05-21-2018 07:14 PM
Nice save on the case hazard for the pipes, DGroves!
And a question for any of you specialists:
the CPU Fan connectors (19 and 20 in mobo diagram) have 5 pins:
What's the Tach2 used for? Redundancy check, since it's the CPU?
If so - looks like it can be tricked by cloning the Tach1 signal, which would explain how Javato connected a 4-pin 3600 RPM case fan to a 5-pin CPU Fan header 😛
05-22-2018 04:45 AM - edited 05-22-2018 05:53 AM
tach 2 does the same thing as tach 1, it accepts input from a fan's rpm signal i don't have a z620 so i can't say if the z620 bios is configured to display information from this sensor. hp may have used 2 fans in a custom OEM case with the z620 board
however a 3rd party program like hwinfo can most likely be configured to read this sensor if hp enabled it on the z620 motherboard
08-23-2018 07:42 PM
Well, dear gurus, here's the first update 😄
Succesful pre-test of pump and chiller fan powering!
For now I keep a single CPU running in the Z620, so that means:
- rear double-fan connector only feeds the Chiller Fan from the H45.
@Dgroves - thanks for the idea to feed both Sense off the same output from the fan!
(after I upgrade to 2x 2667v2 - now that it's thermally feasible 😄 - I will feed from here the second H45 Radiator Fan)
- I opted for MOLEX powering at constant 12v the pump on the H45... Corsair sais so - I do so.
- Cutting up the case, and actually installing the H45.
- Decoupling the CPU fan and using that connector to power the lower Rear Fan (the one that will stay in place, unaffected by the chassis cut)
(I'll do the same here - trick both Sense pins into receiving data from a single fan :D)
Thanks Javato and Dgroves for inspiring me & holding my hand while I grasped the pin layouts and possible options I had in there!
More photos to come when I find a Dremel tool to borrow / rent!
08-25-2018 09:44 AM - edited 08-27-2018 05:28 AM
I tried some ideas for a z620 single and dual CPU liquid cooling solutions and worked out an idea far enough to establish feasibility.
The single CPU solution was external, rear panel mounted, and used a single fan :
This does delete the upper internal rear case fan but is arranged so as to avoid blocking back panel connections access.
As the z420 and z620 motherboards are so similar an duse the same BIOS, after some research, I decided to try the z420 liquid cooler:
> which simply plugged in about 20 minutes without any modification. This works well: 29-33C at idle and during CPU rendering at 61-64C. As the z420 and z620 use the same BIOS, the z420 cooler is controlled by BIOS and the settings have not been changed from the default. This is impressive as the E5-1680 v2 operates at 4.3Ghz on all 8 cores and it's running well under the 85C maximum rating. An additional attractive features is the small auxiliary fan to cool the chipset. This is the best $50 I've ever spent on a computer component!
For the dual CPU configuration, it was more difficult to keep the rear panel connections clear and the four liquid tubes would mean removing one of the case fans. I was also concerned that the liquid cooler would be intaking the hot air from the case, lowering the efficiency.
For the dual CPU design, in order not to block the back panel peripherals connections, have to remove one of case fansas there are four coolant tubes instead of two, and avoid having the hot air extracted from the case going into the intake of the CPU coolers, thisconfiguration places two, single 120mm CPU coolers in an enclosure sitting on top of the case for a clean air stream, and run the tubes for the coolant though the top of the case. As this is a closed-loop system, the CPU cooler units can't be detached and so have to installed and removed through a removable panel cut into the top of the z620 case.
The coolant tube connections run downwards between the power supply and the drive bays. The drive bays are not shown:
The strange triangles are guides to eventually drawing the liquid tubes properly. As I didn't have the precise positions of the components, those were left as a sketch. The riser board is shown in the wrong position.
The pricincipal concern in this design is the difficulty of having to mount the riser board socket cooling block in place. As the CPU1 shroud has to be installed before the riser board. I never tried it, but I think the riser shroud has to be installed before sliding it into the mainboard sockets. Perhaps it can be installed after, but I never tried iadding the riser shroud after the board. If not, it is probably necessary to cut a slot in the top of the chassis to allow the riser cooling block to be installed with the riser, sliding the cooling tubes horizontally.
