Just burned out a RAMBO FET (sigh)

[Greg. L]

Was installing my brand new extruder fans tonight on my RAMBO board. I'm an electrical engineer, so I'm working without fear or worry, happily throwing a new wire/header into the screw terminals on the RAMBO. Got my new fan installed, ready to connect the header, and send the M106 command. I grab my voltmeter to test to make sure I'm getting 12V out of the header when my wife walks in and bumps into me, shorting the two pins of the header across one of the leads of my meter.

There's a brief spark, and a pop.

Now I don't have an extruder 1 FET anymore.

I can (theoretically) disassemble the whole machine, take the board to work, and use our rework stations to replace the FET, or just buy a whole new RAMBO board.

I really liked the monolithic board idea until today. Until today, I said "Yeah, but average users shouldn't be doing anything that would affect the board, and advanced users (e.g. me) shouldn't have problems, so monolithic is the way to go."

But tonight I burned out a FET, and it'll be hours of work to get it uninstalled, fixed, and reinstalled, or a 200$ cost, plus all the install work anyway. Sigh.

Thanks for letting me vent. 

On that note, can I just use Extruder 2 fan output for now? Are they both on the same M106/107 commands in standard Marlin?

[JohnD]

I think you'd have to adjust you fan pin in pin.h.  For what it's worth, I'm an EE as well, and I *hate* monolithic design - I've had the same sort of experience, but didn't have the wife to blame it on.

My favorite board by far is the RAMPS 1.3 - almost all the goodness needed in a through hole design.

I've been thinking about my ideal board, and I still think I'd keep the Pololu footprint and pinout for stepper drivers, but I would add a "heater / high power" board as a separate, stackable shield - keep all the 'fets where I can cool them, fuse them, and in general keep 'em away from everything else.

I'm to the point now where I want heated print bed, dual extruders, and dual fans - so at least 5 mosfets... 

Maybe the real dream should be a 'bot where I don't have to play with the electronics! :-)  Not sure I'd actually enjoy that.....

[Triffid Hunter]

On 21 February 2013 11:34, Greg. L <greg...@gmail.com> wrote:
> There's a brief spark, and a pop.
>
> Now I don't have an extruder 1 FET anymore.

does rambo not have a fuse that should have saved the mosfet?

> I really liked the monolithic board idea until today. Until today, I said
> "Yeah, but average users shouldn't be doing anything that would affect the
> board, and advanced users (e.g. me) shouldn't have problems, so monolithic
> is the way to go."

If you popped a mosfet on RAMPS you'd have exactly the same trouble.
Removing a TO-220 from a board when the pads have no thermals
generally requires similar equipment to removing a smd fet.

the main issue with modular boards is that there's either a lot less
surface area available for heatsinking (pololus), or you end up with
numerous boards scattered everywhere (gen3)

I'm firmly convinced that the high rate of death of the pololus is
directly related to their modular nature. The only ones I've ever had
die on me are the ones inserted into sockets ala ramps and friends. I
still have original A4983 pololus ~2 years old which have worked
flawlessly, they are soldered into a board instead of inserted in a
socket.

The driver chips have a thermal pad on the bottom through which most
of the heat is dumped. Putting a heatsink on top of the package does
almost nothing because the thermal resistance between die and top of
package is substantial. They're designed to use the pcb itself as a
heatsink.

A pololu struggles to drive 1.25A without overheating. I have
personally verified that the A4982 drivers on smoothie (same silicon,
slightly different package) can push 1.8A without overheating, purely
due to having enough PCB area to dump the heat. Smoothie uses
through-hole mosfets for its two high current outputs.

SO yes monolithic has its drawbacks, but it also has its advantages. I
can't comment on rambo specifically since I haven't familiarised
myself with it, but if I designed a whole new board tomorrow it would
definitely be monolithic.


ps: I picked up my SMD rework station (iron + hot air) for $66
shipped, has served me well so far :)

[Dan Tyrrell]

On 21 February 2013 23:20, Triffid Hunter <triffid...@gmail.com> wrote:

On 21 February 2013 11:34, Greg. L <greg...@gmail.com> wrote:
> There's a brief spark, and a pop.
>
> Now I don't have an extruder 1 FET anymore.

does rambo not have a fuse that should have saved the mosfet?

