RAMPS for Due!

Be careful of low voltage zeners. Below about 5V or so they can be very leaky / high slope. A clamp reference may be easier to do accurately with one of the cheap shunt regulators.

Quote
uncle_bob
Be careful of low voltage zeners. Below about 5V or so they can be very leaky / high slope. A clamp reference may be easier to do accurately with one of the cheap shunt regulators.

Will definitely keep that in mind, thanks.

How well does your typical shunt regulator handle current flowing into it from a higher voltage rail? Usually the current is simply flowing out of it and to ground.

Something that is designed as a shunt will handle current into it just fine. The main issue is whether it will handle enough power. A typical setup is to bias it with something like a 10K off of the 12V line so its active and then let it absorb what ever you are going to dump into it. If there's no 12V, it will be at zero volts and you will have your clamps start to act at 0.7V rather than 3.3 (or what ever).

[www.onsemi.com]

Is one of many parts out there. You can get them in TO-92's.

They are commonly available and in the 30 to 60 cent range at 1-100 pieces. You would have to add three resistors and a cap to that cost to compete out the BOM. The result would be a clamp source you could set to (say) 3.0V so your diodes would cut in at 3.3. There are indeed more exotic / higher current devices. Given the limited time I searched, I'd bet there are cheaper parts as well.

Quote
Cefiar
The reason 1K was chosen is that it gives a reasonable response around the critical area, but doesn't put too much current through the thermistor. They all have max power ratings, and putting too much current through them will kill them.

I don't quite get that, power ratings are based on thermal rating, so if you have a thermistor rated to 300 deg it doesn't seem like self-heating would kill it.

Anyway 1k seems like a good compromise, although I don't see generating thermistor tables as a big issue. Firmware will need to be written for the Due, and if people want to use 12 bit ADC will need new tables.

However, this is the disconnect for me, if we are choosing a "standard" part with a typical tolerance up to 5%, then it doesn't seem worth going the extra miles to improve the layout to get below 1% - any improvements are swamped by other errors.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Quote
uncle_bob
Which brings us back to the original departure point of reduction of noise on the Due board....

Is there a (practical) way to give the thermistor "empire" it's own dedicated ground going back to the Due? Have the thermistors floating unless the Due is plugged in. What ever performance you get that way, it's going to be the best you can get.

In my firmware I sample the ADC at 100 Hz to form a filtered average and the control loop runs at 5Hz, so high sample rates are not required.

There are 5 ground pins on the Due shield, in 3 groups. None of them are very close to the thermistor inputs, but we might be able to have a separate ground trace. There is a single ground plane on the Due, the ground pins nearest to GNDANA are by VIN, which unfortunately is furthest away from the thermistor inputs.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

... so we are back to the basic limitations of the Due board.

---------------

Just so it's been said - I have never ever in 40-50 years of doing this done a "perfect" design. There *always* are compromises. The only thing I'm trying to do here is to bring up options. The worst thing you ever hear is "if I'd only looked at that ...". What ever we come up with here will certainly be a good and useful board. Feature creep is indeed a major risk in a project like this. There very much does need to be one (and only one) person ultimately deciding what's in and what's out. In this case, I'm not that guy.....

I am wondering if perhaps we could relocate the thermistor header.

If we relocate the reset switch and reset header, there should be enough room to locate the thermistor header there.

The advantages are:
1. 2x ground pins right there.
2. Reduces trace clutter around the mounting hole near Due pin 21.

I also think it might be possible to provide a dedicated ground from the same area just to the thermistor components. There is also a ground pin near them (the centre header P106 provides a ground to the Due as well - one you missed in your analysis bobc), which could be connected as well. In fact, by distance this is the second closest ground pin to GNDANA (assuming a solid ground plane on the Due).

Would take a lot of redoing of tracks in the middle of the board though. Fortunately we've moved U1 (the '125) out of that area so there is now a lot more room, and adding in the thermistor protection circuitry is going to change everything around there anyway. Might move C204 and P214 as well just to make it easier to run tracks to that bottom corner.

With the reset switch/header. the only thing I'd suggest is we try and keep the reset line away from the FETs. Last thing we want to do is induce a signal in the reset line and cause random issues.

bobc: If you say we should do it, and also put the thermistor protection circuit into the schematic, I'll happily do the layout side.

