Anyone know anything about the QU-BD oneup?

Pros: Laser cut and CNC parts will be in-house. Thats a step in the right direction as far as quality is concerned. They have a proven reputation.

Cons: Only problem I can see at all is they are using the same Geeetech Printrboard that I bought. That board has too small of a 12V trace for powering the heaters. Mine burnt up on the solidoodle. Of course, they dont have a heated bed, but if they do not reinforce that 12V line, there will not be an upgrade in the future. Honestly not a concern for a PLA only machine.

Nope. Setting aside thats ignoring right away the need for 2A + on a stepper motor... and lots else, Relays are inappropriate and prone to failure as well, especially at high current. And then to combat that, you have to reduce the duty cycle, making them again inappropriate for high switching frequencies.  So that moves you to MOSFETS and SSRs and then you need...

ahh dont worry about it. No its not like you outlined above at all.

Theres a reason people spend years studying EE

There is no 'hot work' involved on the card, there is just high current requirements with switching, that continue to exist regardless (as they have to exist in the same light for the steppers etc).

The 'solution' is rather simple, design the tracks on the board for their appropriate peak and continuous load. Its pretty simple and a very well known formula and just comes back to building the boards in the first place correctly...

Moving stuff 'off board' sounds great.. but there is another bunch of issues there. By splitting the ground plane, the common voltage (0V) between stuff becomes 'different' - the ground potential of one PCB can be wildly different to the potential of a separate and isolated PCB. So then you need to start ensuring you have a 'common ground', and in motion control, that usually means setting up a quality 'star topology' on your ground plane. This is why if you have ever opened say an Audio Amplified they have all those cables going back to a screw in the chassis ? thats the 'star'.... And you then want to ensure that you don't create any 'ground loops'... Since any analogue switching (or more importantly, analogue INPUTS) rely on calculating the difference between 'common ground' and the input voltage. If Ground isn't at the expected value (in this application) of 0V, then the 'difference' is actually 'different' to what your tables expect. In a thermistor application, this can mean the thermistor could read 30,40 even 100°C 'wrong'.

Then there's the fab cost. Fabbing a board and its associated costs is a black art unto itself. Essentially - every board design has to wear a 'setup cost' when doing anything more than 2-3 (and even 2-3 carries it, just you are normally 'batched' onto a much larger panel and everyone 'shares the setup cost', meaning for a handful of  boards it gets back to $20-$30 cost instead of $100+$10.... ). This setup cost is because of the fab process to make a PCB - there's actually a hell of a lot involved between getting the substrate (commonly FR4 fibreglass) sorted out ready to take copper, gluing on the top and bottom copper sheets, then you drill the substrate and have to plate it *again* to plate the inside of those holes (and no, you can't predrill the FR4 and do it all at once, because 'build up' means the inside holes wont plate properly and you risk deforming the surface drilling the first time so the copper on the top/bottom wont plate properly...)... Then there's the silk-screening, the 'depanneling' (involving normally v-routed cuts)... the electrical testing...

Anyway, simply moving from 1 big PCB to 2 separate PCBs, design constraints aside, suddenly increases the cost even though 'its the same board area' ...

Then there's the assembly costs... First you have to have 2 sets of Solder Paste Screens now... or maybe you are going to Wave Solder, which means you better have designed for it!... Then populating two separate PCBs means 2 setup of the "Pick n Place" machine (and the associated component sacrifice involved) which populates the components.. setups take 4-6 hours .... Then there is obviously 2 complete runs through the pick and place. Then its onto Wave Soldering or Reflow Oven... once again, more batches, more runs, more potential faulty yield (due to higher number of processes involved for any one complete unit).

So to not have any of the above issues - since moving High Current is childs play as long as you design correctly - just build the board with the correct amount of copper.