> I really liked the monolithic board idea until today. Until today, I said
> "Yeah, but average users shouldn't be doing anything that would affect the
> board, and advanced users (e.g. me) shouldn't have problems, so monolithic
> is the way to go."

If you popped a mosfet on RAMPS you'd have exactly the same trouble.
Removing a TO-220 from a board when the pads have no thermals
generally requires similar equipment to removing a smd fet.

I use destructive removal. Cut off the legs at about 5mm, heat each joint individually and then pull out with small pliers. Once leg is removed remove solder to taste.    

 

the main issue with modular boards is that there's either a lot less
surface area available for heatsinking (pololus), or you end up with
numerous boards scattered everywhere (gen3)

I think the Pololus carriers are a poor design (for the reasons you give and others) and give modular a bad name:-)

[Greg. L]

> Now I don't have an extruder 1 FET anymore.

does rambo not have a fuse that should have saved the mosfet?

From: http://reprap.org/mediawiki/images/7/75/Rambo1-1-schematic.png near the dead center. All MOSFET outputs seem to be directly connected to output pins, as in the case of FAN-0 feeding to X11-3. 

 

> I really liked the monolithic board idea until today. Until today, I said
> "Yeah, but average users shouldn't be doing anything that would affect the
> board, and advanced users (e.g. me) shouldn't have problems, so monolithic
> is the way to go."

If you popped a mosfet on RAMPS you'd have exactly the same trouble.
Removing a TO-220 from a board when the pads have no thermals
generally requires similar equipment to removing a smd fet.

I've even got a hot air station, so it's not so much the work to fix it, it's the work getting it out and back in again. It was all nice and assembled with cable wrap, etc, so unplugging things and getting it /to/ the station (and then back in) is actually more what I'm whining about, admittedly. 

 

the main issue with modular boards is that there's either a lot less
surface area available for heatsinking (pololus), or you end up with
numerous boards scattered everywhere (gen3)

I'm firmly convinced that the high rate of death of the pololus is
directly related to their modular nature. The only ones I've ever had
die on me are the ones inserted into sockets ala ramps and friends. I
still have original A4983 pololus ~2 years old which have worked
flawlessly, they are soldered into a board instead of inserted in a
socket.

The driver chips have a thermal pad on the bottom through which most
of the heat is dumped. Putting a heatsink on top of the package does
almost nothing because the thermal resistance between die and top of
package is substantial. They're designed to use the pcb itself as a
heatsink.

A pololu struggles to drive 1.25A without overheating. I have
personally verified that the A4982 drivers on smoothie (same silicon,
slightly different package) can push 1.8A without overheating, purely
due to having enough PCB area to dump the heat. Smoothie uses
through-hole mosfets for its two high current outputs.

To that end, it's quite possible to design modular PCB 'sticks' that have appropriate packages and thermal sinking to the PCB if you'd like. You can even stick a block of aluminum on the back of the PCB as a heat spreader if you're into that sort of thing. There's no real difference (thermally) between four small PCBs and one large panel of them, as long as they're using appropriate thermal pads on the PCB, good packages, and such. Regarding the Gen3 electronics, I do wonder if there's something to the idea of lots of boards - you can either connect them with edge-to-edge connectors, or run cabling between, but when something goes, it's far easier to replace. If I'm unsuccessful with my RAMBO fix, it'll be nearly a $200 'oops', which seems painful.

Admittedly, the longer-term fix would be to not pop the FET in the first place, which would mean getting a shrouded fan header put with the board at ship, rather than having me install one. Had I not been putting metal near the end in the first place, I wouldn't have this problem.

 

SO yes monolithic has its drawbacks, but it also has its advantages.

I'll agree that it's cheaper, easier to design and inventory, and in the case of non-failure, fits in a box better, but I think you're stretching quite far with regards to thermals. 

 

[Lee]

*Should* be cheaper... but actually you can get a full RAMPS setup for under $150 quite easily. RAMBO is $185.