Edited 1 time(s). Last edit at 12/29/2013 07:26PM by Cefiar.

Quote
Cefiar
I am wondering if perhaps we could relocate the thermistor header.

If we relocate the reset switch and reset header, there should be enough room to locate the thermistor header there.

The advantages are:
1. 2x ground pins right there.
2. Reduces trace clutter around the mounting hole near Due pin 21.

I also think it might be possible to provide a dedicated ground from the same area just to the thermistor components. There is also a ground pin near them (the centre header P106 provides a ground to the Due as well - one you missed in your analysis bobc), which could be connected as well. In fact, by distance this is the second closest ground pin to GNDANA (assuming a solid ground plane on the Due).

I did miss that one, but I wouldn't rely on it because that header is missing on some boards, notably UDOO. On UDOO I think that means no SPI, but since it is hooked to a linux computer that can handle storage and GUI etc it should not matter. But if thermistors didn't work, that would be a problem.

Quote

With the reset switch/header. the only thing I'd suggest is we try and keep the reset line away from the FETs. Last thing we want to do is induce a signal in the reset line and cause random issues.

bobc: If you say we should do it, and also put the thermistor protection circuit into the schematic, I'll happily do the layout side.

Moving the reset looks like a good option, I will leave that to your discretion.

I do like the idea of thermistor protection, I will try to work on the schematic soon.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

I'll look into it once we've got the thermistor protection circuit in the schematic. No use changing anything yet as that will change a lot of things anyway.

FWIW: On the github bug for the thermistor protection, I linked directly to the .PDF of the schematic for the RAMBo which has the protection circuitry as a reference.

Ok, I put in some proposed circuitry for the thermistor protection plus a couple of other small changes. [github.com]

I have put the circuit in for THERM0, and a couple of variants with Zener and TLV431. The TLV431 circuit I just copied from the datasheet. I am straying beyond my normal skill set here, so I am relying on you guys to tell me what I need to do!

It might be nice to test out the circuit on some stripboard first, I'll order some components shortly, I also need to fix my Due

With a certain a amount of trial and error, I determined that a value of about 220R seems to be a reasonable value for the gate resistors. It gives turn on time in the order of 1us or 3, without too much oscillation.

The traces show output at the logic buffer (HC244) and the voltage on the mosfet drian.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Ok, for the TLV431:

1) Bias resistor from +12 to top of regulator.
2) Nice big cap across regulator, a > 10uf electrolytic would do fine
3) Set resistors to get it to 3.0V
4) Clamp diodes going from the thermistors to the regulator

I suspect that's not any different than what you have already done.

Cool, I think that is what I have. I guess that if we set the limit at 3.0V, then any voltage above is clamped there. That would mean we can't read temperatures below 80 degrees.

While my Due is out of action, I thought I would look at the Mega. The pins from the heatbed connector and extruder 0 collide with the top of the USB connector on the Mega. I really wouldn't like to use it like that, even if there is insulating tape. The FD does not sit properly on the headers, and it would be too easy for the pins to puncture the tape.

The only safe way I would use this is if the headers on the Mega were taller to allow clearance, or we leave a large keep out area around the USB connector (which RAMPS almost does).

We have a bit of space which we could move the extruder 0 connector clear, but the heatbed connector would need to move in at least 6mm, which would be a pretty extensive change I think.

Unless anyone can think of better options, we might have to reluctantly say that RAMPS-FD is not suitable for the Mega, at least not out of the box. I've seen some pretty cheap Due clones ($36), so maybe it is not so much of a problem.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Setting the device at 3.0V means that with the additional diode that should be in between it and the thermistor signal, you should get at least 3.3V, and at max 3.6V (depends on the forward bias voltage of the diode of course).

From what I remember seeing, 3.6V is within the Due's max input voltage range. This way we should get full scale readings with protection for the Due from higher voltages.

Regarding the Mega, you would want to do the following:
1. Use additional headers to provide more height. This means either you'd want to use longer pins, or you could use a bunch of stackable headers to extend the existing board.
2. With the thermistor circuitry, you would need to adjust the voltage divider resistors on the shunt regulator so that they're set for 4.7V, rather than 3.0V.

Both of these things are not insurmountable, but most likely can't be done by your average Joe. If we document it all, and point out the issues, then people will at least know what to do if they want to go down that path.

IMO it's not recommended using it on a Mega unless you ABSOLUTELY know what you're doing.

Note: The USB connector on the Mega is annoying even with RAMPS. Admittedly you can at least cover it with tape, but it's still annoying. I'm quite surprised there aren't more Mega clones with a MicroUSB connector. I only know of one off the top of my head, but it also has on-board Ethernet.

Edited 3 time(s). Last edit at 01/01/2014 08:23PM by Cefiar.

Ok, if use 1k5 we get 3.1V reference. Looking at the SAM3X8E datasheet it says the max voltage is 4.0V on any input pin.

Now you remind me I have got a set of stackable headers somewhere, I will try those. ISTR there are plenty of these on ebay quite cheap.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Quote
bobc
Ok, I put in some proposed circuitry for the thermistor protection plus a couple of other small changes. [github.com]

I have put the circuit in for THERM0, and a couple of variants with Zener and TLV431. The TLV431 circuit I just copied from the datasheet. I am straying beyond my normal skill set here, so I am relying on you guys to tell me what I need to do!

It might be nice to test out the circuit on some stripboard first, I'll order some components shortly, I also need to fix my Due

If we go with the TLV431, what component packaging do you want to use? I personally think either TO-92-3 or (more likely) SOT23-3.

Can't say I really have a preference, but SOT-23 seems to be a good option.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

If you already have SMT parts on the board, then the SOT-23 makes a lot of sense.....

I agree

Ok, I am updating the board layout with the thermistor and other changes then.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Just a Q: I looked through the example circuits you created and noticed the 1K resistor change.

I'm fine with that, however I do have one suggestion:

Where the pull-up resistor is 1K, can we change the in-line resistor to 1k as well?

Not sure if it makes that much a difference, but for some reason something is niggling at the back of my head telling me that they're the same value for a reason. Admittedly I've only just woken up and haven't fully absorbed my first cup of coffee, so I may be mistaken.

Sure, we can do that.

I've just pushed a new layout with the thermistor changes, and a few other things. I added the locking connector for E-STOP, and also a external 5V connector. For now the thermistor connectors are in the same place, we can review if we want to move them around.

I am thinking that we have made a number of significant changes, some of which are incompatible with previous version. Perhaps we should call this V2?

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Given the trouble we have seen on the Due board layouts with un-labeled board changes ---- put a rev on the board and the docs !!!!

I often find myself using things like A1 A2 A3 for small changes and go to B for the big changes. If I was smart, I'd probably start at A0, but I never seem to expect things to change.....

Edited 1 time(s). Last edit at 01/02/2014 10:12PM by uncle_bob.

I seem to remember changing the silk on the layout to say V1 Rev B a while ago.

I'm happy to have it be V2 Rev A.

Note: There's still a few labels that could be moved about to make a bit more sense, etc, but nothing ground breaking or huge.

I do notice that the ground plane fill isolation gaps seems to be slightly smaller. Did you alter the rules slightly? I say this because I now see ground fill on the bottom around E0-MOT, seeping in between the round pins on the motor connector. I'm not against the change, but the square pin on that connector seems to stop it being the same everywhere. Interestingly, if I remove the fill then re-apply it, they aren't there. Only if I load it from scratch.

There is also not a huge gap around the top layer 100uF caps under the stepper drivers. There could probably be a slightly bigger gap between the edge of the GND pads and the +V_MOTOR line. Not a huge thing to change though, just a matter of getting the track widths right. Note: This was most likely the same on the previous revision, so I doubt it's much of an issue.

Also notice that an extra thin ground (on the bottom layer) that I put around P802 (the endstop header) vanished, but there should be more than ample grounding in that area now (I was being VERY redundant when I did that, just to be sure).

Only other thing is the 5V rail between JP801 and P801 is thinner than the rest of the track along that area. The only way to fix that would be to re-arrange the tracks again around the hole near Due pin 21. Got an idea on how to do that btw (involves putting Z-MAX around the back of 5V_5, X-MIN between 5V_4 and 5V_5, running N-00000193 on top (like Z-MIN, but moving from top to bottom then back - adds 2 more vias), then moving some other bits around (mainly vias like FD_5V, D32, D43, UART2_RX_LV, and their associated tracks, etc).

Scratch that - Realised that we can move Z-MIN slightly nearer Due Pin 21, and then run N-00000193 around the left of the hole rather than the right. This allows us to move N-00000194 enough to widen the track at that end, but we still need to move a few other bits around (as before, mainly vias like D32, D43, UART2_RX_LV, and their associated tracks, etc).

Edited 3 time(s). Last edit at 01/03/2014 12:27AM by Cefiar.

Quote
Cefiar
I seem to remember changing the silk on the layout to say V1 Rev B a while ago.

I'm happy to have it be V2 Rev A.

Note: There's still a few labels that could be moved about to make a bit more sense, etc, but nothing ground breaking or huge.

Ok, perhaps we can update as we go.

Quote

I do notice that the ground plane fill isolation gaps seems to be slightly smaller. Did you alter the rules slightly? I say this because I now see ground fill on the bottom around E0-MOT, seeping in between the round pins on the motor connector. I'm not against the change, but the square pin on that connector seems to stop it being the same everywhere. Interestingly, if I remove the fill then re-apply it, they aren't there. Only if I load it from scratch.

Interesting, I didn't change any fill settings. I tried refilling the previous version you did and it produces different fill like you describe. So unless there are some hidden settings somewhere, perhaps this is due to different version of Kicad?

Quote

There is also not a huge gap around the top layer 100uF caps under the stepper drivers. There could probably be a slightly bigger gap between the edge of the GND pads and the +V_MOTOR line. Not a huge thing to change though, just a matter of getting the track widths right. Note: This was most likely the same on the previous revision, so I doubt it's much of an issue.

Also notice that an extra thin ground (on the bottom layer) that I put around P802 (the endstop header) vanished, but there should be more than ample grounding in that area now (I was being VERY redundant when I did that, just to be sure).

Ok, we can adjust that. I was trying to merge segments with a global track edit, but I noticed later it seems to completely remove some tracks it doesn't like. Possibly the algorithm thinks some tracks are redundant. The options in Kicad have changed from previous versions, so I am not entirely sure of what it is doing. There used to be an option to delete specific track nodes, I can't find that anymore. It's only a problem because the Drag command refuses to work on multiple segments.

Quote

Only other thing is the 5V rail between JP801 and P801 is thinner than the rest of the track along that area. The only way to fix that would be to re-arrange the tracks again around the hole near Due pin 21. Got an idea on how to do that btw (involves putting Z-MAX around the back of 5V_5, X-MIN between 5V_4 and 5V_5, running N-00000193 on top (like Z-MIN, but moving from top to bottom then back - adds 2 more vias), then moving some other bits around (mainly vias like FD_5V, D32, D43, UART2_RX_LV, and their associated tracks, etc).

Scratch that - Realised that we can move Z-MIN slightly nearer Due Pin 21, and then run N-00000193 around the left of the hole rather than the right. This allows us to move N-00000194 enough to widen the track at that end, but we still need to move a few other bits around (as before, mainly vias like D32, D43, UART2_RX_LV, and their associated tracks, etc).

That would be a useful change I think.

I'll take a pause for the moment while I do some other stuff, but we can collect up any other issues in the meantime.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

never mind hopefully.. my package has been found in philly.. what the heck is it doing there..
now its in harrisburg..


heres hoping its still okay..




speaking of these ramps-fds. anyone got a spare one in the usa? I ordered one on 12/02/13 and it made it to a town 50 miles away on 1/2/14, and then it seems the postal service lost it. and I really dont want to order another one from china. since it took so long.. I really need one badly so I can keep going on my kossel mini build..

Edited 2 time(s). Last edit at 01/04/2014 05:02PM by firehopper.

Quote
firehopper
never mind hopefully.. my package has been found in philly.. what the heck is it doing there..
now its in harrisburg..
....


speaking of these ramps-fds. anyone got a spare one in the usa? I ordered one on 12/02/13 and it made it to a town 50 miles away on 1/2/14, and then it seems the postal service lost it. and I really dont want to order another one from china. since it took so long.. I really need one badly so I can keep going on my kossel mini build..

Philly <-> Harrisburg is typical UPS silly season on their scheculing computers. Seen it a *lot* of times.

------------

If you want to use this on a production printer, I would very much suggest you wait on buying one of these boards. The deisgn is far from fully debugged.

and yay, got my ramps finnaly, and thanks to help in the reprap channel and ambro's work on his aprinter firmware for delta and Due, I have motion on my udoo powered printer
now all I need is extruders and stuff

I have updated the schematic and layout with some thermistor changes and other things Cefiar mentioned. I have routed a separate ground from the thermistor header to two of the ground pins on the Due header.

