MakerBot Industries

Excellent thoughts.

1. As the motor has issues both loaded and unloaded, I'm not sure I need to load it to test. Part of this is I'm just not sure how to load it other than actually extruding, which seems a waste.

2. Is there a datasheet? Do you have any further info?

And, if I may ask, as you seem to be have the right qualifications, in your opinion, what do you think is causing these issues?

Have a read through Ed Nisely's recent posts regarding all the improvements he's been making to his ToM electronics.

Very thorough!

I'd like to help step through the issues with you, but I'm leaving for a week vacation tomorrow morning. I can offer a few hints as to where I would start diagnosing what is going wrong, however.

It sounds like the motor is fine, since it runs OK off of a basic 12V power supply. This makes me think that it is either the extruder controller or the ATX power supply that powers the ToM.

Taking a voltage reading of the 12V line on the PSU during a run should let elliminate that pretty quickly and easily, so do that first. If it drops below 11V or so, that could be the problem.

Next up is the extruder. I haven't taken a close look at the schematic, but it looks like a basic H-bridge being driven by an Atmel uC.

If you have access to an o-scope, take a look at the motor power and the H-bridge control signals to see if the PWM on both of those looks correct. At 255 (assuming the code is correct and sane, I haven't looked) You should see a 100% duty cycle in both, with a possible very short glitch as the PWM timer rolls over. At 128 you should see a 50% duty cycle, etc.

I would expect that the PWM coming out of the H-bridge to the motor is bad below 255, but the question is it correct going into the H-bridge? If so, then the H-bridge is probably the root of the problem. If the signal isn't correct going into the H-bridge, it points to a bad uC or bad code.

If you don't have access to an o-scope, you can kinda do the same by looking at the DC voltage at the motor terminals with a multi-meter. It won't be nearly as good of an indicator, but you should see the average input voltage vary up and down as you vary the duty cycle.

It might also be interesting to see what the motor current is at different PWM levels by using your multimeter inlie in current mode.

max6675 is ic8 right?

So… I would solder a jumper wire across the blown out area to c5. Leave the Thermoresistor disconnected? And see if I get an open circuit 17 C current temperature reading? How do I determine if the max6675 is still working exactly?

After that, I just put a new resistor between the t- connector and the max6675?

Sorry, I am not that versed in doing this. I mean, this looks kind of easy to fix, I can see the broken connection, but I don't have too much experience in fixing electronics in this manner.

Also

I've only had this a week, is there a chance they will send me a new board and will attempting to fix this board like that ruin my chances of RMA'ing it and getting a new one?

It would be nice if I could get it working while waiting for a replacement though.

And also, you are saying that there was some kind of short up by the extruder, and that is what blew it out? Does this happen often? What usually causes the short? It worries me using the JB weld, because if I hook it up wrong in some way, then it is stuck that way and would be hard to fix no?

Edit: Nevermind that last bit, I just realized you said seal it with the jb weld, then clamp it under the washer. My mistake.

Thanks,

Renosis

Edit: Sorry, I knew I should have stayed in my other post about this problem, because it is kind of off topic. Anyway, I just put my multimeter on it, and it reads c5 reads 4.6 OHMs. So I assume that resistor is blown, correct?

Also did the same thing to the red and yellow leads on the thermo resistor cable and it read 5 too.

Ed,

Thanks so much for your help.

I just want to be sure of what you are saying here. I hooked up my multimeter to the red and black wires coming from my extruder (the wires that connect to my power resistors that heat up the block). The reading was ~5.4 OHMs. Does that mean that my two resistors are blown and need to be replaced as well as my extruder controller board?

Also, how can I prevent these power resistors from shorting out like that in the future?

I will do what you said to the tip of thermocouple with JB weld. That is a great idea. It won't effect temperature readings much, will it?

Only one of the two heater resistors is blown; you'd have infinite resistance if both were dead. You'll have to disconnect them from each other to figure out which one because they're hooked up in parallel, or you could try applying a little power to see which one heats up - the working one.

The resistors didn't short themselves out. One of them likely blew out (they aren't designed for this sort of heating application) and made contact with the heater block, which in turn found the path of least resistance up the thermocouple leads to kill the MAX6675. Insulating the thermocouple will at least stop that from happening again.

