(follow up) Would you be willing to cut into your CPU socket?

[bigben2k]

Originally posted by crane
An easier way than cutting your socket is to use enameled wire for your thermistor leads. I used windings from a small motor.... the enamel won't cause shorting ,yet the wire is tiny enough to go between the socket and the CPU without interference.

Just make sure you clean the enamel off before soldering.

That's not a bad idea! What do you think, pHaestus? Since87?

[jaydee]

Originally posted by bigben2k
That's not a bad idea! What do you think, pHaestus? Since87?

As long as your still able to mount the block level with the core it might work. How small is small. Will not take much to make the CPU offset enough to cause a bad mounting. Of course if it is that small you could take a small file and make a little passageway for the wire.

I guess the question is is the wire sheilded enough not to effect the readings.

[crane]

There isn't any "might work" about it. The wire is a single conductor,small as a human hair and the enamel coating is tough enough to prevent a motor from shorting out.

I've been using a setup like this 8 months or more and hever had a problem.

[jaydee]

Originally posted by crane
There isn't any "might work" about it. The wire is a single conductor,small as a human hair and the enamel coating is tough enough to prevent a motor from shorting out.

I've been using a setup like this 8 months or more and hever had a problem.

Well more info is needed. I will leave it at that.

[Joe]

Originally posted by crane
There isn't any "might work" about it. The wire is a single conductor,small as a human hair and the enamel coating is tough enough to prevent a motor from shorting out.

I've been using a setup like this 8 months or more and hever had a problem.

Well Thermistor's mostly are affected by cutting and re soldering the wire, different gauge of wire, and different distances run. I have NEVER seen a thermistor read the same it did at one point again after cutting and resoldering after changing lengths. Now then again if you are using a Thermistor you aren’t concerned about 4 - 5Deg offsets since that’s about as accurate as any Thermistor reader that most people would use for this. Also if you believe enamel is enough to stop a short after a few inserts and removals of CPU's.. you got more balls than me.

Also the gauge you are talking about is so hair thin that I am guessing its increasing the resistance in the line enough to cause your temps to be a good bit off from what they would be with a clean connection.

So I would go along with it "Might work" depending how accurate you want. if you want to go cheap, simple, and "relative" temps, then this would work fine. If you were in it to know actual temps, then I would just go with a good quality Thermocouple thermometer.

[Zymrgy]

I just want to get a full understanding of what we are trying to accomplish before I cut into the socket. (the prying the thing off looks like a winnar to me, BTW) Is the plan to have everyone buy a particular brand of thermistor & call that the "standard"? Then I assume that the thermistor would be attached to the underside of the cpu & have the leads routed out of the relief.

If we are truely after standardized testing, then the enamel wire is not an option for the reasons that Joe stated. To insure consisintancy, the leads would have to be "factory", no modding allowed.

Which brings us to my concern about the whole deal, & that is attaching the probe to the underside of the CPU. Now, call me picky, but I really don't want to be using epoxy to attach things....I would prefer something less perminate. I dunno....maybe we could work out some sort of spring setup. Epoxy also presents the problem of consistancy....you could have the probe contacting not the base but a glob of epoxy instead.

Also, do all mobo's have the same dimensions under the socket? I know mine...a abit KT7A ( yea I know ) has a thermistor already in the socket. my thinking is the more cluttered up the socket, the more chance of the thing not being repeatable.

Just some thoughts...........

[jaydee]

Originally posted by Zymrgy


I know mine...a abit KT7A ( yea I know )

I got a half dead one of those. Droped something on the south bridge and now the hard drive controller doesn't work. That was a good board to. Loved to OC Durons on it. Aww the good ol days.....

No I don't think we are after a standardized testing equipment (although that would be nice) but I think if your going to be taken seriously it at least needs to be a thermocoupler factory built or at very least properly calibrated if self built. And then use a standardized testing method which I assume someone is working on?