You see every PCB 'track' has a resistance. This resistance is a calculation based on its surface area and volume. To move high current through high resistance tracks means a lot of power has to turn into heat. If you dont have enough surface area in your PCB to dissipate that heat, then the tracks heat up and 'lift off' the PCB or simply 'burn up'. The amount of heat generated isn't linear either, so dropping the power by even 500ma can dramatically reduce the heat generated as its almost-sort-of a square product of power. Obviously, where you can't reduce power because you 'need it', then you have to go the other way, and INCREASE the copper on the PCB.

All of this is calculated very easily with established maths, such as over at http://www.desmith.net/NMdS/Electronics/TraceWidth.html .

Basically, you take the Ambient temperature the PCB will live in - for consumer goods this is usually 25°C - and then you apply what you consider an 'acceptable temperature rise' - i.e, how hot it is ok for the PCB tracks to rise to - again in Western Consumer Certified Devices this is usually quite low - around the 35°C mark, meaning that you are ok with the device starting at 25°C and rising to 60°C under load. For 'lab equipment' or 'not so certified or important equipment' that temp rise is often up around 50°C. And because resistance of the copper changes with heat, tracks in a lower ambient can carry higher current than the same given tracks at a higher ambient... so again, designing something to do high current in the Snow is different to designing something to do high current in the Desert.

So taking that into account, then based on the thickness of copper, width of track, peak voltage and amps, you can determine 'how much current can my PCB tracks handle'.

Its then a simple matter of doing standard engineering 'rules of thumb' - take the thinnest track you can, add 20% to it for margins and there you have the 'minimum track size' you should use.

So - when they built say the original Sang - it was designed to run only say a 5-8 amp bed - about what those large power resistors would draw. The tracks were all calculated up and they crammed the whole design onto a board based on that. Then we go and try and shove 10-12Amps through it... 'but its only a few amp more' we cry.. and yes, it is only a few amp more, but the heat has increased by squares and consequently the size of the tracks needed have increased considerably (they 'should' be almost a 1/3rd bigger again jumping from just 8A to 10A using 1Oz copper.... )

The issue then is really - you build a PCB to a certain spec - and at the currents we are running, its important to remain *within those specs*. If you want something bigger - then yes - make sure its run on a PCB with the correct specs. And as badly outlined above - its not a case of splitting things up etc or blaming 'all the heaters are the problem' - its a very very simple matter of using the "Right PCB Board for the Right Application".

This is the whole reason RAMPS and RUMBAs and RAMBO's Boards exist.....

I've more than rambled enough now... but I hope you can see - there is a fair bit involved here and we are just as much to blame (in using PCB's beyond spec)  as the designers (who may or may not have designed a PCB to an appropriate spec) and Retailers (who may be selling PCB designs for use in inappropriate spec'ed solutions). And the 'fix' is 100% not just 'move it off the existing PCB to another PCB' as that not only increases costs, it has nothing to do with the problem as the 'separate' PCB is just as prone to all these issues as the 'single PCB' solution....

<sorry for the long post - but I figured it might help shed some light on what the actual cause/issue/fix is....>

P.s.. re the 'relay' and only using it as implied reference - I hear you, but appreciate that using specific terminology like that when its incorrect is like me trying to describe how an engine works by starting the conversation about using 'rockets'.... when I really just meant 'propulsion' Sorry if it seemed I was jumping on you but its an important distinction

Indeed, that is a poorly designed PCB... but my EE comment was not about obvious design errors if you reread it in the context it was given

If it aint off a MESA Card you're doing it wrong these days http://www.mesanet.com/

Oh!!! I didn't realize they had actually got their pre-orders all shipped out yet, will be interesting to see some reviews of the maki in action from different sources

It looks like people are starting to get their Makibots shipped out now.

http://www.makibox.com/blogpost/items/m … _shipments

http://makibox.com/forum/category/60?pageof=2042

The standard printrbot extruder is rubbish, you will have a much better time with this replacement:

http://www.thingiverse.com/thing:153583

There is also a good mod to do to the bed height screws to stop them vibrating out of place.

FYI Ronsii ; Blog post on a build of a shipment today: http://cantosoft.com/blog/assembling-my … inter.html