The iCUE Commander...
Corsair iCUE H150i Elite Capellix - this is exactly what worked fine in one workstation till "the maintenance job" occurred after which all the fans and pump started to work on a full throttle with all the LEDs blinking red. You may think of an mistake while plugging back the cables, sure that might be the case but after checking and re-plugging everything still nothing changes. Thinking further, perhaps some of the cables broke?
Corsair provides control software iCUE for their sets but unfortunately for windows or mac only so I had to boot into windows to see what it reports. Pump Failure it says, gosh quick check for temperatures and it is normal. I can hear the pump buzzing, fluid bubbling thus I don't think it is stalled. Firmware update poops out, fine lets try this out - problem still bothers. Too bad that warranty is no more here but hey at least it still does its main job - cooling - well until pump truly stops unnoticed this time.
Having nothing to rethink I decided to open it up for measures. There are 2 plugs from the pump to control its motor and lighting: one multipin JST going into the commander controller and three pin molex connector with just single cable used. The last one is plugged into PC's motherboard for CPU fan speed signal. Check with the scope reveals no clock signal on it, then maybe the problem is within pump board? Getting into the pump board is easy not worth a comment.
Tested for continuity between the end of molex wire, pump board and the commander plug (I don't remember which pin it was but that is easy to find) - again everything seems okay they are common line. Pump driver chip is BD69830FV to be exact. Data sheet says the pin 7 is a tachometer output, lets check it with the scope - pretty square wave, geee everything is functional so far? Deeper searching and the signal travels through n-mosfet Q11 with source grounded. Quick test and the transistor looks good, turns out there is no pull-up on the drain output - soldered additional 10k resistor to the 5V line just in case. Still no clock, hmmm. Switching towards the commander socket, tracking down corresponding pin and finally found something - zero resistance between the ground, seems like a short. Gently removed the wire terminal from the plastic connector to isolate it from the controller, plugged it back and now the pump signal is present - nice indeed seems like the problem is with the controller box. Disassembling the pump block was a unnecessary waste of time, pull-up resistor must be only on the controller board - strange but whatever, no longer only after modification.
Pump board and added pull-up resistor.
Controller enclosure halves are tied by latches, no screws - those are too expensive for this class of elite product. If you take a close look into the picture, notice where the hook/catch and loop/latch arm are. Stick a piece of plastic (credit card for example) through a visible slit between the hook and loop to leverage it and bend the hook part outside to disengage the latch. Once it looses put another piece of plastic or bend paper as a spacer and move one to another latch. Be careful, those are a bit fragile to crack and I succeed in breaking one.
Controller casing.
Controller board views.
Investigate all the fuses for continuity (green SMDs). First traces from pump speed pin lead to gate of transistor Q2 (top rightmost corner) which seems to be shorted and it is, why did it happen? FDV301N replace it or with some other ~20V rated N-MOS. According to data sheet Gate-Source is rated to 8V and I don't remember board having resistor divider (or zener diode) on the gate input - this is weird design corner cut, if someone forcefully sticks the speed feedback molex connector to the wrong place and is lucky enough to hit the 12V supply line it will blow the mos. Not that easy to surprise me I think that is possible, still better than the blown microcontroller though.
Replaced failed transistor.
And that's it, luckily nothing else went wrong, after powering-up fans are accelerating then slowing down and their rainbow colors persist. Enjoy fully operational AIO.
About a fan speed problem thingy, this one asks for those Z17 and neighborhood checks...
Summary
- When nothing helps on pump error with commander core sets just a controller replacement may be needed - even after disconnecting the speed sense cable from the controller box you will still get 0 fan speed in BIOS because of missing pull-up resistor. This is less of an issue since pump and controller are integral parts.
- The circuit design is a bit faulty.
- Repair is easy 0.1$ but requires some basic tools.
Bonus note
The copper cooling plate fins are made by scratching its surface with a edge tool. Due to bad manufacturing practices those structures may clog after time by crystallized brazing flux residuals. Get one AIO for cheap, clean, flush out with distilled water, refill and use in your projects.
Closer look on water cooled copper block.