Corsair H60 cooler cycles on and off every 3-4 seconds

[hfffoman]

I just installed a new motherboard and the CPU fan isn't working. Consequently it won't boot. A bit of detective work is needed here. If I explain the exact situation there are enough clues for someone smart to figure it out.

The old board was professionally installed and instead of plugging the CPU fan connector onto the motherboard they plugged the two power wires directly into a power cable from the PSU. That leaves a single wire in the connector which goes into the IN pin when the connector is correctly located on the motherboard. This arrangement always worked fine and I think it's recommended by Corsair. The new motherboard is similar to the old but slightly upgraded (ASUS Z97-P instead of ASUS P8Z68V-LX). It has exactly the same fan socket configuration.

When I first turned on the motherboard the fan didn't work at all. The machine started to boot. Of course the CPU temperature started to rise and I quickly turned it off before it went above 34C.

I found that one of the CPU fan power leads was loose. I then made sure it had a good contact and next time the fan worked. But it cycled on and off every 3-4 seconds and the computer wouldn't boot. When I pressed the manual override on the fan it still turned off after a few seconds.

I then unplugged the connector with the single wire from the motherboard. The behaviour of the machine was exactly as before - the fans runs on and off every few seconds and it won't boot. This suggests the signal isn't getting through from the motherboard to the cooler.

I then got hold of a cheap fan with a conventional connector and plugged this into the motherboard. The fan works as normal and the machine starts to boot. Again I quickly switched it off because the working fan wasn't attached to the CPU.

So it looks like one of several possibilities:

1. The new motherboard won't tolerate the unconventional connectors. I would be surprised if this were the case given the similarity of the boards.

2. The wire is broken. I can't see any sign of damage but it probably got pulled a bit in the course of changing the motherboard.

3. The cooler is bust. However, the last time the cooler was switched on, with the old board, it worked fine.

What's your advice?

 

[Paperdoc]

Some background to help you understand.

The mobo CPU_FAN header has 4 pins. Of them, Pin #1 is Ground, Pin #2 is the + DCV power supply, and Pin #3 accepts the speed signal generated by the fan and sent back to the mobo on this pin. Pin #4 provides the PWM singal to a 4-pin fan.

On a simple 3-pin fan, the voltage supplied on Pin #2 is varied by the mobo to accomplish speed control. In the newer 4-pin system, that voltage is always the full 12 VDC and the fan itself uses the PWM signal to modify that supply and accomplish speed control.

Now, the H60 system needs power for TWO functions, and hence there are two connectors that go to different places. Your post does not include those details - you write only about the "CPU Fan". So let me outline how that system is designed to be used.

The pump module mounted on the CPU has a cable from it ending in a 3-pin female fan connector. The pump needs the full 12 VDC supply at all times so that it runs full speed. For this purpose it can be plugged into a mobo fan header that is configured to keep that power on Pin #2 at 12 VDC constantly. OR it can be connected (via an adapter) to a power output from the PSU, which will always be 12 VDC because that's its design. IF you plug it into a mobo header, the pump's speed signal MAY be sent back to that header IF there are three wires from the pump. Then in BIOS Setup that pump speed can be "seen", but it will be labeled as the speed of whatever fan header you have used for it. BUT in this case you may need to set the configuration of the header used a special way.

Separately, the fan mounted on the radiator needs power and control. The ideal way to do this is to plug it into the CPU_FAN header and allow the mobo to control its speed automatically, depending on the temperature actually measured inside the CPU chip. Thus all control of the CPU cooling is done by the mobo adjusting that radiator fan speed, and the pump speed is just constant. This has another important benefit. Most mobos pay special attention to the speed of the "CPU Fan" (that is, the device plugged into the CPU_FAN header). If it gets no speed signal from that, it assumes the CPU cooling system has failed and overheating damage to the CPU is a real danger. So many such systems will send out an alarm and then quickly shut the system down without waiting from the temperature sensor inside the CPU chip to show high temps. Likewise, if it sees NO fan speed there at start-up it may refuse to let it start. So, if there is NO speed signal being delivered to Pin #3 of the mobo CPU_FAN header, this may happen.

From your post it is not clear to me exactly what connections had been made in your original system. But I suggest you do this.

1. Connect the leads from the pump unit using an adapter to a PSU power output. This will give that unit constant full speed power, but it will not send any pump speed signal to the mobo. Hence, any future failure of the pump will not be detected. There is a way to do this otherwise and set up for failure detection, so if you want that, post back here.

