Rep2's Supid Thermocouple: What's wrong and how to (temporarily) fix.

[David Kessner]

My Rep2's Thermocouple has been increasingly failing, and it died big time earlier this week.  Since I was bot-less and bored, and I've been an electrical engineer for a LONG time, I decided to take the thing apart and see what was wrong and what could be fixed.  Let me preface all of this post by saying that I am basing this off of a sample set of ONE.  Your bot might be different.

A thermocouple is basically two wires of different types of metals.  At the business end the wires are welded or soldered together.  The other end of the wire is connected to a circuit that does special stuff to measure the voltage given off by the business end.  Of course this is a massive oversimplification, but the idea of two wires bonded at one end is fairly accurate.

What's wrong with the Rep2's Thermocouple:

The standard Rep2's thermocouple is made by stripping the wire so the last 1 or 2 mm is bare, and twisting the ends together.  This bare end of the wire is stuffed into a brass-bolt looking thing, crimped down (the bolt-thing is deformed), some heat shrink is put over it so it looks nice and offers a little bit of strain relief.  The bolt-thing is then screwed into the heater block of the extruder.

When a thermocouple fails, MBI tells you to "relieve the strain on the thermocouple cable".  Other people on this forum will say thing like, "the wires got pinched against the motor and broke".  

Let me say that in my case the thermocouple wires were not broken or stressed.  What failed is the twisting of the two wires at the business end of the thermocouple.  Remember the description, above, that a thermocouple wires are welded or soldered together?  Well, MBI just twisted the wires together.  I measured the electrical resistance between the two wires as an open circuit (greater than 20 mega-ohms).  Although clearly the wires were tightly wound together, they did not make an electrical connection!  I am not a metallurgist, but I am assuming that the wire got corroded or otherwise oxidized.  The last 3mm of the wire does have a slightly different color to it, but I cannot say what the significance of that is.

Just for grins, I cut off the last 1 cm of thermocouple wire, stripped it, and looked for breaks.  I found none.  Using the multimeter and my fingers I inspected the rest of the wire for breaks and found none.

My conclusion is that the design of the Thermocouple is faulty because MBI didn't bother to make the thermocouple correctly.  They twisted the wire instead of welding or soldering it like responsible engineers.

How it could/should be done:

Of course it should have been welded or soldered.  But the other failing is that it is just jammed into this brass bolt thing.  There is an air-gap between the thermocouple and the brass that will reduce the response time of the temperature sensor.  Also, the crimped part could damage the thermocouple cable if not done with care.

A better solution would have been to use a thermally conductive epoxy to completely fill the cavity within the bolt.  I have not looked to see if there is a thermal epoxy that can withstand 250+ deg C.  And then put some heat-sink goo in the hole that the brass bolt thing screws into.

Oh, and MBI should never be telling people that "stress on the cable" is involved.  That's just silly, wrong, and a bit of a lie.

How to repair a faulty thermocouple:

Disassemble the thermocouple by removing the cover over the motherboard on the underside of the bot.  Unscrew the terminal strip (green thing with wires and screws on/in it)  that releases the wires.  Cut the zip-ties along the black sheath that holds the wires to the gantry.  Remove the thermocouple wire from the sheath.  Next, remove the extruder from the gantry.  Loosen the two Phillips screws on the left side that hold the filament fan in place, and remove the shroud/air-guide-thing.  Then remove the two screws on the underside of the extruder with the Allen wrench.  After that the extruder just lifts off of the gantry.

Carefully unscrew the brass bolt-thing from the heater block.  Cut off the black heat-shrink tubing.  With a pair of pliers, carefully un-crimp the wire from the bolt.  Do this by applying pressure in the correct areas to make the crimped section circular instead of flattened.  Remove the wire from the brass bolt-thing.  

Cut off about 1 cm of the wire and discard.  Strip off about 2mm of the insulation, leaving the bare metal of the wires behind.  Strip off the outer insulation (looks just like clear Kaptan tape), leaving about 2-4mm of the red and yellow insulation of the wires exposed.  With a pair of pliers (or two), carefully and tightly twist the exposed wires together.  

Check the resistance of the thermocouple by using an ohm-meter from the far end of the wires.  In my unit I measured 30 ohms.  I would not be alarmed if you measured up to 100 ohms, but I would hope yours is close to 30 ohms as well.

Stick the twisted end back into the brass bolt-thing and re-crimp using a pair of pliers.  Then reinstall everything.  Make sure that when you reconnect at the motherboard that you put the yellow wire into the part that says "yellow", and red into the part that says "red".   You can use more heatshrink if you want.  I didn't.

