Apogee from Swiftech...

[Angry_Steel]

Bottom line: No way will I stick a stock apogee in my loop. I've tested it, and it performs well, and has incredible flow... but without a re-machined top, it is useless.

I assume that I got one of the "first batch" blocks... that were rushed out to get distribution channels full. I just ordered another apogee to see if things are the same.

First of all I'm, just trying to understand why in the world you would order a second Apogee if you think so highly of the first?

Second, regarding your phone call with Swiftech. I dont see why you wouldnt send the other block back to let them take a look at it, and maybe see what happened? Maybe he did hit the tops with hammers and nothing broke. All the more reason to send it in and find out what happened.

Having you pay for shipping? 99 percent of everything I ever had to return, I had to at least pay for the shipping back to RMA. I may be wrong, but if it was found to be a manufacturing problem, some companies would credit you the shipping back. Also I think that even offering to take a look at it after you epoxied it yourself is at least generous.

This has nothing to do with my opinion of Swiftech one way or another, I just think you could have handled the whole thing differently.

[Orkan]

Got the other one today. Took it apart:

Only found 3 or 4 shavings this time, instead of a bucket full... but they were still there.

... and along with that... there were some bent pins! I couln't make this up if I tried. lol QC at swiftech is non-existant on this block. period. You can see by the scratches in the pic, that someone has been in there after some shavings though. At least they are trying. Looks like they nailed a few pins with the tool they were using to get shavings out.



::EDIT::

Angry Steel,

My call to swiftech started out with me wanting to help them out in whatever way I could. ... However, after speaking with gabe... I don't really care about them. Gabe had his mind made up that I definately screwed up his product before he even picked up the phone. How dare I post this stuff on a public forum without talking to him... and especially since the apogee is better than any other block out there. It is stronger, performs better, and looks better than them all. <<< that was his attitude. He was not interested in my opinion of the weak spots on the top... nor was he interested in anything else I had to say. Why buy another one?

1) My own morbid curiosity. As you can see by the above pic... this block was screwed up too. Thats 2 for 2.

2) I have interested parties examining them. With a couple of tweaks... the block can be made better... much better.

[bigben2k]

Originally Posted by Annirak

Simple answer: create a bootable CD that performs the same operation with a very small granularity. As in repeates ever milisecond. May not be 100% power output, but it sure is repeatable. If you want to vary the power, vary the V-core.

Insulate the loop, and measure power output at the radiator. This is not hard. Just pretend you're running a chiller in your system and you want to avoid condensation. Ignoring such an obvious solution is silly.

That makes the heat source repeatable; it does not measure the actual wattage.

Do you know the temp difference that you're proposing to measure? (i.e. radiator output)? You also need to have a precise measurement of the airflow, if you're going to measure a dT air-in to air-out. Both measurements at 1% accuracy or better.

[Annirak]

Originally Posted by bigben2k

That makes the heat source repeatable; it does not measure the actual wattage.

Do you know the temp difference that you're proposing to measure? (i.e. radiator output)? You also need to have a precise measurement of the airflow, if you're going to measure a dT air-in to air-out. Both measurements at 1% accuracy or better.

Ok, then better than radiator, use a TEC chiller, and calculate power output based on the variables you have:
Internal loop temp
External loop temp
TEC current

TECs have a firm relationship between applied current and moved heat. I don't have equations, but they are available.

[bigben2k]

...but they depend a lot on the clamping pressure. Again, you're stuck with no actual measurement of power.

We've discussed this on several occasions, and came to the conclusion that water-in water-out (of the block), while hard to measure, is the closest, most practical solution. I'd consider the air chamber and radiator air-in air-out as an alternative, but I'm not sure it'd be cheaper. (need four temp probes on each side of the radiator, and the chamber will be large).

Otherwise a heater cartridge with measured volts and amps is much easier. Combine with the above, and you have an estimate of secondary losses...

[Annirak]

Then by all means, use water in/out of the block.
Since difference in temperature is all that's important, then a pair of thermocouples from the same batch should measure difference in temperature very accurately, particularly if they're coupled to the same isothermal block.

[bigben2k]

Yeah, but there are limits to thermocouples that prevent them from being useful in this setup; the resolution just isn't there, not for the small differences that would be measured here.

I'm still trying to get an RTD setup together: a 24 bit sigma-delta ADC (by TI: AD7713AN) and RTD elements in 4-wire configuration.

[gone_fishin]

Originally Posted by Orkan

Got the other one today. Took it apart:

Only found 3 or 4 shavings this time, instead of a bucket full... but they were still there.