2. The fan on the radiator may have 3 wires coming from it ending in a female connector with 3 holes, OR it may have 4 wires and holes. The second version of the H60 system switched to a 4-pin fan. Plug that fan into the mobo CPU_FAN header. When you boot up, go into BIOS Setup to where that header is configured. If there's a choice, set it to either PWM Mode for a 4-pin fan, or Voltage Control Mode (aka DC Mode) for a 3-pin fan. Set it to the Standard profile, which means it will use its automatic fan speed control functions. IF there's an option to monitor the fan speed, make sure it is NOT set to "Ignore". Remember to SAVE and EXIT to save these settings. The system should reboot and run smoothly.

Having said all that, I find something suspicious in your post. It seems to indicate that, no matter what you do now, that fan constantly starts, runs for a few seconds, and stops and repeats while the entire system fails to start up. If that is still happening with these new connections, then there is something NOT related to the fan and pump. There might be a problem with the way your new mobo was installed. So, if this fails, post back and tell us and we can look for other things.

 

[hfffoman]

Thanks for your excellent explanation. It is always great to discuss with an expert. I think I understand. Sorry for the shortfalls in my description. I can complete the picture.

The radiator has an identical Corsair fan on each side. Each of these has a 3-way lead terminating in a female connector. These are plugged into two adjacent matching sockets on the pump module. The connectors are properly seated in place.

The pump module also has a 3-way lead directly attached, as you described. Instead of connecting this to a 3-way connector this lead has been split. One end is joined to the third position of a 3-way connector. This sits on the motherboard CPU fan header such that the sole lead connects to the 3rd pin. The other two ends are connected directly to the PSU.

I can't get to bios in this configuration. The first time I switched on, with the faulty connection to the PSU, neither the fans nor the pump turned but I got into BIOS which reported a CPU fan failure. If try to reproduce that, by unplugging the power connector between the pump and the PSU, it does nothing. This made me think the mobo had failed. However, I did subsequently get into bios when I plugged my cheap fan into the mobo CPU fan header and that is the only other occasion I have got to the bios. That suggested to me the problem was related to the cooler, not a mobo failure. It may be that the bios changed a setting when it first encountered a failed fan and that may be the reason that reproducing the initial test reproduce the results. However, I didn't save on that occasion but powered off.

Switching to the alternative configuration using the mobo header, instead of direct connections to the PSU, seems like a good idea. Can you give any suggestions how to go about it? Presumably I need to buy some special tiny junctions. Or might I be able to correct the problem by going into bios with the the other fan connected? If this is the solution, it should be possible to do it before the CPU heats up.

 

[Paperdoc]

If your system cannot work when connected as it was originally - 2 wires from the pump directly to a CPU output via some adapter, and a third to a single pin of the CPU_FAN header - then maybe you got the connection mixed up when you changed equipment. See this web page on how a 4-pin connector looks and works

https://allpinouts.org/pinouts/connectors/motherboards/motherboard-cpu-4-pin-fan/

Note that the male connector (with pins), which is what the mobo CPU_FAN header looks like, has a tongue sticking up from the base beside Pins 1-3, whjle Pin #4 has nothing beside it. No doubt the single lead from the pump to the CPU_FAN header is the one that is supposed to take the pump's speed signal to that header, and it MUST be connected to Pin #3. That's the one NOT at the end of the connector base, but at the other end of the tongue. It's right next to Pin #4 (the one not near the tongue).

Look at the colors of the wires from the pump to its connectors. They MAY use the standard color codes for 3-pin fans. As in that web site (first pic under "Notes"), the BLACK wire is Ground and goes to Pin #1. The RED wire is +DCV and goes to Pin #2. The YELLOW wire is Speed and goes to Pin #3, but in your case it is the one that is separated out and taken to the CPU_FAN header.

There's another possibility, especially if your pump DOES use those standard colors for those pins. Your system has the first two wires (Black and Red) connected in some manner to a PSU power output, and you had to re-do that connection. VERY often the output connector from the PSU used is a 4-pin female Molex connector. It is about 3/4" wide with 4 holes in a straight line, and it has four wires into it from the PSU. The two middle ones are black, and the outer ones are Red and Yellow. But here's the confusing part. On a Molex connector, the color codes are DIFFERENT from a fan wires. On the Molex, the Blacks both are Ground, so that's OK. But the YELLOW is the +12 VDC line your pump needs, and the Red of the Molex is +5 VDC. So, if you rigged the connection so that the RED wire from the PUMP connects to the Red wire of the Molex output, that is WRONG! That way the pump gets less than half voltage and will not run decently at all. The RED from the pump (on connector Pin #2) must connect to the YELLOW of the Molex output. So check that. See the photo and diagram under "Disk Drive" on this web page

https://en.wikipedia.org/wiki/Molex_connector

 

[hfffoman]

If your system cannot work when connected as it was originally - 2 wires from the pump directly to a CPU output via some adapter, and a third to a single pin of the CPU_FAN header - then maybe you got the connection mixed up when you changed equipment. See this web page on how a 4-pin connector looks and works

https://allpinouts.org/pinouts/connectors/motherboards/motherboard-cpu-4-pin-fan/

Note that the male connector (with pins), which is what the mobo CPU_FAN header looks like, has a tongue sticking up from the base beside Pins 1-3, whjle Pin #4 has nothing beside it. No doubt the single lead from the pump to the CPU_FAN header is the one that is supposed to take the pump's speed signal to that header, and it MUST be connected to Pin #3. That's the one NOT at the end of the connector base, but at the other end of the tongue. It's right next to Pin #4 (the one not near the tongue).

