all good! thanks for the offer. If I can't find anyone I'll resort to ebay, but even the chinese sellers of the stock fan model want like 15 dollars per fan, yikes
Taming the C6100
- Thread starter PigLover
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
I cut the wires to hookup san ace 1.1A fans. I didn't throw the original ones out. I should have them in garage somewhere. If you cant find a reasonably priced set, i can send them for the cost shipping.
Sent from my Nexus 6 using Tapatalk
So I don't know how many C6100's are still in lab use, but I just turned my 4 year old C6100 into my new homelab system a couple of weeks ago and needed to quieten it down a bit. It's an improvement over my older lab...
Anyway, to that end and thanks to this thread I purchased the Evercool EC8038HH12BP from Amazon and a set of the 4 pin to 5 pin adapters, also from Amazon; https://www.amazon.com/gp/product/B005G50C6Q/ref=oh_aui_detailpage_o02_s00?ie=UTF8&psc=1
I knew that the pinout was wrong but it didn't take long to figure out how to re-wire the 5-pin end so that it carried through the right signals. Basically you just make them "straight-through" adapters as the proper wiring is in the same order as PWM wiring on the fans.
All four are spinning, but only 3 of the 4 are reporting RPM's back. I am not 100% sure if it's the fan or the adapter cable that's wonky but I intend to do a bit of troubleshooting when I get some time (at the weekend). Anyway, these adapters gave me plenty of wire to reach the controller.
Now, I CAN report that my CPU's are running about 10C hotter than before on average, and the "MLB TEMP" are showing about a 15-20C hike... not great but certainly still well within limits of the components. With the PIC16 controller I can also report that the fan speeds are highly variable, which leads to much greater temperature variance than I was getting with the stock fans... again though I consider it acceptable for my lab use. Besides, I only have 2 of the 4 blades actually powered up, but even a full test with all four blades fired up made zero difference to the temperatures or fans on the utilized nodes.
As far as noise though? Wow, this made all the difference in the world. I now can't hear my server stack over the noise of the 10GBase-T switch they're attached to (Dell 8132 I flashed to an N4032). I measured the noise in front of the rack at the drives before the upgrade as around 60db... now I am getting a pretty good 48db but I suspect some of that's from the switch. I will measure it again with the switch off this weekend after I've troubleshot that dodgy fan.
For reference, my CPU's are dual E5530's, so not the coolest of the bunch... with each blade rocking 8 sticks of RAM (with all the baffles in place). Yeah, cooler CPU's might bring my temperatures down a bit but I'm not overly concerned about it.
Hope that helps anyone else still planning to do this mod... it's worth it, but anyone should also be aware that you might end up cooking your server... or reducing its lifespan. Since it's a "play system" for me anyway and is actually just extending my timeframe to buy a new homelab system anyway it's all good. If I kill it, I kill it
Anyway, to that end and thanks to this thread I purchased the Evercool EC8038HH12BP from Amazon and a set of the 4 pin to 5 pin adapters, also from Amazon; https://www.amazon.com/gp/product/B005G50C6Q/ref=oh_aui_detailpage_o02_s00?ie=UTF8&psc=1
I knew that the pinout was wrong but it didn't take long to figure out how to re-wire the 5-pin end so that it carried through the right signals. Basically you just make them "straight-through" adapters as the proper wiring is in the same order as PWM wiring on the fans.
All four are spinning, but only 3 of the 4 are reporting RPM's back. I am not 100% sure if it's the fan or the adapter cable that's wonky but I intend to do a bit of troubleshooting when I get some time (at the weekend). Anyway, these adapters gave me plenty of wire to reach the controller.
Now, I CAN report that my CPU's are running about 10C hotter than before on average, and the "MLB TEMP" are showing about a 15-20C hike... not great but certainly still well within limits of the components. With the PIC16 controller I can also report that the fan speeds are highly variable, which leads to much greater temperature variance than I was getting with the stock fans... again though I consider it acceptable for my lab use. Besides, I only have 2 of the 4 blades actually powered up, but even a full test with all four blades fired up made zero difference to the temperatures or fans on the utilized nodes.
As far as noise though? Wow, this made all the difference in the world. I now can't hear my server stack over the noise of the 10GBase-T switch they're attached to (Dell 8132 I flashed to an N4032). I measured the noise in front of the rack at the drives before the upgrade as around 60db... now I am getting a pretty good 48db but I suspect some of that's from the switch. I will measure it again with the switch off this weekend after I've troubleshot that dodgy fan.
For reference, my CPU's are dual E5530's, so not the coolest of the bunch... with each blade rocking 8 sticks of RAM (with all the baffles in place). Yeah, cooler CPU's might bring my temperatures down a bit but I'm not overly concerned about it.
