Board size: 115x153mm
---------------------------------
PCB after eatching

Edited 1 time(s). Last edit at 01/25/2012 08:56AM by codec.

Bryanandaimee, (are you bryan or aimee ? )
anyways how far along are you with your design because I think it would be great for my second reprap (after I finish my first and print all the parts I'm making out of wood now it's the first thing on my agenda. )
One suggestion when you design the final revision of the board and, try to design and etch it in milling style and don't spare a few wires on the flip side for low current lines if it simplifies things. because if you use a milling style etch you'll lower the resistance and this is something I saw on codec's board.

@ Codec it seems like some of those traces are in danger of melting with the current the bed and extruder need as well as the toshiba steppers the traces you got there will have to be taken to the third dimension with solder coverage or they will melt. so redoing it as a milling style pcb i.e. only etching the holes and a fine trace around all of the much bigger traces might help with this.
terramir

I'm just about done, but I'll be gone for the next week or two so it will be a couple weeks most likely before I have an alpha design to test. Once I have tested one with toner transfer I'll try to get a milled board tested as well.

terramir - this board is for test's only. when everything will work fine, I will build new pcb. I have idea how to design new small PCB. now waiting for TB6560 and atmega

Well here's the alpha board. I have one etching now and I hope to do some actual printing tests with it in the next week or so. The earlier board was able to move motors so I'm pretty sure this one will work. The only changes are position of the jumpers to get them out from under the drivers and added a pull up resistor to ensure the ATX supply stays off if USB is disconnected. Oh, and I powered the drivers 5V off the standby 5V from ATX in a further effort to ensure the power on sequence isn't violated. I guess it's about time for me to figure out github.

Oh, and it's currently set up for TT, so if you want to do routing you would need to do a global line width -10 mil and hole size +10 mil.

Edited 1 time(s). Last edit at 02/11/2012 06:24PM by bryanandaimee.

Well done, bryanandaimee.

Quote

added a pull up resistor to ensure the ATX supply stays off if USB is disconnected

As the ATmega runs all the time, this could be done in software easily. I never had a PSU turning on unexpected, though.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

is it just me or are the power carrying traces a bit thin? Especially the motor power supply line to the drivers me thinks a etched milled style board would be more appropriate for the setup. but well that can probably wait till this goes from alpha to beta. also the second ground line to the drivers seems a bit on the cut close side.
but let us know how the testing goes great work brianandamiee
terramir

I have had a supply turn on unexpectedly, it happens if you run the ATmega on USB power and turn off the computer or unplug the usb cable. And since that would likely smoke the drivers I thought I'd try to prevent it if possible.

Yes the inter trace spacing is close. I did that to maximize the width of the traces carying power. It came out pretty well with toner transfer and I'm getting ready to test it with the 2A motors I have on my Prusa. I don't reccomend using these drivers at higher amperages anyway though I'm sure some will try. I'll let you know if I melt any traces. I do plan on having some milled sometime soon to test that method out.

I have successfully jogged all 4 axes at 2A and didn't burn the board. So it looks like I have a pretty good candidate for a beta board. Though as I got it working I noticed I had swapped the step and direction pins on the X axis. Not sure why I did that as it will route just fine the other way, so I'll likely swap them back so it's pin compatible with Gen7.

At 2A and 100% torque setting the drivers get hot slowly over a few minutes (but thermal protection did not kick in.) so likely some heat sinking will be required at 2A. The power traces do get warm at this setting as well but there didn't seem to be any real hot spots under the board. Unfortunately there's not a whole lot of room once the stepper plugs are inserted, maybe 1/4 inch. I'll have to look at ZIP-25 heat sinks if I can find some and see if they fit the board as is. I think a simple square of aluminum sheet bolted or glued on would work if there was a fan there too. I'll try that next.

At 2A and 75% torque the drivers barely warm up at all and the board (foil side) is just above room temp to the touch.

Pull up resistor on the ATX power-on control works as desired, so hopefully this design is at least smoke resistant.

All in all it looks like this is doable as designed.

I'll be testing thermistor and heater stuff next, (replacing my hot end) but I don't expect any issues there since it's basically directly copied from Gen7.

I lied about the power-on control. It still allows the ATX power to turn on if the computer is off and USB is plugged in with the USB power jumper set. So I guess I'll take Nophead's advice and drive it with a transistor.