Use Artic Silver thermal epoxy if your concerned about the thermal transfer. Should be able to remove the epoxy if needed although you may need a new thermocoupler. You might even get away with silicone sealant. I will have to experiment with it. Should have my new thermocoupler/meter Monday.

[crane]

Was just a suggestion, the enameled wire. Works just fine for me and seems as accurate as it was before I altered the leads. Take note... I made sure both thermistor leads were identical in length and used a heatsink on the thermistor when soldering to the leads. I still have the original wire and it doesn't seem to be special... just insulated braided copper of some small qauge.

I use it to monitor temps and also compare different homemade blocks I make.
Sure ain't gonna mention temps though..... I might not post alot,but I read plenty.

[Althornin]

why not just make a thermal diode reader?

[Since87]

I'm with crane.

Find some fine magnet wire (enameled wire used to wind transformers and motors.) and avoid the need to hack the socket all together.

[jaydee]

Originally posted by Since87
I'm with crane.

Find some fine magnet wire (enameled wire used to wind transformers and motors.) and avoid the need to hack the socket all together.

Ok so now we go from an easy mod to the socket to having to solder tiny wires and do it to were it doesn't mess up the accuracy of the probe? Remember most people can bearly solder at all let alone to perfection on tiny parts. And now we have to add a light dudy solder gun and solder to the parts list if they don't already have one (witch most people probably do not).

I would stick with the simple socket mod with factory made thermocoupler and not worry if I messed something up. I thought we were going after simplicity aswell? Are we going to have to hunt down all this stuff? Not all people are electronic wizards.

[pHaestus]

thermocouples are ok if type T but the K series are not so great. Take a look at that article I posted in my worklog a week or so ago for guidance on choosing a temperature probe.

Ben wasn't willing to even spend $30 on a Digidoc; doubt he will pay for the enameled wires much less a decent TC reader.

[bigben2k]

One does not exclude the other.

The poll clearly shows that 1 in 3 are not willing to chop up their socket, so if we can present an alternative, then why not?

If they don't have the skills or tools, and can't find anyone else to do it for them, then they don't have any options, period. We can only do so much, and Crane's suggestion addresses the needs of the other 1/3 of DIY'ers, which will cover most people.

Crane, you have PM.

[Since87]

Originally posted by jaydee116
Remember most people can bearly solder at all let alone to perfection on tiny parts.

What? You mean not everyone has been doing electronics soldering regularly for the past 27 years?

Ok, maybe not the best solution for everyone.

pHaestus,

The ONLY way to eliminate error with thermistors resulting from lead length, is to use a four wire (Kelvin) connection, where one pair of wires is carrying the current for the thermistor and another pair of wires brings out the voltage across the thermistor.

A four wire connection using 40 gauge wire, is likely to be better than a two wire connection using 24 gauge wire. Of course this doesn't do much for you if you are hooking up to something with no provisions for four wire readings...

[jaydee]

Originally posted by bigben2k
One does not exclude the other.

The poll clearly shows that 1 in 3 are not willing to chop up their socket, so if we can present an alternative, then why not?

So your saying the one's not willing to make a simple socket mod are going to be more inclined to put together this probe? And by some miracle we are going to automatically accept that they did it right?

I think they would be better off just using the on board probe and soldering the wires from the reader to the output of it on the back side of the socket and push the probe to where it touched the bottom of the CPU. Probably be just as accurate and easier to do.

Or if people are that good at soldering they can remove the current onboard probe and use the one they would be using with the thin wires and solder it in place of the old and then use the holes the original left to run the wires through. But that would be considered moding the board which some seem not to want to do for whatever reason, in which case they need not be bothered to be worried about temps anyway because their running a stock AMD air cooler at default mobo settings.

[MMZ_TimeLord]

pHaestus,

I was looking to purchase a set of thermocouples to do my measurements... can you point me to that article... I can't find your worklog. :shrug:

[Joe]

Originally posted by pHaestus
thermocouples are ok if type T but the K series are not so great. Take a look at that article I posted in my worklog a week or so ago for guidance on choosing a temperature probe.