I consider this solution to be temporary because the twisted wires are just going to go bad again.  But at least now you can fix it in the middle of the night and you don't have to wait for 1+ weeks for MBI tech support to tell you to relieve the stress on your cables.

[lassikin]

what I've done on mine on replicator 1 is to just twist stripped ends together, stick some kapton over it and stick it under the holder. that's the thing, it's just two different metals touching each other.

and as far as I understood you couldn't solder them together with solder without messing up the function? I suppose welding would be the correct thing to do but that sounds really tricky. what sucks about the rep 1 design is that the thing you're supposed to stick it under is just a washer that rotates when you're tightening the screw(would be much better if it had a square washer and rectangular indententation in the alu-block (so it wouldn't rotate and would lock properly and keep tight).

but I don't think it's too far fetched that it's a possible fault state that the wires get stressed internally and break, it's single strand after all and in general the link out from the extuder head should rather be digital.. but you know, a one dollar micro costs too much to place there.

-lassi

[Z LeHericy]

+1!!! Great tutorial, and spot on for the instructions on the repair. Also, your super oversimplification, well, isnt... that's pretty much exactly how TCs work. The red and yellow wires are two different materials (Chromel and Alumel) and at the point they meet, they generate a voltage that is related directly to the temperature of the junction. That voltage is then measured by a special chip on the motheroard (it's a very very very small voltage)

I'll also add, lead or tin solder is NOT ok to use on a thermocouple, and it's best not to solder them at all (but soldering is cheaper than welding) The introduction of more metals to the junction can sometimes change the thermal characteristics of the junction. While it's true that both sides of the new junction (chromel to lead and lead to alumel) will be at the same temperature, and therefore *shouldnt* affect the reading, lead and tin solder have marginally higher resistance than a weld or TC solder would have, and can mess with the measurement.

-Zeno LeHericy

//((=:Z:=))\\
INVENTIONS
Technologies
zinventions.com

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to makerbot+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[TobyCWoods]

 Yet another reason why I am here all the time. Thanks David!

[Dan Newman]

On 30 Mar 2013 , at 8:46 AM, David Kessner wrote:

> My Rep2's Thermocouple has been increasingly failing, and it died big time
> earlier this week. Since I was bot-less and bored, and I've been an
> electrical engineer for a LONG time, I decided to take the thing apart and
> see what was wrong and what could be fixed. Let me preface all of this
> post by saying that I am basing this off of a sample set of ONE. Your bot
> might be different.
>
> A thermocouple is basically two wires of different types of metals. At the
> business end the wires are welded or soldered together. The other end of
> the wire is connected to a circuit that does special stuff to measure the
> voltage given off by the business end. Of course this is a massive
> oversimplification, but the idea of two wires bonded at one end is fairly
> accurate.
>

> *What's wrong with the Rep2's Thermocouple:*

> *How it could/should be done:*

>
> Of course it should have been welded or soldered. But the other failing is
> that it is just jammed into this brass bolt thing. There is an air-gap
> between the thermocouple and the brass that will reduce the response time
> of the temperature sensor. Also, the crimped part could damage the
> thermocouple cable if not done with care.
>
> A better solution would have been to use a thermally conductive epoxy to
> completely fill the cavity within the bolt. I have not looked to see if
> there is a thermal epoxy that can withstand 250+ deg C. And then put some
> heat-sink goo in the hole that the brass bolt thing screws into.

As posted on this list several times in the past, Omega CC High Temp,

CC HIGH TEMPERATURE Cement for Attaching and/or Insulating Thermocouples for Temperature Measurement

It's available retail and wholesale directly from Omega and intended for
this purpose. My Makerbots have the t/c's potted with the stuff. It
has a limited shelf life, but is easy to use and doesn't require any
special equipment. Welding the t/c ends, OTOH, is best done with a
welder intended for the purpose. I've used a DCCC HotSpot in the past
but for my Makerbots, I started with already manufactured Type K t/c's.

Dan

[Dan Newman]

>> A better solution would have been to use a thermally conductive epoxy to
>> completely fill the cavity within the bolt. I have not looked to see if
>> there is a thermal epoxy that can withstand 250+ deg C. And then put some
>> heat-sink goo in the hole that the brass bolt thing screws into.
>
> As posted on this list several times in the past, Omega CC High Temp,

Sorry if that came across as critical to you David. It wasn't meant to
sound that way. It was my exasperation with the fact that potting t/c's
has been understood in the industry longer than I've been on this planet.
And people with far better street cred have blogged about this and posted
about it in the past as far back as 2010. While there's a high SNR on this
list -- welcome to the Internet, eh? -- there's also a lot of good wisdom
and it's always disheartening to me to see people continuining to fumble
around with t/c mounting.