Look at the colors of the wires from the pump to its connectors. They MAY use the standard color codes for 3-pin fans. As in that web site (first pic under "Notes"), the BLACK wire is Ground and goes to Pin #1. The RED wire is +DCV and goes to Pin #2. The YELLOW wire is Speed and goes to Pin #3, but in your case it is the one that is separated out and taken to the CPU_FAN header.

There's another possibility, especially if your pump DOES use those standard colors for those pins. Your system has the first two wires (Black and Red) connected in some manner to a PSU power output, and you had to re-do that connection. VERY often the output connector from the PSU used is a 4-pin female Molex connector. It is about 3/4" wide with 4 holes in a straight line, and it has four wires into it from the PSU. The two middle ones are black, and the outer ones are Red and Yellow. But here's the confusing part. On a Molex connector, the color codes are DIFFERENT from a fan wires. On the Molex, the Blacks both are Ground, so that's OK. But the YELLOW is the +12 VDC line your pump needs, and the Red of the Molex is +5 VDC. So, if you rigged the connection so that the RED wire from the PUMP connects to the Red wire of the Molex output, that is WRONG! That way the pump gets less than half voltage and will not run decently at all. The RED from the pump (on connector Pin #2) must connect to the YELLOW of the Molex output. So check that. See the photo and diagram under "Disk Drive" on this web page

https://en.wikipedia.org/wiki/Molex_connector

 

[hfffoman]

Thanks for this clear explanation. I did the wiring again, hoping there might have been a mistake in the way it had been. There is no question with the wire to the CPU header. It is soldered to the 3rd position of a connector so it automatically goes in the right place over the speed pin of the header. The power leads do plug into a Molex. I plugged them again, making sure they were in yellow and black as explained. But no, sadly, I still get the on and off behaviour.

 

[Paperdoc]

So let me review some details and make sure I've got it right. It is the wires from the PUMP unit that are split up and plugged into the PSU output and soldered to Pin #3 of the CPU_FAN header, right?

Next: your post says you tried an experiment of plugging a different fan into the CPU_FAN header and it worked just fine. Two details to ask. Did you un-solder the wire coming from the pump unit to do this? AND, did you disconnect the pump from the PSU power source when you did this? I'm thinking about whether the pump unit has a fault.

Next: the Molex connector from the PSU that the pump unit is plugged into probably is one of two or three such female Molex outputs on a common set of wires. What else is plugged into the other Molex females?

 

[hfffoman]

So let me review some details and make sure I've got it right. It is the wires from the PUMP unit that are split up and plugged into the PSU output and soldered to Pin #3 of the CPU_FAN header, right?

Next: your post says you tried an experiment of plugging a different fan into the CPU_FAN header and it worked just fine. Two details to ask. Did you un-solder the wire coming from the pump unit to do this? AND, did you disconnect the pump from the PSU power source when you did this? I'm thinking about whether the pump unit has a fault.

Next: the Molex connector from the PSU that the pump unit is plugged into probably is one of two or three such female Molex outputs on a common set of wires. What else is plugged into the other Molex females?

Thanks for your post.

Your first assumption is correct.

I didn't unsolder the wires. I just unclipped the connector from the header and replaced it with the identical 3-pin connector from the cheap cooler.

There are 4 Molexes on the cable but none of the others have anything on them.

To answer your other question I repeated the test but now I can no longer get to bios using the cheap fan (regardless of whether the radiator fans are plugged into the molex). The cheap cooler also turns on and off in sync with the radiator fans. I didn't imagine it. The first time I tried it, I got to the bios. I don't understand why it should work once only. I didn't leave it on long enough to damage the CPU. I doubt it got above about 35C.

I also noticed that the fans on the GPU are doing the same thing, also in syn. They are so quiet that I didn't notice them before but I remember the tiny noise that they do make and I think they were doing the same thing from the very first time I switched it on, except the times I got to bios.

So maybe this looks like a mobo problem. If so I must eliminate other possibilities before I return it.

 

[hfffoman]

Correction: the cheap cooler has 4 wires so all 4 pins on the header are being used when I connect it.

 

[hfffoman]

I have some more information.

1. Pressing my ear to the PSU, I can hear its fan is turning on and off in sync with the others.

2. With the H60 cooler connected, the cycle is 6 seconds on, 4 seconds off. With the cheap cooler, the cycle is 4 seconds on, 4 seconds off. With the cheap cooler I can hear the faint double beep that you usually get just after powering up a PC. It is audible at the beginning of each on-cycle. I don't hear this with the H60 connected.

3. The pump light flashes on for an instant at the beginning of the on-cycle. It then stays off and turns on for half a second at the end of the on-cycle. This might suggest the pump is failing. However, since I can't boot using the cheap cooler either, it can't be just the pump.

4. I have disconnected everything except the boot drive, fan and ATX. The only difference is that when I remove the Graphics card, the radiator fans give a little kick at the end of the on-cycle, as though they are having another go at starting, just for a second. I thought at first this meant the problem was with the PSU, but it may be the motherboard is reacting differently when it's not asked to direct power to the GPU.

5. The case fans only have two leads and are connected directly to the PSU. This means the mobo case fan headers are empty. The case fans cycle on and off like all the others. Disconnecting them from the PSU makes no difference to the rest of the system.

 

[Paperdoc]

What this really sounds like is that there is a severe drain on one part of the power supply, causing it to re-start. I suspect the pump unit of the H60 system. If I read this right, in all cases that unit is plugged into the Molex PSU output, so IT may be the cause of excess power load.

Now, with a liquid cooling system. the liquid in the heat exchanger / pump module fastened to the CPU does absorb a fair amount of heat for a while at start-up, but eventually it becomes necessary for all that heated fluid to be pumped out to the radiator for cooling. So you CAN afford to do a short test with the pump unit completely DISconnected from everything. But DO connect your substitute "cheap fan" to the CPU_FAN header so that it gets a valid signal indicating the CPU cooler is working and will not generate an error. If I am right, in this configuration with no pump connected, the machine will boot properly with no trouble, but you should only allow it to run this way for a couple of minutes tops, then shut it down.

If this allows a clean boot, you know the pump unit is the problem. If not, then we look elsewhere for another source of trouble.

 

[hfffoman]

What this really sounds like is that there is a severe drain on one part of the power supply, causing it to re-start. I suspect the pump unit of the H60 system. If I read this right, in all cases that unit is plugged into the Molex PSU output, so IT may be the cause of excess power load.

Now, with a liquid cooling system. the liquid in the heat exchanger / pump module fastened to the CPU does absorb a fair amount of heat for a while at start-up, but eventually it becomes necessary for all that heated fluid to be pumped out to the radiator for cooling. So you CAN afford to do a short test with the pump unit completely DISconnected from everything. But DO connect your substitute "cheap fan" to the CPU_FAN header so that it gets a valid signal indicating the CPU cooler is working and will not generate an error. If I am right, in this configuration with no pump connected, the machine will boot properly with no trouble, but you should only allow it to run this way for a couple of minutes tops, then shut it down.

If this allows a clean boot, you know the pump unit is the problem. If not, then we look elsewhere for another source of trouble.

 

[hfffoman]

Thanks again. That suggestion was certainly what I was thinking, except that I had tried that experiment and it still wouldn't boot.

I now tried something different. I opened up another computer and connected its ATX and EAXT to the motherboard. I also ran one of its SATA power connectors to the boot drive. This PSU is a Corsair HDX620, easily powerful enough to run a MOBO plus one SSD with everything else removed. I also disconnected the suspect cooler and connected instead the cheap cooler.

So, guess what happened? Exactly the same thing. The fan cycles on and off and it won't boot. The case fan on the other computer also cycled on and off because the power supply on the other computer was directly connected to the other computer's case fan and must have been receiving on and off commands from the new motherboard.

So we know it is nothing to do with the PSU, nor the cooler, nor any of the peripherals since I disconnected them too. the case doesn't even have a mains lead. That only leaves the mobo and the CPU. Given that it looks like I am going to return the mobo (which I just bought) I might as well break the thermal seal and try a different CPU just to be sure that's not the problem.

Is there anything else I can check before returning the mobo? Although it's an old model it is brand new. It's damn annoying as I have spent hours and hours on this.

I wonder whether to get a modern mobo and CPU. The reason I went for an old model is that this is my second computer and with the spec I have, it's performance is really not bad (GTX 760, huge SATAs, one of the best CPUs you can get for £250, and 16GB of good quality 1600 DDR3). Also, if I get a newer mobo I need to get new memory as well as a new CPU. I'd appreciate any thoughts.

 

[hfffoman]

What this really sounds like is that there is a severe drain on one part of the power supply, causing it to re-start. I suspect the pump unit of the H60 system. If I read this right, in all cases that unit is plugged into the Molex PSU output, so IT may be the cause of excess power load.

Now, with a liquid cooling system. the liquid in the heat exchanger / pump module fastened to the CPU does absorb a fair amount of heat for a while at start-up, but eventually it becomes necessary for all that heated fluid to be pumped out to the radiator for cooling. So you CAN afford to do a short test with the pump unit completely DISconnected from everything. But DO connect your substitute "cheap fan" to the CPU_FAN header so that it gets a valid signal indicating the CPU cooler is working and will not generate an error. If I am right, in this configuration with no pump connected, the machine will boot properly with no trouble, but you should only allow it to run this way for a couple of minutes tops, then shut it down.

If this allows a clean boot, you know the pump unit is the problem. If not, then we look elsewhere for another source of trouble.

 

[hfffoman]

Correction: I don't have a spare CPU that will fit this socket. Both the CPU and the motherboard are new. Is it a fair assumption that the problem is the mobo?

Is it possible that the cooler has a short and this has fried the mobo? Remember that it did start to boot a couple of times?

 

[Paperdoc]

Since you're considering disassembling things for possible replacement, do an extra type of test process called "breadboarding" AFTER you have removed the mobo and major components from the case. The concept is to assemble the bare necessities for booting outside the case and place on a safe insulating surface, like a dry wooden board. Hence the name "breadboard".

1. Assemble the motherboard suitably supported and electrically isolated, the CPU chip, its cooling system (in your case, the H60 pump module on the CPU plus the radiator and its fan) and the PSU with connections to the mobo and the special CPU power cable to its socket. Do not connect any drives; do not install any memory sticks; do not connect the keyboard or mouse. If your mobo has its own built-in video output, use the back panel connector from the mobo to connect its output to your monitor for viewing, and turn on the monitor. If not - if you can only use a third-party video card in a PCIe slot - do not install that. For this step, forget the video output system and we'll deal with that soon. Connect the pump unit to a PSU output, and its separate speed signal line to the CPU_FAN header Pin #3. In your mobo manual, check the pinout diagram for the Front Panel header on the mobo (usually at bottom front corner) and locate the pair of pins that are used to connect the case's front panel ON / Off power switch. With the switch on the back of the PSU turned on, use a small screwdriver blade to short together those two On / Off header pins for just a second or so. This will trigger the start-up sequence. If it works properly, it will go through some initial self-checks and then beep a couple of times (IF your mobo has a "beeper" installed - otherwise you won't hear that) and then stop. But it will not shut off and re-start, as your system had been doing. It will just sit there with fans running. This is the symptom for a failure to POST because there is no memory installed. If you get that behavior, that's good. BUT if it goes through its cycle of re-starting constantly, that confirms the one of those components is faulty. And you appear already to have eliminated the PSU, although you could repeat that test with the second PSU system to verify. Likewise, you could disconnect the pump unit for a brief test as I outlined earlier.

2. IF that first step works without a reboot loop, and IF you have to use an add-in video card, shut down and install it now. Then connect its output to your monitor. But if you have already been using the mobo's video output socket, skip to the next step.

3. IF that first step gets you a partial success (no reboot loop), you proceed with the next step. From here on, the process is to shut down and add ONE component, then try again. In this step, install ONE memory stick and try to boot. If this works, it may complete the POST sequence to the point of telling you it cannot boot because there is no bootable drive to use.

4. Next step is to install all of your memory and boot again. Success will look like the previous step.

5. Next items to add are the keyboard and mouse.

6. IF you have an add-in video card BUT have been working all this time using the mobo's video output instead, this is the time to install the vid card and move your monitor connection to that. IF you do this, it may not show you anything on the monitor. In that case, you have to shut down and re-connect the monitor to the mobo video output, then boot into BIOS Setup and tell it to use the add-on card for video, and not the mobo system. Then you shut down, move the monitor cable back to the card, and do your boot attempt. If this still succeeds, you can proceed further.

7. Next comes your boot drive device. I don't remember whether that device has had Windows installed on it successfully or not yet. If not, you'll still get the message about no bootable device. If yes, it may actually boot into Windows.

8. Continue on adding one component at a time if you continue to get successful boot-ups.

This is an organized way to identify the faulty module causing boot failure. Of you actually get through all of your system able to boot cleanly without those reboot loops (except for no real bootable drive) you can be confident your hardware is NOT the problem. Let us know what you find. If all this says you have no hardware problem outside the case, then there are a couple other things to check. But if you did encounter failure in the process, you will now much more about which item is the likely culprit.

 

[hfffoman]

Since you're considering disassembling things for possible replacement, do an extra type of test process called "breadboarding" AFTER you have removed the mobo and major components from the case. The concept is to assemble the bare necessities for booting outside the case and place on a safe insulating surface, like a dry wooden board. Hence the name "breadboard".

1. Assemble the motherboard suitably supported and electrically isolated, the CPU chip, its cooling system (in your case, the H60 pump module on the CPU plus the radiator and its fan) and the PSU with connections to the mobo and the special CPU power cable to its socket. Do not connect any drives; do not install any memory sticks; do not connect the keyboard or mouse. If your mobo has its own built-in video output, use the back panel connector from the mobo to connect its output to your monitor for viewing, and turn on the monitor. If not - if you can only use a third-party video card in a PCIe slot - do not install that. For this step, forget the video output system and we'll deal with that soon. Connect the pump unit to a PSU output, and its separate speed signal line to the CPU_FAN header Pin #3. In your mobo manual, check the pinout diagram for the Front Panel header on the mobo (usually at bottom front corner) and locate the pair of pins that are used to connect the case's front panel ON / Off power switch. With the switch on the back of the PSU turned on, use a small screwdriver blade to short together those two On / Off header pins for just a second or so. This will trigger the start-up sequence. If it works properly, it will go through some initial self-checks and then beep a couple of times (IF your mobo has a "beeper" installed - otherwise you won't hear that) and then stop. But it will not shut off and re-start, as your system had been doing. It will just sit there with fans running. This is the symptom for a failure to POST because there is no memory installed. If you get that behavior, that's good. BUT if it goes through its cycle of re-starting constantly, that confirms the one of those components is faulty. And you appear already to have eliminated the PSU, although you could repeat that test with the second PSU system to verify. Likewise, you could disconnect the pump unit for a brief test as I outlined earlier.

2. IF that first step works without a reboot loop, and IF you have to use an add-in video card, shut down and install it now. Then connect its output to your monitor. But if you have already been using the mobo's video output socket, skip to the next step.

3. IF that first step gets you a partial success (no reboot loop), you proceed with the next step. From here on, the process is to shut down and add ONE component, then try again. In this step, install ONE memory stick and try to boot. If this works, it may complete the POST sequence to the point of telling you it cannot boot because there is no bootable drive to use.

4. Next step is to install all of your memory and boot again. Success will look like the previous step.

5. Next items to add are the keyboard and mouse.

6. IF you have an add-in video card BUT have been working all this time using the mobo's video output instead, this is the time to install the vid card and move your monitor connection to that. IF you do this, it may not show you anything on the monitor. In that case, you have to shut down and re-connect the monitor to the mobo video output, then boot into BIOS Setup and tell it to use the add-on card for video, and not the mobo system. Then you shut down, move the monitor cable back to the card, and do your boot attempt. If this still succeeds, you can proceed further.

7. Next comes your boot drive device. I don't remember whether that device has had Windows installed on it successfully or not yet. If not, you'll still get the message about no bootable device. If yes, it may actually boot into Windows.

8. Continue on adding one component at a time if you continue to get successful boot-ups.

This is an organized way to identify the faulty module causing boot failure. Of you actually get through all of your system able to boot cleanly without those reboot loops (except for no real bootable drive) you can be confident your hardware is NOT the problem. Let us know what you find. If all this says you have no hardware problem outside the case, then there are a couple other things to check. But if you did encounter failure in the process, you will now much more about which item is the likely culprit.

 

[hfffoman]

The results of this test are amazing and completely unexpected.
I am using the onboard display output.

First test, fan runs steadily.
Add a single RAM module to slot A1 - power cycles on and off
Add a single RAM module to slot A2 - power cycles on and off
Add a single RAM module to slot B1 - fans run steadily, the pump is working, but doesn't beep or display anything on the screen
Add a single RAM module to slot B1 - fans run steadily, the pump is working, but doesn't beep or display anything on the screen
Add RAM to B1 and B2, ditto
Using a different RAM module, ditto

That looks to me like a duff mobo, and most importantly, it confirms the PSU and cooler are OK. Thanks for all your help. Any final thoughts?

 

[hfffoman]

If your system cannot work when connected as it was originally - 2 wires from the pump directly to a CPU output via some adapter, and a third to a single pin of the CPU_FAN header - then maybe you got the connection mixed up when you changed equipment. See this web page on how a 4-pin connector looks and works

https://allpinouts.org/pinouts/connectors/motherboards/motherboard-cpu-4-pin-fan/

Note that the male connector (with pins), which is what the mobo CPU_FAN header looks like, has a tongue sticking up from the base beside Pins 1-3, whjle Pin #4 has nothing beside it. No doubt the single lead from the pump to the CPU_FAN header is the one that is supposed to take the pump's speed signal to that header, and it MUST be connected to Pin #3. That's the one NOT at the end of the connector base, but at the other end of the tongue. It's right next to Pin #4 (the one not near the tongue).

