Project

General

Profile

Forums » Hardware development »

Rev6 ideas

Added by jegb about 9 years ago

Hi

I am new to this forum, glad to contribute. Here there is a list of topics for fuelling HW dev in next rev:
-adapting footprints of newer computing modules, moving from NXP to other alternatives, should not be hard to decouple main board dev time from the mbed or other alternative, you may iterate faster the main pcb.
-advance on smd version, cutters do not have space issues, but overall cost may be reduced greatly, sticking to 805 or 1206 parts should be fine.
-PWM with faster optocouplers.
-DC/DC converter for PSU logic 5V (single input 24V)

Smoothieware and Duet are a good examples of the modular approach to outputs handling, something should be learnt from this.


Replies (3)

RE: Rev6 ideas - Added by jaap about 9 years ago

Hi Jegb,

I agree with most of this:
  • through hole -> SMD: check! That would make production much easier
  • better opto's -> check!
  • dc-dc -> most small lasers have 5v in their power supply already. If not, I recommend using a TRACO power TSR 1-2450, which fit the 7805 footprint.

foot prints of newer computing modules

Replacing the module is one thing, but that would also require changes in the firmware to adapt for a new module. What kind of module do you have in mind, and what are it's benefits?

Making the computing module not modular, but replacing it with a onboard LPC1768, would also be an option.

RE: Rev6 ideas - Added by jegb about 9 years ago

-power supply from laser may be used for auxiliary dc elements (pumps, lights, evac extraction), seems too noisy and would be great to fully isolate ground paths and leave control outside.

I have attached a dc/dc converter to the VMOT input and fed back on 5v rail. This module is barely 1.5eur cost. 7805 or other linear regs are out of question, with a drop of 19v the power disipation is not acceptable.

Mcu units, some ideas:
-Pi zero
-Arduino due
-lpcxpresso

Stabilizing development on an existing footprint like due (mega 2560 has been around for some time), makes easier cycles of development. Eventually you may end with a single board that includes the mcu, but unless you produce over 500 units that wouldnt make sense.

RE: Rev6 ideas - Added by Springuin almost 9 years ago

My recommendation: don't change the microcontroller to something else than an LPC1768, because it will require rewriting of the firmware and will lead to two different platforms ('the old' and 'the new'), will result in bugs and just wastes time on something not neccesary.
It would however be interesting to switch from mBed to something cheaper, like the lpcxpresso or just a plain lpc1768 on the board. For the latter option, offering boards with a presoldered lpc1768 is a neccessity, because not everyone can solder 0.5mm pitch TQFPs. Also, with a plain lpc1768 you would need an option to update firmware without requiring a separate programmer. The mBed has a chip on board that handles that, the lpcxpresso also has a programmer, but a plain lpc1768 needs something on the side.

I only see an advantage for switching to SMD parts when you want to go for machine assembly. The price will not be different and while 0805's and 1206 parts are not that hard to solder, through hole parts are still easier.

Raspberry Pi's are (imo) out of the question, because they cannot offer the required realtime performance (pwm, motorcontrol, limitswitches).

    (1-3/3)