Dan

[John Armbruster]

OK, this couldn't be more timely for me, for after two 11 hr Cell Bowl prints, I'm now having touble with my extruder. I'm assuming now it's the thermocouple, but before I attempt this fix, can someone please comfirm my symptoms are correct?

After reloading a different filament, my stepper starting clicking (Wingcommander upgrade installed), filament was dotted across the plate on the perimeter. I assumed a clog, and proceeded to go through the standard toothpick/ .4mm drill clearing process. Three times. A small piece of burnt filament was extruded, but the stepper still clicks, and still filament curls out the nozzle. I am still assuming a clog, but can this be that the heater block is not coming up to temperature? What attempted to print was very stringy and thin, as if there is not enough pressure or the filament is too cold. The LCD is giving me a 230C/230C reading.

Thanks in advance.

[John Armbruster]

And before I get into this repair, am I to understand this is somekind of homemade thermocople by Makerbot, not some off the shelf K type thermocouple to be replaced?

[Dan Newman]

On 30 Mar 2013 , at 4:20 PM, John Armbruster wrote:

> OK, this couldn't be more timely for me, for after two 11 hr Cell Bowl prints, I'm now having touble with my extruder. I'm assuming now it's the thermocouple, but before I attempt this fix, can someone please comfirm my symptoms are correct?
>
> After reloading a different filament, my stepper starting clicking (Wingcommander upgrade installed), filament was dotted across the plate on the perimeter. I assumed a clog, and proceeded to go through the standard toothpick/ .4mm drill clearing process. Three times. A small piece of burnt filament was extruded, but the stepper still clicks, and still filament curls out the nozzle. I am still assuming a clog, but can this be that the heater block is not coming up to temperature? What attempted to print was very stringy and thin, as if there is not enough pressure or the filament is too cold. The LCD is giving me a 230C/230C reading.

Likely not this problem. From what I know of both the MBI and Sailfish firmwares,
you'll see something entirely different for the reported temp. It may be "1024",
"24" (1024 truncated to 3 digits and the leading "0" dropped), or NA ("not available").
If it shows up as 230/230 then 230C is the set point and 230C is being measured.

When the t/c fails, the firmware generally knows that something is wrong and will
report either a wildly wrong number or 1024 which is it's way of saying that it
cannot read the temp. (RepG may show 1025 for a 1024 reported. I don't even
want to look at the code to see why it does that, but people have said that it
does.)

Dan

[Dan Newman]

On 30 Mar 2013 , at 4:28 PM, John Armbruster wrote:

> And before I get into this repair, am I to understand this is somekind of homemade thermocople by Makerbot, not some off the shelf K type thermocouple to be replaced?

It sounds that way, doesn't it? It's not at all unusual for manufacturers
to buy spools of chromel and alumel wire and make their own t/c's. Labs
do this all the time which is why there's a market in portable and benchtop
t/c welders. It's far more cost effective and when done properly is just as
good as buying them from a third-party.

Dan

P.S. Ideally, the manufacturers even label the leads as per the standard colors.
Hah! Standards are so awesome and useful, that having several of them is a Good Thing,
right? As a result we have many of them just for t/c leads: ANSI, IEC 584-3, BS 1843,
NFC 42-324, DIN 43710, JIS C 1610/1981, etc.

[David Kessner]

I was already looking at Omega for stranded TC wire, but I didn't think to look at them for potting compound.

No feelings of being over-critiqued here.

[Dan Newman]

On 30 Mar 2013 , at 4:54 PM, David Kessner wrote:

> I was already looking at Omega for stranded TC wire, but I didn't think to
> look at them for potting compound.

My current batch is out-of-date. It has a shelf life of about 1 year.
If you order some, it may be worthwhile to call them on the phone and
ask that they not send you stuff with under 8 months left to go. (I once
received some with ~5 months left. I called them and they exchanged it.)
And you will get far more than you will need.

It's a two part chemical: powdered zirconium silicate, and liquid
sodium silicate. The latter is something many of us played with as
kids: it was the clear liquid component of something sold as "Magic Rocks".