Hope that helps anyone else still planning to do this mod... it's worth it, but anyone should also be aware that you might end up cooking your server... or reducing its lifespan. Since it's a "play system" for me anyway and is actually just extending my timeframe to buy a new homelab system anyway it's all good. If I kill it, I kill it
You might consider going on eBay and picking up dual L5520 CPUs, or possibly dual L5630 CPUs if your boards will support it. Lower power usage and will run cooler too. Costs will be pretty small, as these CPUs are going for less than $5 each.
Right... that's the PIC16 board, the PIC18 FCB is quite different. Still hackable but more complicated.
So yeah, for anyone with the older systems like me the adapters work like a champ with a little modding.Good thoughts... yeah I'll probably do some hunting around when I have some time this weekend. Given my workloads are pretty small (couple of small VM's and some Docker containers) I haven't been overly concerned about power. This is already better than my i7-950 that ran my previous lab... which at full tilt burned some watts. I still have that 950 running as I migrate services to the 6100. You're right though, while a single blade is about equal for power to my 950-based system, two blades is more expensive on power.
Having said that many of us are pretty spoiled living in the US where electricity is cheap. Maybe I'll think differently when the temperatures start going above 90F in the midwest
EDIT: L5630 CPU's ordered for one blade... let's see how well it goes now
If it works out, I'll order more.
Last edited:
Wanted to add a bit of an update for everyone.
I have confirmed that one of the Evercool fans is actually broken... it's working (spinning) but seems that it's not reporting any PWM information back. As a result, I've RMA'd it... hopefully should get my replacement next week or so.
Also, I have installed the L5630's in my blade and they seem to be working great. Going to give it a few hours while everything runs but so far the temperature profile looks a LOT better than the E5530's I was rocking before. I'll give it a bit of time, but so far I'm showing no perceptible performance difference for my workloads, which is awesome. Bonus, I was able to replace a dodgy DIMM with a good replacement... ECC had disabled one of my 4GB DIMMs dropping me to 20GB of usable RAM. I actually have a stack of quad-rank DIMMs sitting around here (amazing what one accumulates after years in this field) so I swapped the dual-rank DIMMs in there with the quad-rank DIMMs as well and they work like a champ... and I'm back to 24GB of usable RAM.
So all good, and more L5630's and a replacement fan on the way... I would say I'm pretty happy with this thread for making my C6100 usable again.
Now I've just got to figure out how to quiet down that damned N4032 that is now WAY louder than anything else in the rack...
I have confirmed that one of the Evercool fans is actually broken... it's working (spinning) but seems that it's not reporting any PWM information back. As a result, I've RMA'd it... hopefully should get my replacement next week or so.
Also, I have installed the L5630's in my blade and they seem to be working great. Going to give it a few hours while everything runs but so far the temperature profile looks a LOT better than the E5530's I was rocking before. I'll give it a bit of time, but so far I'm showing no perceptible performance difference for my workloads, which is awesome. Bonus, I was able to replace a dodgy DIMM with a good replacement... ECC had disabled one of my 4GB DIMMs dropping me to 20GB of usable RAM. I actually have a stack of quad-rank DIMMs sitting around here (amazing what one accumulates after years in this field) so I swapped the dual-rank DIMMs in there with the quad-rank DIMMs as well and they work like a champ... and I'm back to 24GB of usable RAM.
So all good, and more L5630's and a replacement fan on the way... I would say I'm pretty happy with this thread for making my C6100 usable again.
Now I've just got to figure out how to quiet down that damned N4032 that is now WAY louder than anything else in the rack...
Need some feedback on UPS with the C6100.
I have a dual power supply C6100. 4 sleds, dual L5520 with 48G.
One PS is powered from the wall, and one PS is plugged into an OPTi-UPS TS2250B. The UPS says it is Active PFC compatible. (OPTI-UPS TS2250B UPS - Newegg.com)
The issue is that unless it is a very short outage (flicker) the C6100 will reboot. I don't think it's an overdraw issue on the UPS. I have other devices plugged into the UPS that do stay up.
Any thoughts or recommendations? The C6100 is the only thing that reboots. I have a Dell 2950 with the same power configuration (one PS powered from the wall, the other from a UPS (a different UPS from the C6100)) and the 2950 never goes down.
I have a dual power supply C6100. 4 sleds, dual L5520 with 48G.
One PS is powered from the wall, and one PS is plugged into an OPTi-UPS TS2250B. The UPS says it is Active PFC compatible. (OPTI-UPS TS2250B UPS - Newegg.com)
The issue is that unless it is a very short outage (flicker) the C6100 will reboot. I don't think it's an overdraw issue on the UPS. I have other devices plugged into the UPS that do stay up.
Any thoughts or recommendations? The C6100 is the only thing that reboots. I have a Dell 2950 with the same power configuration (one PS powered from the wall, the other from a UPS (a different UPS from the C6100)) and the 2950 never goes down.
