... and its just the 'usual' LED behaviours (it comes and goes.. I could go into a long winded explanation but'll bore everyone to tears...) ...

Please do -- or, a couple of quick leads for further searching would also do (casual search in forums or google didn't turn up anything). I'd like to eventually understand what's going on, even if it's "normal". Thanks!

Finally, fully remote operation! Now I can leave printer running, and use cellphone to check webcam and properly switch off printer wherever I am.  If anyone else cares about details, I can post more about circuit and minor firmware mod (basically a #define and 3 lines of code in main loop). Although at ~$15 BOM (plus labor) this is admittedly not the most cost-effective approach unless if you're also doing it out of personal curiosity, and can probably be avoided if you also plan to do further upgrades in future (eg, ATX PSU, electronics, etc -- but if I reach that point, I'll get another printer instead or perhaps build, say, a Kossel.. if only I had the time ).For now, it's nice, though!

Ambient room temp is ~25C (+/- 3C).  If it goes close to 30C, with the humidity in this place, *I* suffer (nevermind Pololus), so I switch on A/C.   Now that I'll probably be doing longer/unattended prints, we'll see.

Thanks!

-----------

Just ran printer (after removing the board), didn't see any LED flickering/dimming (certainly not to the extent it happens with the board inserted -- I'm pretty sure I'd have noticed it before if it was that severe).

Also, here is an example of symptoms I started having:

Exact same G-code, the one on the left printed before connecting the board, the one on the right printed after *dis*connecting the board.  I did only one print in-between these two (the one with the board connected), and didn't touch anything else on the printer (other than re-flashing firmware to change PS_ON_PIN etc).  The "wiggly" layers are filled (box bottom), the decent ones are 2mm thick walls.

Never seen this before.  If anyone has ideas where to start my debugging, that's be great.  Is it possible that, if the board was not providing the necessary current, it has thrown something out of whack?  Argh...

In meantime, I'll make a simple harness so I can easily connect board and break circuit to measure current, so there's some numbers...

Also, wrt electronics cooling fans: no, but I've run up to 6hr prints and never had issues before, so it didn't seem necessary.  Plus, these problems appear from the very beginning of the print (with printer switched off before printing).

So far, sounds like the board gave you no issues, and its just the 'usual' LED behaviours (it comes and goes.. I could go into a long winded explanation but'll bore everyone to tears...) and potentially overheating steppers...

Here is the prototype, back:

and front:

(I have no training in EE, but I had etched a simple analog mux PCB in the past just out of curiosity for process -- this is my second ever PCB)

The actual components are:

Power MOSFET FQP27P06
Control MOSFET VN2222LLG
Reverse input polarity protection diode VFT2045BP-M3/4W

Regulator for RasPi works fine, board responds to M80/M81, and everything is fine with printer in idle just heating the bed.  However, under full load (when printing), I get what seems to be like brownouts: LED strip visibly flickers and (I think) the steppers occasionally sound funny.  I ran a 2hr print which completed, but the walls came out wavy and, in certain parts, "squished" on the Z-axis (but no layer shifts or anything).  Interestingly, the problem remains after re-connecting directly to the PSU (still investigating that).

I was concerned at first about how much the diode and MOSFET heat up, until I touched the Sanguino heated bed MOSFET tab and blistered myself (even hotter, lesson learned ).  My PCB's traces get hot, but not too hot (I can comfortably hold my finger on them for any amount of time), definitely not as hot as the MOSFETs.  For size reference, the copper pour at bottom right (+12V output) is ~400mil wide.  Also checked wires (using 16AWG, tinned) didn't see anything signs of overheating or improper connection.

Any ideas?  I'm out, other than blindly trying higher rated components and/or removing the protection diode. If you see an easy fix, I might revise, otherwise I think I'll just admit my ignorance when it comes to circuits and shelf this.

1. Do I *have* to use a P channel MOSFET for high side switch?
2. Is there any way that a logic-level (5V) signal is sufficient to turn on the MOSFET on high side (it's switching 12V), or is a separate control circuit necessary?
3. How does "logic level" translate into specs? A particular Vgs(thresh) value? Or..?
4. How do you prototype high-current circuits? Anything that doesn't involve soldering/desoldering (which is what I do now)?

Thx!

Edit: FWIW, attaching what I'm going with (pls ignore component names, mostly picked randomly to match package types). Basically the regulator circuit is copy-pasted from datasheet, and the high-side switch is pieced together based on Google results: a P-channel MOSFET with a pullup resistor to 12V, controlled by a smaller, secondary logic-level N-channel MOSFET connected to the Sanguino, with a pulldown resistor. Components won't be here till next week, so if you happen to see anything horribly wrong till then...

Edit: shorting the 5V digital out pin to 12V is obviously a bad idea (doh!), but other than that...?

1. Most serious: is it even possible to do this when the circuit being switched is also the one which controls the MOSFET?  In particular, when the Sanguino digital pin (connected to the gate) goes low, then basically the gate is shorted to the *drain* and, based on googling, this is not good.  (Of course, having the control circuit be the one switched makes no sense, unless it also gets power from elsewhere: in this case, through USB from the RasPi -- my Sanguino is always on in that way, and what I'm trying to achieve is make a digital out pin have the same effect as manually plugging/unplugging the 12V SD PSU, which is what I currently do).

2. Wrt combining an always-on regulator (for the RasPi) in the same circuit: problem is that all USB ports share the same ground, so the RasPi closes the ground circuit (in attached schematic, the RasPi connects USB GND to pin 2 on PS_OUT -- pls ignore the 10K resistor permanently shorting gate to drain, I realized the stupid mistake after taking screenshot, pulled out resistor but was too lazy to take new screenshots -- still, flaw noted above remains even w/o that resistor).

Any hope of making this actually work (with or without the regulator, preferably with)?  Any help would be appreciated, as I'm completely clueless when it comes to even relatively basic circuits...  thanks!

MOSFET it is.  Do you think a max current draw of ~0.5A (based on this, not self-measured) or ~3.5W waste heat(?) justifies the extra trouble for a switching instead of linear regulator (e.g., this -- does the inductor really have to be rated at 5A, btw?).

Use an N-MOSFET to switch ground off on the to Pi.

The above schematic has the filter circuit to remove ripple..

M80/81 a mosfet will do the trick - Mosfet and 2 resistors (1k for inline with the Gate to limit current to the Mosfet, and a 10k on the Gate to Gnd after the 1k to ensure the mosfet really is off when its off (and gate isn't floating)). But if you're pulling 12.5 Amps - make sure to get a High On Resistance, Logic Level MOSFET thats capable of say 60V/30A , but 100V/75A is probably more preferable...

Using a relay is possible - but you need to be careful as 12V high current relays will draw a quite reasonable amount of current just to turn on - often you'll need a transistor to drive them from a Micro pin.. they also will 'crack' lots as the physical contacts make/break... so much better to use a MOSFET... if you do end up with a relay, make sure its Coil-Resistance is very very high and use Ohms Law to calculate that at 5V you wont draw more than 20ma through the Micro pin... For example, 120Ohm is not unusual to see for the coil-resistance, and at 5v, that will draw 41ma - more than you can safely provide via an Atmel... so you'd need to use a transistor to switch the relay, and the micro switching the transistor... at which point, you may as well have used a MOSFET

1. MOSFET or relay? Anything more to pay attention to, beyond current rating?
2. If regulator is in parallel (RasPi always on), what do I need to filter noise (been bitten by this last time I tried semi-blindly on a different project, still not sure why)?

Any pointers to relevant info appreciated.