Identify the blown heater resistor and that's you're most likely justification for a simple RMA/replacement request to MakerBot for both a new extruder and 5 Ohm power resistor (request two just to be safe). Marginal parts die, especially when stressed; it happens.

Thanks alot clothbot. I really appreciate the help. You too Ed.

Well, I am gonna have to take it all apart anyway, so I'll test both resistors tomorrow.

That's a load off my mind that I probably get an RMA for this. I checked the price on extruder controller and I wasnt looking forward to dropping 90 dollars plus 8 for the resistor plus shipping for a part that didnt last a week. I mean, i just dropped 1255 for the TOM not long ago.

I am definately sealing that tip with the Jb weld, so if this happens again, i won't blow out my controller again. 8 bucks is reasonable, 98, not so much.

JB weld is only rated for 500F for a few minutes. After tens of hours it turns to dust at extruder temperatures.

MBI is sending me out new parts! I wanted them for the weekend so I offered to pay for overnight shipping. They were able to send it out today and everything. I knew MakerBot was good to their customers, but I am floored, I am one happy loyal customer now :)

Oh and apparently they are going to change the build manual to include electrical isolation of the thermocoupler!

The motivation for using JB Weld epoxy is that it's cheap, readily available, and will get you over this particular hump.

You really want high-temperature cement from the likes of Omega, but that's twenty bucks for a small can… and you only need a little dab.

Pot the thermocouple in epoxy, wrap that in Kapton, snug it under the washer against the Thermal Core, and it'll work fine for long enough…

Actually they say 600F for ten minutes, 500F continuous but I have used it to glue thermistors and it always failed pretty quickly. Makergear.com sell suitable high temperature ceramic and epoxy glues from Cotronics. I haven't tried those myself yet.

How about electrically isolating the resistors from the heater block? They use thin mica insulators on TO3 power transistors to transfer heat to the heatsink. They also prevent the case from grounding out against the metal.

We'd have to figure out how to attach the resistors to the block since the screws would also complete a connection…

Ok, well I am willing to spend 20 bucks on any kind of glue that will prevent my 90 dollar extruder controller from getting damaged. This happened already once in five days of operation, I am willing to bet that 20 that it will happen again and soon.

I don't have much knowledge in adhesives. You say cement from Omega, is there any way you can point me to the exact stuff I would want to use?

Someone (I've lost the link) referred me to this page:

Omegabond Chemical Cure High Temperature Cement

and it looks like the right stuff…

Here's where we had that discussion in the MakerBot Google Group: Thing-O-Matic Hackage

Thanks Ed and Nophead! I got some Omega on the way and fast.

I didn't have to tweak things at all, but for bonus points I set oozebane to turn off the extruder motor early to match the 600ms delay.

This would also explain the motor behavior I've seen. I would not recommend taking apart the motor, though, unless you have some skills. I butchered one of mine (but took pics so that others could see):
http://wiki.makerbot.com/forum/t-296624/inside-the-mk-5-extruder-motor
If the relay circuit helps with the missing voltage, that also explains why that solution works.

Very good points. Let me offer some additional thoughts.

I observe plenty of failures with the motor un-loaded. If it was truly "bad" and suffering the bad segment issue you are talking about, why is it intermittent and shouldn't it be more likely to occur under a heavy load? I have not observed this to be the case. Just playing around with the motor on the dc supply last night and it seemed happy to chug along at anything from 2 volts all the up to 24. Note: it would run at 2 volts "with a load".

2nd, this does not address the extruder boards failure to PWM below 255, which it fails to do with nothing hooked up at all, though it will show 12.3v at 255, it drops to 0 at 254 or any other setting I've tried. It repeats this behavior with any motor or resistor load I have tried.

If "some" of the extruder boards will PWM under 255 they all should, or something is wrong.

Ok, while some of my original assumptions may have been wrong, I have identified an issue on the new extruder boards and have a fix:

http://www.thingiverse.com/thing:5630

It's likely the bad motor issue can contribute to this and make the problem worse, but the cure is simple-take out the stupid ESD diode array that wasn't on the version 2.2 boards and everything works.

I just received unofficial confirmation that non-255 PWM values were disabled because they were strongly linked to an unusually high number of A3949 failures.

