P870 melting power plugs and sockets

I am exceedingly displeased with my Clevo (OriginPC) P870MK1. While playing a game today in my hotel room the power plug fell out of the back of the computer and when I picked it up I received a very nasty burn from the shell of the plug end. The plug would not reseat in the back of the machine so I took a look inside of the connector.

http://www.picpaste.com/Clevo_P870_melted_power_plug.jpg

http://www.picpaste.com/Clevo_P870_melted_power_socket.jpg

Clearly the power plug/socket combination is inadequate for the power demands of the P870 running under near normal conditions (GPU overclocked only). One of the power pins actually melted the surrounding plastic and pushed back in to the shell of the connector in the laptop.

This STUPID AC-200 adapter has such a short cord to the laptop that you have to be very specific in how it seats in the back of the machine and even under normal conditions the slightest bump will make it fall out.

Right now I am running on a single 330 watt brick instead of the dual 330's that are recommended for the
NOW I get to send the laptop back to the integrator for them to sit on it for a month before fixing the problem.

As I had ordered a Eurocomm 750 watt brick with a single plug that does not rely upon the STUPID AC-200 I will limp along and try to pull the power pin forward in the connector shell and see if the 750 watt adapter with its longer cord is friendlier. (the 750 watt supply arrives this week).

BUT, I get to send my laptop back to OriginPC for warranty repair and let them sit on it a month before I get it back. Right now this machine is rating on a scale of #2 (out of 10) for reliability.

 

When it happened were you using both 330s? That's an unusual problem as far as I know.

 

probably knocked the ac200 enough to make 1 of the pins disconnect / not properly fit

 

Yes, I had both 330's plugged in but as you know those connections are not too great either. I have had one power supply cord fall out of the AC-200 and the first notification I would get would be a few minutes later when the other power supply would trip out on overcurrent. Then the P870 reverts to battery and falls out of overclocking. While in the middle of the game what I would observe would be a performance hit and then look over to see that one of the feeding power supplies in to the AC-200 had disconnected and the other had tripped out on overcurrent.

The dual 330 supplies/ AC-200 is such a bad kludge that premier sellers of the P870 should make it an option to not even buy them but to use the Eurocomm 750 watt supply.

Additionally, from an electrical engineering standpoint those twinned up electrical connections and the small pin sizes is a marginal design. Any sort of imperfect connection will cause significant overheating and melt the connector shell in the laptop.

 

Hasnt happened to me yet, or to quite a few people using the same combination here.

But so far you are the second person who has faced this issue. The first one i think was a Sager and using P870DM-G.

 

It is possible to melt almost every connector. My friend managed to melt psu ac inlet connector. We just have to make sure all cables are seated properly and that there is no way it could move around and make sparks. Stay safe people!

 

With such a high throughput it is very important to keep the connector fully inserted, especially under heavier loads.

 

I haven't really seen a lot of this with the P870s or the dual plugs, I'm inclined to think it was related to a loose plug from the pics. That said I'm completely with you on the cord being so short, it should be a lot longer, the current form factor is a bit ridiculous.

 

We've gotten this a few times on the 330 watt adapters, also, and I'm inclined to agree. Although "brick" is an umbrella term for AC adapters, this thing resembles an actual brick more than most, and the weight tugging on the cord is no bueno.

 

I haven't been able to find anything in my admittedly short searching, but I wonder if someone makes an extender for that kind of cord/plug.

 

Problem is that it's difficult to even find these plugs on eBay. Otherwise some of us would just solder up our own extensions. Heck I would even go as far ahead so as to make an extension board with multiple male to male ports and have the PSU'S housed into it / feed in. Mainly because I had situations where I was running running both my DM1 and DM3 at the same time.

 

If you happen to find them, make a post, I'm certain we're not the only ones would would love to get their hands on a few.

 

With multiple power supplies feeding a common power bus it gets tricky to balance out the voltages from the bricks. They are not each exactly the same on output voltages and even a tenth of a volt difference will leave one supply handing the majority of the load. It will continue to provide most of the power until the voltage drop (due to I squared R losses in the conductor) bring it down to where the supply with the slightly lower output voltage can begin to contribute.

