Blowing 5 stepper drivers in one brain fart

[Sean Mitchell]

I had a frustrating evening lastnight of my self-source build, first a minor wiring mistake on my part caused me to blow a stepper driver.  Fortunately I had a replacement on hand which after 8 hours of building I installed it upside down and took out all the other drivers.  Fortunately the arduino and ramps are all fine (as is my laptop which instantly shut down when the 24v ps came on), though my build is now on hold until I can get motion in the axes again.

Question for the community, does anyone have any experience with soldering the A4988 chips onto the driver boards?  I now have 6 dead boards sitting here.  I can get 4988 chips (not the pololu driver boards, but the actual chip on the board) for 2.70€ instead of the 10€ cost of the entire pololu board.  Any EEs out there comment on the skill of learning SMD soldering (and by trying QFN packages as a starting point?)

I have ordered replacement pololu boards, but should I want to pull a stunt like this again and install one of them upside down it would be handy if I could get my 6 dead boards replaced for cheap plus a bit of labour and a learning experience. 

Also, for anyone interested here are my build pics of the process until this point.

Barring any other egregious errors, I should be up and running by next weekend

--

========
Sean Mitchell

echo "zvgp...@tznvy.pbz" | tr '[a-m][n-z][A-M][N-Z]' '[n-z][a-m][N-Z][A-M]'

[Jean-François Talbot]

I have to congratulate you on your build.

I do have few comments however.

- Dog Bones: you must change them now. Get them made in alu asap;

- General: have you use locktite in all your t-slot nuts ?

- Back Skirt: make sure you have at least 1mm clearance between the Y axis motor and the back skirt;

- Backlash nut: you should try to fasten the anti-backlash nut with the screw head on the bottom, nut on top, and long enough screw to hold the X axis. Those who have gotten the best prints yet, are the one with a "floating X axis" that comes down only by gravity;

- Gantry plates: cut those, to fit the 60mm gantry post, you are going to hurt yourself one day :D

- Are those 8mm or 10mm smooth rods (if 10mm where did you source the clamps?)

- You should ditch the budda nozzle and order a MetalMagma. Go all out man !

- I hope you won't run a torrino in there. Go smoothie, or Azteeg X3, or go 12v.

Nice work.

Jeff

[John D]

Hey Sean - ouch - I feel your pain!  Yes, it's entirely possible to replace the A4988's on the pololu's.  It's unfortunately a damned pain in the arse unless you have a proper rework station.  I've *never* had good luck trying to get them off using hot plates or other DIY methods, but since the boards as dead, it's not like you have anything to lose by trying....

[Sean Mitchell]

Thanks for your comments, it's still under construction so some of the pieces are temporary.

As for the dog bones, it was my understanding that the plastic ones are ok, other than the fact that they might melt.  I followed the thread of aluminum dog bones not being perfectly flat so I'm a bit weary to try aluminum because I don't have a machine shop to get it to perfect tolerances.  My top plate was designed a bit differently than the aluminatus ones, so I shouldn't need that 20mm spacing correction in the dog bones (if my measurements are correct)

Lock tite, I have it in about 50% of them as I wasn't sure which pieces I could secure in place, and which pieces would need loosening and sliding a few mm in either direction.  Goal is to have it in all of the screws when I'm done.

Back skirt, I shortened my Y support rail by 5mm, so I have quite a bit of clearance back there (I have lots of wires running behind the motor, so it should be good to go.

Floating X axis, yes I might print something like what Ezra was proposing to let it float perfectly.  I want to see just how bad the problem is first before trying out the floating axis.  I am using a bearing placed on top of the motor shaft and the leadscrew sits on top of the bearing. 

Gantry plates: 100% correct, just not sure how to easily cut them.  Now that I know how cheap aluminum lasering is, I might get them laser cut like the original.  My first quote I sent in for cutting came in at 250€, where the raw aluminum would have cost around 45€.  At this point I opted for the self-drilled aluminum plates that I have now.  Afterwards I found laser cutting which is cheaper but requires a bit of sanding to remove sharp edges which only approx doubles the cost of the sheet of aluminum.

