Waiting for delivery

[jdurand]

I didn't take a picture but I just held the RMax build platform next to my old printer that's busy printing... I think the entire old printer would fit on the RMax.

Have to place it there and take a picture when it's all together, something like this one from when I was modifying my lathe:
[img]http://interstellar.com/photos/Lathe/La ... dstock.JPG[/img]
(it's an entire mini-lathe bolted to a milling machine bed)

[jdurand]

i opened up the Diablotek power supply and it definitely looks like one voltage is regulated and the others like 12V are "quasi-regulated". This is a great money saver in building a supply.

Something else I noticed is groups of the wires coming into the supply are soldered together and then a single wire goes down to the PC board. Obviously not enough holes in the PC board. So...you REALLY want to get wires from multiple groups to spread the load out, if you happened to grab all from the same group I could see the possibility of frying that one solder joint.

[jdurand]

Got all the masking peeled off and setting up a space for the Afinia to go to the RMax can sit where the Afinia is.

Running the Afinia 21 hours/day to try to catch up on orders. Got a bum spool of filament, it's on its way back to Afinia for an exchange but I lost an overnight print.

REALLY looking forward to the RMax having a fixed bed. Printing tall/large things on a moving bed is problematic. I had a damaged print yesterday when the nozzle caught on one of it's high-speed Y movements. The customer agreed to a discount so at least it wasn't trashed. But, I do NOT like hearing the occasional sound of the nozzle bouncing along the part.

[Mac The Knife]

It can happen with the Max also. Really bad when it removes the glass off of the bed.

[jdurand]

I would imagine it should be a lot less likely. The problem here is the bed is on a spring with 3 leveling screws so when a heavy part (these are 105 grams) is accelerated, the platform tends to tilt and jamb the part against the head.

Can the RMax be set up (firmware? Slicer?) to raise the head more during the long moves so it can't hit anything?

[Mac The Knife]

slic3r has a setting to raise the head during retracts. The incident with the glass came about with an evil model i drew up in solidworks, another twisted Savonius turbine. Tried slicing in Mattercontrol, and Slicer. It still fails. Even "dry" printing.

[jdurand]

Here's how our Max looks so far.

[jdurand]

Progress! Getting it wired for two hot ends from the start and using that alternate power supply. Wish we had more time on weekends but we do pickups for a food bank and getting up before the chickens makes us tired for the rest of the day.

And, of course have our regular customers wanting their stuff done. Oh well...

[Mac The Knife]

cute xmas tree.
I ran heavier gauge wire to the hotbed,, 12 gauge, but I had to clip a few strands to get them to fit in the connecter.

[jdurand]

I didn't think of it at the time, have to see how this works.

The tree was a gift from NXP back when I was contracting to them. It's an office nightlight and we decided it will go on top of the Max, I'm installing a USB charging socket up there to power it.

[jdurand]

It's ALIVE!

Well, I haven't loaded the firmware yet but no smoke. The case cooling fan is just zip-tied to the frame until I make a side panel for it to fit in.

I'm still waiting for the water pump and radiator to show up so the upper bits are functional but not prettified. I have the second extruder but waiting on that pump to know where I need to put holes in the melamine that Astroboy will be cutting up for me.

It's all wired for dual hot ends, the spares are just taped off for now.

[jdurand]

Got the printer powered up, firmware loaded with re-mapped pins, head and bed heat up just fine. Starting the calibrations on the towers now but broke for dinner.

[jdurand]

Just did some testing of heating. This is all with the $30 12V, 360W power supply.

Starting room temperature right about 25C.

I did "ABS Preheat" and the nozzle got to 200C in about 2 minutes, the bed got to 90C in about 9 minutes. I then set it to 100C and it got there in about 4:45 minutes.

I've now set it to 110C for the bed and it's rising but not super fast. Since I'll be building an enclosure for ABS, the extra 10-20C of ambient around the bed should help it quite a bit.

This is with no enclosure. I measured the power supply voltage at the Rambo board and it's 12.03V and the power supply is only warm. So far all is well.

I do notice that the flickering of the LEDs indicates a pretty funky PWM signal, either the algorithm that calculates the PID is hunting around a lot or the PWM generator is pretty sad. It isn't by any chance done in interrupt, is it? I sure hope not for anything designed in the last decade or so. In any case it seems to bumble onto the right temperature.

[jdurand]

It timed out and powered down around 109C, almost there.

