Mick

    My Solidoodle 3's Heat bed lasted 1 day. I had been testing settings trying to get slicer dialed in, so none of my prints exceeded 1 hr (I only ran a few prints, single layer to get my extruder, x and y axis dialed in, as well as my Z-stop). So it didn't see any high duty cycles during the prints, not sure why the heat bed failed so quickly, but it might lend some credence to your theory. I'll try to put a fan back there and see if that doesnt help. I'll look into resistors, but I'm not an electronics guru, so I'll need to get some help there. Machine seemed to be performing solidly, and we all know that these kit printers require tinkering an improvements, so I dont view solidoodle in a dim light (still waiting on a reply from support). I think a topic like this just helps us all with our machines. I know it opened my eyes to what I might be able to improve upon and what needs extra cooling.

Mick

any more then that probably get a more expensive better built unit if needs arises.

and the boards are all at over limit without cooling!

The only issue I see is this assumes that the PCB and driver are kept at ambient of 70F.
Put that unit in a room with someone that likes to jack up the heat and uses no infrared heaters or put it in the back of an IKEA box compartment and the unit is no longer operating at 70F.

Fans are cheap and easy to get these days.  After I finish turning my SD2 inside out it'll be getting one.

I would be so inclined to redesign the electronics for this entire thing if it wasn't my whole point to buy a system that was RepRap like.  If the drivers are seriously an issue I might be inclined to make a larger capacity module.

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

So:
X axis: 1.11A
Y axis: 1.3A
Z axis: 1.25A
Extruder: 0.48A

is correct but that is then driving the coils at 70% in full step mode

X motor: .77A
Y motor: .91A
Z motor: .875A

so they are witihin the specs of the motor or close to the 1A. But the the A4988 is rated only to 1AM without cooling! so hence its over driving and full duty cycle should higher loads at the motor.

so they are not being over driven per-se buut right at max and with full duty cycle yes on hairy edge of motors.

and the boards are all at over limit without cooling!

So to clarify they did not totally miss the boat at 30% over load LOL

I use them extensively in other projects, and if you run them continuously above their rating without cooling the a4988 will current limit themselves. There is next to no risk running it at an elevated temp that's below the destructive temps. Silicone in some obtuse ways runs better at a somewhat elevated temp, but I understand that's more in the case of computer architecture and it's something I simply heard from my EE friend (suggested it was because it's a better conductor at higher tempteratures, as semiconductors need more energy to jump their  band gap into conducting).

I WENT ON A TANGENT, sorry.

Anyway, my point is in the past 10 hours of printing I've not heard the sound of the drivers current limiting themselves, which would totally fuck up a print, so I think the firmware must be appropriately dealing with things.

As a design complaint, I would point out that on my X & Y axis, the printed piece that holds the belt bearings (in the corner), is printing layer by layer up. The tension is therefore applied on the Z-axis. From what I understand, the weakest bond in extrusions is between the layers, so they are printing it in a way that applies forces on the part in it's least tolerable way. It would be more reliable (and almost no different in costs) if they rotated the part 90 degrees.

If this thread is going to be repeating the same thing over and over and over. then there is not a lot of point in writing anything else.

Perhaps the 4layer board for stepper driver is something that they'll consider in future, but saying the same thing over and over isn't any more likely to make that happen.

exaggerating and saying that the machine nearly burned someone's house down won't make it happen either.

for the electronics, the only real advice is add a fan.
the bed resistor, kind of is what it is, not guaranteed to fail, in most cases not failing, but probably prone to premature failure.

What I meant was, if there is a problem then by all means let's suggest a fix, but getting fixated on one small part isn't really helpful to anyone.

How about we approach it a little more positively .. ie I would certainly like to add heatsinks to those drivers and had already considered it before building a case .. with fans as well, I touched one fairly one without thinking at one point .. will not be doing that again.. anyone found the ideal heatsink ?

elmoret wrote:

danny wrote:

if this thread is to bitch and moan about the pololu drivers, well that's not going to get fixed.

The point of this thread is to point out design issues so that people are aware of them and can choose to fix them if they wish. I am sure Solidoodle is aware of all these already, and if they choose not to fix them, that is their prerogative.

Fix it how?

Did you get your printer yet?

If not then why not just cancel your order, or as if they'll send you a frame and belts but no electronics or motors?

1amp @35v is the power rating.
The solidoodle runs at considerably lower voltage.
And you can wind the current back as much as you like.

Feel free to dial it right down and just print slower if that's what you want?

As it is. Nothing is actually being over driven. Being over driven is something that's just not happening. Sure equipment is being driven at or close to limitations, and this is showing up lesser quality components, it bad batches.

There's well over a thousand solidoodles shipped now and stories of two failed bed heaters. Maybe three?

It's really not the problem you're making it out to be.

The only point I really agree in is that 2 or four resistors would have drastically improve the uniformity if the heating.

Yes, I have my printer.

Fix it by putting cooling on the electronics, upgrading the bed resistor once it fails so it doesn't fail again, using motors that are rated for the necessary torque, instead of overdriving smaller ones.

You are correct that the SD doesn't operate at 35v, but that does not mean you can push the Pololu controllers harder. Let me explain why.

The primary source of heat on the A4988 chip is from the mosfet on resistance. It is roughly 400mohms. The drivers modulate such that average current is whatever is specified, in this case 1.33A.

If you drive with a higher voltage, the power dissipated while the mosfets are on is greater, but the duty cycle is signifigantly less to achieve 1A average. Thus, heat dissipation is not a function of supply voltage. That is why there is no derating in the specifications with regard to supply voltage. It does not work like a LM7805 or other linear power chip. It works a lot more like a switching mode power supply. I can explain further if you are interested.

You are absolutely correct that one could print slower at less current. However, Solidoodle's own recommendations are beyond factory spec. So, if I desire to operate the product I purchased at the level for which it as advertised to perform, parts are being stressed beyond design max, over 30% in some cases. I feel like that is worth mentioning on a forum so people can be aware of it, and turn it down or add cooling if they so choose. It's worth mentioning also that Solidoodle provides no warranty, so when you smoke a bed resistor or A4988 or stepper motor down the road, you might be out of luck.

Something like adding more bed resistors, that's not a design deficiency, because it works as is, and will work indefinitely as is. No specifications were exceeded, and a tradeoff was made between uniformity and cost.

How about the manufacturer of the Sanguinololu specifically saying the traces are too small for heated beds, and SD putting the heated bed on there anyway without reinforcement in the way of jumpers? Ian's practically caught fire... You can't say this isn't a big deal!

Maybe it's just me. I know I couldn't sign off on some of these decisions, from an engineering ethics point of view, but maybe that's unreasonable. Who knows.

Don't get me wrong, I don't regret the purchase. I think it's great and I've enjoyed it a lot. I just think that with a little more thought, and $20/unit, it could be designed such that everything is within spec.

danny wrote:

if this thread is to bitch and moan about the pololu drivers, well that's not going to get fixed.

The point of this thread is to point out design issues so that people are aware of them and can choose to fix them if they wish. I am sure Solidoodle is aware of all these already, and if they choose not to fix them, that is their prerogative.

Fix it how?

Did you get your printer yet?

If not then why not just cancel your order, or as if they'll send you a frame and belts but no electronics or motors?

1amp @35v is the power rating.
The solidoodle runs at considerably lower voltage.
And you can wind the current back as much as you like.

Feel free to dial it right down and just print slower if that's what you want?

As it is. Nothing is actually being over driven. Being over driven is something that's just not happening. Sure equipment is being driven at or close to limitations, and this is showing up lesser quality components, it bad batches.

There's well over a thousand solidoodles shipped now and stories of two failed bed heaters. Maybe three?

It's really not the problem you're making it out to be.

The only point I really agree in is that 2 or four resistors would have drastically improve the uniformity if the heating.

I haven't revisted the data .. but from memory the only thing the specs indicate regarding temperature is that spot heat does not exceed 200C.

the polulu driver say 'but to supply more than approximately 1 A per coil' .. seems a bit vague to me .. not really set in stone that one. As I said, mine has not skipped a step using the default settings, not twiddled any pots. not even looked at the currents. again .. at the limits but exceeded ?

not looked at the motor data yet .. i will later.

if this thread is to bitch and moan about the pololu drivers, well that's not going to get fixed.

The point of this thread is to point out design issues so that people are aware of them and can choose to fix them if they wish. I am sure Solidoodle is aware of all these already, and if they choose not to fix them, that is their prerogative.

see, now you have made me check your homework.. and you are simplifying the driver design decision too much. awfully impressive first post, i am sure made people feel disappointed in their device based on your opinion. heres mine..

pololu recomends the 1A max along with a max supply voltage of 35v. we are using 12v, the reduction in supply voltage by 300% should more than compensate an increase of current by 30% over a recomendation by a module maker.
the actual chip maker does not mention 1A anywhere, 32C/W on a 4 layer. how about hanging panelolu out to dry for not putting it on a 4 layer pcb as per manufacturer recomendations .. at the end of the day, thats all it is .. recomendations. no maximums have been exceeded anywhere.
You then build it, test it and base decisions based on how many units overheat, mine has never skipped a step yet that I know of. that would be a symptom of excess current, not a number in a pdf.

i would certainly chuck in fans to assist cooling though, probably add some sort of heatsink to boot if I go the rear cover.

resistor spec .. same deal, you are over simplifying the data, max rating is 50W at 25C .. big deal, that is a continuous power, it is regulated in this case, the power is not 50W but 0W well before resistor is compromised. not the same thing as 50W continuous, no where does the data state how the device is to be used, i am sure that having a PID controller and a thermocouple on it would make the manufacturers happy.
manufacturer specs are spot heat not to exceed 200C. they don't .. spec not exceeded.

. there is such a thing as over engineering something,  its not going into space or doing life support though, if it works .. do it.
What would NASA charge if they made this ?

disclaimer, I am not an engineer and do not claim to be. these are my opinions based on 40 plus years of professional electronics.

Plenty of people are seeing 100% duty cycles on the resistor to hit 105C. Then that is 48W at 105C, far beyond max spec. (50W at 25C). This has resulted in bed failures, at least 3 that I know of so far, in less than 3 months.