They are 10mm smooth rods and I bought the holders here, along with the 10mm rods, 2mm pitch leadscrews, and brass bushings for the x ends.

I have two metal magma hotends here but due to a miscommunication no extruder motor for them.  The extruder will be one of the new TL ones that are really light.  The budaschnozzles are what I have on my MMax and what I have laying around that I know I can get working right away, waiting on all the parts to get the proper hotend implemented.

It is currently powered by my mendelmax controller which was a cheap chinese arduino clone, ramps board and pololu drivers (which I just ordered pololu black ones to replace the ones I fried...$H|@().  This board has worked really well on my MMax til now, I probably have 1000 hours of operation on it without a stopped print. I only run on 24v, everything I have is setup for 24v.

It will be eventually powered by a smoothie.  I ordered a smoothie mid-march and my family received it and forwarded (march 27) to me in Germany, but it still has not shown up.  I am beginning to think it might be lost in transit, but customs here have stepped up inspections on things and some of my worst items from ebay china have taken 52 days to arrive so there is faint hope it might still arrive.

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

[Triffid Hunter]

On 28 April 2013 23:16, Sean Mitchell <mitc...@gmail.com> wrote:

Question for the community, does anyone have any experience with soldering the A4988 chips onto the driver boards?  I now have 6 dead boards sitting here.  I can get 4988 chips (not the pololu driver boards, but the actual chip on the board) for 2.70€ instead of the 10€ cost of the entire pololu board.  Any EEs out there comment on the skill of learning SMD soldering (and by trying QFN packages as a starting point?)

I'd say don't start here. you'll need hot air at minimum, preferably a reflow plate of some sort. you CANNOT succeed with soldering iron alone- the A4988 has a large thermal pad underneath and it's critical that it's soldered correctly along with all the pins.

The procedure is basically, heat board until the solder melts and pull off the old chip. let it cool, clean up the solder on the pads with iron and wick, then lay down solder paste and place the new chip. then back on your reflow plate to melt the paste, and hope you don't get any shorts underneath or you'll have to pull it off and try again.

You can't just plonk the new chip down in the old solder as you need flux to clean the oxides from the metal, without which the solder will not stick, or will not stick reliably. The old solder may also be contaminated with copper, nickel, gold, oxides, etc which will degrade its ability to bond to the pins properly.

I have soldered QFN and MLC and similar packages and I can say with confidence that it's not particularly difficult, but is definitely not for the faint of heart, poor of sight, ill-equipped or the inexperienced.

If you want to give it a try, you'll need some solder paste. 10g will last you almost forever, and needs to live in the freezer. Usually I'd suggest leaded because it's significantly easier to work with, however in this case we have a chip that may spend most of its life very close to the melting temperature of leaded solder (~183c) so I must suggest lead-free.

Your reflow plate must be able to hit 260c in 3-5 minutes, and cool down to room temp also in only a few minutes, without blowing air across the pcb as that would simply send all the smd parts flying. I've seen clothes irons and electric skillets used for reflow but can't in good conscience recommend these. You need at least ~220c to melt lead-free solder, but most of the components degrade above 120-150c so you want to swing it through that hot phase as fast as possible.

You'll also need some isopropyl alcohol to remove the flux afterwards- it's mildly corrosive to metals and will eat your board slowly over months or years, so should be removed afterwards for best results.

A good set of tweezers will be invaluable, and possibly some magnification for post-operation inspection. You already have a multimeter I hope!

You should also let the board soak at ~90c for a while before starting- if the parts or the board have absorbed moisture, that moisture boiling will certainly break something! This is called "popcorning" for a reason ;)

experimentation required, good luck

[temo]

Sean, nice work!

If you want a set of aluminum dog-bones I will send it to you for free. I had them made just when TL sent out new ones that extend your bed forward. Mine does not extend the bed forward but it is precision beveled at 30 mm height. Make sure that 30 mm get you over the screw-heads, I think the TL ones are 32 mm or so.

If you want, give me your email and I will get in touch with you offline.

Terje

[ChrisHS]

Looking at the pololu website gives another driver option: The 8825 based drivers.