Dan

[lassikin]

99.9% of the time the symptom would be the bot showing a message along the lines that "my temperatures are dropping check the thermocouple connections".

you could be just having some gunk in the barrel/nozzle. you could try a higher temp for a while, and do reload/load couple of times.

-lassi

[Eighty]

John, if you're reading temps ok, and you're getting failures far into prints, you probably have hear creep issues. There's another active thread about this right now...
https://groups.google.com/forum/m/?fromgroups#!topic/makerbot/-HnE2vwBHEY
Your issue is more about using 230C on a print that only extrudes intermittently. Lower your temp to 210-220, and are what happens. Also try an external fan across the print...

[John Armbruster]

thank you, Dan, that's why I am assuming a clog; I wasn't getting the 1024/24 status. But I have taken the extruder/heater unit off the gantry to examine it, and for the first time took the cover off the motherboard to get familiar with the repair and I have two impressions:

1) without the expertise of all of you on this forum; there is no way I would have been happy with this product - and there is no way I would feel comfortable diving into repeated disassemblies of a $2100 machine with no real owner's manual.

2) the more I disassemble it, the more concerned I am with unintentionally breaking something I can't fix. Robust design does not describe it. Just like the X-axis control wire/rollpin issue, the thermocouple wires are bent at a 90 degree without support - and both wires were barely holding in the buss on the motherboard.

While the heater block is loose, I'll remove the nozzle and burn it out; and tighten the thermocouple wires. I'd like to post a picture of the heater block, though, because looking at it, I'm not sure (should I have to down the road) how the brass fitting works. Do you loosen it at the hex head?

[John Armbruster]

Eighty: no, literally went straight from a successful 80% scale Cell Bowl which finished perfectly, to a change in filament and massive stepper clicking. As hard as it is even for me to believe, the printer (with the extruder upgrade and all the preemptive fixes posted on this forum) has given me amazingly great service for three months. Wingcommander: actually, yes, it probably is slightly warmer today; I hadn't considered that. But it so acts like a classic clog that this is what I addressed first. Used your technique, and then went so far as to go to Harbor Freight to get a .4mm drill that was recommended, that it had me start to wonder about the heater block temps being correct. I'm hoping that perhaps it is a clog too large to push through the nozzle, and that burning it out will do the trick. I'm hoping it was just because I used the Matterhacker gold for it's first long print that perhaps it left something major burnt in the nozzle.

I know I've read some folks have just printed something at temps up to 245C to clear out clogs...if that is recommended, how long a print is safe before overheating the extruder upgrade?

And again, as this is one of my first troubleshooting posts after lurking for 3 months, let me take the opportunity to thank everyone for all the support. I'm sure I speak for a large, silent crowd of fellow lurkers - this forum has been the absolute best part of the Makerbot experience. Ya'll are some bright people!

[Eighty]

And since you mentioned Matter Hackers filament:  I've had some odd experiences with it.  I like their stuff, but their formulation needs temperatures down in the 190-200C range (for me).  If I extrude their filament at my normal 220C temp, I get chunks and stringiness.

[John Armbruster]

Yes, I noticed that, and I've been very happy with the print quality and surface finish of their more translucent colors (especially the red and dark blue), though they have a slight stringiness in the bridging of objects like the cell bowl. I had only used the more opaque white and gold to do much more simple, generally solid objects for work. Today was the first time I noticed the gold do the chucks you describe, but indeed, it was there. I need to use Henry's new bridging calibration tool to adjust accordingly, especially, as he says, ambient temps are increasing.

[John Armbruster]

thank you, everyone. Apparently is was just a major burnt filament clog. Disassembled the nozzle and burnt it clean with a torch, everything is reasonably fine. Did lower the temperature of a calibration block print to 220 on Matterhacker white, and the surface finish is noticeably better. I'm going with Wingcommander's observation that ambient temperatures are up, and am getting a digital thermometer to stick on the frame.

[John Armbruster]

I have to go back to the thread that discussed the effectiveness of the cooling fan / heat sink. Now that I've had it apart a few times, I have to ask the same question: what is the fan cooling and what is the stock heat sink supposed to be drawing heat from (especially with the upgraded extruder)? If the goal is to cool the filament at the drive gear, the heat sink pretty much blocks all the fan's air flow to that area. The heat sink doesn't actually touch much of the extruder upgrade to cool it, and that does nothing for the gear and the filament. Surely, instead of a heat sink, wouldn't some kind of duct/nozzle from the fan to the drive gear be better? With a shroud to make sure the cool air isn't directed down toward the heater block?