[jdurand]

I restarted the heating and it got there.

[RocketMagnet]

Like your PSU solution.. very tidy.

It's pretty simple to add your % Loads to Hotend and Bed as the variables are already there in the code.
Anyway here's what I edited mine to display so I could keep an eye on exactly how well it was controlling.. I also moved the Home Towers away from the Speed adjusters (who hasn't accidentally homed their towers when altering the speed ) that reminds me I was intending it to display the filename
it was printing on the status line of the display when using SDCard... hmm looks like it's %os.. but how to reference the filename to it when you select the file?

Current Temp > Target Temp [ %Load/Power Output]

So in uimenu.h I changed it to the following for the above:
UI_PAGE4(ui_page1,"Noz:%ec>%Ec [%oC]","Bed:%eb>%Eb [%ob]","CF%Fs%%% SP%om%%% FL%of%%%","%os");

%oC = Nozzle Output
%ob = Bed Output
%Fs = Layer cooling Fan speed

It's around line 171 but I always remark out code i'm replacing so I can flip it back if it doesn't work so the line may be a different number as a result: anyway original
code was:
//UI_PAGE4(ui_page1,"Noz:%ec Target:%Ec","Bed:%eb Target:%Eb","Speed:%om%%% Flow:%of%%%","%os");

[Eaglezsoar]

RocketMagnet , thanks for the changes to the menu.
I will try to implement your code changes.

[BenTheRighteous]

(who hasn't accidentally homed their towers when altering the speed )

BLARRRGH, I hate that!!

[jdurand]

I made those changes to the menus and modified the printer name some more. Found out that even though the display is 20 columns wide, you can only have 16 character strings. So, I took off the READY and just centered the name Crowfoot.

Looking at the % output numbers for the bed and nozzle, it confirms my suspicion that the PID algorithm is hunting*. Guess I should take a look at that some point and fix it so it actually works. Could be that the auto-tune values were way off, that could do this. Seems like it needs a lot less D and maybe more I, but that would only work with having it two stage like I'd mentioned before.

* proper PID heating will bring the temperature up rapidly and then slow to a crawl just below the set temperature. It then creeps up the last little bit to lock in. You should NEVER see the %power jumping around, it should start off at 100% when cold and smoothly go down to whatever level maintains temperature. Jumps would normally be in no more than a few %.

For things like the bed, just PI without the D would work well if set right. The nozzle needs some D to kick in once it's in the control band to allow for rapid flow changes.

[jdurand]

A customer was coming over to pick up a board I designed for another project and since I have the Max together but not fully calibrated yet, I loaded a part and started it "printing" with no filament loaded and the nozzle temp turned down low. He was here looking at the new board and then noticed it running. "WHOA!" Good first impression. Then he said another customer he's doing contract work for is going to need some larger items 3D printed and he was wondering where to get that done. Then he saw the MAX v2.

Of course our Afinia is also printing away in its box a few feet away.

[RocketMagnet]

(who hasn't accidentally homed their towers when altering the speed )

BLARRRGH, I hate that!!

Presumably you've already moved it.. (kind of inspiring when you do it half way through a 12 hour print).. if not here's my code where I've simply moved it further down the same list after the Preheat ABS option in the list with all the other non printing options.. Line 381 in uimenu.h
The variable it uses is -> ui_menu_home_all

#define UI_MENU_QUICK {UI_MENU_ADDCONDBACK &ui_menu_quick_speedmultiply,&ui_menu_quick_flowmultiply UI_TOOGLE_LIGHT_ENTRY ,&ui_menu_quick_preheat_pla,&ui_menu_quick_preheat_abs,&ui_menu_home_all,&ui_menu_quick_cooldown,&ui_menu_quick_stopstepper DEBUG_PRINT_EXTRA}
UI_MENU(ui_menu_quick,UI_MENU_QUICK,6+UI_MENU_BACKCNT+MENU_PSON_COUNT+DEBUG_PRINT_COUNT+UI_TOGGLE_LIGHT_COUNT);

Arguably you could still accidentally hit Preheat PLA if you wanted to adjust Flow rate... I considered a space but I personally rarely touch flow. None of the other options are "safe" and accidentally clicking any of them will stop the ongoing print instantly so other than a space or repeating speed or flow will put a buffer between you and accidentally killing your print.