I don't think it's exactly wise to drive some parts of a board with 5V over USB and other parts of the same board with 5V coming from the PSU. These two 5Vs might be slightly different, to unwanted, possibly high currents flow.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Yes it's probably a bad idea, I did it once when I was having communications issues and found it to be somewhat more reliable when the chip was powered from USB. That's when I found out that unplugging USB while configured that way would allow ATX to turn on. I would probably recommend against it too for users, but I can see that some might set it up that way to program it and then just forget to change it back and leave it jumpered as they move on to testing motors and stuff, so even though it's kind of an edge case, it seems like something worth worrying about, at least for these particular drivers.

Well I thought about what you said about heatsinking and I'd recommend you make something like this I built it in blender the z axis should be 3mm and the x should be 36 the y should be 40 and it seems like the uping the scale by 1000 worked from blender if you print this in abs you'll have a model use 3mm think aluminum or 2mm or 1mm if that's what fits use the print as a drilling mask and cut the fins with a blade that will cut about a mm wide or make two cuts at the edges and drill out the bottom part of the blanks between the fins with a 1mm drill bit it's a small drill smaller than 1/16" for sure. the holes are 3mm which should be enough for a pilot hole or as a mounting hole with a washer on the other side I dun have one of those toshiba's at hand so I'm just going by the data sheet you could use a 1/8 drill to drill out the holes on this mask and then drill 1/8 holes because the diameter of the mounting notches is 3.9mm so 9/64th would be ok too 5/32nd however would be a hair too big. unless your in the over tolerance category because the radius of those notches can be up to 4.1 mm according to the data sheet.

I think 1 mm thickness would do for the toshiba's even if you use 2.5A for the motors, but I'd recommend going 3mm if your gonna blast 3.5A at full duty cycle through these chips.
as for the pcb along the power lines I would go 3D with solder to reduce resistance if your going current happy
otherwise you might end up like this LOL

But on another note can someone verify that the stl came out alright in some of the printing software I looked at the gcode and it's just gibberish to me. (used slic3r macosx version) e I'd like to see if my first thingy came out alright
should be 36 x 40 x 3 mm respectivley and the holes should have a center at x=3 y=5 and x= 33 y=5 with a diameter of 3mm respectively this should be a perfect model for a TB6560AHQ Heatsink but I'm learning here let me know.
terramir

Successful print. I still have some bugs to track down but everything is working well enough to print stuff. It resets when starting the motors if the power is coming from 5V stby. Maybe these drivers are a bit harder on the power supply and need a bigger capacitor to keep the line clean. Or maybe running the driver 5V off standby isn't going to work. We'll see.

Bryan
whatcha think of my heatsink design
terramir

Looks like that would work nicely. Should be easy to make out of bits of sheet aluminum. It comes up fine on RepG so I would guess the STL is in good shape. I haven't needed a heatsink yet but I would probably do something just like that if I wanted to run the motors full out to test max speed. At 2A and 75% torque I don't really need heatsinking, so I'm not in too much of a hurry. Plus I have other bugs to squash first.

at 3 mm thickness it should be sufficient for passive cooling at 3-3.5A X2 (both coils draw this max) because these chips don't burn off that much (i.e. waste in heat) @ 1mm sheet I would not push it past 2.5A without some airflow though and if you make a nice pla or abs case for the electronics make sure you either add a fan or let the heatsinks poke out
Well might want to add heat-sinks on the final draft either way because lower temps mean longer silicone life. (PS it's not the high temps that gets the chip as long as it stays under junction failure temps it's the cold warm cycles so heat-sinking anything that get's hot or even warm is a good idea for longevity
terramir

So it looks like it's just a brown out detect reset. Seems ATX 5V standby is not very reliable as has been mentioned in several other places. Maybe we should have some bootloaders available with brown out detect turned off or down.

The earlier version of Gen7 had BOD turned off and some of them didn't even survive shipping.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

How could BOD being off brick a chip? Maybe corrupt a firmware, but I can't imagine anything serious happening with BOD off.

Quote

How could BOD being off brick a chip?

It easily corrupts not only firmware, but also the bootloader. The chip isn't bricked then, but you have to re-program it.

As I'm currently experimenting a bit on how to improve (lower) the MOSFET heat I've built sort of a heated bed and get these resets, too. Any PWM above 75 (at 78 kHZ PWM frequency) leads to a reset. Not always immediately, sometimes after a minute or so.

So I suspected some interference between PWM and the PSU's stabilisation and tried by lowering the PWM frequency. This made things a lot worse, the maximum possible PWM goes quite close to zero. Connecting the bed to the PSU directly, without a MOSFET in between, heats the bed very quickly, so it's not a straight overload. PWM 255 works sometimes as well, as this is pretty much always-on.

