Dell H200 versus H310 heat

Status
Not open for further replies.

wblock

Documentation Engineer
Joined
Nov 14, 2014
Messages
1,506
I've recently reflashed a couple of Dell H200 controllers to IT mode. They work fine. In a normal desktop, that little heatsink gets too hot to hold a finger on, so I added fans to both.

The H200 connectors are in a more convenient location for a desktop than the H310, which was why I picked them. For variety, I recently converted an H310, and noticed that it runs less hot, cool enough to hold a finger on the heatsink.

Is this just a random variation between cards, or has anyone else noticed the H310 running less hot than the H200?
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,194
That's odd, they're both SAS 2008 cards, so I'd expect them to be similar. Even the newer SAS 2308 stuff seems to produce the same amount of heat.
 

joeschmuck

Old Man
Moderator
Joined
May 28, 2011
Messages
10,994
My H310 is hot, too hot for my liking but I understand that it is normal. I don't have an H200 to compare to.
 
Last edited by a moderator:

wblock

Documentation Engineer
Joined
Nov 14, 2014
Messages
1,506
It occurs to me that I have not yet replaced the ordinary wax heatsink compound on this card like I did on the others. Maybe it just isn't transferring all that heat to the heatsink. Will do that and report back.

Incidentally, these really do need fans. 40mm fans fit right on those heatsinks, and running a 12V fan at 5V seems to be enough air to keep it at a reasonable level. It does make the card much thicker, enough to potentially interfere with the card in the next slot. A larger fan blowing from the front of the machine could be effective, more for the H200 where the cables aren't in the way.
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,194
They're really designed for a traditional server layout with a row of fans across the chassis, implying that everything gets airflow.

On the X10SL7-F, the SAS 2308 can get rather hot as well, when there isn't any airflow going across its heatsink - but people with even a little bit of airflow say it's alright.
 

Mirfster

Doesn't know what he's talking about
Joined
Oct 2, 2015
Messages
3,215
In my 2U Servers, the H200 is fine. But as mentioned by EricLowe there is appropriate airflow passing over them to keep it decently cooled. While I don't have a H310, the LSI 9211-8i is about the same for me as a H200.
 

Spearfoot

He of the long foot
Moderator
Joined
May 13, 2015
Messages
2,478
It occurs to me that I have not yet replaced the ordinary wax heatsink compound on this card like I did on the others. Maybe it just isn't transferring all that heat to the heatsink. Will do that and report back.

Incidentally, these really do need fans. 40mm fans fit right on those heatsinks, and running a 12V fan at 5V seems to be enough air to keep it at a reasonable level. It does make the card much thicker, enough to potentially interfere with the card in the next slot. A larger fan blowing from the front of the machine could be effective, more for the H200 where the cables aren't in the way.
Wax heatsink compound? I had no idea! Now I'm gonna pull my H200s and reseat the heat sinks, 'cause they do run hot.
 

wblock

Documentation Engineer
Joined
Nov 14, 2014
Messages
1,506
Just that ordinary gray waxy heatsink compound, the kind the gamers always say "oh, it's all dried up!" I'm pretty sure it's supposed to be that way, that's how it's made, but it's not the best at transferring heat.
 

wblock

Documentation Engineer
Joined
Nov 14, 2014
Messages
1,506
Semi-interesting: the BGA processor on the H310 is in a diamond orientation versus a square orientation on the H200. Can't see that making any difference, but I could imagine some revision of the card or chip being better at power management.
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,194
Semi-interesting: the BGA processor on the H310 is in a diamond orientation versus a square orientation on the H200. Can't see that making any difference, but I could imagine some revision of the card or chip being better at power management.
I don't think the SAS 2008 has had any major hardware revisions, since they rolled out the 2308 in the middle of the SAS2 product cycle.

The card is a custom Dell design and not just an LSI rebadge, like everyone else does, right? Power delivery may be different, but that wouldn't affect the SAS controller directly...
 

wblock

Documentation Engineer
Joined
Nov 14, 2014
Messages
1,506
The card is a custom Dell design and not just an LSI rebadge, like everyone else does, right?
Not sure. It looks like the other cards with the SAS connectors on the end. At least based on a reference design, if maybe not identical.

The card is Dell P/N 0HV52W. There is an e1 in a circle near the LED connector, repeated on the chip. Serial number and SAS ID blurred because all the cool kids do that:

card.jpg


chip.jpg
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,194

wblock

Documentation Engineer
Joined
Nov 14, 2014
Messages
1,506
Some numbers measured with a digital non-contact thermometer that is probably not too trustworthy for absolute values. Also, the heat sinks are silver and reflective, so it would not surprise me if actual temperatures were 50% higher or more. Room temperature was about 23-24C.

Both tests with the heatsink alone, Gelid "GC Extreme" heatsink compound, one drive connected, approximately thirty seconds after booting FreeBSD (mfsBSD), no drive activity. Heatsink temperature:

H200: 88.5C
H310: 26.8C

H310 after read-only speed test ( diskinfo -tv da0) on WD 750G drive (approximately 35 seconds): 31.8C Felt warmer, uncomfortably warm to a finger but not instantly too hot. Tilting the thermometer input one way gave 35.8. The highest temperature seen in a repeated test was 38.2C. It felt far hotter than 100F. (Donations of a Flir thermal imager cheerfully accepted. Or anything else, really.)

Adding 40mm, 12V, ball-bearing fan to H310 heatsink showed that the heatsink was one row of fins larger. With the nominally 12V fan at 5V... it had to be manually started. Once running, the same disk speed test went to 33.0C. I was able to hold my finger on the heatsink without it being anything more than a little warm.

Lessons learned:
1. $20 non-contact thermometers are not particularly useful to measure spot temperatures of shiny components.
2. Don't count on 12V fans to start at only 5V.
3. Given the choice, take the H310 over the H200.
 
Status
Not open for further replies.
Top