Question DeepCool LT720 randomly decreases pump speed ?

Bluefish42

Reputable
Mar 27, 2019
9
0
4,510
The pump starts making noise at less than 3000 RPMs. I cannot get the pump to stay at 3000. After an hour or so, it will decrease to 2200. Fan control software, and BIOS, are set to 100% power at all times. When the RPMs decrease to 2200, the fan monitoring software still claims 100% power is being delivered, while the RPMs are correctly reported as 2200. I can reset the RPM back to 3000 by going into the fan control software and turning the pump off and on. Obviously this is an annoying workaround.

The motherboard is Gigabyte Aorus B650M. The pump is connected to CPU_OPT. I have tried both control modes - Voltage and PWM. These random slowdowns happen in both modes.

Does this sound like a motherboard problem, or a pump problem? Any possible solution?
 
Your MoBo has dedicated header for pump: SYS_FAN4_PUMP (listed in holy bible of PCs, page 19). Connect your AIO pump wire to that, instead CPU_OPT. Look if your issue goes away.

Dedicated pump header usually runs AIO pump at 100% regardless the temps within system. CPU_FAN and CPU_OPT, in the other hand, are forced to follow CPU temp and fan profile in them isn't that flexible. Nor i have seen either of the two headers capable of running connected hardware at 100% at all times.
 
I agree with Aeacus, but it is possible that OP actually set the CPU_OPT header to a fixed full speed, rather than letting the header change that speed. Not clear from the post.

OP reports that the pump starts at full speed, but after an hour it reduces to less even though the header claims it is trying to get full speed. (That means it is already sending the pump full 12 VDC voltage and full 100% PWM signal. It can't do more than that to force speed higher.) Further, this pump speed can be pushed back to full speed if it is shut off and then re-stated.

My suspicion is one of two things:
(a) Some of the cooilng fluid has been lost and the system has an air bubble that moves around until it lodges inthe pump and causes cavitation, leading to poor fluid flow and increased workload in the pump cavity, thus lowered speed. The shut-down-restart sequence may move that bubble out of the pump temporarily.
(b) The pump bearings are wearing out so that, when it warms up, they cause excess friction slowing the pump down. In this case, though, I would NOT expect a brief shut-down to let it cool enough to resolve the issue temporarily.

For either cause, the solution will be to repalce the pump / radiator closed system completely. In almost all AIO systems that set is sealed and you cannot add fluid to replace what has been lost; nor can you replace the pump unit only.
 
Thank you both for your help. The header is set to provide 100% power at all times, and claims, though monitoring software, to be doing so. The reduction in RPMs is not associated with any reported decrease in power.

The CPU_OPT head is documented in the manual as being for pumps, same as SYS4_PUMP:

(5) SYS_FAN4_PUMP (System Fan/Water Cooling Pump Header
(6) CPU_OPT (CPU Fan/Water Cooling Pump Header


I will try switching the pump to SYS_FAN4_PUMP. Otherwise I will pursue a replacement for the cooler!
 
Deepcool makes you jump through hoops to return an item, and it would have given me a week of downtime. So I bought a new one from Amazon and returned the old one. New pump is connected as the previous one was, and is holding 3000 RPM steadily and quietly.

Before I replaced it, the old one began showing different symptoms as it settled into its failure mode. No longer did it go from 3000 to 2200 RPMs and begin making noise. Instead, it would wildly swing from 1000 to 4000 every few seconds, and make bubbling noises every few seconds. Turning it off and on no longer gave a temporary fix.
 
another update, the replacement started showing the same symptoms. The RPMs would reduce, and the pump would start making noise. Needed to stop/start the fan through mobo control software every time. Returned and got a different brand, which seems fine so far.

Further google searches indicate my issue is not unique to me. This deepcool model has this issues, maybe related to motherboard compatibility, or maybe it's just the cooler. Anyway, never again will I get a deepcool.

The new one is a coolermaster, which goes up and down in its RPMs in response to the mobo controller. Which is fine with me. I guess it's a deepcool thing to require a constant RPM for the pump.
 
Glad to hear you have things working. BUT I see something wrong in your post.

In an AIO system, TWO items impact the rate of heat removal from the CPU: the speed of the pump (how fast the heat can be moved from CPU to rad), and the speed of the rad fans (how fast the heat can be moved from rad to room air). As a control system changes the speed signals to these, each item has a different response time and settling time. So if BOTH items' speeds are being altered in response to a CPU temperature change, they "chase" each other. The one that responds more quickly does its thing, then the other tries to do the same job and causes the total result to be too much. So they both start to reverse that to get to the target, and the result of their combined actions again over-shoots the target in the opposite direction. This just keeps getting worse.