Last edited:
@Dale McKay - Check that UPS. Its likely junk. Tons of reviews on Amazon indicating the exact same problem. Seems that on power loss, its output power drops to 90v for a small window, and then bounces back up. Small devices are likely not affected, because the power is still high enough, but the server needs a lot more then that. I'd dump that UPS and switch to an Eaton, APC, CyberPower or TrippLite UPS (in that order of best experience to least but known to be decent). Just make sure your UPS can provide true Sine-Wave power.
I realise that this is a very old thread at this point, but I was wondering if anyone had any luck finding a 7Pin for the PIC18 to 4 pin adapter cable? My searching hasn't been able to find any (currently thinking I'll be going with some Noctua fans, but might go for something else), and I don't really want to start cutting up the onesones got in there currently if I can avoid it.
Last edited:
I purchased the Noctua's with the extension cable, and I flipped the pins on the extension cable. This way it worked without issue, but also left the fans in stock form so I could re-use them in the future. I just labeled the extension as re-pinned so I would know the future not to use it as a stock fan connector.
Wanted to post this here before I go out to Ebay. You might remember my original post. (Reposted below).
My C6100 is ready for new home. Specs are below. The fourth sled is a little shaky. No BMC, the power button doesn't light and the video is barely usable, but once it boots it's all good. The other three no issue. All four boot from USB to ESX 6.5 (around update 2, don't remember exactly.) I'll leave the USBs with the sleds. No onboard disk but three trays.
Price is very negotiable to anyone on this forum. If you are local to central Texas even better.
"My C6100 experience.
I wanted to upgrade my home lab. I am a consultant and run a variety of systems in a pretty extensive home lab. MS 2012 R2 domain, ESXi cluster, Hyper-V, Cisco, and iSCSI storage via Synology RS812 platforms. I was using Dell 1950 servers for a lot of this, so I now have three of them for sale,,,just sayin.
I recently purchased a DCS (no valid Dell service tag at Dell support) C6100 from a seller on Ebay. Final agreed price was $700, which included 4 500GB drives as well. The C6100 (3.5" format) has 4 nodes, dual L5520s, (only 60W power) and 48 GB in each of the nodes and dual power supplies. I pushed the seller to update BIOS and BMC to the latest versions before shipping. (BIOS at 1.71, BMC at 1.33).
I added the network cards to three of the nodes that I intended to use as ESX nodes. (DELL NetXtreme II Dual Port Gigabit PCIe Network Interface Card G218C). When I installed these, the top of the metal bracket extended too far into the airflow for my liking and I used tin snips to cut the metal bracket right at the top of the PCB. The network cards don't actually attach to anything other than the PCI riser card and associated metal bracket, but they are surprising steady and quite small so no issues with the lack of the bracket. (Total cost for three network cards, $30). ESX had no issue with these cards and I run both a VDS and a standard switch in each ESX host connected via LAGs to my core 1Gb switch."
My C6100 is ready for new home. Specs are below. The fourth sled is a little shaky. No BMC, the power button doesn't light and the video is barely usable, but once it boots it's all good. The other three no issue. All four boot from USB to ESX 6.5 (around update 2, don't remember exactly.) I'll leave the USBs with the sleds. No onboard disk but three trays.
Price is very negotiable to anyone on this forum. If you are local to central Texas even better.
"My C6100 experience.
I wanted to upgrade my home lab. I am a consultant and run a variety of systems in a pretty extensive home lab. MS 2012 R2 domain, ESXi cluster, Hyper-V, Cisco, and iSCSI storage via Synology RS812 platforms. I was using Dell 1950 servers for a lot of this, so I now have three of them for sale,,,just sayin.
I recently purchased a DCS (no valid Dell service tag at Dell support) C6100 from a seller on Ebay. Final agreed price was $700, which included 4 500GB drives as well. The C6100 (3.5" format) has 4 nodes, dual L5520s, (only 60W power) and 48 GB in each of the nodes and dual power supplies. I pushed the seller to update BIOS and BMC to the latest versions before shipping. (BIOS at 1.71, BMC at 1.33).
I added the network cards to three of the nodes that I intended to use as ESX nodes. (DELL NetXtreme II Dual Port Gigabit PCIe Network Interface Card G218C). When I installed these, the top of the metal bracket extended too far into the airflow for my liking and I used tin snips to cut the metal bracket right at the top of the PCB. The network cards don't actually attach to anything other than the PCI riser card and associated metal bracket, but they are surprising steady and quite small so no issues with the lack of the bracket. (Total cost for three network cards, $30). ESX had no issue with these cards and I run both a VDS and a standard switch in each ESX host connected via LAGs to my core 1Gb switch."
