the PCB for it. I just finished the board layout for the single switches, the ones that dont require a kit (Front Panel, Rear Panel, Mainboard, etc.)

[nlancaster]

Falcon cable length has always troubled me too. I will go out sometime this week on my CRP4848 and get some measurements of what cable lengths I currently have for my mechanical limit switches.

[Falcon69]

It looks like on the CNC router Parts page, they are using 5 meter cables, and 10meter cables for the Y-Axis for the larger machines. So 16ft and 32ft.

That's gonna add alot to the cost of wiring. :/

[nlancaster]

i can tell you i went with the 12 and 20ft wires, from cncrouterparts for my motors and they are over 4 ft longer then needed, some even longer then that.

[Falcon69]

thanks nlancaster.

but for bigger machines, 10meters (32ft) may be needed.

That is a big problem. the cost of wire alone is about 150 bucks just for 8 switches. that may put the kit well over $250. this may not be feasible. most people overlook the cost of wiring until they need it. all they will see is the kit and think it will be too much. maybe i can put together a website and make the switches available in 3ft increments and make them separate from the other parts of the kit.i know that the first two kits i finish will be mine! for my machines i plan to build. maybe, i can make it so people can do the soldering themselves and purchase wiring as they like. but, i think most people would have a problem soldering a part that has 0.50 mm leads on it.

on the single standalone switches. i got the circuit boards and components ordered. i also ordered stuff to remake the molds. the resin isn't coming out clear enough (i'm a perfectionist) so going to do a better job of the molds and see what happens. maybe i'll make a video of doing that! i think if i sand (5000 grit sandpaper) the mold to a gloss finish, then pour the rtv to make the mold, then the parts should come out with a near gloss finish once the resin dries. should look pretty good.

[nlancaster]

sell it both ways. with and without wire? Then someone could buy thier own wire in bulk and keep you from having to source and ship it.

[Falcon69]

The problem with that though,

if they dont do a good enough job of soldering the wire and creates cold joints, i'll get phone calls all the time by p'd off people saying my kit dont work. That, plus, there's a possibility of a slight voltage drop as well if the wire is spliced.

[nlancaster]

yes true that would suck.

Wouldn't it be nice if people understood that buying something DIY means an inherent amount of Troubleshooting on the part of the purchaser?

I do tech support for a living for a very large tech company. And we get people that think our products can perform magic all the time.

[Falcon69]

I tried sending you a private message nlancaster. Your box is full.

I noticed on your profile you are in Portland. I'm in Vancouver! Very close.

[Falcon69]

Well, I managed to get some of the switches assembled, the ones that DON'T require the kit. I have also sold a few, but, I have not heard a response from the buyers yet. I am sure they have not had a chance to place them on their machine yet. They did say that if they work as good as they look, it would be awesome!

Also, I got the boards ordered and they arrived two days ago. I spent yesterday assembling the parts. One part is super tiny, like a grain of sand. It's about a 0201 size SMD chip. It's one of the ESD TVS Diodes. I've added Electro-Static-Discharge protection to this kit to help protect the components. It won;t protect it completely, understandable, but, it will help greatly. most of the chips or 0603 size and smaller. Really tedious work to hand place them all. But the stencil I got for the solder paste works great! Makes it easier.

As you can see from the pictures, it is not completely assembled yet. I mis-ordered one part. I layed it out wrong on the circuit board using the wrong footprint size for the chip. Luckily, they make that identical chip in the size I laid it out for on the PCB. Got lucky there. Just waiting for it to arrive before I can finish them up.

The main-board is much bigger then the last one. Almost 4 times as big, however, for the added features, it's worth it.

I've taken the suggestions here, and went even further. The board now features opto-isolated mosFETs for each switch. In addition, I've added jumpers to allow the user to select normally open switch operation, or normally closed switch operation. In addition, I now have added a jumper to select +5volt pull-up instead of sinking to ground. From my understanding, BoBs like the C10 or C32 require the limit switches to be pulled up to +5volts, instead of sinking to ground. So, that feature has been implemented for each switch.

Also, as you can see, the switches will now be pre-wired with cable.

The standalone switches work beautifully now. And can be selected as normally open or normally closed operation. If the wire is accidentally cut, for example, it activates the software to shut the machine down (only works that way in Normally Closed operation). :banana: Just as in the video above on the prototype board.

I may later, get better opto-FETs to allow more current switching. It's an optically isolated mosFET, but they actually call it an SSR (Solid State Relay). Currently, what's on the switches and boards only allow for AC/DC operation at a max of 60v @ 100mA. That one chip alone is $0.76 when ordered in multiple of 100. Pretty spendy. I did a search the other day, I can get those in the same chip size for up to 40v @ 2A. But, that raises the price of a single chip to $3.00 each. I have to do alittle research and see if those chips would require some type of a heatsink to dissipate the heat that could result from switching a 2A load. If it requires a bigger footprint on the circuit board, then I can't use the chips. If a simple little heatsink placed on top of the chip, via thermal grease or something, then they should work. I'll do some research on that.

As most break-out boards use an optocoupler for the limit/home switching, the opto-FET will only be switching about 10-20mA anyway. But, I could make it an option to upgrade that chip, which could allow separate switching for other applications requiring up to 2A loads. As these are digital Hall Effect Sensors, there's so many applications these can be used for, other than Limit Switches for a CNC.

I am hoping I don't run into any problems once I fire this kit up and test it. In simulation, it works, but in reality it could be a different story. Wish me luck!

If the kit works as intended, I will post a video of it working.

FYI, I'm still working on the schematic, I mean, I have it drawn up, and converted it to the PCB software, but I need to tidy it up and make it look good before posting it.

If the kit works as intended, then I'll post the schematic later, for those who want to attempt this kit themselves. As you can see from the boards, there's alot of chips; almost all of which are logic chips. those can be cut down, or more added, depending on the number of switches you want to use. I decided to go with 8 switches in the kit. But, with this new design, unlike the old design using ULN2803 chips, I am not limited in how many switches I want to use with the kit. By making the board bigger, or smaller, I can and or subtract the amount of switches and just have new boards printed up. The ULN2803 chips, as in earlier design, only allowed me to have 8 switches total.

For those who will ask, 'why didn't you use a micro-controller and shift registers?' Well, I could have, yes, but, I'd be looking at the same size board, or maybe even bigger, as I would have to add transistors to switch the LED's on and off. The logic chips can sink enough current to turn the LED's on and off. Cost wise, is probably about the same. And from what I've been told, Logic Chips will switch faster then a Microcontroller/Shift Register/Transistor layout.

Jason

[kiwibrick]

Wow, I'm buying some of these tonight once I've measured my machine.
I haven't looked at this thread for a long time!