The solution is to have only ONE of these items change its speed according to the CPU temperature, while the other is set to a fixed speed that provides the right cooling rate overall. By far the most common way to do this is to have the PUMP run at a fixed full speed all the time, and let the RAD FANS control the temperature via varying their speed. (Fixed FULL speed of the pump means the system will never be limited by poor circulation rate even at high workloads).

In most Cooler Master AIO systems the instructions say you have two cables for cooling power - one each from the pump unit and the rad fans. (Several rad fans may be connected together by a Splitter to a single mobo header.) This does not include cables for lights in the pump or fan frames. The one for the PUMP will have only THREE holes in its connector. This should go to the CPU_FAN or CPU_OPT or PUMP header of your mobo. Since you have one, the PUMP header is the best choice. Then you MUST set the header in its configuration options to behave as a PUMP header, and NOT as a FAN header. That guarantees that the pump will always receive signals to run at full speed, AND that its speed speed signal is monitored for possible failure. The RAD FANS should be plugged into the CPU_FAN header, and it should be set to use the new PWM Mode of control. The default system will control their speed automatically according to the CPU's internal temperature sensor, plus their speed also will be monitored for failure.

Alternatively you can connect the PUMP to CPU_FAN or CPU_OPT (these two do the same things) but you MUST set that header to use the newer PWM Mode of control, and NOT DC or Voltage. Because of the wiring design of the pump, even though those headers WILL try to change the speed of what is connected, it will NOT be able to do what the PWM signals say and WILL run full speed as intended. IF you make your connection that way, then the RAD FANS are to be plugged into the other CPU_xxx header. They ARE designed to work properly with a PWM signal set and they WILL have their speeds controlled by the header according to the CPU temperature.
 
Last edited:
Thanks for the good info. It makes sense that only one thing should need to change, when there are two parts of the system. Holding one constant will be more stable and predictable. In fact, after i posted the previous update, I did go ahead and change the power setting to the pump to be 100% at all times, and there it has stayed. 2700 RPMs for the coolermaster, as compared to 3000 for the DeepCool. But the coolermaster is able to stay there indefinitely, and it doesn't make noise in any case.
 
Thanks! Since the DeepCool attempts to stay at a constant maximum speed regardless of the mobo's intended and reported power feed, and regardless of whether it's set to Voltage or PWM, I guess this bug is in the pump firmware, as the pump itself is the only thing capable of affecting its own speed. Or maybe I was extremely unlucky and got two bubbly trapped air coolers in a row.
 
Well, not quite. The key here is that pumps almost always are designed and wired the same as older 3-pin fans. The speed of those can be controlled ONLY by varying the Voltage supplied to it on Pin #2. Newer 4-pin fans work differently. They ALWAYS receive from Pin #2 the full +12 VDC supply, and then also receive the new PWM control signal from the added Pin #4. This fan type has a small chip inside that uses that PWM signal to modify the flow of current from the fixed 12 VDC supply line through the windings to achieve speed control.

Most mobo fan headers now have a configuration option to allow you to specify whether it behaves as an older 3-pin fan in Voltage Control Mode (aka DC Mode), or as a newer 4-pin PWM Mode header. If you plug a 3-pin fan OR pump into a header that is set to use the 4-pin PWM Mode, then that fan will get a full constant 12 VDC from Pin #2 and always run full speed. It does not receive the PWM signal from Pin #4 and it has no chip so it could not use that signal, anyway. IF you set that header instead to use the older Voltage Control Mode, then it WILL change the Voltage sent to the fan or pump and change its speed. That is exactly what NOT to do for an AIO system pump. So if the header MODE is set that way, then you must also set it to Manual Profile and a fixed full-speed (12 VDC) output. But the simpler way is to use that quirk of design and set the header MODE to PWM, so the fan can NOT have its speed reduced.

By the way, many such headers have a "Auto" Mode option that supposedly detects which thing is plugged in and sets itself. BUT for this situation it does NOT. It tests and finds that the PUMP cannot be controlled by PWM but CAN have is speed changed in Voltage Control Mode, so it sets that way to give it ability to control. Exactly what is NOT desired for a pump. So ensure the header is set to PWM.
 