I have not opened one up yet but I suspect that the PS-200 just passes the two supplies separately on different pins in to the laptop, where the power is there combined. That puts the electrical stress/heating on the laptop plug. If the combining/balancing was done inside of the PS-200 (even using .1 ohm resistors in series and high power diodes to prevent backfeeding) then the power to the laptop could be doubled up on the four pins an no single pin would see excessive current (the two positive pins would sum up to the current of the two negative pins with no single positive or negative pin seeing excessive draw).

If it did not matter to me and if the laptop was out of warranty I would just install an "Anderson Power Pole" connector on the back of the laptop with 50 amp contacts to handle the power. That is an industry standard DC connector that even has sizes all the way up to things that would run a battery operated forklift.

 

Manufacturer is Singatron, part nrs. are '2MJ-3432-*' and '3MP-3432' for the jack and plug, respectively. Won't help much in finding a supplier for them, though. Much easier is simply to cut the adapter -> laptop cable and solder two KPJX-CM-4S female sockets to each end and make the extender cable with two KPPX-4P male plugs. Could make two of those male-male cables, depending on the distance required; short version for stationary and a 1m+ for the road.

Normally I just solder in an extension between the socket -> adapter section (preferable due to the high voltage + lower current), but yes; the extremely short length of the 330Ws adapter -> jack cable does warrant a mod. Not sure why you'd want to make it detachable, though? Just splice in 1m of cable (using solder and heat shrinks) and it will be much more manageable. Also no jack required and it won't have two extra joints where the connection has a possibility of being severed.

How would this work to balance out the draw over the two adapters?

 

I think what @bloodhawk and I were referring to is a simple extender so you can put the brick on the ground since most users aren't going to be comfortable cutting up a short but otherwise working adapter cord to splice something in.

 

This. And of course a homebrew (and improved) version of the AC-200. That would offer the flexibility of a longer cable.

 

Ah ... yes, that .

Ok, then there is no alternative; you need a compatible female plug and only the Singatron part will work, but it might very well be made from unobtainium.

Spoiler: Well ...

... you could swap the jack to a more common model. But guess that if soldering cables is out of the question then soldering a working motherboard will be an even harder sell .

 

I might actually do it on my DM1, since its pretty much out of warranty now. IF i dont sell it first But instead of splicing the cables, i would rather replace the female end on the laptop. Soldering has never been a problem, acquiring the parts is more or less the issue.

 

The circuitry inside the original box is fairly simple:



Slight differences in the brick are regulated by the LT chip.

 

Oh! So there are actual components in the package, that is good to know! Those two white components labeled as R002 are 2 milliohm resistors to create a slight voltage drop. There will be sense circuitry (U1) across those two resistors to measure the voltage drop and to infer the current value from each supply. (U1) will then control the two banks of FET's for voltage regulation from each supply (PQ1 to PQ5 and PQ5 to PQ10).

 

Did you see the contact rating on the Kycon connectors? They are rated for a maximum current of 7.5 amps each.

Even doubling up the pins (you bet they do) you can get up to 15 amps (if every pin connection is perfect, every time). If you do end up hitting somewhere upwards of 500 watts power draw that is 25 amps of current. Way over the rated specs, those connections will get hot and if you have one that is even less than perfect things will get melty-hot.

We have all had notebooks that have been through the ringer after a few years of portable use. In almost every machine the power connector is beat to H#ll and sloppy like... well, I am not going to say in polite company (I use that term sparingly around guys) <j/k>

That is bad design, that is a manufacturing defect due to incorrect component selection by the design engineer. Likely the engineer was given a certain specification for what current rating to expect and designed to that. There was not an overall design review committee who looked at system-wide implications.

Those connectors would be OK for a single 330 watt supply like a "normal" notebook but not for any machine running dual GPU's. When they went with high end GPU's the connector should of been replaced for something with a higher current rating (upwards of 30-50 amps).

I doubt anyone looked up the specs on the connectors after things like SLI NVIDIA 1080's were considered. That happens much more often than you can imagine, the weakest link keeps moving around as a platform evolves.

Going with a Power Pole retrofit is looking better every minute.

 

So what is the main culprit that is causing this type of damaged on the unit. Is it the connector itself or the 2 power supply set up or an accidental situation like the connector is not inserted properly?
This 1080 sli laptop setup had been released for almost a year now and we just heard this type of fault on only 2 people. Maybe isolated case?

