[mcm_xyz]

sorry for the late answer

I have been very busy doing some other stuff lately so the boards are not ready yet. I am going to try not to let u folks down.

About the custom servos... Some advices...
1. Buy a cheap DC motor...
2. Look for and old brokem mouse in the house...
3. Measure the diameter of the stick of one mouse wheel...
4. Drill a hole on the back shaft of the motor as straight and centered as possible... It will help a lot if you remove the rotor from the motor...
5. Put the wheel from the mouse in the hole and glue it.

So much about the mechanics...
A little about electronics. It will be of great help if you will use single capsule double phototransistors. There are old mouses that had 2 separate phototransistors for obtaining a quadrature signal. It will be very hard for you
to find the right position for two separate transistors. Use single capsules.
I have used A4Tech mouses and they all have a single transistor.

The IR LED must be powered from 5V using a 1K resistor in series or it will be destroyed.

Best to you all

[mcm_xyz]

Hmm.
The modified board should work with of the shelf servos as well. It is not important if you connect A and B or !A and !B of the servo...

Grretings from Romania

Mike

[hesham morsy]

Hi Mike


have you finished your boards I have 4 hitaci dc motrs 19 w 24 v + encoder 100 ppr

[Dom B]

Hi Mike,

Thanks for posting the pics, and the information on your drives and servo's. I'm still coming to terms with the simplicity.

You mentioned some minor software changes, what did you end up changing?

I'd be happy to share my board design too for those interested, but it will be a while before I have anything worthy. As in my first post, mine is based on IRF540 fets actually mounted below a CPU heatsink and fan assembly.

I've attached a preview of my board.

Pat

Hi Pat

I have been trying to design a H bridge for a while learning all the time, it is actually for a welder but the same principles apply I guess. I was just wondering do the fets north of the bridge have to be p channel fets and the fets to the south 'n' channel fets reapectively?? I see you have specced IRF 540s on both the north and south. Will this work will they stturate fully etc. I don't know enough about all this at the moment but I keep seing both types of designs, could you shed any light on the benefits of all 'n' channel and a mixture of 'n' and 'p' channel??

Many thanks for your time

Dom

[joe plasma]

hi, new to cnc have built stepper controllers (linstepper)
and would like to build servo controllers would you have the shecmatic and SOURCE code for the elm dc servo controller went to org.,and didn't find anything on them
thankyou
joe plasma

[patndel]

Hi Dom,

I started my H bridge design based on the notes related to the OSMC (Open source motor controller) here's a link http://www.robotpower.com/osmc_info/
It's worth dowloading the documentation, which has a wealth of information on the design philosphy undertaken.

The fets are driven using a dedicated driver, which ensured the fets are fully saturated, and minimising the chances of shoot-thru where by both legs of the H bridge could be conducting and leaving the fets legless.

From my understanding, using all N channel fets simplifies the design and layout of the components, and that there are a larger range of fets to choose from in terms of capacity (though I could be wrong). You maybe be better directing the questions to some more knowledgeable.

I'm by no means an electronics engineer, but I can follow a schematic. I'm a mechanical engineerI initially started looking at the design to retrofit a manual mill (Sieg X3) to CNC using servo's. The conversion has been completed, but I took the easy way out and opted for Gecko drives, only because of time constraints, and the need to get the mill up and running in a short a time as possible.

Regards, Pat

[pastera]

P-channel FETs have a larger Rdson per unit channel area than does an N-channel.
The reason to use a PFET is you don't need a secondary supply rail. The N channel fet must have the gate positive to the source to turn on, but the source is driven to the +rail when it does turn on so it needs a bootstrap or isolated supply to work. The p-channel gate needs to be negative so the gate drive is simpler.

With the availabilty of high side drive IC's there is little reason to use a P-FET.

Aaron

[smarbaga]

if anyone has a basic working layout of the elm system i will draw a nice pcb layout for it. ( gerber, excellon, pdf, bitmap etc.).
i would need the schematic at minimum , preferably a picture of a prototype on perfboard or something like that, to go by.

[Halfnutz]

This is an aweswome thread, I am going to follow everyones progress and would love to build one (or 3-4) of these myself. You have my attention, I mayl try one of my own and post as time permits. That mouse encoder is ingenious. If anyone etches extra boards I would be interested in buying 3-4 if the cost is resonable.