And this hows that the there is full access to the peripherals slots and the internal case fans are not affected.
This gives an idea of the way that the socket-mount components are installed and removed through the removable top panel. The top access panel has slots for the collant tubes and would slide in towards the back.
What is not shown is the fan control is by a front panel 5.25" bay fan controller, for example:
The enclosure panels are 1/8" Aluminum, plasma cut. I drew a 140mm version as well that made the total height of computer / cooler 27.5" / 70cm.
Proprietary systems are so specifically engineered, that they are difficult to modify and especially to derive a tidy solution that is also easy to service. This I think would provide effective cooling but is still a bit of hot rod / chop shop solution. As a single CPU may today have 28-cores and a V2 Z-420 or z620 may use to a 12-core cores, I'd suggest the z420 liquid cooler and for dual CPU systems under high, continuous loads, consider instead a z820 with the HP z820 liquid coolers. The cost may be the same and certainly the time and effort is much less being able to use a well-enigneered, proprietary system. A liquid cooled z820 would have a higher resale value that a modified z620 as well.
05-21-2019 10:16 AM
Hi Javato and whole forum 🙂 !
Im modding my Z620 and i decided to change all my stock fans with noctuas, their silence being main reason and also i got them really cheap 😄 But i run onto problems with system detecting its low speed ( low fan speed ) and i simply cant find an solution to this. Also when booting i got fan error messages. How did you managed this ?
I changed front intake fan, no problems only during boot.
But i wanted to change a stock fan on my z420 liquid cooling kit for noctua, because the stock one is only good at it slowest speed, otherwise its very high pitched noise its driving me nuts. And this is where i run into problems, that fan will just pitch itself into 100% and system cannot identify it. Im out of ideas ...
Thanks for any help !
08-10-2019 08:28 AM
I saw once some fake rpm modules that were reading fake rpms to the motherboard... that could help...
but, honestly, i would just buy one of these fan controllers for the 5.25" bay, connect all the noctuas there and not use the motherboard ones. That means bearing with the boot warning and pressing F1 each time...
11-01-2019 07:06 AM
Yes i can do that, but honestly , its not anywhere near the elengancy i want 😄
Im now experimenting with rewiring noctua fan extender cable to HP 5pin for Cpu, as i intend to connect an noctua fan to this modified extender, i dont wanna destroy new fan 🙂
11-01-2019 08:10 AM - edited 11-01-2019 08:15 AM
This is a comment on the wiring of the CPU heatsink fans.... above in this thread is a diagram from the technical manual showing the pinout to be 1->5 as GND +12VDC Tach1 PWM Tach2
In more detail:
GND... pin 1. A "Performance" heatsink/fan will have a ground jumper wire running the short distance from this pin to pin 5.
+12VDC (this stays at a constant value.... the fan RPMs are not controlled by the motherboard changing this voltage as needed. However, one can use a resistor in line and drop the constant voltage in this line to, for example, +8VDC and the fan will run slower and the PWM speed control will still work. If you drop it too much however the motherboard may detect that RPMs are too slow and respond by boosting the RPMs as has been noted in a post above.
Tach1... this really is simply RPM speed feedback from the rotor to the motherboard along this lead.
PWM.... this really is simply the standard PWM control signal path from the motherboard out to the rotor. If this line is cut there will be no PWM braking and the rotor will spin at its full speed at 12VDC (or slower if you have used an in-line resistor).
Tach2... this appears to be incorrect information. This is simply where a ground jumper from pin 1 to pin 5 lands if the fan is part of a high cooling capacity heatsink/fan unit. This method had been used by HP for years to differentiate a "Mainstream" lower cooling capacity fan from a "Performance" higher cooling capacity fan. The motherboard has been programmed to detect the individual CPU attached and if it is a high power CPU the motherboard demands to see ground at that pin 5 as proof that a Performance heatsink is in place. If a Mainstream heatsink/fan unit is in place the 5th hole in the fan plug is not populated.