Hello, I just registered an account to reply you and I an from China. I meet the same issue (3100 rpm to 2800 rpm) and someone else in the China forum meets the same (his pump 3100 rpm to 2400 rpm). It seems to be a normal issue happens in recent LT520s because I just bought it on Nov.12
 
The pump will decrease its speed in both DC mode and PWM mode. The pump uses 3-pin connecter which means in PWM mode it receives full voltage and should not drop speed. It must be something wrong in the pump. In DC mode, even I set to 60%, the pump still keeps in 3100 full rpm, but will drop to 2800 after several hours, bringing big niose.
Well, not quite. The key here is that pumps almost always are designed and wired the same as older 3-pin fans. The speed of those can be controlled ONLY by varying the Voltage supplied to it on Pin #2. Newer 4-pin fans work differently. They ALWAYS receive from Pin #2 the full +12 VDC supply, and then also receive the new PWM control signal from the added Pin #4. This fan type has a small chip inside that uses that PWM signal to modify the flow of current from the fixed 12 VDC supply line through the windings to achieve speed control.

Most mobo fan headers now have a configuration option to allow you to specify whether it behaves as an older 3-pin fan in Voltage Control Mode (aka DC Mode), or as a newer 4-pin PWM Mode header. If you plug a 3-pin fan OR pump into a header that is set to use the 4-pin PWM Mode, then that fan will get a full constant 12 VDC from Pin #2 and always run full speed. It does not receive the PWM signal from Pin #4 and it has no chip so it could not use that signal, anyway. IF you set that header instead to use the older Voltage Control Mode, then it WILL change the Voltage sent to the fan or pump and change its speed. That is exactly what NOT to do for an AIO system pump. So if the header MODE is set that way, then you must also set it to Manual Profile and a fixed full-speed (12 VDC) output. But the simpler way is to use that quirk of design and set the header MODE to PWM, so the fan can NOT have its speed reduced.

By the way, many such headers have a "Auto" Mode option that supposedly detects which thing is plugged in and sets itself. BUT for this situation it does NOT. It tests and finds that the PUMP cannot be controlled by PWM but CAN have is speed changed in Voltage Control Mode, so it sets that way to give it ability to control. Exactly what is NOT desired for a pump. So ensure the header is set to PWM.
 
  • Like
Reactions: Paperdoc
alcatrazeone, you are right. As long as the header is configured to deliver only full 12 VDC power to the pump, it can only run full speed. So when it does NOT, there is a problem with the PUMP, not the header. Now, for any FAN when it runs too slow and makes noise, the problem almost always is that the bearings are worn and the fan needs to be replaced. For a PUMP liquid cooling system that also can happen. However, the much more common the problem is that an air bubble is trapped in the pump causing "cavitation". That is a mixture of air and fluid splashing around inside the pump so it cannot cause smooth flow of fluid out. This causes excess noise and also slows the pump down due to extra work it is doing for no useful purpose. And, of course, poor cooling because the heated fluid is not moving out through the radiator.

An air bubble that causes this may be because the arrangement of components in the fluid system is poor, allowing any air in the loop to accumulate in the pump as a high point in the loop. As a temporary solution, tilting the entire system around while running may allow that bubble to be expelled from the pump and move to another point where it causes less problems. However, it may move back to the pump again. A better solution is to re-mount the components of the system so the high point is well above the pump and any small bubble does not stay in the pump.

A more common cause of this, especially in AIO systems several years old, is that very small leaks in the system have allowed significant quantities of the original fluid to leave, creating a large air void. Most AIO systems are sealed so that you cannot open them and replace the missing fluid, then re-seal. For that, the only real solution is to replace the entire AIO system. SOME such systems, however, are NOT AIO ones. That is, they are custom fluid cooling loops assembled from several components, including an open-topped fluid reservoir that you CAN replenish if needed.
 
Hello, I just registered an account to reply you and I an from China. I meet the same issue (3100 rpm to 2800 rpm) and someone else in the China forum meets the same (his pump 3100 rpm to 2400 rpm). It seems to be a normal issue happens in recent LT520s because I just bought it on Nov.12
Thanks for the reply. It happened in two pumps in a row for me, so I think their whole batch is defective. Probably they don't like talking about it, and will still make you jump through hoops to get a replacement.
 
I recently purchased an LT720 - same issue for me. I didn't want to rip it out of my rig so I am using a combo of HWiNFO64, Fan Control and a custom batch job to manage it on my Windows 11 install.

I found that Fan Control could spin the AIO pump down and up which I also noticed would essentially "reset" the AIO pumps back up to 100% and the silence I wanted. (This fan needs to be set to Voltage mode in bios)

Also in Fan Control is the option to use a custom trigger that could set the fan/pump to any value at any specified temperature. So if I set the fan/pump to 0% at 10 degrees and 100% at 50 degrees, then I could force it spin down and back up as long as I could set those values.

Fan Control also offers a File-based sensor, so I can point my custom trigger to look at the file and have a way to reset the AIO pump.