 

DRV8825 Stepper Motor Driver Carrier, High Current

http://www.pololu.com/catalog/product/2133

1.5A or 2.2A with fan, Max 45V, 32x Stepping

 

A4988 Stepper Motor Driver Carrier, Black Edition

1.2A or 2A with fan, Max 35V, 16x Stepping

http://www.pololu.com/catalog/product/2128

Both have same copper in pcb but 8825 has a bigger thermal pad to spread the heat out.

5.18 * 3.1 mm 16.06 mm2 8825 versus 3.15 * 3.15 = 9.92 mm2 4988 (62% bigger pad area on 8825)

Also the thermal pad can be soldered & then the pins seperately (with a sharp iron) ie easier for future 'accidents'.

 

The module pinout is close enough to just plug in as a replacement for the 4988 with ramps.

Fault signal available to show limits hit.

 

They do need 2nS pulses rather than 1nS for stepping.

 

Anyone considered these & ruled them out for some reason ?

 

Chris

~~~~~

[Triffid Hunter]

On 29 April 2013 22:24, ChrisHS <chris...@googlemail.com> wrote:

Anyone considered these & ruled them out for some reason ?

Their fixed off time is almost twice that of the allegro drivers, so your steppers won't run quite as smoothly. apart from that they reportedly work fine

[Sean Mitchell]

I know my favourite supplier here for pololus (I'm rapidly becoming their best customer) had those boards on clearance for a while.  If I recall correctly, they said they normally work as a drop in replacement however there were some differences that would cause them not to work under special circumstances.  I suspect this has to do with the timing (as you and Triffid have pointed out)

The fact that they were on clearance makes me suspicous.  I have ordered 3x black editions and 3 normal ones as replacements for testing.  I am also going to put a big ass arrow sticker on them which shows me how to install them upright, so I don't do that again :/

As for the SMT soldering, I am going to keep the blown drivers and work towards attempting a replacement.  I have done a bit of reading on using the (ill-advised) hot plate methods and it seems people can have success with it.  I may try to hack something together using a hot plate and PID controller to do both a reflow station and an ABS vapour bath station in one shot.  For the amount of reflow work I want to try I dont think it's worth investing too much into it (and I also found the sparkfun article that had quite a bit of success with hot air, toaster ovens and hotplates)

--

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

[Sean Mitchell]

An update to my build, and some tips.  Since my last post I have: 

• Received replacement drivers and had it up and printing its first parts with RAMPS
• Decoupled the X axis from the Z rods (using a floating design as proposed by Ezra and suggested by Jeff)
• Replaced the moose brain with a Smoothie board
• Replaced the printed ABS dogbones with aluminum (Thanks Terje!)

I was amazed at the quality of the first prints, I printed a T-Rex head (see here) at 300% size.  This was the sixth thing that I have printed on the printer, and you can see details in the skin pattern. There are two mistakes in the head, both are user error, as well as some overhang calibration is required.  I noticed the leadscrews were rattling against my rod clamps causing quite a bit of noise, so I applied a bit of rubber tape.  I bumped the leadscrew(first mistake), and later the tape slid down so I had to peel it off, causing a full rotation of the leadscrew (second and more noticeable gap).  As with most things, had I left it alone, it would have been perfect.

I have also printed some of Ezras idea to float my X axis on the Z and let it fall by gravity.  This has caused a huge improvement in the Z ribbing, which I could see when I tried the single wall calibration cube from thingiverse originally.  The T-Rex head was printed post-upgrade and I can see no sign of any ribbing or repeatable pattern (as far as I can tell, this is a "perfect print").  Once I get smoothie tuned in I want to print some of these tall models that have large flat surfaces (I think I will join the rocket club, or do one of the waiting for battle girls).

One note, post upgrade and my leadscrews have the worst wobble that I have seen yet.  Because my leadscrews are only constrained at the bottom, the tops can now move around 1-2 cm, and do this during a print.  This doesn't seem to be affecting my print at all, and I'm hoping that some replacement couplers can help with this (if my strange size is available Jeff).  I also have some 6000zz bearings (10mm inner diameter) on hand, so I can also try constraining it at the top and bottom but at this point it is more of a cosmetic problem that they wobble so much.