 

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

From my observations and research allow me to summarize;

The connector on the back of the P870 is not a very secure connection; If you have a machine that you realize that there is not a great deal of mating pressure or connector engagement between the laptop and the cord. It is very easy to get a partial connection and this is much worse when using the PS-200 adapter with the very short cable length between the adapter box and the computer.

The pins on the connector are rated at 7.5 amps each. With the doubled-up power connection the current rating of the four pin connector should be no greater than 15 amps with a lifetime limit on connections at 1000 insertions. Fifteen amps of current comes out to 300 watts of electrical demand (20 volts times 15 amps equals 300 watts).

The connector choice may have been adequate for a single-supplied notebook when the P870 first came out. Once the power requirements increased to the point where dual power supply bricks were needed that are capable of providing 660 watts of power. Then the connector is running well beyond specification and the pins heat up, the metal loses temper (becomes softer and distorts) and the plastic connector melts, misaligning the pins even further.

If you bump the connector and it becomes partially unseated during high power utilization (like when gaming with overlocked GPU's like SLI 1080's) then the connector pins will heat up very rapidly (in seconds).

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

The proper "fix" will be for Clevo to select a different connector that is capable of handling total currents approaching 30-40 amps; this likely will not happen on the P870-x platform. Connectors capable of 30 amps (600 watts) to 40 amps (800 watts) are substantial and I did not find a pre-existing connector package that would be readily applied. Clevo may need to work with a manufacturer to design a connector capable of handling such loads or to completely change the connector form-factor (to something like the 45 amp Anderson Power-Pole).

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

For existing owners I would suggest that you never allow the machine or the connector to move around while operating in a high performance (gaming) or overclocking mode where currents are the highest. Even the internal electrical arcing from make-break action of the connector will rapidly shorten the lifetime of the connector contacts.

 

Not so much inserted improperly but pressure on the power plug itself where it plugs into the laptop is my guess. It's not a common enough problem to really call a defect IMO, more like they didn't account for how the system might be used. The cord isn't long enough and doesn't plug in securely enough to be good for frequent plug/unplug, which the mfg likely didn't consider since it's a desktop replacement and those typically (many exceptions of course) don't get moved that much after setup. I think you'd have to force it to insert the connector improperly if it's possible at all because of how its shaped. Compare to a defect like the power pin issues in the ASUS G73 where the pins were made of an alloy that weakened under long duration of heavy current causing them to become brittle and snap off (ASUS stepped up and replaced those even out of warranty with a better parts), that was very widespread and affected I would guess hundreds of machines.

 

If the pin is rated to 7.5 amps and you are doubling-up pins (two positive, two negative) then the sum of the current that can be provided to the device is 15 amps. For 15 amps at 20 volts that would mean that the power supply should be limited to 300 watts.

We know that the dual 330 watt supplies is require for SLI GPU's because the machine draws much more than what a single 330 watt supply can provide. Some of the power consumption numbers put out there for the P870 are upwards of 500 watts.. that indicates the current is greater than 25 amps at the power connector.

If you are that far-off of what the connector is rated for then that is a design defect. Maybe not from the original design but as the power requirements changed the entire design was not reviewed for weak-points.

I was being charitable in saying that the design engineer probably chose the connector based upon the original estimation of power maxing out at 300 or so watts. In that case the connector would be adequate.

When the P870 was first designed nobody knew what dual 1080's were going to require, or a 7700K that can draw 91 watts just for the CPU. Now on a full build out with SLI 1080's, 7700K, a full load out of four drives and moderate overclocking on the GPU's/CPU you are well over the rated capacity for the plug.

I bet nobody ever went back and said "hey, I wonder what all of this extra current is going to do to the poor little power connector?"

Engineers designing PCB's or figuring out how to wedge SLI GPU's in to a machine tend to look down on the newbie engineer that was given the "easy" task of finding a power plug for the back of the computer.

At one time I was a newbie engineer (back in the 1980's) and was given the "easy" task of finding a power supply for a product we were making. It sounded easy until that product had to go for FCC Part 15 certification testing and it turned out that the power supply I had chosen generated a tremendous amount of electromagnetic interference. That was not listed as a requirement at the beginning but it turned in to a major hassle that delayed product launch for six weeks while I had to re-design the power system and then nursemaid the device through a retest at an FCC certified lab.

