I'm working on a build using Yaskawa servos with SGD7S drives, and I have some questions. I plan to use a 6XD mainboard, and I'm starting design of a PCB (my first one, so bear with me...) to streamline connections between the 6XD and the drives. Assuming I keep all the magic smoke in, I'm happy to share the final product with anyone looking to use similar drives.

Question 1: "Servo ON". The "Servo ON" inputs are 24V, opto-isolated, w/4.7kOhm resistors. So, they should be 5milliamp. Can the 6XD's EN(-) pins handle this, or are they only 5V tolerant?

If they do tolerate 24V, could they also handle 3 parallel inputs; 15milliamp at 24V? There are separate forward/reverse prohibits which also need to be pulled on, otherwise the servo will be enabled/on, but not permitted to move. It would be easiest to just link them all, and control all three with EN(-).

Question 2: Step/Dir voltage. The drives want 3.3V inputs for step/dir, not 5V. Based on the wiring schematic, 100ohm resistors on DO_Step(-) and DO_Dir(-) would correct the voltage. Would adding these resistors adversely effect rise/fall times? The drivers want ≤ 0.025μs rise/fall on step/dir signals. I realize the step/dir signals could be sourced from 3.3V, but I'd prefer the resistors if they work. (I started a thread with this specific question earlier, but the questions are already branching out and may continue to do so as this build progresses, so I thought a generalized thread would be more appropriate.)

Thanks for any assistance!

Well, it now exists.

All truly necessary functions can be controlled with the 6XD driver inputs and +24V. With two opto-outs, alarm reset and torque-limiting can also be controlled, and two additional opto-outs can jumper-select between addressing P-Control mode or one of two free breakout pins for user-selected functions.

A GitHub repo has been set up with readme, schematic file, and pcb file. There are a lot of used Yaskawa drives out there; maybe this will help some people who want to try the servo route.

Thanks @dc42 and @chrishamm for the help!

Looking great! Keep the pictures coming if you've got 'em. Boy, your electronics section is so much neater than mine, making me think I may need to go knock mine over and see what I can do.

Original config is interesting; mine did not have the second 24v power supply when purchased in mid 2019. Otherwise, looks very similar, and I see they're still just duct-taping the capacitors together.

One of your images reminds me of that little L-beam under the x-axis rail. I have switched to stainless-endcap, high-temperature rails, so it's not 100% comparable to your setup, but still, pretty sure you can remove that. Looking at it, I couldn't imagine it actually added any useful rigidity, and indeed it doesn't seem to. Removing it drops about 120grams from your y-axis load, and also gains you additional clearance if you ever do sequential/multi-part printing. The rail itself is more than enough to carry the gantry, in my experience.

Getting an air pump in the spare bay as you have is currently high on my to-do list, currently modeling a TPU mount for it. I've just finished a new watercooled Dyze extruder with cooled toolboard, but it will be a Berd setup, so I can't make the swap until I get the pump setup.

@ageis I suspect you'll need to provide more info before someone who really knows the board can provide good help. Presumably you're running the bed on DC? What voltage & wattage? Have you taken a close look at the board for signs of a damaged component? If you don't know what you're looking for, high-res photos of the area in question would help. Do you have access to a decent multimeter? Confirming that the bed heater is truly fully off when off would probably be a good check, as would gate voltage of its mosfet when on.

@Dizzwold I didn't mean anything aggressive/argumentative by my statement; I'm not put off in any way, it's all good. Just when you said that the I/O switch missing might explain some of my other points, I wanted it to be clear that it didn't.

I have trouble with left/right (not dyslexic in any way, just don't process those two particular words correctly for some reason!). I was careful through the post, but then got it wrong in my edit comment. Apologies for the confusion!

I do not agree that switching should be on neutral. In DC circuitry it's often sensible/preferable to switch to ground, but in dealing with mains voltage this is not good practice. If you're cutting off/on mains voltage to an appliance (which the printer is, for all practical purposes) switching should be done either on line/load or on both line and neutral, and you should do it as close to mains entry as possible.

Imagine the following scenario; SSR2 faults closed. The Duet detects a heater fault, and turns off the SSR2. Faulted SSR2 continues to overheat the bed. Eventually Duet detects that things are now way out of spec, and shuts down SSR1. Ok, the bed is now "off". However, the bed heater is still energized to line voltage; you have a potentially defective component (it did just potentially overheat) that has 110/220V running to it, possibly distributing that potential to the metal bed or elsewhere. Additionally, if it has a good enough (high-current enough) short, then it may not even be off if there isn't a ground-fault interrupt in the system.

That's the biggest issue, but similarly, the line and neutral terminals of both PSUs are always live when the mains entry is plugged in/switched on (not needing an SSR failure, just by default with this wiring, they're live). The PSUs are not "on", because their neutral line is open, but not being on is not the same as not being dangerous; they're still at 110/220v potential.

If SSR2 fails closed and the left is off/open, there is NO terminal or wire on your mains circuit that isn't energized. The same is true if SSR1 faults closed and the other is off/open. If both SSRs are functional and off... there's no terminal or wire on the mains circuit that isn't energized! The top blue wire is safer than others, but it's not safe.

When dealing with mains voltage, we want "off" to mean both off and deenergized to the extent possible, which means switching the line voltage as early as possible.

As one additional note; personally, I would prefer not to have two SSRs as my safety cut, simply because they're similar devices and so will have similar failure modes. I.e., if something causes one to fail, it may well cause the other to fail as well. So, I would vote the SSR plus mechanical relay route as safer. Or, thermal fuse on the bed. This is down in the weeds level though; if I also already had two SSRs I'm not sure I would take the trouble to switch to an SSR and relay. The switching line instead of neutral is the low-hanging fruit here.

Edit: P.S., IMO the best under-bed insulation is carbon-felt welding blanket material.

@curieos I'm only just starting to play with the 6XD, but triggering the 6XD local drivers' "ERR" pins certainly does seem to raise an event. I get an Event Notification popup: "Driver 0.1 Error: Exernal driver error". I've yet to try actually doing anything with this event, but I assume it would run the macro if I had one.

Remember the 6XD does have an external driver error selection jumper for pull-up vs pull-down. Also, are you sure the external drivers themselves actually triggered an error? From what you describe it sounds like they may have just happily continued on.

@dc42 I did also test with drivers already powered first, but I only had two drivers connected, so based on this majority-rules approach it wouldn't have decided on NC configuration anyway.

I'll test later and confirm whether I get the response you expect. Either way, I do think the ability to set this logic would be a significant improvement, both for the delay in driver boot-up, and for the added assurance that the firmware will not make an incorrect assumption if something goes screwy at startup.

@lazy_mosquito said in how to enable out6 on 6xd for M569.7 command for brake:

when i later use the command M569.7 P0.0, it seems it doesn't do anything on port out6.

@lazy_mosquito I could be wrong about this, as I haven't used the brake function, however I would expect the brake output level to change when you run M17/M18, not when you run M569.7 again.

@Bad-Joker Whether a pelt can do what you need it to do without a fan is not a simple question that someone can confidently give you a yes/no answer on.

If you are allowing your sensor to hit 80C and your chamber to be 110C, your first thought might be that you need a peltier that creates a 30C difference. That is not the case; you need a pelt that will sustain a higher temperature difference, because if the hotside of the pelt/heatsink is at 110C, it is not releasing any heat to the chamber. I.e., your hotside must be higher than chamber temperature. Additionally, the pelt itself is a heater to the extent of it's inefficiency, which will also vary with temperature and temperature difference. The balance will be determined by the °C/W rating of your heatsink, which varies heatsink-to-heatsink, as well as with airflow. So, you are chasing a multivariate equilibrium point, and the answer is therefore "maybe, maybe not."

Another option would be to put the sensor on a waterblock and watercool it, in which case the answer to whether it is cool enough will almost definitely be "yes."

@Electriceye

Since you have done the basic electrical checks, I expect that your M307 parameters are far enough away from your bed's actual response that the Duet is assuming a fault. Immediately jumping out to me is a dead-time ("D" parameter) of only 1.53 seconds. Based on your 1000W heater, I assume you have a big, heavy aluminum plate for your bed. If your thermistor is in the bed and not just sitting on the heater itself, dead time could be much more than this. I have an aluminum bed with a 1200W heater, thermistor placed into a drilled hole on the perimeter, and my PID-tuning calibrated dead-time is just over 30-seconds. Don't remember what the "C" parameter is exactly, but mine is over 3000. (Note; C parameter is deprecated--though still read--in latest firmware)

If you're confident you're not burning anything down, you can also play with increasing M570 parameters to delay error-triggering and see what happens. Once calibrated, remove or re-reasonable-ize M570 for safety.

@Electriceye A 1000W heater should not trip a 15amp breaker; it only draws about 8.5 amps. If your heater matches your bed mass appropriately it should not be melting instantly, either. Everything is pointing to the SSR not working.

But as @Phaedrux notes, you can attach something else to either side of the relay to test function, so long as that something else is rated for the appropriate voltage and won't pull too much current.

@Electriceye If you are confident in doing so, I would bypass/short your SSR and see if the bed heats. BE READY to kill it. Current is clearly not flowing or there would be a kilowatt of energy showing up somewhere. Obviously this is not a solution, but could be diagnostic.

@Electriceye

Since you have done the basic electrical checks, I expect that your M307 parameters are far enough away from your bed's actual response that the Duet is assuming a fault. Immediately jumping out to me is a dead-time ("D" parameter) of only 1.53 seconds. Based on your 1000W heater, I assume you have a big, heavy aluminum plate for your bed. If your thermistor is in the bed and not just sitting on the heater itself, dead time could be much more than this. I have an aluminum bed with a 1200W heater, thermistor placed into a drilled hole on the perimeter, and my PID-tuning calibrated dead-time is just over 30-seconds. Don't remember what the "C" parameter is exactly, but mine is over 3000. (Note; C parameter is deprecated--though still read--in latest firmware)

If you're confident you're not burning anything down, you can also play with increasing M570 parameters to delay error-triggering and see what happens. Once calibrated, remove or re-reasonable-ize M570 for safety.

@Dizzwold
@Phaedrux said in Dead Heater?:

There's no way to easily remove a bonded heater short of cutting. I've always considered it a one way operation.

I second that; no easy way. Flat-razor scraper is probably your best bet, but it will be a pain.

I would not feel comfortable suggesting an RCD type to you. Central RCDs generally trip below 30 milliamps. At-wall GFCIs are about a fifth of that. So, should be way less ground current than would be necessary to create your smoking-bed scenario. The only thing I feel comfortable suggesting is having an electrician investigate whether your protection is working.

@Dizzwold

Glad this did not end worse.

It does leave a question though. A thermal fuse shouldn't smoke when it trips. So, where did the smoke come from? Seems your theory is the damaged insulation let the fuse short to the adjacent foil. But I recall from your re-wiring thread that you've set up a grounded chassis with ground-fault-interruption protection; so, why did your GFI protection not trip? Worth looking into.

Seems like replacement of the thermal fuse will restore functionality. I would not rely on typical heat shrink to wrap those connections; cheap heat shrink is polyolefin and does not have a service temperature high enough for reliable use on a bed heater, in my opinion. You can get high-temperature heat shrink with service to 250C or so. But in any case, I would probably use a fiberglass sleeve as the final layer over these connections.

As a final note, I prefer both my thermocouples/thermistors and bed thermal fuses to be touching the aluminum bed itself, not the heater. If the bed is being heated to 130C, the heater itself should get significantly hotter than this during heating phases, and that's fine. It's possible--repeat, possible, not guaranteed--that there was never any real control fault here, but that the fuse tripped when the heater got it too hot, even while the bed was in expected range.

@T3P3Tony I would suggest either removing or updating the 130mm dimension up the left side as well, as this dimension is in fact 129.5.

Splitting the proverbial hair, I know!

@marco_meier Those drives can use a 24v signal, however they will also accept 5v Step/Dir signals on the proper pins, so you can simply run 6XD 5v to CN1 pins 3 and 5, and Step- and Dir- to CN1 pins 4 and 6, respectively.

All you really need to concern yourself with voltage-wise is the 24v enable signal, which the Duet team has confirmed the 6XD CANNOT sink on the EN- pin. The enable circuit on my breakout board would do what you need, and you could easily coble something like that together without making the whole board. Really, any simple relay/electronic-switch style setup that allows you to trigger a 24v connection with the 5v-tolerant EN- pin from the 6XD will accomplish what you need.

Bit of a weird one here...

I printed a mounting plate for the 6XD a while back and found that it didn't quite fit right. Recently I made a new one, built in assembly from the STEP on GitHub, and it fit perfectly. That got me curious.

My original part was made from the physical dimension drawing on the 6XD Hardware Overview page. This drawing indicates a 5mm offset from the board edges for the mounting holes; hole coordinates (5,5), (5,135), (110,5), (110,135).

However, based on the CAD files on Github, (both v1.0 and v1.01) and my real-life genuine Duet 6XD v1.01, the mounting hole near the VLC fuses actually has a 5.5mm offset from the edges; it does not match the other holes, and the holes do not form a rectangle, as this hole is at (5.5,134.5).

The dimension drawing on the overview page is for v0.1 and v1.0, and I have v1.01. However, the STEP for v1.0 also has this asymmetric hole, so it does not seem to be the case that this is just an oddity in v1.01 (the only board I can physically measure), and the dimensional drawings are accurate for v0.1 and 1.0; at the very least, the drawing and the CAD for v1.0 are discrepant.

So, it seems likely that the original error here is in the CAD, as there doesn't seem to be any reason to asymmetrically crowd a perimeter hole into component space. However, the physical boards exist, so at this point the relevant error is in the dimensional documentation, which does not indicate real-world hole-spacing.

@Amagatth Ok, good luck. Sorry I interfered.

@Amagatth Yes, you can certainly move to a 12/24volt with a relay/mosfet. I would suggest making things very simple for yourself first and just jumping the drive's enable pins from 12/24v to ground (NOT going through the enable pin on 1XD) and see if that activates the drive as expected. If it does, then you know what you have to do to finish the setup. If it doesn't, you have a bigger problem.

Generally, if you already know from the jump that you are going against a manual specification, that would be a good place to start a troubleshooting request! Rather than someone else considering possibilities, going and looking it up, and then you basically saying "Yeah, I know".