HWiNFO64 is what controls when to update the file and values that Fan Control is looking at. I have an alert set on the AIO Fans (CPU_OPT is where i have mine plugged into on the mobo) so that if it drops below 2600RPM I want it to log it and execute a custom batch job (running as a minimized shortcut) that just writes out the number 10 to the file, waits 10 seconds, and then writes 50 back out to the file.

So far it's been working pretty well - whenever the AIO pump drops and starts whining my batch job kicks off and within 10 seconds it spins down to 0, and then I assume back up to full speed but in total silence.
 
  • Like
Reactions: Deecope
I recently purchased an LT720 - same issue for me. I didn't want to rip it out of my rig so I am using a combo of HWiNFO64, Fan Control and a custom batch job to manage it on my Windows 11 install.

I found that Fan Control could spin the AIO pump down and up which I also noticed would essentially "reset" the AIO pumps back up to 100% and the silence I wanted. (This fan needs to be set to Voltage mode in bios)

Also in Fan Control is the option to use a custom trigger that could set the fan/pump to any value at any specified temperature. So if I set the fan/pump to 0% at 10 degrees and 100% at 50 degrees, then I could force it spin down and back up as long as I could set those values.

Fan Control also offers a File-based sensor, so I can point my custom trigger to look at the file and have a way to reset the AIO pump.

HWiNFO64 is what controls when to update the file and values that Fan Control is looking at. I have an alert set on the AIO Fans (CPU_OPT is where i have mine plugged into on the mobo) so that if it drops below 2600RPM I want it to log it and execute a custom batch job (running as a minimized shortcut) that just writes out the number 10 to the file, waits 10 seconds, and then writes 50 back out to the file.

So far it's been working pretty well - whenever the AIO pump drops and starts whining my batch job kicks off and within 10 seconds it spins down to 0, and then I assume back up to full speed but in total silence.
Is there any way I can contact you to understand your steps better? I have been trying to use your fix but I do not know how to do it.
 
Is there any way I can contact you to understand your steps better? I have been trying to use your fix but I do not know how to do it.
I took some screenshots to try to explain it better.


1. In HWiNFO64 open up settings
2. Find your AIO pump (for me I plugged into CPU_OPT header on my mobo) and add an alert. I want to detect when the pump drops below 2700 RPM (normally it runs around 3100), as this is when it makes noise. When it sees it drop RPMs then I want it to log to a file and run a program (see below)
3. In Fan Control I want a custom File sensor called lt720.sensor.
4. I created a custom Trigger Curve using the File sensor from (3), that when it sees 10 degrees it will drop the pump to 0%, and at 50 degrees will set it back up to 100%.
5. Finally I set my AIO pump to use this custom curve.

In order to control the file sensor I setup created a batch file in textpad to set the sensor to 10, wait 15 seconds and then set it back up to 50 - which looks like:

Code:
@echo off
>C:\Users\Rob\Documents\AIOFanControl\lt720.sensor echo 10
timeout /t 15 /nobreak >nul
>C:\Users\Rob\Documents\AIOFanControl\lt720.sensor echo 50

This window will pop up so I also created a shortcut to run minimized. This is what is being set in Step 2 above.


And that is all there is to it. It's a bit of a hack but it does the job although I wonder if this pump will last the 5 years it's warrantied for.
 
** Update - i got tired of the pop up window so changed the shortcut to a vbs script instead.

reset-lt720-pump.vbs :

Code:
Set WshShell = CreateObject("WScript.Shell")
WshShell.Run chr(34) & "C:\Users\Rob\Documents\AIOFanControl\reset-lt720-pump.bat" & Chr(34), 0
Set WshShell = Nothing

Now it all happens in the background without any annoying window popups.
 
I'm experiencing this same issue. Already had the first pump returned, and the replacement does the same thing. No sound at 3100 RPM, but after some minutes the RPM drops to 2300 and makes the sound.

Fan Control is unable to control the pump speed for me, so I can't use the hack above ^. I've tried, but It just says "The control doesn't respond or is being overriden" and enabling Force apply makes no difference.

Has anyone found any other solution? About to just return this and get a different cooler.
 
I would suggest return and replace with another brand. The hack that borictor above devised is a slightly complicated work-around for a pump problem. Of you don't get all its details right it may not work for you. Meanwhile, several people have reported in this thread the same problem with that particular AIO system. It is not clear whether the system has a mechanical problem in the pump, or is especially sensitive to component placement and development or air bubbles. Some above have reported success by replacing with an AIO system from another maker, and I will note that I have seen elsewhere in forums few such problems reported from other similar systems.
 
I have the same issues as well with my Deepcool LT520. Bought 2 weeks ago. runs at 3100 RPM connected to CPU_OPT then reduces to 2400 RPM after a while making a loud humming noise.