My experience with smoothie so far is mixed.  On smoothie it seems much faster and more importantly quieter (still louder than my mendelmax, but it is now an acceptable loudness).  I'm currently struggling with smoothie tuning and hope to be able to print with it soon.  

-sean

kl. 15:16:02 UTC+2 søndag 28. april 2013 skrev Sean Mitchell følgende:

I had a frustrating evening lastnight of my self-source build, first a minor wiring mistake on my part caused me to blow a stepper driver.  Fortunately I had a replacement on hand which after 8 hours of building I installed it upside down and took out all the other drivers.  Fortunately the arduino and ramps are all fine (as is my laptop which instantly shut down when the 24v ps came on), though my build is now on hold until I can get motion in the axes again.

Question for the community, does anyone have any experience with soldering the A4988 chips onto the driver boards?  I now have 6 dead boards sitting here.  I can get 4988 chips (not the pololu driver boards, but the actual chip on the board) for 2.70€ instead of the 10€ cost of the entire pololu board.  Any EEs out there comment on the skill of learning SMD soldering (and by trying QFN packages as a starting point?)

I have ordered replacement pololu boards, but should I want to pull a stunt like this again and install one of them upside down it would be handy if I could get my 6 dead boards replaced for cheap plus a bit of labour and a learning experience. 

Also, for anyone interested here are my build pics of the process until this point.

Barring any other egregious errors, I should be up and running by next weekend

--

========
Sean Mitchell

echo "zvgp...@tznvy.pbz" | tr '[a-m][n-z][A-M][N-Z]' '[n-z][a-m][N-Z][A-M]'

--

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

--

========
Sean Mitchell

echo "zvgp...@tznvy.pbz" | tr '[a-m][n-z][A-M][N-Z]' '[n-z][a-m][N-Z][A-M]'

--

========
Sean Mitchell

echo "zvgp...@tznvy.pbz" | tr '[a-m][n-z][A-M][N-Z]' '[n-z][a-m][N-Z][A-M]'

--
Med vennlig hilsen

Terje Moe
Arendalsgata 14
0463 Oslo
Norway

Mobiltelefon: +4792299480
E-post: terj...@gmail.com

--

========
Sean Mitchell

echo "zvgp...@tznvy.pbz" | tr '[a-m][n-z][A-M][N-Z]' '[n-z][a-m][N-Z][A-M]'

--
Med vennlig hilsen

Terje Moe
Arendalsgata 14
0463 Oslo
Norway

Mobiltelefon: +4792299480
E-post: terj...@gmail.com

--

========
Sean Mitchell

echo "zvgp...@tznvy.pbz" | tr '[a-m][n-z][A-M][N-Z]' '[n-z][a-m][N-Z][A-M]'

[Triffid Hunter]

On 6 May 2013 08:38, Sean Mitchell <mitc...@gmail.com> wrote:

My experience with smoothie so far is mixed.  On smoothie it seems much faster and more importantly quieter (still louder than my mendelmax, but it is now an acceptable loudness).  I'm currently struggling with smoothie tuning and hope to be able to print with it soon.  

what troubles are you having with it?

perhaps join us in irc://freenode.net/#smoothieware or http://webchat.freenode.net/?channels=smoothieware

[Jean-François Talbot]

Yes Sean, the supplier will produce your weird size :)

[Sean Mitchell]

Hi Triffid, 

You have already helped me out, I'm Muzik` on IRC.  Most of the issues I have had so far is just lack of familiarity with smoothie and learning how it works.  Also the board showed up with a dead driver on the X axis (given the subject of this thread, I swear I didn't kill it.  You cant plug 'em in upside down so I'm safe here! ...and I know not to disconnect the motor while power is applied, as far as I can tell this one driver did not work from the get-go), and I had some trouble with noise causing skips on the pololu carrier that I built (you were helping me online with this - I found the problem too... noise on the 3.3v line which I was using to power the logic side and pull up the microstep pins).  

A second smoothie showed up today which I am up and running with.. this went much faster than the first setup, I was printing within an hour of having all the headers soldered.  Without jinxing it I will have my first print done in the next little bit here.

As for the original, my first try at solder rework is planned for next weekend (I may be hitting up IRC for some help with rework - as you suggested I have lead-free solder on order, some flux and a hot air station)