... and along with that... there were some bent pins! I couln't make this up if I tried. lol QC at swiftech is non-existant on this block. period. You can see by the scratches in the pic, that someone has been in there after some shavings though. At least they are trying. Looks like they nailed a few pins with the tool they were using to get shavings out.



::EDIT::

Angry Steel,

My call to swiftech started out with me wanting to help them out in whatever way I could. ... However, after speaking with gabe... I don't really care about them. Gabe had his mind made up that I definately screwed up his product before he even picked up the phone. How dare I post this stuff on a public forum without talking to him... and especially since the apogee is better than any other block out there. It is stronger, performs better, and looks better than them all. <<< that was his attitude. He was not interested in my opinion of the weak spots on the top... nor was he interested in anything else I had to say. Why buy another one?

1) My own morbid curiosity. As you can see by the above pic... this block was screwed up too. Thats 2 for 2.

2) I have interested parties examining them. With a couple of tweaks... the block can be made better... much better.

Dude, your 15 minutes of fame are up, but keep buying them blocks you hate.....

[Annirak]

Thermocouples are analog devices. They don't have resolution. Do you mean the accuracy isn't there? If that's the case, then I'll agree with you.

The reason I said two thermocouples from the same batch is that they'll both be inaccurate. But the point is that they'll both be inaccurate by the same ammount. Or am I missing something here?

Are there other devices (I see you mentioning RTDs) that are very thin thermal sensors?

[Joe]

Originally Posted by gone_fishin

Dude, your 15 minutes of fame are up, but keep buying them blocks you hate.....

[Orkan]

Originally Posted by gone_fishin

Dude, your 15 minutes of fame are up, but keep buying them blocks you hate.....

I'm not out any money... My fame? lol

Nice productive post... troll.

[Swiftech]

Originally Posted by Orkan

These guys want me to ship the block back to them at my cost, then "if we find that it is a manufacturing defect, we will send you a new block."

I specifically told you (at least two or three times during the course of the conversation) that we would replace the block whatever the outcome of the inspection, and despite the fact that you had epoxied it. We agreed. Then you spoke to Michelle to sort the RMA process, and she quoted general Company Policy (inspection first then replacement if found to be defective) without knowing that I had approved the replacement "no questions asked".

This is a simple miscommunication. We still need to inspect this block, and as a good faith gesture we will pay for the return.

You might be interested to learn that we have since conducted severe stress tests on a number of housings with 100% negative results (no failures). I actually sent a response to Lee's (Robotech) concerns, including videos of the stress test procedure, featuring yours truly hammering these blocks like a madman It's actually funny, specially the part when I hit my fingers with the hammer. I hope they post it.

Seriously, and for the record every single block is pressure tested. How your block escaped QC is still under investigation. Our organization is composed of people. People sometimes make mistakes. We learn from them, and try not to repeat them.

The bottom line is: we do stand for quality; we always have and always will. When this fact is questioned, I do tend to be somewhat defensive. I am sorry if I came across like I didn't want to hear what you had to say. The reality is: I do. Swiftech owes its success to the support provided by users worldwide, and I will never forget this fact.



Best Regards,



Gabriel Rouchon
Chairman, CTA

Swiftech Inc.,
1703 E. 28th St.,
Signal Hill, CA 90755
T. (562) 595-8009
F. (562) 595-8769

[stev]

Originally Posted by Orkan

Got the other one today. Took it apart:

Only found 3 or 4 shavings this time, instead of a bucket full... but they were still there.

... and along with that... there were some bent pins! I couln't make this up if I tried. lol QC at swiftech is non-existant on this block. period. You can see by the scratches in the pic, that someone has been in there after some shavings though. At least they are trying. Looks like they nailed a few pins with the tool they were using to get shavings out.



::EDIT::

Angry Steel,

My call to swiftech started out with me wanting to help them out in whatever way I could. ... However, after speaking with gabe... I don't really care about them. Gabe had his mind made up that I definately screwed up his product before he even picked up the phone. How dare I post this stuff on a public forum without talking to him... and especially since the apogee is better than any other block out there. It is stronger, performs better, and looks better than them all. <<< that was his attitude. He was not interested in my opinion of the weak spots on the top... nor was he interested in anything else I had to say. Why buy another one?

1) My own morbid curiosity. As you can see by the above pic... this block was screwed up too. Thats 2 for 2.

2) I have interested parties examining them. With a couple of tweaks... the block can be made better... much better.

OK, the lady doesn't have a dental degree to clean out between those pins.

Did you test the cover with a yet?

[bigben2k]

Originally Posted by Annirak

Thermocouples are analog devices. They don't have resolution. Do you mean the accuracy isn't there? If that's the case, then I'll agree with you.