IMO it's also worth the $4 to buy heatsinks and a fan for your stepper drivers. My drivers don't even get warm. I prefer the modular solution. The heat problem is one that I can easily fix. If I did something dumb like forgot to turn of my power supply before unplugging a stepper and ended up blowing out a driver... that would be a lot more effort.

[Greg. L]

Should the M106 code also turn on extruder 2 on a RAMBO/Marlin setup? The RAMBO design has two fan outputs, on separate FETs, but I can't find a different M-code to turn the second one on vs. off.

[Greg. L]

Did some digging into Marlin, so I thought I'd post here, both in case I'm correct (in which case someone might someday benefit) or in case I'm incorrect (in which case someone will hopefully tell me before I do something dumb).

M106 in Marlin.ino sets the variable FanSpeed

FanSpeed is regularly updated in planner.cpp, which analogWrites to FAN_PIN

FAN_PIN on a RAMBO (Motherboard 301) is by default set to 8 (meaning Digital Pin 8, which is a PWM-enabled pin, and thus, can be analog written to)

Digital Pin 8 on an Arduino Mega/AVR2560 is real pin H5

H5 on the RAMBO schematic does correspond to FAN-0, which controls a PSMN7R0 FET, and an output

Thus, that's the natural state.

To get it to drive FAN-1, one needs to update pins.h to point at pin 6, which drives the FAN-1 output. As a note, it's actually a very different FET on this output - only a 3A or so FET, compared to the much higher current FET on FAN-0. Not sure why that its, but the FAN-0 output FET is the same FET used for bed heating and extruder hot-end heating, but FAN-1 (and FAN-2) get smaller FETs. 

I sent an email to Johnny over at ultimachine to see about the FETs, but that seems to be how to fix it. I'll likely make the change myself once I get a reply from Johnny, and then hopefully this thread can help out the next person who hears that dreaded spark.

[Greg. L]

And to seal off this discussion, I should finalize by noting a few things:

- My FET never did burn out. One of the SMD fuses gave its life for it (yay!) so replacing the fuse with a new one fixed things right up.

- The customer service at Ultimachine was more than happy to send one of those fuses out, so that was kind of them.

[Greg Link]

Not a problem, and sorry you're having issues. There are two nice fuses on the board, both Littelfuse SMT mount fuses. I got my replacements from Newark, Littelfuse 435005's. Newark stock # 69H6763, Manufacturer PN 0453005.MR . They were 86.3 cents a piece, so I bought a few, as shipping was a few bucks more. 

If you look at the RepRap page for RAMBO, you'll see this generic picture:

http://reprap.org/mediawiki/images/5/5c/Rambo-conn-all.jpg

The fuse is clearly visible just above Ext2 and to the left of the Heated Bed inputs. Thing is, that's an old RAMBO. There are newer revisions with another fuse, but the general rule is that they look like that, and pop out easily with a pair of tweezers (as long as you unplug power and USB first!). 

That should get you started at least. Let me know!

- Greg

On Mon, Apr 15, 2013 at 3:30 PM, Anthony Willey <adjw...@gmail.com> wrote:

Hi, Greg, I was wondering if you could tell me which fuse you had to replace (and possibly the specs on the fuse, if you have them). I just made the same mistake...bridging the two pins on the Fan-0 port, taking out the fan and the extruder. I did not purchase this direct from Ultimachine, but got it as part of a kit, so I think I'm a little more on my own on this.

--
You received this message because you are subscribed to a topic in the Google Groups "MendelMax Support" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mendelmax/Un3wSwHSvM0/unsubscribe?hl=en-US.
To unsubscribe from this group and all its topics, send an email to mendelmax+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Anthony Willey]

Thank you for that. It was indeed the F3 fuse, which incidentally is the same one as the F2 fuse for the heated bed. I ordered a replacement. In the mean time I invented a solution of my own so I can continue printing. I pulled the fuse out, soldered some wires to it (using the solder as both mechanical and electrical connection) and attached the wires to an automotive 5A fuse that I already had. Ghetto, but it allows me to continue printing, hopefully with the same level of protection. Very nice of them to put a fuse in.

Thanks again for your response! This thread saved me a lot of time and troubleshooting.