Look at the colors of the wires from the pump to its connectors. They MAY use the standard color codes for 3-pin fans. As in that web site (first pic under "Notes"), the BLACK wire is Ground and goes to Pin #1. The RED wire is +DCV and goes to Pin #2. The YELLOW wire is Speed and goes to Pin #3, but in your case it is the one that is separated out and taken to the CPU_FAN header.

There's another possibility, especially if your pump DOES use those standard colors for those pins. Your system has the first two wires (Black and Red) connected in some manner to a PSU power output, and you had to re-do that connection. VERY often the output connector from the PSU used is a 4-pin female Molex connector. It is about 3/4" wide with 4 holes in a straight line, and it has four wires into it from the PSU. The two middle ones are black, and the outer ones are Red and Yellow. But here's the confusing part. On a Molex connector, the color codes are DIFFERENT from a fan wires. On the Molex, the Blacks both are Ground, so that's OK. But the YELLOW is the +12 VDC line your pump needs, and the Red of the Molex is +5 VDC. So, if you rigged the connection so that the RED wire from the PUMP connects to the Red wire of the Molex output, that is WRONG! That way the pump gets less than half voltage and will not run decently at all. The RED from the pump (on connector Pin #2) must connect to the YELLOW of the Molex output. So check that. See the photo and diagram under "Disk Drive" on this web page

https://en.wikipedia.org/wiki/Molex_connector

 

[hfffoman]

Further update: I took the cooler off and there was one very slightly bent pin. THese sockets are far to fragile. All I had done was seat the CPU once, carefully, without ever touching the pins. I managed to straighten the pin and now I can get the fan to run with all four RAM sockets installed. However, it will not beep or send any signal to the display.

 

[Paperdoc]

Very interesting info. The bent pin might have caused the issue, but not clear yet. Two things I can suggest.

1. Whether or not you hear any "beep" signal depends on whether your mobo has a any mini-speaker mounted on it. Some do, and some have only an empty pair of pins where you can plug one in. Such a device is not really used as a speaker for anything beyond allowing those diagnostic "beep' signals to be heard during the POST process. So, unless you know that you do have such a mini-speaker, the lack of a "beep" means nothing.

2. Try a mobo reset to default conditions, in case some BIOS setting somehow has been corrupted and prevents video output. Follow this sequence.

(a) Find on the mobo the BIOS backup battery. This is about the size of a quarter, slipped into a plastic holder. NOTE which side of it is up so you can put it back the same way.
(b) Look close to the battery for the Reset pins. Usually this is a set of 3 pins in a row, with a small plastic jumper slipped down over two of them. SOMETIMES, however, it is a pushbutton.
(c) Unplug the system from the wall so there's no external power.
(d) Remove the battery.
(e) Move the Reset jumper to the alternate pin pair and leave it on for 10 - 15 sec. Them move it back to its original position. IF you have a pushbutton, hold it down for 10 - 15 sec.
(f) Re-install the battery. Plug the PSU into the wall.
(g) Boot and see whether this has made any difference.

Doing a Reset like this will restore most BIOS settings to factory default. It is not yet complete, but it might get you to the point of being able to see video output. If you still get nothing, I'd certainly suspect mobo failure.

If you DO get video output to your monitor, proceed though the breadboarding process, adding one component at a time and testing. If you get past the connection of keyboard and mouse, you can enter BIOS Setup. In there, go to the last screens where you can find the options to Load Default Settings or Load Optimized Settings. Use one of those options to ensure that a reliable basic and COMPLETE set of defaults is loaded, then SAVE and EXIT. This will save those settings and reboot. Then you can continue the process.

 

[hfffoman]

If your system cannot work when connected as it was originally - 2 wires from the pump directly to a CPU output via some adapter, and a third to a single pin of the CPU_FAN header - then maybe you got the connection mixed up when you changed equipment. See this web page on how a 4-pin connector looks and works

https://allpinouts.org/pinouts/connectors/motherboards/motherboard-cpu-4-pin-fan/

Note that the male connector (with pins), which is what the mobo CPU_FAN header looks like, has a tongue sticking up from the base beside Pins 1-3, whjle Pin #4 has nothing beside it. No doubt the single lead from the pump to the CPU_FAN header is the one that is supposed to take the pump's speed signal to that header, and it MUST be connected to Pin #3. That's the one NOT at the end of the connector base, but at the other end of the tongue. It's right next to Pin #4 (the one not near the tongue).

