// #eevblog — 2019-08-05
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
00:57:28 <DHess_US> <steve30> Now it's really kicking the heat out. Heatsink is 60C
00:57:29 <DHess_US> <Ampera> 60C isn't much
00:57:44 <DHess_US> 60C on the *heatsink* is a big deal. It means the junction temperature is a hell of a lot higher.
00:58:15 <DHess_US> For reliable operation of logic, a jucntion temperature of 60C is close to max.
00:59:07 <DHess_US> nVidia gets away with 100C CPUs because they are only intended to have an operating life just exceeding the warranty period of a year.
00:59:40 <DHess_US> <Halcyon> Another shooter dead in the US
00:59:52 <DHess_US> Which one is that? The El Paso shooter was taken into custody alive.
01:07:59 <DHess_US> Oh, this guy? https://heavy.com/news/2019/08/dayton-ohio-active-shooter-shooting/
01:43:15 <Ampera> DHess_US, I've found 90C to be max for most CPUs, but I'm just a fucking retard, so what do I know
01:43:38 <Ampera> that's the thing about trying to learn about stuff, at any point someone can come over and shit on everything you thought you knew
01:43:44 <Ampera> might as well remain stupid
01:47:19 <DHess_US> It doesn't mean that they won't operate at that junction temperature. It just means that that junction temperature is not compatable with reliablity over a longer time period.
01:48:03 <DHess_US> https://en.wikipedia.org/wiki/Batesian_mimicry
01:48:15 <DHess_US> err, wrong link.
01:48:22 <DHess_US> https://en.wikipedia.org/wiki/Arrhenius_equation
01:49:17 <DHess_US> Going from 60C to 90C lowers the operating life to 1/8th.
01:49:45 <DHess_US> operating life is limited by junction difussion and electromigration.
01:50:08 <Ampera> My CPU is currently running at 82 degrees.
01:50:11 <DHess_US> Power semiconductors can operate at higher junction temperatures but not small signal logic.
01:50:23 <DHess_US> 82C or 82F?
01:50:39 <Ampera> F
01:50:46 <Ampera> Just to confuse you extra, 27.8 C
01:50:48 <DHess_US> None of my systems operate at more than 55C when fully loaded.
01:50:59 <DHess_US> Right, that is typical for idle.
01:51:08 <Ampera> Idk what it goes up to fully loaded, let me try to stress it
01:51:10 <Ampera> somehow
01:51:50 <DHess_US> The Radeon single slot 5670 I had for a while operated at 90C and died after about a year. I replaced with with a 5770 double slot which has worked fine for years now.
01:52:35 <DHess_US> My older 4850 works fine also except of course no more windows drivers for it. :(
01:52:54 <Ampera> well it appears to not want to change from 27.8C so I'd imagine this program isn't accurate
01:53:03 <DHess_US> Or it is measuring the MB temperature
01:53:27 <Ampera> I'm using screenfetch, so who knows
01:54:06 <DHess_US> Anyway, one of the tricks nVidia did to raise clock rates was to not derate their GPUs for 10 to 20 years of operation allowing a 90C specification.
01:54:48 <DHess_US> Which goes along with all of the packing failures which led to lawsuits against them.
01:54:56 <Ampera> around 35C idle, but my fan curve isn't set well
01:55:13 <Ampera> It hits 90C very readily though
01:59:12 <Ampera> changed some stuff, 30C idle, full load is 80C for all cores.
01:59:25 <Ampera> That being said I am running an overclock to 4.6Ghz
06:00:01 Quit: narayan ([email protected]) Ping timeout: 121 seconds
06:03:41 Quit: Halcyon ([email protected]) Ping timeout: 121 seconds
06:25:06 Join: narayan ([email protected]) to #EEVblog
09:24:09 Quit: narayan ([email protected]) Quit: leaving
12:44:21 Join: T3sl41 ([email protected]) to #EEVblog
12:44:31 Mode: (by ChanOP) +o T3sl41
12:45:17 Quit: T3sl4 ([email protected]) Ping timeout: 121 seconds
14:38:42 Nick: T3sl41 is now known as T3sl4
15:38:56 <T3sl4> coppice_: Interesting. I'm looking up faster switching controllers, and finding very little of anything above 2MHz.
15:40:03 <T3sl4> I'm currently pondering a planar, resonant 2-3MHz design, which looks like it should work given the other parts available
15:40:45 <T3sl4> seems like i'll end up spending more space on the [discrete] control circuit, than the actual power circuit and magnetics
15:41:35 <T3sl4> the push to 2-3MHz is really only due to the ELP14 transformer i'm looking at, and it seriously looks like i should be using ELP18 or so, just because no one makes resonant controllers this fast
15:41:51 <T3sl4> what a silly reason
15:49:18 <T3sl4> BD9851 seems to go up there
16:05:39 <T3sl4> god, they make 555s up there
16:05:46 <T3sl4> must i resort to using a 555?
19:00:41 Join: Halcyon ([email protected]) to #EEVblog
19:42:27 <Halcyon> -
19:42:36 <Halcyon> I'm thinking of moving my server rack to the garage
19:42:38 <Halcyon> Thoughts/
19:42:39 <Halcyon> ?
19:45:00 <K> hello
19:45:21 <K> is the garage mmmmoist?
19:48:44 <Halcyon> Hehe Moist *giggles*
19:48:47 <Halcyon> I don't think so
19:48:53 <Halcyon> Probably should log data first huh?
20:06:37 <Halcyon> Ampera, cards?
20:06:43 <Halcyon> T3sl4, ?
20:28:06 Quit: DogWomble ([email protected]) Ping timeout: 121 seconds
20:34:17 Join: DogWomble ([email protected]) to #EEVblog
20:34:27 Mode: (by ChanOP) +o DogWomble
20:51:34 <coppice_> T3sl4: have you looked for switching controllers specifically aimed at the GaN and SiC market?
20:51:57 <T3sl4> i was search digikey, and a bit of TI
20:52:18 <T3sl4> TI doesn't even put their GaN controllers in the main list, almost like they feel they're still too new and experimental or something
20:52:35 <T3sl4> but i saw one or two that wouldn't be bad, if they weren't so special purpose (I2C programmed, ack)
20:53:09 <coppice_> TI have been pushing low voltage GaN parts for a few years. They shouldn't be that special/niche now
20:56:18 <T3sl4> they've got those nice integrated driver/inverter chips
22:31:39 * steve30 purrs at Halcyon
22:40:08 <steve30> I went swimming earlier :)
22:40:33 <K> how did it go?
22:40:47 <K> (swimmingly! ((badum-tiss!))
22:47:58 Quit: coppice_ ([email protected]) Ping timeout: 121 seconds
22:50:17 Join: coppice ([email protected]) to #EEVblog
Search this day's buffer: