Molex connector issues

kazolar

Over the course of calibrating my machine, I've had molex connectors on my z axis (4 independent lead screws) connected to duex5 melt on me multiple times, it basically singes one pin, I've now learned to recognize it, the z axis starts struggling, as soon as it does I check each of the bed steppers and replace the singed connector, re-crimp a new pin. I'm running 4 nema 23s rated at 2.8a, current is set to 2.4a. which is within the driver capability. I've never had the same issue with the lower current nema17s, of which I have all of duet2 populated, so I'm wondering if it has to do with duex5, the connector?
I'm I have the boards mounted where the duex5 is on the bottom, but there is enough strain relief on the cables, so pins are making proper contact. I'm really getting close to just desouldering the molex connectors and replacing them withe screw terminals.

A Former User

The connectors are rated for 4A, so if they're melting there must either be excessive current or a bad connection/solder joint somewhere. Internmittent problems could suggest something that moves or expands with heat shorting out.

nophead

Yes I have always replaced Molex connectors with screw terminals on my previous boards because vibration and thermal cycling causes fretting corrosion. I was going to ask if it was a problem on Duets before replacing them on mine but I see it is, so I will.

All the machines I have built without screw terminals have failed like this after a few years.

kazolar

@bearer I have 2 50mm fans blowing across the duet+duex5 combo and duet doesn't go over 30 c, and I checked connector temperature and it's barely above ambient. Of all the wire runs on this machine, and there are a lot with 16 steppers, 4 extruders, etc, the z steppers have the easiest wire runs since the steppers are on the bottom of the machine, and so is the board. These are the only connectors I've had issues with - fairly certain it's because they're the larger nema23s. I guess if it happens again I'll bite the bullet, pull the duex5 out and replace the molex connectors with screw terminals. Curious that it started happening after I switched to 4 lead screws and single start lead screws. It is basically like clockwork, after about a couple of weeks a connector which was perfect before is now fritzing out and I check and see it has a brown mark on the plastic. Have gone through 4 or 5 molex connectors already, since I don't reuse them after they've had an issue, initially I thought it was just a random glitch, because a reset cleared the issue, but then it would stop moving one corner or start shaking, and by then the connector was really melted, and really hard to remove. So now as soon as I see a glitch, I shut it down and check all z connectors. Like I said have happens at least one time on each corner, and I had confirmed the connectors were perfect after the first one failed.

nophead

@kazolar Are they subject to vibration? If the wires move in the slightest the connector will eventually burn. Clamping the wires to something solid can help but I eventually decided to switch to screw connectors.

kazolar

@nophead no, since this is z, the steppers are stationary the duet is stationary. Nothing moves there, all my x steppers move (4 x axis), so do the extruders, but y (and u) and z steppers are stationary. This is a quad idex config. Hence 16 steppers. So I'm maxing out the duet onboard, plus I have external drivers. I'm considering maybe switching to Dupont style connectors. Won't be locking, but might be tighter.

Phaedrux

@kazolar said in Molex connector issues:

single start lead screws

Curious if you adjusted your Z axis acceleration when you switched to the single starts.

kazolar

@phaedrux i did. Everything was dropped significantly to account for 1600 steps/mm

dc42

The common reasons for Molex connectors failing like that are:

Bad crimp connections. You are running the connectors fairly close to their rated current of 3A (for the Wurth connectors that we use on the Duet), so a good crimp is essential.
Crimp pins not pushed fully home into the shell. Look in the slot of in the housing to check that pin is fully home and the barb engaged. Do not use the crimp pins that we supply with the Duet (which are made by Wurth Electronik) with shells made by Molex, because the pins won't lock in place.

nophead

@kazolar Is there significant vibration or wide temperature swings? The wires don't have to be on moving parts, if the machine vibrates they can vibrate relative to the connectors unless they are tied down close to it.

It is a recognised problem in the connector industry that these types of connectors will fail in machines and I have experienced it first hand on my first three 3D printers until I got wise to it and switched to screw terminals and ferrules. The current doesn't need to be high, even 1 Amp will do it and I have had signal connectors with virtually no current lose contact until re-seated. Most notably a thermocouple amplifier output and power connector gives me grief every now and then.

I lost faith in my crimping so I soldered them but they still failed. It can take years of continuous printing before they fail though and my early machines do vibrate quite a lot with x8 and x10 microstepping, heavy axes and high acceleration.

kazolar

@dc42 I have really clean crimps. Like I said. After the first one melted, I triple checked all others, now after the 5th, I'm pretty sure it's an issue with the connector. The 6th, I'll bite the bullet and pull the board and swap for screw terminals. I'm using the connectors and pins provided with the duet.
There are no vibration or movement at the controller. The machine weighs 150lbs, top of it may vibrate as the gantries are large and doing small movements does shake things, but the base of it is solid. I initially had the current set to 2.6, to maximize the stepper torque, but lowering to 2.4 hasn't lengthened connector life at all.

dc42

@kazolar, are you using the Wurth shells and crimps that we supply, or genuine Molex ones, or some unknown Chinese clones?

Is it just the shells that are melting, or the plastic headers on the Duet too?

How thick are the wires going into the crimps?

btw on Duet 3 we are changing to connectors rated at 8A - 10A, because of the higher stepper motor currents supported.

kazolar

@dc42 I am using everything supplied with the duet. The connector is getting singed, not really melting, since I don't let it get that far anymore, and troubleshoot at first sign of trouble. I'm using 24 gauge wire, I've had it happen with the stock wire which came with the nema23s, and also had it happen with the 24gauge silicone insulated extension wire. Once I remove the connector, there is just a little discoloration on the duex5 housing, no real damage there, but connector is not fusing with the housing, typically one pin ends up charring and the z starts acting funny since it's 1 of 4 lead screws it's pretty obvious when it happens and the machine tries to twist the bed. When I removed it last time it happened, I could have probably reused the connector since it really wasn't charred much, but I didn't, I still have spares from my duet/duex5 packages. All I had to do was remove the charred crimped pin (female side) and crimp a new one and fit it into a new connector making sure everything locks in properly. I got my duex5 2 years ago, have the design been updated? My duet2 is newer has the blade fuses. Curiously enough no more i2c issues.

dc42

You could try using genuine Molex shells and crimp pins, because they are rated at 4A instead of 3A. I don't know what the difference is, but the dimensions are obviously slightly different because Wurth crimp pins won't lock into Molex housings. I've just measured the width and depth of the male connector pins on a Duet, and they are almost exactly the same as on a genuine Molex KK header.

kazolar

@dc42 I have used a number of genuine molex connectors and pins on other steppers, but it's a bit of a tighter fit, they lock in differently. I can try it on the z, I have a bunch of molex connectors. It's either I go all in and remove the molex headers off the board for screw type connectors which would be non-trivial surgery, run it until it has a problem, and swap for genuine molex at that point, btw duet supplied pins fit perfectly fine in the genuine molex connectors, I have that combination in a few fan headers. Or proactively open it up and swap all the beige duet provided connectors for molex with molex pins?

kazolar

@dc42 ps - I don't use any chinese clones when it comes to electrical components, and really nothing else in this machine, learned the hard way when I tried Chinese linear rails, now all hiwins. All the silicone wires are imported from Hong Kong. Last thing I need is something to go poof because I cheaped out on a connector or a wire. Kinematics is hard enough to get right on a quad idex, then to complicate things with connector problems.

A Former User

Regular 24 gauge is only rated for 2A at room temperature, the silicone stuff is rated higher, but that is just because the insulation has a higher melting point not because the same amount of copper can carry more current; the connectors might need to be de-rated if running at elevated temperature?

Still weird its just one pin unless I read it wrong? Might be worth while re-soldering that one pin, wick up old tin, get some good flux on there, solder it and clean up.

kazolar

@bearer The steppers are rated at 2.8a.
And maybe I wasn't clear, but STOCK wiring that came with the stepper had this happen. Yes one pin get singed, on the stepper female connector, not the board. The board mating connection housing only has some mild discoloration. The rating of the wire is interesting, and would make sense if only the steppers which had extensions on them had a problem, but that's not the case. I may add a current meter in line with one of the steppers to check what the actual current draw per phase is, I took multiple temp readings with an ir thermometer, and the wire is not even warmer than ambient (did this while the bed was doing a long move), so it's not an issue of current exceeding the wire capacity, it's an issue with the connector. I know this because when I took the last one that failed apart I pried the singed pin open and the wire was perfectly fine inside. Such that I was able to crimp a replacement pin. I've had this happen just one time on another machine which used Dupont connectors and it happened after over 2 years of operation. In my experience this usually only happens when the connector is not making a clean connection with the pin, and my concern is that these molex connectors - by design, can't, the spring action required to keep it secured is not enough for it not to arc at some point.

kazolar

@dc42 genuine Molex shells and pins have worked and I have not had a single blown connector since switching to them, and have done a lot of calibration routines and started doing test prints again. I also have switched a couple of other steppers to use molex shells -- next time I have to service the controller I'll switch them all. Shame that the stock ones have this issue.

dc42

@kazolar said in Molex connector issues:

@dc42 genuine Molex shells and pins have worked and I have not had a single blown connector since switching to them, and have done a lot of calibration routines and started doing test prints again. I also have switched a couple of other steppers to use molex shells -- next time I have to service the controller I'll switch them all. Shame that the stock ones have this issue.

Thanks for the feedback. Can you confirm that you used the same crimping tool and technique when you crimped the Molex pins as you did when you crimped the Wurth pins that we supply? How the pins are crimped makes a lot of difference.