The reason I said two thermocouples from the same batch is that they'll both be inaccurate. But the point is that they'll both be inaccurate by the same ammount. Or am I missing something here?

Are there other devices (I see you mentioning RTDs) that are very thin thermal sensors?

I'll defer to Bill for the TC. (I'll try to dig up the reference).

Very thin, no, not many out there. I found an RTD element from Omega.com that's 10mm long, and 1mm in diameter, but it's a bare element (i.e. 2 wires). A bit pricey, but definitely available. The application is still a bit fuzzy to me (i.e. mounting of the element within a heat die). Not important though, there are many larger RTDs available for water in and water out.

[Orkan]

Apology accepted gabe.

Agrivates me when a company thinks it cannot mess up from time to time. I am glad to hear that you are taking the complaint seriously. I'll ship the block back asap.

[pHaestus]

You can get an output from a TC out to 0.01C, but it's just noise.

[BillA]

moot

[Annirak]

Shielding doesn't work?

[bigben2k]

Got it:

Originally Posted by BillA

...the TTVs use a TC so resolution is limited to 0.1°C...

(from post#248 in this thread:http://forums.procooling.com/vbb/showthread.php?t=12376 )

Originally Posted by Annirak

Shielding doesn't work?

Unfortunately, no.

[Albigger]

If you want to know the heat ouptput from a cpu, measuring the dT water in-out of the block is not too practical (I gather) for the reasons stated above.
Either of these seem practical?

1)slow the flow rate so there is a larger dT that can be measured with more confidence. Problems: at slow flow rates 'our' meters become inaccurate (would have to buy a new low flow meter probably) and as flow slows, the block heats up and more heat is lost to secondary sources (mobo) and so this wattage would underestimate the heat output under normal flow conditions.

2) How about making a 'thermal bath' - insulate the entire loop and pump water through the processor into some reservoir (recirculate the water) and keep measuring the temp increase. Need to be careful - many variables, also pump power input.
3) Better yet to put a copper slug with a round tube full of water (think the LN2 containers the benchers use but with less mass) to rest on top of the cpu die. keep sides and top well-insulated, with a stirring stick and temp probe and measure increase in temp of copper slug and water - caculate wattage.

--Jay

[bigben2k]

I admire your persistance, Jay.

1) that would work, but we really want to keep the flow rates to a somewhat typical rig, so 0.5 and 1 gpm are a must.

2) Possible. You could do that without any heat power applied, as a comparison point, to account for heat induced by the pump and pumping action.

3) That might work too. You'd need a good measurable temp increase which means a long time sample; if you can do that without frying the CPU, then all is well. I have a mercury thermometer with 0.1 deg C graduations, ranged from 19C to 27C that would be perfect for this. I think I prefer #2 though.

[Annirak]

well, if we can accurately measure the power output of a CPU once, then the bootable load CD should be viable. The only question I have is whether the onboard CPU diode can be trusted.

[Albigger]

Thanks ben. I didn't know if I was being ignored because I had stupid ideas before or just no one bothered to read it.

The only thing about 2 is that there are more losses / variables to consider, with the pump and all.

what really needs to be done is look at the current data available (I haven't really done this yet) for the dT water in-out, and the flow rate and see what wattage and what error margins we are talking about with the current method - has anyone with a test rig looked at that? (I'm sure they have).

anyway I know you guys have been over things like this a thousand times (kinda feel like I'm coming into the middle of a 3-year (or more) conversation), so if I'm out of line and need to read up on stuff or just need to shut up then tell me so...

[Albigger]

Originally Posted by Annirak

1)well, if we can accurately measure the power output of a CPU once, then the bootable load CD should be viable.
2)The only question I have is whether the onboard CPU diode can be trusted.

1) agreed. on THAT particular setup and cpu.
2) doubt it. that's why pH goes through the trouble of soldering directly to the temperature pins on a AMD cpu - so he can calibrate and read the core temps without the mobo. interferring.

[bigben2k]

Originally Posted by Albigger

Thanks ben. I didn't know if I was being ignored because I had stupid ideas before or just no one bothered to read it.

The only thing about 2 is that there are more losses / variables to consider, with the pump and all.

what really needs to be done is look at the current data available (I haven't really done this yet) for the dT water in-out, and the flow rate and see what wattage and what error margins we are talking about with the current method - has anyone with a test rig looked at that? (I'm sure they have).

anyway I know you guys have been over things like this a thousand times (kinda feel like I'm coming into the middle of a 3-year (or more) conversation), so if I'm out of line and need to read up on stuff or just need to shut up then tell me so...

The only person I know that does water-in water-out is Derek (pHaestus).

If you're more confortable exchanging ideas over email or the WBTA forum, feel free (I'm there either way)