Now I'm scratching my head what might be a proper solution. If I have the PSU on standby and pull the 220V plug, the standby LED takes several seconds to go dark. So, the capacitor flattening 5V Standby seems to be more oversized than undersized. Hmm. Buying a more modern PSU would solve the problem for me, but not for anybody else.

Edited 2 time(s). Last edit at 02/26/2012 05:42PM by Traumflug.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I'm starting to wonder if ATX supplies are worth the trouble. I guess I could add a voltage regulator and pull 5V from the 12V line, but then it seems like there's not much point in using an ATX supply any more. On the other hand they are basically free if they work, since many hobbyists have some old computers laying around that they can part out.

you guys do know the 644p are rated for 2.7V we have brownout set for 4.3V I think why can't we use a different brownout value? or is running it at 20mhz do we need at least 4.3V?
just curious because I think there are other values we can use. PC PSU's should at least supply 3.3V on the 5V standby even under the worst conditions, here's one of those worse conditions we are not supplying enough load for the 5V line, this is why I was suggesting a dual voltage bed that is triggered with the psu this way when the atmega tells the ATX PSU to turn on the bed will immediately zap about 20W or 4Amps from the 5V line this should stabilize out voltages which in turn will cause the whole thing to run a bit better.
Of course this would mean less load on the 12V line for the heated bed as well. this could also improve things.
terramir

Quote

you guys do know the 644p are rated for 2.7V

This depends on the clock frequency, see figure 25-2 on page 329 of the data sheet. For 20 MHz, 4.5 volts is the minimum safe operating voltage, 2.7 volts is sufficient for just 10 MHz.

That said, Arduinos indeed use 2.7V BOD while running at 16 MHz, so they run out of specification for short times.

Quote

I'm starting to wonder if ATX supplies are worth the trouble.

Do you have a better solution, other than using $150+ supplies? So far, many people apparently solve this problem by vastly oversizing the PSU. RAMPS people suffer the same problematic: [forums.reprap.org]

BTW., additional load on the 5V rail doesn't help. Tested this as well.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I found a 1000uF capacitor in my scrap box and soldered this in parallel to the existing 100uF one where 12V comes in. No noticeable change, unfortunately.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

12V 30A supplies from China are cheap, that's what I ran my Gen6 on. [www.ebay.com]
I'm going to use it on my Gen7 once I have the heated bed hooked back up to it. I have a small voltage regulator board that supplies the required 5V to a standard disk drive style molex connector. I'm taking the heated bed mosfet off my Gen7 v1.2 board, as I don't want the board being driven through the 100kk headers - they are only rated for 3A. I see this has been changed on later boards already for a more robust connector.

I don't think it is a PSU issue because all PSU's go out of voltage spec when they are starting up, or turning off. That is why you need the brown out set to match the CPU. Otherwise it can crash during those transitions and run amok. Most CPU's have a very specific sequence to write to flash though, so I am surprised it corrupts the code, but I am not familiar with Atmels.

---

[www.hydraraptor.blogspot.com]

Today I googled almost the entire day to find solutions. It's apparently a well known fact switching power supplies don't work well with PWM applications. The learning experience was low, though, about everybody is experimenting without plan or changes to another supply until the issue goes away somehow. No magic formula like "when doing PWM, you have to put in a cap with X uF * current required".

What I'd like to see is either some sort of circuitry (coils, capacitors, etc.) to enable a switching power supply to deliver reasonably what's on the PSU's spec sheet, even when using PWM.

OR, a well working formula like "If you need 10 amps for PWM and 5 amps non-PWM, your supply must be specified for at least 2 * 10 amps + 5 amps at 12 volts".

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I doubt there is a simple formula because it depends on the control loop of the PSU. Clearly some just work.

Why do you need PWM? I use simple on off control for my bed and it has a very small temperature swing.

---

[www.hydraraptor.blogspot.com]

Quote
Traumflug
Now I'm scratching my head what might be a proper solution. If I have the PSU on standby and pull the 220V plug, the standby LED takes several seconds to go dark. So, the capacitor flattening 5V Standby seems to be more oversized than undersized. Hmm. Buying a more modern PSU would solve the problem for me, but not for anybody else.

This is what I did. I didn't realize my resets were PWM-related. I just assumed my first two PSUs were faulty, which is reasonable since they were in my pile of replaced parts.

Now I have a $50 650W CoolerMaster PSU, and it seems to be doing ok.

Since the poorly tuned PID is effectively working like bang-bang anyway, bang-bang might be The Solution to all this.