So this project i'm working on is a hobby closed loop cnc controller with all accompanying electronics. Reason i'm working on closed loop is that i can use very cheap an powerful DC motors. Also the entire electronics part of the project will cost me less than 100$.
On the way i'm developing microelectronic(atmega) software.
It seems like a shame to develop and perfect all this and not share it with other hobbyists. Do you think this project has the perspective, or is it worth sharing and making an open source? And are there many similar projects posted online already?

Its an ambitious project, for a complete PC based CNC control that has to compete with the Hobby style Mach and Linux etc, you require a HMI front end that accepts and interprets G,M,S,T commands as well as a multi axis trajectory controller, if the loop is not closed back to the trajectory engine then outboard drives that close the PID loop are required.
You also require a PLC style control for the MST code processing and a suitable I/O interface.
It sounds as though you are moving close to the lines of Dynomotion.com
Max.

 

Thanks for the link. This is quite affordable and high grade.

Yes this is similar but low grade(for hobby application). Loop is closed trough the main controller for all the axis. No additional controllers are present.
I'm testing with 3 axis and 8Mhz chip, but all this is easily portable to another chip. High frequency controller and high resolution encoder would assure higher precision.

I don't see the point of open HMI for hobby use; The telemetry is periodically sent back to pc via serial link(or usb if we switch max232 for some usb chip). G-code and machine controls are sent via this link also.

So PC is necessary but only for monitoring and as G-code feeding device.

I'm targeting at very low price and very light solution when it comes to electronics.. something suitable for DIY project, that can be build from of the shelf electronics. The cnc controller board is worth less than 10$ in parts. The motor drivers cost about 20$ for two axis(available online).

I don't think this solution is better than step motors and mach3 in any way except at the price of attaining the motors. Though this brings the issue of PSU that can give these start currents. Problem i solved with cheap car battery and PSU in parallel. Battery acting as buffer giving houndreds of Amps of starting current.
But since i'm developing it (for myself) i'm wondering weather it would be usefull to someone else also.

 

One other problem I see is obtaining small enough resolution when using a simple slotted disk for the optical encoder, you will not be able to get any where near the approximate minimum of 1kp/rev that is usually required for practical positioning, without going to the present method of using the Moiré effect disc.
Max.
.

 

Well for the project i'm working on, this shouldn't be a problem. It is for routing acrylic and high tolerance is acceptable.
But for someone who would want to use it for metal working it would be an issue. My guess is they would have to get their hands on some professional optical encoder, rather then making an slotted disc. But these cost around 50$.
And with 1024 detents per rpm, they would have to use 64Mhz chip at least.

Ok i see where this is heading. There is limited hobby application for this.
Probably only one application(that i'm making it for) would benefit form this. High feed rate, large machine, hobby made, brushed dc-motor powered, low precision cutter.

 

Max, do you have a link for moire effect disc? Some DIY for rotary encoder.
I'm not sure i understand the physics of this well enough, to reproduce. I understand the concept, but don't have idea how to practically apply this to double slotted optical sensor.

Hey what about using ball mouse foto diode/double transisitor like this with bit larger slotted disc ?
http://www.romanblack.com/onesec/DCmotor_xtal01.jpg

 

The Moiré effect type is a bit difficult to reproduce for a DIY project, it is essentially a grating the same pitch as the main disc, a small piece of grating is placed in front of the reading head and angled slightly so that it produces an even wide shutter effect with up to 10x to 100x the pitch of the scale itself.
I only have experience from some old Ferranti documentation on the principle.
The slotted opto heads come all in one such as H22A1 or the double OPB822s.
Max.

 

Ok theb i'll use classic 100 slots disc 45mm diameter with old ball mouse diode and double transistor. They have very small distance between receiving transistors.

Thanks for information.
Regards

 

Hi guys, here's a little update.
IR2110 drivers are workin perfectly after your help and they are driving the CNC on different PWM frequencies without an issue. I've programed PID control and it is working fine.

Now i'm having a dillema about PWM frequency, and i'm looking for some PRO opinion. I'm using an atmega328, and 14,750Mhz crystal, and i can set motor pwm to 28,8Khz, 3,6Khz, 451Hz, and 56Hz.

