Gen7T or Toshiba drivers for Gen7?

I've tried adding capacitance, and extra load on 5V as well and it doesn't seem to help. I also tried a spike suppression diode and no change.

For Gen7T the problem doesn't seem to be a PWM issue because it happens when the drivers are enabled, not when the heaters start up. I guess there is likely some PWM going on internal to the drivers. I can change the clock frequency of the drivers by swapping clk capacitors out. Maybe I'll try that next. But it may be like I said before that these drivers are just harder on power supplies.

As for solutions the $30 12V supplies that are becomming popular around here might be the way to go, though I guess there's not much difference between supplies once you put a regulator on board, you could still use 12V off the ATX and just ditch the 5V stand by. At that point though it seems like a waste of circuit board to have all the ATX specific connectors and such.

After looking at the 3 scope pictures at the bottom of this page: [www.sprut.de] I think these flattening caps are pretty useless, as they're sucked empty by the power supply its self when a voltage drop happens. A solution may be to run 5V through a Schottky-Diode. This would reduce supply voltage to 4.7 volts, but also prohibit electric load to flow from the flattening caps back into the PSU, so it continues to supply the CPU instead.

Assuming all 5V loads sum up to 300mA (plenty) and the drop time (judging by the scope pictures) is 2ms, the required cap would be: c = t * I / dU = 0.002s * 0.3A / 0.4V = 1.5 mF or 1500uF. Two of these, for example: [www.reichelt.de]

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

does all these problems only apply to the gen7T or also to the gen7? I bean the heated bed and brownout issues, and if so what are the current fixes regarding these issues right now?
just curious since I have not even programmed my 644P yet LOL.
Been busy trying to finish the build and the hotend right now I'm filing a square whole for my sand resistor LOL.
Anyways just curious
terramir

Quote

what are the current fixes

Ha ha ha. Possible fixes are exactly what we discuss here.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

yeah but are these problems happening on the regular gen7 as well that is the question?
terramir

I've brought up the BOD issue on mikrocontroller.net, a german electronics forum: [www.mikrocontroller.net] They were very nice, helpful and quick.

Four solutions were found:

1. Use a wall-wart PSU for the ATmega, the PC-PSU for the high current things. Not exactly elegant.

2. They think the Schottky-Diode thing mentioned above should work.

3. Using an LM7805 or similar to produce 5V from 12V should work as well.

4. One guy remembered the 5VSB rail on a PC-PSU is pretty much a supply distinct from the main supply, so it shouldn't bog down on current bumps on the 12V rail.

I think 3. is fine for generic PSUs, 4. to be prefered for PC-PSUs. That means getting rid of the power supply jumper on the Gen7 board, routing everything 5V to 5VSB and put in an optional LM7805.

bryanandaimee, if I see your work correctly, you already have everything connected to 5VSB. You could simply pull the 5V (red) strand from the Molex 4-pin to test case 4.

I'll try to re-route my test board as well as finding a proper LM7805 solution. These LM78xx need some special considerations, like a backslash-diode.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Quote

are these problems happening on the regular gen7 as well that is the question?

Yes. If the power supply is weak and the heated bed is strong. You'll notice it without doing any harm and as you get there, we might have found a solution (other than buying a bigger supply).

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I think I've found the issue. The problem aren't voltage dropouts on the 5V rail, but voltage spikes on the GND rail.

Look at the (light blue) GND track of v1.2 on this picture:


Look at where GND comes in from the power supply, where the MOSFETs with their current bumps sit, how thick these tracks are and where the ATmega gets his ground. If you know it, it's pretty obvious all bumps coming from the MOSFETs directly hit the ATmega.

I've a fix already which requires cutting two tracks and soldering a cable or two, I'll document that later today.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Cheers for the link to this topic (backlink: [forums.reprap.org]).

I've tried option 3 independently (since I figured out I needed a 5V line, so my 12V power supply alone wasn't enough) with a 2200uF cap to stabilize the power line - it did not help.

I'll gladly try your new solution as I already need to solder in a new MOSFET; this time with a heat sink attached...

Re option 4. Nothing but the limit switches is connected to the 5V rail on Gen7T unless you jumper the 5V supply for the ATmega, which I never do. And I have run the board multiple times with no limit switches plugged in so the 5V rail was completely disconnected from the rest of the board, the reset issue did not change. I think we can rule out option 4 as the primary solution though it may be part of the solution later. So your theory about ground issues is looking more and more likely. I'll try the schottky diode fix as well though when I get back home.

Here we go: Improving Gen7 v1.2 Heated Bed Tolerance.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I may be able to put some input in here.

I am running a 7805 on my Gen 7 1.3.1 and have it supplying power to the 5v rail through where the disk headers are supposed to be. It works quite well, Atmega powers up properly and there's no problem with voltage drop through the longer traces.

I don't have one, but I would suggest rather than using an LM7805 or the like, substitute it for an Oki 78sr series.
Price is about 5x the cost of a 7805, but being a switching regulator it is far more efficient, and it's reported 20mv ripple should be cleaned up effectively by the caps on the Gen 7 board. (it's directly pin compatible to the TO-220 package as well)

Though the cost wouldn't be prohibitive, I suppose since we're using power supplies it may not be justifiable. It may be worth it just to some holes for a to-220 vreg onto the board as an option for the builder.

