Bug: fan Custom curve graph ignores Point 5

Couldn't find any sort of bug-reporting related email address or bugtracker, so I'm posting this here. The support link in LINK leads here, too, so I guess it's okay.

When creating a custom fan curve for my H80i, the graph visualizing the curve ignores "Point 5" and cuts off at the RPM of "Point 4".

The actual fan control seems to honor this invisible point, though, and the fan spins at the correct RPM.

http://i.imgur.com/8CtPsOV.png

Bug existed ever since I started using Corsair LINK back in version 2.something. It's present right now in version 3.2.5676 on Win7 64bit.

I think this bug still exists. I have an H100i GTX and I noticed that my fans would rev up to max pwm % at the last control point (Point 6 for my cooler).

http://i.imgur.com/bhygT0L.jpg?1

This behavior seems to also present itself for all the other non-custom fan profiles as I initially noticed this behavior when had the fans set to quiet mode. It would mysteriously rev up to max at 40 deg C. I have this issue with the stock fans, non-stock fans, and have repeated results with another H100i GTX which I got to see if it is a hardware issue. I am using windows 10 pro with corsair link 3.2.5695.

You may want to set Point 6 to 34 or 35.

I have already tried point 6 at 35 deg C before even uploading that picture, using initial custom curve settings from a fresh CL install. It exhibits the same behavior where the fans revs up to max rpm at 35 deg C instead of the 32 in that pic.

Actually, now that I look at it, the other issue is that you have all these points set at 20% when their minimum is supposed to be 40%.

You seem to be quite fixated at that one fan curve picture that I have posted, when I have mentioned that it affects ALL fan curves including the non adjustable ones such as quiet, balanced, and performance modes. If it takes more images to convince you that there is an issue then here are some more:

http://i.imgur.com/QxzCVp7.png?1http://i.imgur.com/pNxGcin.png?1

Also if the minimum is supposed to be 40% then why does quiet mode appear to go below that? I don't have the percentages but just from looking at the two different curves it clearly is below the 40% line from the custom curve. I also do not see why being below 40% should cause issues. In fact the current curve I have now seems to work fine as long as I don't make point 6 a contributing part of the curve and that curve has half of its points below 40%.

http://i.imgur.com/hHeeGkk.png?1

I also do not see why being below 40% should cause issues.

Some fans below that % are subject to stalling and if that happens they would go to full speed to synch again.

Some fans below that % are subject to stalling and if that happens they would go to full speed to synch again.

Based on my current setup the fans just seem to stall if I push the pwm % low enough.

http://i.imgur.com/lLETfY0.png?1

Anyways, I think that is besides the point with the current issue that I am having, which is the last temperature point not being followed and causing the fans to rev up to max RPM.

Actually, now that I look at it, the other issue is that you have all these points set at 20% when their minimum is supposed to be 40%.

Why is the minimum 40% please? Below you can see for the H80iGT 25% to 100% works well.

I would like to use PWM 0% to totally stop the fan.

There are some custom curve shenanigans with Link 3.2.5676 right now and a new version is en route. I'd ask that you wait for the next version and if that doesn't solve the problem, ping us again and we'll file the bug.

I'm on 3.2.5695 now and Point 5 still doesn't seem to be displayed. However, I cannot confirm the other wonkiness others have meanwhile pointed out in this thread.

I would like to use PWM 0% to totally stop the fan.

That's something I've been missing on my H80i, too, and I've yet to see if the Commander Mini can stop 3-pin fans (since it's still in the mail). But it's more of an inconvenience than a dealbreaker, I can live with it.

Thank you for the chart dude !

The minimum is 40% to prevent fan stalling.

40% makes sense for CoolIT 3-pin fans, but from the plot the Asetek firmware seems to allow a low limit of 25% and I suspect the same will be the case for 4-pin PWM CoolIT.

Honestly I suspect it's probably a global minimum implemented as a safeguard, since we can't predict how non-Corsair fans will react to low PWM signals/voltages.

I suspect the same and my opinion this should be changed. Some time ago I posted the following which shows a 5% PWM works for some fans so feel is low limit of 40% is inappropriate and should be changed.

With PWM fans the CLCC firmware allows fans to be set below 40% and it's CL that has the 40% lower limit. If you write your own CL control software you can set them to 12 of 255 (~5%). I have only managed to do this for PWM mode but it should be simple enough to change the firmware so lower speeds worked for custom curves should Corsair choose to do so.

 

Just because 5% works for some fans doesn't mean it works for all fans. I do think we can probably safely reduce the minimum PWM signal below 40%, but not too far.

Why not have a step from 0% to 40% for all fans except those connected to a hydro series cooler. That way we could set a curve so the fan is off at 0% until a certain temperature is reached then the fan spins up to 40%. That way the fans that don't go below 40% will still work and the long requested ability to turn off fans would be fulfilled.

There are really two issues with control of a fan. There is a speed that below which, the fan will not reliably start from a stopped condition, which I'll term the StartSpeed. There is a different speed, to which the fan can be slowed once it is rotating, which I'll call the LowestSpeed. The LowestSpeed always falls below the StartSpeed. This is just good 'ole physics at work, because there are a couple components to friction, and once the fan is started, it takes less energy to keep it moving, than it took to start it moving.

With PWM fans, it is pretty easy to characterize the values for StartSpeed and LowestSpeed, because the PWM signal provides feedback which says if the fan has started moving or not. So it is possible to programatically apply several test values for StartSpeed for short periods of time, and determine at what point the fan starts up. Then it is a simple manner to move the speed down and determine at what speed the fan stops spinning. Typically a small design margin is added to these values so that as the bearings in the fan age, the values chosen will still be adequate.

So then under program control if a fan has a StartSpeed of 35%, and a LowestSpeed of 20%, If a user were to ask for the fan in a stopped state to go to 25%, it would apply the StartSpeed of 35% for about a second, and then immediately change to the requested value of 25%. If the fan is already moving, it can change to any speed from 20% to 80% without going through the StartSpeed step.

It is possible to do this under control, given a processor, and the feedback of knowing when the fan is spinning, and when it isn't. Corsair Link is designed to appeal to the more technically savvy users. Most non-tech savvy users will never bother to even try and use Link. So dumbing down the interface for people not likely to use the product, seems like a dubious value propositon.

I do find that in systems with lots of cooling capability, that when such a system is very lightly loaded, it may be unnecessary to run fans. For example, in a system I did with 120.9 of radiator, when not under load, there was more than enough cooling with simple convection through the radiators, that fans were not needed on the radiators, and running two exaust fans at slow speed was plenty, and made for a nearly silent system. So I also think that a fan control solution which won't allow me to turn a fan off, is lacking control I want.

Since the point of this thread has already moved from point 5 curve bug to zero fan implementing,I feel its safe to post regarding this.

I posted in another thread regarding this

I wanted to show users the ups and downs of having radiator fans turned off and how systems react to it.

First pic is having one exhaust and fans on a 240 rad running at a quiet 500 rpm's

notice in the lower graph how temps remain steady

http://i39.photobucket.com/albums/e152/wytnyt123/34f383b5-149e-44e1-96f3-07e117099e36_zps7y4m7vk3.png

now notice in the next pic with the fans turned off how the temps are like a yoyo.

The graph has a 30 minute span so for those that appreciate a quiet system,it aint happnin with zero rpm fans

http://i39.photobucket.com/albums/e152/wytnyt123/b6e14d63-d472-4d3f-acb4-7629072ff5ce_zpsun6oqulp.png

and more to the point of the ops question,while point 5 is a flaw or bug,having 4 working points is more than sufficient as actually only 2 points are needed since fans operate in increments between 2 points...

Since the point of this thread has already moved from point 5 curve bug to zero fan implementing,I feel its safe to post regarding this.