Ben wasn't willing to even spend $30 on a Digidoc; doubt he will pay for the enameled wires much less a decent TC reader.

Well... If people are goign to test with something thats many degrees off in some direction and not at all calibrated ( like all this thermistor over motor wire). a Cheap K type T/C would still be better than any Digidoc or Compunurse.

K Types are easy to locate, T types seem to be a bit more expensive depending where you look.

Also depends what your thermometer will read. Most are only J/K type for cheap ones. I know mine will support like 8 - 10 types of T/C's including T Type.

[jaydee]

Originally posted by MMZ_TimeLord
pHaestus,

I was looking to purchase a set of thermocouples to do my measurements... can you point me to that article... I can't find your worklog. :shrug:

Staff worklogs: http://forums.procooling.com/vbb/for...?s=&forumid=33

[MMZ_TimeLord]

Thanks Jaydee... I don't care what enyone else says... you're okay.

Excellent Document on Temperature Sensing Technologies!!!

Saved that one for my library!

[#Rotor]

just so everybody know, I voted NO... I would not cut into a socket! reason...... I don't have to.....

[jaydee]

Originally posted by #Rotor
just so everybody know, I voted NO... I would not cut into a socket! reason...... I don't have to.....

That not bad, but I would rather the CPU set flat on the socket to maintain a repeatable mount. Not to mention if I used my Air cooler as a reference it would add to much height for the clip.

[Since87]

Originally posted by jaydee116
That not bad, but I would rather the CPU set flat on the socket to maintain a repeatable mount. Not to mention if I used my Air cooler as a reference it would add to much height for the clip.

I think that's the point of the blue wire shown...

To provide equal spacing around the CPU.

However, even though you can make electrical connection that way, you are moving the CPU farther away from decoupling caps on the motherboard. At gigahertz frequencies, every bit of length there matters. Doing that probably drops max overclock a hair.

[jaydee]

Originally posted by Since87
I think that's the point of the blue wire shown...

To provide equal spacing around the CPU.

However, even though you can make electrical connection that way, you are moving the CPU farther away from decoupling caps on the motherboard. At gigahertz frequencies, every bit of length there matters. Doing that probably drops max overclock a hair.

Yeah I understand how it is setup, but like I said I want repeatablility and consistancy. Does the plastic coating on the wires not compress after a while? What if it compresses while on the test bed? And I still couldn't use a Air cooler as the CPU would be to high above the socket lugs. Unless the socket lugs magically raise the same thickness of the wire? I have a Volcano 6Cu+ and it is hard enough getting it on with the CPU at stock height. And what if the blue wire compresses differently than the other wires? Lop sided mount anyone? Could all be avoided by the simple socket mod. Maybe nit picking but why worry about it and not just do it right.

[#Rotor]

you do have a good point there, yes, the wire does compress a tad bit over time ( 3 months or so). The compression is finite, though.

I was however able to fit a STD AMD HSF to said CPU with no more or less fuss than usual, Sorry I don't have any experience with other after-market heatsinks, I'm just not the HSF kinda guy. I skipped that era completely nun the less, I think the repeatibility of this method is just as good as say with a cut-socket method. Definitavely unrivaled in it's reliability as opposed to having the probe on top of the CPU, next to the core.



as for the OCability.... hmmmmm.... that might be, I am however not able to sense it, and as with all good things in life , there are always taxes to be paid........

[Zymrgy]

here is a thought...Honestly I do like the idea of not having to modify the socket. just plain less work for everyone. now...how bout if someone takes some measurements (several anyways) with some of the "stock" leads on the thermocouple & then try several with the thin enamaled wire. If the readings are within an acceptable margin...then no more arguement. Just set a standard that if you do do the enamled wire route then they should be cut a "X" length, using "Y" size of wire.