I wondered if we should be using AREF? It might be a cleaner reference for the thermistors, and I think it would be quite easy to route.

I've also been looking at the D102 crowbar diode. It's supposed to protect against reverse polarity by blowing the fuse. But the 1N4004 only has a continuous current rating of 1A, and a peak current of 30A for 8.3 ms. That does not seem nearly long enough to blow a fuse, which might take 0.5s.

So for that to even work, we would need to change it to something nearer 10A rating. Call that Option A. A possible part is : [uk.rs-online.com]

However I'm still not sure that provides much protection, since -12V is still applied for maybe 100-500 ms. It would seem to make more sense to use a blocking diode in series, which would need to be rated 10A+, there are some available at 15A in TO220 packages. (Option B ) We might be able to get by without a heatsink on it, there is also the voltage drop to consider.

I am fairly sure the 1N4004 is a waste of space though. I think we should at least change it to Option A. Thoughts anyone?

Edited 1 time(s). Last edit at 01/08/2014 05:33PM by bobc.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

I'm not a big fan of series diodes. You paid good money for that power and now you are just throwing it away .

There are some pretty cheap D2PAK dual diodes out there with big current ratings. The big problem is the area they take up on the board.

[www.onsemi.com]

okay I have video, and a image of my ramps-fd

version 1 revison A

[www.flickr.com]

and

[youtu.be]

Quote
bobc
I have updated the schematic and layout with some thermistor changes and other things Cefiar mentioned. I have routed a separate ground from the thermistor header to two of the ground pins on the Due header.

I wondered if we should be using AREF? It might be a cleaner reference for the thermistors, and I think it would be quite easy to route.

As long as we can pull the appropriate amount of current from it, and it'll be stable, I guess it makes sense. I'll have a proper look and leave a better opinion soon.

Quote
bobc
I've also been looking at the D102 crowbar diode. It's supposed to protect against reverse polarity by blowing the fuse. But the 1N4004 only has a continuous current rating of 1A, and a peak current of 30A for 8.3 ms. That does not seem nearly long enough to blow a fuse, which might take 0.5s.

So for that to even work, we would need to change it to something nearer 10A rating. Call that Option A. A possible part is : [uk.rs-online.com]

However I'm still not sure that provides much protection, since -12V is still applied for maybe 100-500 ms. It would seem to make more sense to use a blocking diode in series, which would need to be rated 10A+, there are some available at 15A in TO220 packages. (Option B ) We might be able to get by without a heatsink on it, there is also the voltage drop to consider.

I am fairly sure the 1N4004 is a waste of space though. I think we should at least change it to Option A. Thoughts anyone?

Quote
uncle_bob
I'm not a big fan of series diodes. You paid good money for that power and now you are just throwing it away .

This is the reverse-biased diode across the supply (ie: get the supply into the board backwards, it forward conducts so the fuse will blow).

Usually what happens is that the diode here fails as part of it's job of protection. I've seen a number of posts recently (eg: about Rumba) where people have connected the PSU backwards and the crowbar diode dies, going short circuit. This protects the rest of the electronics at the cost of a diode and some soldering. Unfortunately I don't know if the failure mode for large currents is ALWAYS to go short, or to go open. Of course, this also depends on things like the tracks on the board (ie: Will the failure be the diode or the circuit around the diode).

I don't think there is any perfect answer here, but something slightly better than a 1N4004 would always be a good idea.

Re: The forward diode on VIN - Unfortunately a lot of the Arduino boards I've seen end up with ~5V on VIN if you connect them to a USB port (ie: no other voltage on VIN). If there is no forward diode between the supply and VIN, then you'll get +5V back onto your supply, which means things like a fan on an aux connector, etc could try and run. Draw too much current backwards through the on-board reg on the Arduino and you'll end up with a fried reg on your Arduino when you connect via USB with no PSU connected/on but with other devices still connected (or if your PC isn't great a dead USB port - or both). This also happens with 3 pin pin regs, so the 12V reg doesn't save you either if you're using that (for > 15V supply).

Edited 1 time(s). Last edit at 01/08/2014 07:30PM by Cefiar.