Not had the time to look into it but I'm intending to replace the Totals page with more useful "current" info (e.g. XYZ position, current filename printing if from SD Card etc).. I just don't need instant access to "historical" totals , it's stored on the Eeprom and is viewable in Repetier or MC anyway so seems like a waste of top level menu real estate to me.....

[RocketMagnet]

I made those changes to the menus and modified the printer name some more. Found out that even though the display is 20 columns wide, you can only have 16 character strings. So, I took off the READY and just centered the name Crowfoot.

Looking at the % output numbers for the bed and nozzle, it confirms my suspicion that the PID algorithm is hunting*. Guess I should take a look at that some point and fix it so it actually works. Could be that the auto-tune values were way off, that could do this. Seems like it needs a lot less D and maybe more I, but that would only work with having it two stage like I'd mentioned before.

* proper PID heating will bring the temperature up rapidly and then slow to a crawl just below the set temperature. It then creeps up the last little bit to lock in. You should NEVER see the %power jumping around, it should start off at 100% when cold and smoothly go down to whatever level maintains temperature. Jumps would normally be in no more than a few %.

For things like the bed, just PI without the D would work well if set right. The nozzle needs some D to kick in once it's in the control band to allow for rapid flow changes.

I think it's restricting the size of the variable strings to 16 when you reference a "fundamental" variable to ensure compatibility with 16X2 LCDs. You can build up to 20 (or more with a bigger LCD) if you want by combining variables, well that's the logic I think it's using. Any extra beyond the screen size are simply lost, but the SD Card section does allow >20 and scrolls the text.

Yes it's essentially ON/OFF but it does the job and holds the temps very stable in my experience +/- 1 DegC MAX but normally much much better (see your temp logs in repetier). Ultimately once up to temp and stabilised then to be considered in control you would expect the output to be In the 20-80% range. I have seen stuff around calibrating the PID so it does control better however and there are settings in Repetier I believe you can tweek.. but I don't recall where.

Not sure I totally agree with your controller logic, in this instance there is little harm in a slight overshoot especially on the Nozzle if you preheat your bed, once the flow starts the nozzle temp will immediately be challenged anyway. Get to temp as rapidly as possible seems to be the main aim so you don't want it slowing down and creeping there.. slam it on get it there and then tune it. In reality it doesn't overshoot anyway at full speed 100% heat up in my experience.. especially the heated bed way up at 90 DegC and it tends to creep there anyway. Yeah if it was flying up and overshooting to 95 or 100 DegC with a target of 90 then I'd agree it need s to slow as it gets there... but it doesn't and it just flops over the line

[jdurand]

Proportional bang-bang control? Is that some sort of Fuzzy Logic?

Mine is overshooting by a lot more than a degree (even the bed) so I'll run auto-tune again and see if that helps. I see there's a setting for control type, have to look into what they have and what can be tweaked.

My Afinia printer has lock in issues at times. I talked to a factory tech and he confirmed they sometimes see it there, but since I'm printing so much one after the other I seem to have found the trigger. If the nozzle/bed is already fairly hot when I start the next print, it may overshoot quite a bit and then ring for a while until it aborts with TOO HOT or TOO COLD part way into the print. I'm talking about swings of +/- 10 degrees C and more.

[RocketMagnet]

OK my interest was piqued so I've done some testing and IMO it does control pretty well.
I heated my bed and as can be seen it overshoots by about 1 Deg C then homes in pretty quickly.. Watching the % Output can be very misleading... as can be seen the graphs below It does control
and it doesn't just slam on/off, although I think the output values are a continuous average of the previous 30 seconds. Still once at temp it's absolutely rock solid at +/- 0.1 DegC.
It's definitely doing some calculation to hold temps once close as I'm watching it and its sitting at say 57% output for long periods. To me looks like it controlling around rate of change as it's not immediately
cutting or increasing power and it's definitely not just controlling to 1 decimal place as even when it's holding 90.0 exactly it tweaks the power output ..

So yes it's PWM controlled so its fundamentally on/off just really really fast .. but now I've looked at it I've got no issues with it at all .. works and works extremely well IMO.

10 Mins - You can see how its controlling on the green output trace at the bottom .. looks good to me.

5Mins

1min after some stabilisation time.. pretty rock solid control at target temp.

[jdurand]

But note the power levels, large jumps. That shouldn't happen. Mine isn't holding that closely, I have to re-run the auto tune before I make final judgment.