[Eighty]

Well, it's supposed to draw heat from the aluminum block below, that houses the nozzle. It's just an unfortunate arrangement, as it tends to cook the PLA passing on the other side of the heat sink.
Someone on a Rep1 reported a while back that they chopped down the heat sink to make for more airflow at the drive gear. Said it works great. I haven't had the cojones to do it myself, as I'll void my warranty. Oh, wait. I forgot...no warranty. Guess I'm just too lazy to try it, I guess.

[John Armbruster]

so it draws heat from the block through the approximately 3/16" inch it touches across the bottom? Just for fun, I'm going to make up a small 90 degree flange at work tomorrow that replaces the heat sink and seals the hole in the trolley in front of the aluminum block (in lieu of the business card and similar to http://www.thingiverse.com/thing:31872). The heat sink is 15mm thick, I don't see why I couldn't just mount it on the back of the stepper motor as Henry did with his 40mmx40mmx10mm. 

[Eighty]

Aluminum is very good at conducting heat. So it only needs a little contact to draw the heat away.

Not sure I'm picturing your idea well enough. I'd love to see pics and hear how the experiment goes.

[John Armbruster]

Oh, I agree. And the flange I make will be aluminum as well, with the same contact area. With access to the same cooling air from the fan. Forgive me, as a 3 month n00b to both printing, Makerbot, and this forum, I'm trying not to sound as ignorant as I know I am, but I'm really trying to understand the design of the extruder now that I've had it apart. Without rehashing everything mentioned in the other thread about the heat sink, the common problem seems to be a fear of heat soak from the heating element up the filament to the drive gear. There's a cooling fan. It cools a heat sink. The heat sink blocks the things we want the fan to cool: the filament, stepper motor, drive gear, and the plunger assembly or extruder upgrade.   The heat sink does touch a small part of the aluminum block that is basically the common mount to the trolley for the stepper motor/extruder/drive gear above and the heater block/nozzle assembly below. The indirect cooling of this block probably cools the hole in the block where the filament passes through after the drive gear to the heater block. But this hole would most likely be cooler if air could pass on it directly from the fan if it weren't blocked by the heat sink. 

Do I have this correct?

[John Armbruster]

Henry - when discussing the stepper motor heat; is this concerning the motor primarily becoming overheated by heat creeping up through the assembly from the heater block, or the heat generated by the motor itself? Is there a thought that the motor heat itself is heating the drive gear through the axle? I can almost believe this might be the real issue of filament softening.

[John Armbruster]

No, no, that was one of my questions. I was wondering why you had the heat sink on the motor, other than it makes sense to keep all the components above the cooling block as cool as possible. Up until this thread, I hadn't even considered that the motor might generate sufficient heat and that could be contributing to the problem. I guess what's new to me after looking at it is the reference to the common block of aluminum that ties all the components together to the trolley as the "cooling block." To me it's a chassis piece that just happens to need to be cooled. But for the life of me, the use of the heat sink to cool it indirectly through a narrow contact patch rather than cool it directly didn't make much sense. Especially when the heat sink blocks air flow to the extruder and stepper motor.

On Monday, April 1, 2013 12:19:50 AM UTC-4, Wingcommander whpthomas wrote:

There's a cooling fan. It cools a heat sink. The heat sink blocks the things we want the fan to cool: the filament, stepper motor, drive gear, and the plunger assembly or extruder upgrade.

John, I am going to step in here and attempt to correct your mental model of what you think is going wrong here.

One has to learn to be forensic - i.e. you start without a theory and attempt to rule out possible causes systematically. If we start with a theory, that we try to confirm, we risk confirmation bias in our diagnosis.

I have observed that the stepper motor, even when its is too hot to touch, has never been hot enough to cause the filament to actually soften and fail - except when I had a problem with KISSlicer turning the heatsink fan off - which was the only time the filament has gone soft all the way up to the drive gear.

In my experience, heat soak causes problems further down in the heat chamber, somewhere between the heating block and the cooling block, causing the filament melt to happen higher up than the tip, causing it to push up the sides and jam.

[John Sondericker]

Being a EE (for more than I to admit) as well I appreciate your eloquent explanation of your problem and solution. Great info. 

On Saturday, March 30, 2013 9:46:10 AM UTC-6, David Kessner wrote:

My Rep2's Thermocouple has been increasingly failing, and it died big time earlier this week.  Since I was bot-less and bored, and I've been an electrical engineer for a LONG time, I decided to take the thing apart and see what was wrong and what could be fixed.  Let me preface all of this post by saying that I am basing this off of a sample set of ONE.  Your bot might be different.