Pololu isn't to blame here, they clearly state that due to the PCB choice they made, they recommend 1A max. That is what I am operating off of, not the A4988's spec sheet. Not to mention there's half a dozen people that I know of with skipping steps unless the electronics are cooled. From Pololu's site:

These are being configured to draw 1.33A. That's beyond spec. Plain and simple. Pololu makes no allowances for supply voltage variation permitting more current.

It's just poor design, man. These things are being designed with factors of safety less than 1! How about the steppers being overdriven? No comments?

It's just not good engineering practice. I'd be fired in a heartbeat, even if it wasn't "for life support or going into space".

At the end of the day, the Solidoodle 2 could have been redesigned such that all parts were within manufacturer spec for less than $20 more. I for one would have happily paid that to know that everything was operating within manufacturer specifications.

The USB cable, you need to remember its there and its all good, only move case back a small amount and lift. I agree it could be better protected .. again though its a printable solution. I was thinking of a small frame that screwed between the two board mounts and provided protection under the USB cable so the metal case will hit that before touching the USB cable.

I thought at the time that this was the wrong place to post the ideas about the board.

just to clear things up...

This is not a design deficiency, this is, in my opinion, To be honest, just the same as the orientation of the spool holder previously was not a design deficiency, it could be solved with guides etc.
that was talked about in this thread: https://groups.google.com/forum/?fromgr … R46L6jPDJU
and in the end they did change the design.

What I was saying was, I've been an idiot and broken my machine, I didn't think I'd be the only person to make this mistake, and a simple way to make the machine more idiot proof might be to shift the electronics 2" to the left!

I tried this last night, unfortunately, (I imagine it's the same for all machines) the Y stepper lead travels to the bottom of the case the to the outside and to the electronics, moving the board 2" to the left means that the y cable becomes a physical barrier and causes the carriage not to be able to move all the way to the front.
I'll just have to be careful!!
I may re-route cables another day so that I can move the board because I don't really want to have to fix it again. by making the same stupid mistake!

Again, not exactly a deficiency.
Just something for the design team to consider.
You're right, $2000 options are available, but at the same time, this is a really simple suggestion, a tube of epoxy based glue is like $1, and takes 30 seconds to apply, then about 30 minutes to set. and a single tube would probably have glued the magnet to every door produced so far!!

Yes it adds time, yes it adds cost (the time is probably the biggest cost), but, it also adds reliability to one tiny aspect of the machine.
just like adding an elbow and a 4" piece of pipe into the box added cost, but added reliability.

I agree with you regarding the electronics case, and the door knob. I can't help but feel that a cool thing to do might be for the solidoodle guys to add all necessary mounting points for the case, (where they uploaded a design) and then make this a suggested first print.
something that is well known, that will test all areas of the bed, something that is easily measured and something that's useful!!

I guess the from my point of view...

if this thread is to bitch and moan about the pololu drivers, well that's not going to get fixed. -the drivers come as a product there is nothing that Solidoodle can do about this. (and besides which, as you say this is not actually so much of an issue)
if this thread is to bitch and moan about the fact that the sanginolou (I'll learn how to spell that one day) board traces are not rated properly for the current drawn by the heated bed. then again that is no going to get fixed.

If this thread is to be of any use at all, then it needs to be saying things like.
a 90 degree bend in the spool feeder leads to cleaner feeds.
a stronger glue won't get hot and fail.
it the electronics could move if only an inch to the left the machine would be more idiot proof.

Small things, that can be easily integrated to the machine without completely re-designing it! that improve either efficiency, or accuracy or reliability, make it more fool proof.

On the other hand, you are right to say that 1500 $1 improvements will make this a $2000 machine!

pololu recomends the 1A max along with a max supply voltage of 35v. we are using 12v, the reduction in supply voltage by 300% should more than compensate an increase of current by 30% over a recomendation by a module maker.
the actual chip maker does not mention 1A anywhere, 32C/W on a 4 layer. how about hanging panelolu out to dry for not putting it on a 4 layer pcb as per manufacturer recomendations .. at the end of the day, thats all it is .. recomendations. no maximums have been exceeded anywhere.
You then build it, test it and base decisions based on how many units overheat, mine has never skipped a step yet that I know of. that would be a symptom of excess current, not a number in a pdf.

i would certainly chuck in fans to assist cooling though, probably add some sort of heatsink to boot if I go the rear cover.

resistor spec .. same deal, you are over simplifying the data, max rating is 50W at 25C .. big deal, that is a continuous power, it is regulated in this case, the power is not 50W but 0W well before resistor is compromised. not the same thing as 50W continuous, no where does the data state how the device is to be used, i am sure that having a PID controller and a thermocouple on it would make the manufacturers happy.
manufacturer specs are spot heat not to exceed 200C. they don't .. spec not exceeded.

. there is such a thing as over engineering something,  its not going into space or doing life support though, if it works .. do it.
What would NASA charge if they made this ?

disclaimer, I am not an engineer and do not claim to be. these are my opinions based on 40 plus years of professional electronics.