I am going to start with the smoothie because the chip is bigger (and if it didn't have the heatsink pad on the bottom, I could do it with my iron - I have soldered this size before successfully)

[AllThumbsTom]

I don't know what size SMD chip your replacing.
I've done a lot of rework. Mostly I'd use hot air for removing a part.
I've always used a low power microscope and a low power iron that has been sharpened to a fine point. (and tinned)
I would first make sure the pads were tinned and lightly fluxed.
For a chip, I'll hold it down with my finger and solder one pin on the end of a row, adjust position slightly and then solder the opposite pin.
By solder I mean just touch the pin to the pad and don't try to add solder, if it's a properly tinned pad, it will stick just fine.
After the opposite leads are connected the chip will be aligned properly and now it's easy to solder the remaining pins one at a time.
Take you time, don't create any solder bridges.
Sometimes I'll add just a smallest touch of solder to the hot iron tip. You can buy solder in very small gauges, works better and a tiny roll lasts about 10 lifetimes.
Wash the board with flux remover.
I do prefer to use leaded solder, it seems to wet better.

Tom M.

[Sean Mitchell]

Hi Tom,

Thanks for the response.  For my intro to smd soldering I'm going into the deepend, the smoothie uses 24tssop which I have done by hand before, but unfortunately these have a gigantic heatsink on the bottom which also needs to be well connected to the heatsink vias on the board below.

For my next trick I have six 28qfn (I ordered 10 for a cheaper price, I can waste a few learning) to give a go at, again with the solder pad below as a heatsink.  I think I'm going to have more trouble with these, even with a sharp pointed tip the leads are not available, hence the rework station.  The chips were only 2.70 each and if it works I'm repairing an 10.00 piece and learning along the way, so never bad!

I am going with lead-free on Triffid's advice, because this is used to bond the hot part of the driver chip to the board - which could melt the solder if it has too low a melting point

--

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

[djam]

Hi Sean,

I would advise against lead free, the absolute maximum operating temp for the A4988 is 85° and
if runs hot enough to melt leaded solder, the device will be long dead.

Because leaded solder has a lower melting temp it will be much easier to solder and it wets up so
much better then lead free solders.

Note, I never heard of keeping solder paste in the freezer, I keep is in the fridge. It does last well past
the use by date if kept in the fridge, but can dry out eventually. I have a liquid flux that I add to the paste
from time to time to improve printing (screen printing the paste).

The flux I use is IF8300, http://www.interflux.com/en/node/71 use this and SMD soldering will be a
walk in the park, though a good tip tip helps.

This is my favourite rework tip  http://www.hakko.com/english/tip_selection/type_bcm_cm.html


Darren

[Triffid Hunter]

On 7 May 2013 10:08, djam <darrenja...@gmail.com> wrote:

I would advise against lead free, the absolute maximum operating temp for the A4988 is 85° and
if runs hot enough to melt leaded solder, the device will be long dead.

that's max ambient temperature.

Tj(max) is 150c, and I've had parts unsolder themselves from the board while still operating fine frequently enough for the type of solder to be a concern. See the xbox 360 red ring of death debacle for example- many folk reflowed it back to working condition by wrapping it in a blanket!

Also, all modern devices are rated to spend at least a minute at 260c so that they can be soldered properly. They can't handle it for long though, so if your operation doesn't quite take the first time, let everything cool down and rest before trying again.

The A4988 does have internal overheat protection, but considering how frequently they die I'm not sure of its efficacy...

[djam]

Quite right, I should have realised, 150° was more what I was expecting. Still well below the
re-flow temp for leaded solder. I hope people have enough cooling to keep them under that,
even if there is a chance they will survive. Thermal cycles is the biggest killer of integrated
circuits. There are strict temperature profiles for re-flowing, and most devices are ok with
at least 3 passes.

Standard FR4 is not happy with extended time above around 120° unless it is one of the
higher temperature substrates.

I don't think using leadfree solder is a way to deal with the drivers running too hot.