 

So for you the cause of the damaged is the rating of the connector. Now my question is why the issue only occurred in two instances including yours. If this is a defect, I think that this type of problem will be more common especially on those power users and we have a lot here in this forum.

By the way I'm not debating your findings and I hope if there are clevo representative here reading your post can get something from it to improve the overall design of this unit in the future.

So the best way to avoid this issue is to make sure the connector power is securely fitted and maybe put the power supplies in the table or at much as possible near the unit to to avoid movements on the connector if accidentally the unit is moved.

 

Unless we start seeing a lot of similar cases (like the ASUS example from my earlier post) this is my position. That or get the crazy high powered Eurocom adapter, which should also wok fine.

 

Have been using dual power bricks for over a year now, first with the DM1 (Using AC-100) and now with the DM3 (AC-200). Never had the connector come loose, i always make sure that all the connectors are inserted properly into the laptop firmly. Specially when and after i move the laptop around or when im in bed.
But i do feel that the adapters cable is a bit too short and probable cause in your scenario.

 

I understand,

That is what I am looking for as well, so power users who push the limits are aware of just how fragile that connector is if it gets disconnected under load and arcs or is not fully seated and generates a great deal of heat.

On other connector types you see manufacturers doing a few things differently if they have to stay with the same form-factor;

Using resins instead of plastics for the connector body that do not soften and distort under heat
Using more durable base metals with silver plated contacts
Increasing finger tension at the contacts for a more secure connection

 

This is a big one. Not a full locking plug, but something that required a bit of a squeeze or something more than just a pull to disconnect would help immensely.

 

I think the newer ASUS laptops have that ginormous plug. And its truly painful IMO. But i have only tried it 3-4 times.

 

I sacrificed a wimpy Clevo 240W AC adapter to steal the cord to mod my AC-100/200 convertor. The mod works flawlessly and makes the use of dual 330W AC adapters far more pleasant than the stubby little joke of a cord between the convertor box and laptop. I ordered the KPJX and KPPX adapters @t456 mentioned earlier in the thread a very long time ago and they do not fit. Brother @j95 and I were searching high and low for these jacks and hoped they would work well, but they did not. Maybe soldering a new jack/port to the motherboard to match them would have worked, but that was too much of a hassle and would have made the finished product a mod too proprietary to use on other systems with a standard Clevo or MSI charging port. The male/female portion of these fittings are the exact opposite of what they need to be to fit the existing jacks, and modding the mod to cobble things together was not something I wanted to deal with.

The cord is thinner, but never gets hot and has no issue handling all of the dual 330W output during extended extreme overclocking sessions. It was totally worth it because it made a super clean factory-looking mod with the cord soldered to the connection on the PCB inside of the AC100/200 converter. The 240W AC adapter was totally worthless to me otherwise because it was not able to keep up with demand for the level of overclocking I was doing with the 4790K and hard-modded 980M that I had in El Cazador.

 

They are not Kycon parts, unfortunately. The manufacturer is Singatron, which is why people have such a hard time finding replacements. We'd need a customer-side supplier that carries their stuff.

The rating for either model is specified at 48V, though. So we can safely assume ~700W will do fine (that value also matches the 500VAC rating), provided all four pins make contact, that is. This explains why this is such a rare issue; most users won't draw far north of 360W, so half-a-connection wouldn't be a problem.

Well ... it is the Singatrons that are obscure, the Kycons are pretty much run of the mill.

Could keep the female Singatron plug after cutting it off and replacing it with the male KPPX-4P plug. Then use the Singatron plug plus a KPJX-CM-4S to make a short adapter cable. That way the modded adapter can power both modified and stock systems.

 

I can't think of a supplier in the states at least that carries connectors only, sadly.

 

Mouser and Digi-Key are two and there are others. It's really the proprietary Clevo/MSI 4-pin DIN type connectors that are problematic to source. All of the other brands with round barrel connections are easy to come by.

Exactly. You know when something is not even available on web sites that sell everything imaginable, like Alibaba, something is way wrong, LOL.

 

The same will it be with MSI. The bigger Acerboook Predator 21x use two connections for psu on MB. A rather crippled solution. And not wanted.

 

At least it has barrel connectors

 

Yeah, would be nice with two of those on the MB. And you must buy a new designed dual power convertor. Only with opposite design for use with the new 780w psu Of course with the middle pin in the barrel like Dellienware's Many ways to Rom.

 

There are as many issues with barrel connection as there are with 4-pin DIN connection like Clevo and MSI are using. The barrel connectors also get sloppy loose and prone to easily fall out. I have had that issue more with barrel connections getting loose and coming unplugged than I have with the Clevo type to be honest. And, yeah... the signal ID wire on the Dell/Alienware, and some HP notebooks, is truly retarded. There's no excuse other than obsessive greed for implementing that kind of proprietary filth from any brand.

 

True, but with the barrel ones, you can get a full adapter that will work and not need to source a tip only.

 

Yes, that is definitely a plus. Other than ubiquitous availability of the barrel connectors in just about every size under the sun, there is nothing special or particularly redeeming about the design versus the 4-pin DIN connector.

 

Too bad surface magnetic connectors haven't been refined enough to make it to high power systems. It's my favorite feature on the Surface.

 

Interesting idea. I have no favorite anything with respect to the Surface.

Come to think of it, I would be hard pressed to identify a favorite anything about Micro$loth or any of their products at this point in time.

 

Forgot, all tablets are BGA so you might break out in hives or something.

I just feel so spoiled by the mag plugs that I want them on everything.

 

Nah, I have a tablet and a smartphone. I believe there are good uses for BGA... but none are expensive, and none are misrepresented as "high performance" products.

My contempt in the last post is entirely toward Micro$loth and the unbridled proliferation of control-freak filth that they represent as a company, not BGA this time.

The latest example of Micro$loth's digital crimes against humanity is captured here.

 

I received the Eurocomm power supply last Friday, I am very pleased with the supply and it is much easier to carry instead of the dual bricks, PS-200 and all of the cords that make up the dual supply setup. I cannot rate it highly enough.

I also found a supplier who sells the original connector that is on the laptop motherboard; Since I have a hot air rework station I will replace the connector once the parts arrive.

http://rjtech.com/shop/index.php?dispatch=products.view&product_id=30477

Looking more closely at the pin that failed in the connector it appears that the tin pin broke or melted off, leaving a more durable inner core that is undersided to mate with the power supply connector. I could try to find a gold sleeve similar to what is used in a wire-wrap socket or for precision RF connectors (like an SMA center pin) but the correct fix is to replace the entire connector.

 

Not a bad price for that connector even.

 

When I get back from a business trip next week I will take apart the machine to replace the connector on the motherboard. It is definitely going to need the hot air rework station to heat up all of the pins at the same time so the connector will drop off the board. I do not want to tear out a via or rip off a trace; the removal process will be the most critical stage.

Once I get that far I am still conflicted about using the factory-appropriate component or in upgrading the connection to something more robust.

I have some six pin XLR connectors (male sockets, female plugs) that are six pin with each contact rated at 15 amps (gold plated contacts). This would have the advantage of being a locking connector (depress a button on the plug to unlock). They have solder cups so I could add a lead to the circuit board and the socket can be mounted on a piece of metal and attached under the plastic trim on the back of the machine.

Doing a direct-replacement would keep it warranty-fresh, so that is an advantage and it would also be the easiest to do without any fabrication work. I will probably go that way and if it ever breaks again then I will go with the XLR connectors.

If Clevo does a future re-design they should consider something like a six pin XLR as you can get contacts with an appropriate current rating if some of them are doubled up. (double up the primary power pin and the ground pin). The way it is now it does not appear that one of the power pins does anything at all, if it did then either both pins would of melted off or the laptop would not have a total lack of charging when one pin opened up.

You can get XLR plugs in a variety of mating styles to keep people from accidentally plugging in a microphone (commercial mics use XLR connectors) or some other piece of sound-stage gear (light controller boxes use them too and they also supply power across the connectors). It would be industry standard, very robust, locking/latching, keyed, sourced from multiple vendors and have 3-4 times the current rating of the existing plug/socket set.

http://www.mouser.com/ds/2/393/switchcraft_sealed_xlr_npb_604-315827.pdf

 

I like that idea, more robust connectors from all manufacturers would be welcome.

 

That connector reminds me of some of the retro 40s/50s teardowns of lab equipment