I posted in another thread regarding this

 

I wanted to show users the ups and downs of having radiator fans turned off and how systems react to it.

First pic is having one exhaust and fans on a 240 rad running at a quiet 500 rpm's

notice in the lower graph how temps remain steady

 

<pictures graphs deleted for brevity>

 

now notice in the next pic with the fans turned off how the temps are like a yoyo.

The graph has a 30 minute span so for those that appreciate a quiet system,it aint happnin with zero rpm fans

 

<pictures graphs deleted for brevity>

This is true for a marginally cooled system. In a system that has lots of radiator capability, like that I mentioned, turning radiator fans off is easily possible, and still results in plenty of cooling given a case fan or two exhausting, which in my case results in air pulled in through 120.9 of radiator area.

and more to the point of the ops question,while point 5 is a flaw or bug,having 4 working points is more than sufficient as actually only 2 points are needed since fans operate in increments between 2 points...

I utilize a couple different plateau levels in my fan curves, as they correspond to different utilizations of my CPU and GPU's, and allowing for multiple points allows me to trade off temperature/performance for noise. So I disagree that only 2 points are needed. I will agree that a straight line can be implemented with two points, but there are lots of reasons not to want a straight line.

This is true for a marginally cooled system. In a system that has lots of radiator capability, like that I mentioned, turning radiator fans off is easily possible, and still results in plenty of cooling given a case fan or two exhausting, which in my case results in air pulled in through 120.9 of radiator area.

 

 

 

 

I utilize a couple different plateau levels in my fan curves, as they correspond to different utilizations of my CPU and GPU's, and allowing for multiple points allows me to trade off temperature/performance for noise. So I disagree that only 2 points are needed. I will agree that a straight line can be implemented with two points, but there are lots of reasons not to want a straight line.

Well considering you've been a member here well over a year and yet have posted build specs,I can only assume you must have a fairly cool processor to keep the cpu cool with only case airflow.The graph I showed was in a complete idle state,it is a 4.5 Oc but even so the fans come on every 7-8 minutes and I did try the case fan scenario with no joy.

Im still scratching my head as how your getting case flow to go thru the rad ,even more so aiding in cpu cooling since the fans must be moving warm air in the case area,

Well considering you've been a member here well over a year and yet have posted build specs,I can only assume you must have a fairly cool processor to keep the cpu cool with only case airflow.The graph I showed was in a complete idle state,it is a 4.5 Oc but even so the fans come on every 7-8 minutes and I did try the case fan scenario with no joy.

Im still scratching my head as how your getting case flow to go thru the rad ,even more so aiding in cpu cooling since the fans must be moving warm air in the case area,

I have an extreme processor... since I get a nice employee discount. The difference in power consumption, and thereby dissipation in the processor is huge between idle, and heavy processing.

Note that my processor is not air cooled, hence the 120.9 of rads. Keeping the processor cool when idle isn't too hard with the equivalent of nine 120mm rads, as that is a lot of water to warm up, and a lot of copper to heat as well, so it takes quite a bit to get it warm. I have all my rads normally pulling air into the case, and two fans in the back exhausting. So if I just run my exhaust fans, it slowly pulls air across the rads, but not fast enough you could tell by feel alone. But turn all the airflow off, and yeah, you will eventually see the CPU get warm.

There are some custom curve shenanigans with Link 3.2.5676 right now and a new version is en route. I'd ask that you wait for the next version and if that doesn't solve the problem, ping us again and we'll file the bug.

I'm on 3.2.5742 now and the bug described in the very initial post is still present.

Just because 5% works for some fans doesn't mean it works for all fans. I do think we can probably safely reduce the minimum PWM signal below 40%, but not too far.

How about adding a checkbox in the "Options" tab that includes a warning message but effectively lets us unlock the full PWM-range (0%-100%). I'm aware there's probably more to that than meets the eye, but it's a feature that's been requested a lot and and a feature that might raise the value of CL a notch, since fan controls that allow you to do this are quite rare.

What do we have to do to make this a reality?