I gave up using 28,8 khz, because it required too high duty cycle to get things going, so not much room was left for closed loop control

I've tuned PID sucesfully on 3,6Khz and it's working fine and is responsive and has low error. But i've noticed that machine has some difficulties moving small distances. For example if it undershoots/overshoots 0.5mm it has trouble getting to position without overshooting again. I solved an issue by allowing for 0.3mm error on final position to exit the loop.

I tried the 451Hz last night and it seem to be able to start moving at even lower duty cycle(i assume more current is flowing), and it seems to be able to hold even lower speeds before stalling. Also it seems to be able to make very small moves. It is manifesting some sort of stepper like vibration(especially at these low speeds). Slight jittering. I'm guessing this is what lowers the stall speed. An it seems to remove the grease of the lead screw pretty fast.

I haven't even tried the 56Hz. i've read somewhere that driving frequency should be not less than 5x the rotating frequency, and i've got 21Hz(1300RPM) at 100IPM feed rate.

So what do you guys think is the optimal frequency to drive brushed DC motors for such application. Any experience with custom made servo systems?

Btw thanks for the optical encoder tip. Ball mouse optical encoder modification is proving very usefull and cheap for this application.

 

Picmicro claim that to reduce audible noise on DC PWM control, it is not necessary to go much higher than 4Khz.
I tend to keep around 5Khz for DC PWM.
Max.

 

Thanks Max,

Which application do you use them for? as servo motors?

Regards

 

DC spindle motor.
Max.

 

What are you using for a lead screw? From my time on cnczone.com, I learned that people make the mistake of choosing to fine of a pitch thinking you will get better resolution of movement. Many industrial cnc's use 2 or more start threading on the lead screws.

 

The starting point for the majority of Industrial CNC is a lead of 0.2".
The coarser the pitch, the more torque required for an equivalent load.
Max.

 

The starting point for the majority of Industrial CNC is a lead of 0.2".
The coarser the pitch, the more torque required for an equivalent load.
Max.

That 0.2" is true for manual machines, like a Bridgeport. Because the dials are easier to make/read with a 0.2 lead. But cnc doesn't rely on a dial. They use high pitch multiple start threads ,mostly ball screw type. This allows fast movement and low torque at the same time.

The rails of the bed support the load of the table and part to be machined. The main 'load' the screw see is the cutting force.

 

I have been integrating CNC controls for 30 yrs now, I was an integrator using Mitsubishi controls for some time, but also used Fanuc on occasion.
The majority of B.S.'s I have come across is 0.2" to 0.5" lead, the highest was 1" for CNC punches etc, where the load inertia was relatively low.
The ball screw is a class 1 lever, where the dia is the input side and the pitch is the output side, the B.S. centre is the fulcrum.
The highest demand on the servo system is not usually in the cutting force but occurs during the acceleration/deceleration period, these requirements are usually decided by the motor to load inertia ratio, which if too high will degrade performance drastically.
IOW, a lower B.S. pitch value will decrease the inertia ratio.
I also have a couple of shop CNC XLO knee mills, similar to the B.P. which are 0.2" pitch.
Max.

 

Hi guys back again with another issue. Still same ir2110 controlled H bridge design. Only i switched to better MCU - dsPIC. It allows me to lower the PWM frequency and i've programmed it to adjust pwm frequency according to motor speed. So minimum frequency now is 56Hz and it allow for very low RPMs - thus Feed rates.

But i've got issue of lower flyback diode on H-bridge overheating and glowing red and then burning out, if the motor is runing on high load and low speed. I'm guessing too much current due to low frequency or? The diode us UF4007. Mosfets ar IR3205 and they seem to handle this current without problems. Any suggestion?

Only other problem is overheating of brusshes but i will fix this by getting geared motors with high input RPM and increasing pwm frequency. Meanwhile i have to work with these.

Thanks

 

Checked and the frequency running is 160 Hz. I will buy and put in some UF5408 this afternoon, and report back. Still opened for suggestions.

 

Hi guys i've put UF5408 but still after 5 seconds of running the diode starts to glow red. Anyone has idea?

 

The current rating for the diodes should be at Least 50% of the maximum motor current when stalled.

 

The current rating for the diodes should be at Least 50% of the maximum motor current when stalled.

Thanks, so i's the diode rating. I've switched to 3A but still it ignites after about 10 seconds of low speed movement. From my experience these motors(cordless drill motors) draw from 5 -15A when stalled. So i should do something to get them sooner away from stall speeds or find 5-10A fast diode. Any sugestions on diode?