Quote

substitute it for an Oki 78sr series

You mean this one: [de.mouser.com] ? That's 25 times the price of the LM7805: [de.mouser.com]

Gen7 needs just 200mA on the 5V rail, so a regulating resistor is manageable. Preparation for such a thing is a good target, though. Or a small board inside the connector.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Before I start cutting: will the ground rewiring also work if I do not have an ATX cable plugged in?

Traumflug, might this help hot end temperature reporting be more stable as well? I'm not using the heated bed on my Gen7 1.2, but the hot end temperature reported by the thermistor does seem to fluctuate a lot, dropping when the hot end heater is on, then jumping up high when the heater is switched off (this may be the other way around).

Quote

will the ground rewiring also work if I do not have an ATX cable plugged in?

Uhm, no. But you can try to lay wire 4 from the ATmega to the molex 4-pin. Not sure wether this helps, though, as there is only one path back to the PSU. I've put in a note into the wiki article.

Quote

might this help hot end temperature reporting be more stable as well?

Likely, yes.

In any case, if you cut a track and things don't get better, you can alwas reconnect that track witch a short piece of wire.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

To fix the temperature reading problems cut the 5V and ground lines that go to the thermistor connectors and link them direct to AVDC and VSS pins on the Atmega. Analog power and ground should never share traces with digital or power circuits.

---

[www.hydraraptor.blogspot.com]

Traumflug Wrote:
-------------------------------------------------------
> will the ground rewiring also work if I do not
> have an ATX cable plugged in?
>
> Uhm, no. But you can try to lay wire 4 from the
> ATmega to the molex 4-pin. Not sure wether this
> helps, though, as there is only one path back to
> the PSU. I've put in a note into the wiki
> article.

I'll test it as soon as I find a soldering iron that gets hot enough to solder the new (thick) ground wires in place. As a plan B, I'll use a seperate wire to ground the heaters at the PSU directly. And in the same fashion, cut the +12V to the heaters, add a cap between GND and +12V and solder a direct +12V to the heater supply track.

That way, the heaters and the rest of the Gen7 are not sharing +12V and GND directly anymore.

If I understand nophead's suggestion correctly, it would look like this, 2 cuts, 2 wires:

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Yes that looks right although I would star point at the actual CPU pins. You still have short pieces of shared track. The noise through the decoupling caps and the CPU current will add a small offset voltage to the thermistor. With such short sections the effect should be negligible but I am a perfectionist!

---

[www.hydraraptor.blogspot.com]

I've added a 2nd molex connector for the heater + alternate ground on the back of the PCB.

Even though I currently have it all connected to the same loom, it seems that the initial separation of the ground did the trick: the Atmega no longers seems to reset while the bed is on.

On a similar note: how hot is the MOSFET switching the bed supposed to become? I currently have it on bang-bang control and when the MOSFET is on (during heat up), I saw a puff of smoke. My IR-gun shows that the heatsink on the MOSFET is reaching 190 degrees celcius...

Quote

how hot is the MOSFET switching the bed supposed to become?

See [reprap.org] and [forums.reprap.org]

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Cheers for the info.

The only thing I am missing is that these tests are run with PWM. In my case, the MOSFET is overheating with no PWM at all. If I understand your tests, you are trying to determine how the temperature increase due to PWM is minimized. (correct me if I'm wrong)

Alternative solution:
The internal resistance of the MOSFET coupled with the current causes the heat. I want to try an external board with 4 MOSFETs in parallel (with 4 heatsinks) to see if in that case I can run the 10A longer without creating a second heated bed :-)

As an added bonus, I can run a seperate molex connector to this new board.

What type of MOSFET are you using? You can probably lower the ON resistance by increasing the gate drive voltage using a small transistor. Cheaper than putting 4 in parallel. Alternatively simply use a better MOSFET.

Edited 1 time(s). Last edit at 03/03/2012 07:31PM by nophead.

---

[www.hydraraptor.blogspot.com]

I'm not near my printer atm but it cost me 4 euros and can switch 50 amps. Perhaps the problem is indeed a lack of voltage (from the atmega)... I'll measure it later on.

A lot of MOSFETs need more gate drive than 5V and most give the lowest RDSon at about 10V.

---

[www.hydraraptor.blogspot.com]

Quote

The only thing I am missing is that these tests are run with PWM.

Simply ignore everything between the leftmost and rightmost data point. PWM 0 is identical to a non-PWM off and PWM 255 is identical to a non-PWM on.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I actually meant the smaller version sold at digikey [parts.digikey.com]

But you are correct, the linear Vreg is quite manageable.
What did you mean by the 78xx types requiring a backlash diode?

Quote

What did you mean by the 78xx types requiring a backlash diode?

It's shown in the LM78xx data sheet.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Traumflug, your ground track fix worked for Gen7T, I now no longer get resets on motor driver enable. I think I'm 99% there for a beta board. I'm wondering if I should add the schottky diode in anyway just for kicks. Also I think I'll leave the power-on pull up resistor off since it did absolutely nothing, and just add a warning about USB power. Any thoughts?

Edited 1 time(s). Last edit at 03/06/2012 12:19AM by bryanandaimee.