[Arm-netbook] designing a low-cost decent 3d printing board

[Luke Kenneth Casson Leighton]

On Tue, May 23, 2017 at 8:16 AM, mike.valk at gmail.com
<mike.valk at gmail.com> wrote:

> 2017-05-23 3:27 GMT+02:00 Luke Kenneth Casson Leighton <lkcl at lkcl.net>:
>>
>> hi all,
>>
>> ok so i've been looking around and the practice of creating a
>> "modular" 3d printing electronics board is extremely common, thanks,
>> many years ago, to the stupid, stupid decision to use prototyping
>> (evaulation) plugin boards with 1.3 to 2.0 *amp* stepper ICs mounted
>> onto micro-postage-stamp-sized PCBs.  the problems these cause are
>> endless.
>
>
> You're being a bit brief here.

 sorry.

> But essentially, the community has been using
> "default" boards, which are cheapish and fairly documented/understood, but
> too generic and thus not able enough?

 no, look up "polulu stepper boards".  the entire home-grown 3d
printing industry is suffering from ignorant "copy copy copy" itis.
the first person to use the prototyping stepper driver boards *didn't*
read the documentation, and does NOT understand the consequences of
their ignorance.

 in this case it's that the heat generated by a 2A driver IC can't
escape out of an 18x18mm PCB so it overheats and stops working.
sometimes temporarily, and sometimes permanently.

 it's shit design gone badly wrong, copy-copy-copy-copy.  no actual
thought involved.

> But how about the ESP32? (Successor to the ESP8266). Reasonable beefy CPU
> with build in WiFi. The Duet is using the ESP8266 as the WiFi bridge.

 the idea is to be able to get off-the-shelf low-cost arduinos for $10
as opposed to having a larger PCB custom-made that costs $40 because
you're using a USA-based factory and component pricing which has a
4-5x markup.

 plus, porting to the ESP32 is a lot of effort.

 splitting out the processor board into a "stock" arduino leaves very
little to do in the way of software porting.


>> now, debatable is whether to split out the MOSFETs, endstops and
>> thermistors into their own separate shield as well, which i feel might
>> be sensible.
>
>
> Why not place all controllers separately and direct to their HW and connect
> them via a bus? This gives you freedom to expand and replace. And use a EOMA
> card as the "master"?

 lots and lots of pins.  nice idea, but no.

> Or would that kind of modularity up the costs too much?

 too many pins: the duet-wifi requires an ATSAM4 with 100 GPIO pins.
maybe 64.  some of them ADC.  that's too many.  now it turns out that
the cost of those "ADC-to-I2C" or "ADC-to-SPI" ICs (etc) are actually
just as expensive if not more than an embedded controller!

 so by the time you've added in a GPIO extender IC (I2C or SPI based),
an ADC extender IC, you might as well just.... drop in an STM32F or an
ATSAM in the first place.

 cheapest way to do that in a modular fashion?  surpriiise! use an
STM32xxx-NUCLEO, or an Arduino Due.

 l.