[Greg Link]

John -

  Which version of the RAMBO do you have? It should be silkscreened on the board as a "RAMBO 1.0.A", .B, etc, (up to .E, I think? Maybe even .F?) 

The fuses are generally Littelfuse 435005's, which I originally bought from Newark, but can't seem to find in stock right now. Digikey does have them ( F1618CT-ND ). The fuses are mounted into SMT fuse holders, so once your RAMBO is disconnected from power and USB, you can use a pair of tweezers or very small pliers to grab the fuse out of the holder, pull it straight out, and replace it with another. They are orientation independent, so it doesn't matter which way it is oriented on the way in. 

The fuse holder locations vary from version to version, but there are generally two holders - one for the "+12V" rail, and another for the "12V2" rail. Hot end 0 is called "HEAT-0" on the schematic, and runs off of "+12V2" (as does most every other heat/fan output except for the bed heat). 

More generally, /how/ did you burn out the MOSFET? I did pretty much a worst-case scenario (short with a thick conductor for a non-zero period of time) and the fuse saved me. Did you actually burn out the MOSFET? How are you sure?

On Wed, Jun 5, 2013 at 11:54 AM, John Stack <jacks...@gmail.com> wrote:

So, I did burn out a mosfet for Hot End 0. I replaced the fuse but still am not getting any current to it.

Which one is it?

 How do I pull the old one out?

Where can I find one?

Thanks in advance!

[John Stack]

I have the B, All in One.

I replaced the fuses. They were gone.

No voltage. The Mosfets are gone. Just don't know which ones or how to replace them...

[Greg Link]

Sadly, you've really only got two options right now. One, is contact Ultimachine (or the vendor you bought your system from) and hope for support/RMA/replacement - perhaps at a fee. The other is to flat out buy another RAMBO. In theory, I could tell you how to resolder on new MOSFETs, but it's quite possible that it's not the FETs that are burned - it could be that you burned out a trace, or a via, or something else entirely. Unless you can truly diagnose the board, you'll destroy it trying to peace-meal repair it - especially if you don't have a good knack at soldering.

What were you doing that killed it? It should be fairly hard to kill one, so I'm interested. 

[Gert van den Berg (MoHaG)]

On Friday, June 7, 2013 1:37:55 AM UTC+2, John Stack wrote:

I have the B, All in One.

I replaced the fuses. They were gone.

No voltage. The Mosfets are gone. Just don't know which ones or how to replace them...

You can try finding the MOSFET with a multimeter.... They often fail shorting between drain and source (At least the ones I destroyed though overheating, in a non-3D printer application) 

Look at http://reprap.org/wiki/Rambo for schematics....

Also check that the output pin on the microcontroller actually toggle in the way that it should... (If it does, try to check it in several places along the path with a multimeter) (Replacing the firmware with a test program is probably the easiest to get predictable outputs...)

Once you find the problematic component, replacing it might be another story, I'll leave that explanation for someone that have some kind of surface mount soldering experience...

Gert

[vlad]

if you have a dead Rambo still, I am tempted to try and fix it myself to donate to the highschool here.  Don't have the re-work tools, but as long as there is no base thermal pad(there probably is) I can try and re-work it.  If there is I can always go buy a hot air gun and crossmy fingers...

[Filip]

All,

Don't mean to dig up an old thread, but I appreciate the discussion. I did the same on my board, and this thread helped me get back and running. Anthony, I appreciate your workaround too, I wouldn't have thought of using the dead fuse as a holder.

Thanks again!

[Larry Knopp]

Great to hear Filip.

And, actually, that's why these threads are here!  A catalog of *mostly* useful information.  Dig through 'em...  you'll likely find an answer.

--
You received this message because you are subscribed to the Google Groups "MendelMax Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mendelmax+...@googlegroups.com.

[Guy Snover]

Thanks for this thread. Just had the same, and I found an old rambo board I had and swapped the F3 fuse. Back in business! Phew! What a relief!

Guy

[Greglink]

Glad we could help!

Sent from my iPhone

To unsubscribe from this topic, visit https://groups.google.com/d/topic/mendelmax/Un3wSwHSvM0/unsubscribe.

To unsubscribe from this group and all its topics, send an email to mendelmax+...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.