A thermocouple is basically two wires of different types of metals.  At the business end the wires are welded or soldered together.  The other end of the wire is connected to a circuit that does special stuff to measure the voltage given off by the business end.  Of course this is a massive oversimplification, but the idea of two wires bonded at one end is fairly accurate.

What's wrong with the Rep2's Thermocouple:

The standard Rep2's thermocouple is made by stripping the wire so the last 1 or 2 mm is bare, and twisting the ends together.  This bare end of the wire is stuffed into a brass-bolt looking thing, crimped down (the bolt-thing is deformed), some heat shrink is put over it so it looks nice and offers a little bit of strain relief.  The bolt-thing is then screwed into the heater block of the extruder.

When a thermocouple fails, MBI tells you to "relieve the strain on the thermocouple cable".  Other people on this forum will say thing like, "the wires got pinched against the motor and broke".  

Let me say that in my case the thermocouple wires were not broken or stressed.  What failed is the twisting of the two wires at the business end of the thermocouple.  Remember the description, above, that a thermocouple wires are welded or soldered together?  Well, MBI just twisted the wires together.  I measured the electrical resistance between the two wires as an open circuit (greater than 20 mega-ohms).  Although clearly the wires were tightly wound together, they did not make an electrical connection!  I am not a metallurgist, but I am assuming that the wire got corroded or otherwise oxidized.  The last 3mm of the wire does have a slightly different color to it, but I cannot say what the significance of that is.

Just for grins, I cut off the last 1 cm of thermocouple wire, stripped it, and looked for breaks.  I found none.  Using the multimeter and my fingers I inspected the rest of the wire for breaks and found none.

My conclusion is that the design of the Thermocouple is faulty because MBI didn't bother to make the thermocouple correctly.  They twisted the wire instead of welding or soldering it like responsible engineers.

How it could/should be done:

Of course it should have been welded or soldered.  But the other failing is that it is just jammed into this brass bolt thing.  There is an air-gap between the thermocouple and the brass that will reduce the response time of the temperature sensor.  Also, the crimped part could damage the thermocouple cable if not done with care.

A better solution would have been to use a thermally conductive epoxy to completely fill the cavity within the bolt.  I have not looked to see if there is a thermal epoxy that can withstand 250+ deg C.  And then put some heat-sink goo in the hole that the brass bolt thing screws into.

Oh, and MBI should never be telling people that "stress on the cable" is involved.  That's just silly, wrong, and a bit of a lie.

How to repair a faulty thermocouple:

Disassemble the thermocouple by removing the cover over the motherboard on the underside of the bot.  Unscrew the terminal strip (green thing with wires and screws on/in it)  that releases the wires.  Cut the zip-ties along the black sheath that holds the wires to the gantry.  Remove the thermocouple wire from the sheath.  Next, remove the extruder from the gantry.  Loosen the two Phillips screws on the left side that hold the filament fan in place, and remove the shroud/air-guide-thing.  Then remove the two screws on the underside of the extruder with the Allen wrench.  After that the extruder just lifts off of the gantry.

Carefully unscrew the brass bolt-thing from the heater block.  Cut off the black heat-shrink tubing.  With a pair of pliers, carefully un-crimp the wire from the bolt.  Do this by applying pressure in the correct areas to make the crimped section circular instead of flattened.  Remove the wire from the brass bolt-thing.  

Cut off about 1 cm of the wire and discard.  Strip off about 2mm of the insulation, leaving the bare metal of the wires behind.  Strip off the outer insulation (looks just like clear Kaptan tape), leaving about 2-4mm of the red and yellow insulation of the wires exposed.  With a pair of pliers (or two), carefully and tightly twist the exposed wires together.  

Check the resistance of the thermocouple by using an ohm-meter from the far end of the wires.  In my unit I measured 30 ohms.  I would not be alarmed if you measured up to 100 ohms, but I would hope yours is close to 30 ohms as well.

Stick the twisted end back into the brass bolt-thing and re-crimp using a pair of pliers.  Then reinstall everything.  Make sure that when you reconnect at the motherboard that you put the yellow wire into the part that says "yellow", and red into the part that says "red".   You can use more heatshrink if you want.  I didn't.

I consider this solution to be temporary because the twisted wires are just going to go bad again.  But at least now you can fix it in the middle of the night and you don't have to wait for 1+ weeks for MBI tech support to tell you to relieve the stress on your cables.