Just got more info. PWM was disabled to reduce issues with the A3949 chips and so it would be compatible with controlling relays for driving the DC motor and/or the heater. Relays and PWM definitely do not mix.

PWM should be reenabled in the next Gen4 firmware release.

I'll be trying Ed's MOSFET Supply Bypass and 5VDC Supply From Molex hacks first.

The older board didn't have to supply as much 12VDC current directly to the heaters as the Thing-O-Matic/Gen4 does. My Batch 1 Cupcake's still chugging along happily with its original DC motor and nichrome-wound heater, in spite of me having soldered the board myself. ;-)

Ok, so just desoldered both of my $90 extruder controller 3.6 boards and the SMS12 (IC3) came off intact. The trick was to add solder so that it bridged all three pins on each side and allowed thermal contact to the large tip of the radioshack desoldering tool. The IC simply slide off and left the board intact without damaging the solder pads. I'm pretty confident I could solder the SMS12 back on if I had too. Looking at the circuit schematics for the old 2.2 boards and the new 3.6 boards, this is the only difference. I'm also curious that I was thinking the diode array was shottkys but it might be 12 volt zeners and thus further explains why it would be brought into conduction by the motor (the SMS 12 is a clue-probably 12 volt zener) The design theory was probably that the H-bridge would not let higher than 12 volts-1.4 volts (diode drops) out of the outputs-and thus never conduct across the 12 volt zener. If the power supply is a little high-13.5 volts, it would in theory be high enough to "trigger" the zener into conduction and thus short out the A3949. It also could be the ringing of the motor indcutor such that the frequency causes the diode to conduct at the same time the next pulse is coming from the A3949. This does appear to be an easily correctable flaw by just removing the diode array-and ends up being the same as the version 2.2 board.

Now, who knows what extruder firmware 3.6 allows for PWM so I can try and blow this $90 board up?

TVS diodes are like Zener diodes but less accurate and without the hard-stop at their reverse-bias voltage, neither of which short all the way to ground - you wouldn't be able to use Zeners as voltage regulators otherwise. A regular zener is designed for accuracy and burn out more readily if the voltage*current goes too high; TVS diodes clamp more gradually and can survive significantly higher peak currents.

In the case of the SMS12, it only starts to reverse bias at 13.3V. The voltage continues to rise but there's a knee in the curve such that the voltage is clamped proportionally to the current flowing thru it. That's what the Clamping Voltage vs. Peak Pulse Current curve is all about on pg 4. in the datasheet. I.e. to a max of 19V when 5A is passing thru and 23V when 15A is passing thru. The max peak pulse power is 350W.

As it relates to the A3949, the A3959 FAQ mentions connecting Schottky diodes between outputs and VBB and GND. The SMS12 acts like a simple Schottky when forward-biased (i.e. clamping voltages on the output lines less than GND). At higher voltages the TVS aspect kicks and clamps the outputs to a softer GND+Vclamp instead of forward biasing a Schottky to VBB=12VDC. The clamped voltage is still going to be well below the max 36VDC supply rating for the A3949.

I think I'll order a couple of A3950 to swap one in and monitor NFAULT signal on pin 1. That'll at least flag when there's a motor fault (bad motor), undervoltage condition (bad supply) or the internal junction temperature (Tj) gets too high - she canna take much more'o'this!

Yes, I did over simplify the spec sheet and made it sound like the diode would clamp at 13.3 volts which is still somewhat true. What I think we should be saying though is that this type of device is intended for an input- not an output. Why is it that the exact same A3949 works all day on my old extruder controller V 2.2 and even drove a steppper with some decent inductance without any external clamp diodes yet this 3.6 board craps out with the same chip? The only difference is the addition of the diode array. I know that PWMing the motor can produce at least 19 volts or more back towards the controller. At that point, it doesn't matter what the voltage the A3949 is trying to put out- it's shorted back through the diode to ground. This would overheat the A3949 and go into shutdown, thus the motor stops. A build that had a lot of starts and stops would cause this to happen. This explains why people saw the same stop point in a given file. The bottom line is that there is nothing to loose by removing the diodes.

Well, since you're removing the SMS12 anyway, make sure you don't let PWM speed drop into the speed=10 range without at least adding old-school Schottky diode protection of some kind on there. You'll see <50ns pulse-width spikes around 29V high which do have the side-effect of frying the A3949.