Look at the colors of the wires from the pump to its connectors. They MAY use the standard color codes for 3-pin fans. As in that web site (first pic under "Notes"), the BLACK wire is Ground and goes to Pin #1. The RED wire is +DCV and goes to Pin #2. The YELLOW wire is Speed and goes to Pin #3, but in your case it is the one that is separated out and taken to the CPU_FAN header.

There's another possibility, especially if your pump DOES use those standard colors for those pins. Your system has the first two wires (Black and Red) connected in some manner to a PSU power output, and you had to re-do that connection. VERY often the output connector from the PSU used is a 4-pin female Molex connector. It is about 3/4" wide with 4 holes in a straight line, and it has four wires into it from the PSU. The two middle ones are black, and the outer ones are Red and Yellow. But here's the confusing part. On a Molex connector, the color codes are DIFFERENT from a fan wires. On the Molex, the Blacks both are Ground, so that's OK. But the YELLOW is the +12 VDC line your pump needs, and the Red of the Molex is +5 VDC. So, if you rigged the connection so that the RED wire from the PUMP connects to the Red wire of the Molex output, that is WRONG! That way the pump gets less than half voltage and will not run decently at all. The RED from the pump (on connector Pin #2) must connect to the YELLOW of the Molex output. So check that. See the photo and diagram under "Disk Drive" on this web page

https://en.wikipedia.org/wiki/Molex_connector

 

[hfffoman]

Thanks again. I am hoping this is now solved. The bent pin was definitely the main problem. It is a miracle that I managed to fix it considering that in the process I bent several more pins and had to find the thinnest needle in the world together with the strongest magnifying glass and a huge amount of patience to fix it. It really is an awkward operation, despite the youtube videos that make it look simple (I don't think they are using the 1150, which is particularly nasty).

Then it turns out my processor doesn't deliver onboard video so a lot of my tests were false. However, the first couple of times I hooked it up with the pins straightened and using the GPU, it didn't work. Then I swapped the monitor and it did work, so I thought there must be a problem with the monitor. But I tested the monitor on another computer and it does work so I haven't got to the bottom of this. Anyway I am now booting and getting into Windows with the machine fully rebuilt. It crashes after a few seconds and I am hoping that's because the Windows installation is from my previous mobo and it just needs a reinstall. I will do that tomorrow and post back how it goes.

 

[Paperdoc]

That is very encouraging. Regarding the latest problem, I think there is a really good chance that a new clean Windows install will fix that. Normally it does NOT work to just try to boot and run from a version of Windows installed on a different system with different hardware. Although we often don't think of it, the motherboard itself contains many devices that need the correct device drivers installed in Windows. A fresh install on the actual machine in use, with all of its devices present, is the best way to have everything installed correctly.

I have one important suggestion for that. I do not recall what storage devices you have - SSD, HDD, etc. There is a special feature of Windows designed to recover automatically from corruption of its system files at some future time, and it involves placing backup copies of them in a safe place. The safest place is a second storage device, if you have one. But what so many have discovered is that, if you do it this way, at every boot-up Windows will search for those backups and a couple related files and, if the second storage device containing them is not present or malfunctioning for any reason, it will not boot! There is a simple way to prevent that problem that you can do at the time of your first Install process, which is what you are about to do. You can force the Install process to place all those backups on the SAME storage device that your main OS is installed; that way they are always found,, as long as your boot device is working. This is not quite as secure as having the backups on a completely separate device, but it does prevent that problem. IF you want to do this (and of course IF you actually have a second storage device), follow this process.

1. Connect only your main storage device that Windows will be Installed on (an SSD or a HDD) and the source that holds the Install system - an optical drive or a USB memory stick. Have NO other storage devices connected.
2. Run the Install process normally. Since its survey of hardware will find only one storage device, it will Install to that AND place the backups etc. also on that device.
3. After the Installation is complete and your system is running, remove the Install source and shut down. NOW connect any other storage devices to your machine, close up and boot.
4. Windows will find all those new storage devices. However, for the most part they each will not be available for Windows to use, because they contain nothing. For a HDD or anything similar, you must go into Disk Management and find each, in turn (if you have more than one), in the lower right panel. RIGHT-click on it and choose to Create a New Simple Volume. For the most part, the default settings of this process are correct. One thing you might want to change: since you will not be BOOTING from this device (you already have Windows installed elsewhere), the device does not need to be made Bootable. In the case of HDD units, the default choice for the Format operation is to do a Full Format, which will test every Sector of the HDD unit. This takes a long time, so just be patient. Once you have set your Volume Creation options, just let it take the time it needs. When all added storage devices have been so prepared, back out of Disk Management and reboot. They all should show up now in My Computer ready for use.

 

[hfffoman]

And finally, it stops working again. I rebuilt the computer and it turned on ok the first 4 or 5 times and the next time, it just went back to turning on and off.