[mcm_xyz]

hello again!

Sorry for not designing the pcbs. Unfortunatelly I have no time for doing that.
I have worked with somebody at a document that contains lots of useful info about schematics, programming and other stuff regarding elm controller. I will ask for it's permision to publish that document. I think that will help a lot in making a pcb.

Best regards

[smarbaga]

hello.
- i have downloaded the schematic for the elm.
first thing i see is if the parameters are going to be set fron a distance , say greater than 10 feet ( 3 m) the serial port driver should be changed as there will be to great of a voltage loss.
- i have made several uhu servo controllers which work real good but the source code is not available.
- a servo output error signal could be taken off of the error led and brought to an output pin.
- does anybody know how the motor torque is detected
- i don't see any current limit, but this could be handled in the motors power supply.
- what function does the isp have , other than not having to take the 2313 out of circuit for re-programing.
and you say these little beauties work, it sure is a simple diagram and a perfect diy'er
any particular order requested for the serial, encoder, error, isp and dtp/dir pins on a pc board, other than them being grouped?
i like the screw terminals at 0.2 in. centers, except for the standard isp header.
thanks

[smarbaga]

hello
attached is the elm servo redrawn ( in the rough) with some uhu features attached. component values and part ref.# to follow.
i could lay out a pcb if anyone was interested, the fet driver and output section is already in place.
any changes sugestions or recomendations are always welcome.

[pastera]

The motor torque is detected by backEMF - The controller figures out the motor current with the P5 parameter. The problem with this method is the use of unregulated supplies.

Aaron

[smarbaga]

attached is a fet drive ( from the uhu servo )
this is proven, works well, easily expandable
it can be attached or encorporated into the elm cpu
hummm ... noticed a via out of place
- i can add to or change this layout any time.
- i can generate the gerber and excellon files as well.
if anyone wishes?

[hesham morsy]

:wee: :wee:

[smarbaga]

has amyone here compared the elm's operation and preformance with that of the uhu's operation and preformance.
i see the elm's mcu opperates at 16 MHz where the uhu's is at 24.
the atmel 2313 chip came out in the early 1990's, would converting the code
(functions) to the pic format be an imposible task as i am farmiliar and set up for pic opperation, and have only done one project with the 2313 about 8 years ago and don't feel comfortable with its format and software devleopement packages.
there are so many different feature rich pics out there i am hopeing one would fit the bill.
i would be willing to layout and encorporate any features (serial 232,usb, current sense, or whatever , if anyone has the desire and expertise to do a software conversion.
thanks

[Vlady]

Hi, Smarbaga,
Use FT232BM for USB. No software conversion. http://www.ftdichip.com/
Vlady

[smarbaga]

a complete onboard eithnet i/o system would be the way to go , i think
just one plug from your controller board to your
computer, router, or hub.
place a router or hub in or near your controller for multiple
axis opperation or
have the eithnet interface input/output 3 to 4 axis motor controller opperating signals.
so you would have 1 eithernet cable ( mac address) controlling 1 machine.
the hardware would not be hard to set up,
its kinda like join the dots,
i lack the software experience to implement a task like this.
like visual c++ computer and screen controller setup
with cnc software (mack, or printer port) handler/redirective.
which would include,:
1. drive parameter setting,
2. step, direction encoding/decoding.
3. error detection
4. home switch detection.
5. near home and near limit slowdown. (parameter settable)
and any other necessary feature i do not know of at this time....
i will submit a proposed schematic in the near future
open for your sugeations.
- but i guess what i am saying is, why have separate connections to
the same computer like:
step, direction, parameter set/change, home ... etc
when they can be implimented in 1 communication cable - protocall.
- you plug a network cable in and go, just like that ( maybe in my dreams )

[NASR1]

HI
I have a hex file of smc3 for 15Mhz cristal .
-----------------------
Note that it is compiled for a 15Mhz cristal and with initial servo mode set
to 3
.equ INI_MODE = 3 ;Initial Servo Mode (0-3) for no serial control
.equ SYSCLK = 15000000
---------------------
DO someone have 16Mhz hex file of SMC & SMC3 for serial and lpt control . ?

[smarbaga]

elm2pic
see attached: