Fan operation in the U4

[raymondjpg]

I understand that the issue of high pitch noise has been addressed through a software update.

My questions relate to the operation of the fan in the U4.

1. How frequently is the fan employed? Is it most of the time, or more during particular periods of PVR activity? If so, which activities?

2. Does anyone have any experience with whether or not fitting a SSD rather than a HDD, by perhaps cutting down on power requirement, has any bearing on operation of the fan?

No doubt I could find out for myself by purchasing a unit, but I would appreciate any comments users might have before doing that.

TIA.

[Paul_oz53]

I don't know enough about the fan to comment on your questions.

I can say that the U4 fan noise is trivial. I'm very sensitive to noise, so that's a big tick from me.

What I can say is that the ticking of the WD HDD is the sound that I do hear and wish I didn't. An SSD would be immune to that problem.

If it's intended for the bedroom, it would not have to be shutdown at bedtime with an SSD. Really comes down to how you intend to use it.

[peteru]

You have control over the fan speed on the U4. There are two settings - one for normal operation and one for standby. You can adjust the speed of the fan for each mode independently. The speed can be varied in a number of steps between 0% and 100%.

[raymondjpg]

I can say that the U4 fan noise is trivial. I'm very sensitive to noise, so that's a big tick from me.

Thanks for that opinion, but I would be introducing a fan into a fanless home theatre setup so I would likely be sensitive to any introduced noise.

You have control over the fan speed on the U4. There are two settings - one for normal operation and one for standby. You can adjust the speed of the fan for each mode independently. The speed can be varied in a number of steps between 0% and 100%.

That is handy to know.

Would anyone be in a position to elaborate on what might happen if fan speeds were set to zero for both normal operation and standby? Also, whether use of an SSD might mitigate in any way towards heat generation by the unit?

I guess that there is a good reason why a fan has been introduced to this unit. However, I have three NUCs in fanless cases which dissipate enough heat to enable them to run trouble free - at least for the time being. My question really is: is there enough heat dissipation to risk running the U4 fanless in normal operation with, say, only three tuners operational i.e. without the fourth USB tuner, and possibly with an SSD fitted which would also likely reduce heat generation.

[prl]

...
Would anyone be in a position to elaborate on what might happen if fan speeds were set to zero for both normal operation and standby? Also, whether use of an SSD might mitigate in any way towards heat generation by the unit?

The U4's fan is mounted on the SoC's heat sink, which is probably a good indication of the priority area for ventilation. The fan exhausts through the top of the case.

I don't have an SSD available to comment on how that affects heat generation and ventilation needs.

I guess that there is a good reason why a fan has been introduced to this unit. However, I have three NUCs in fanless cases which dissipate enough heat to enable them to run trouble free - at least for the time being. ...

The NUCs appear to have cases designed for fanless operation. The U4 doesn't.

The U4's printed specs say 65/40/< 1W for peak/normal/shutdown power. The power supply brick is rated 12V/3.5A. Measured consumption is 20W with the 2GB 2.5" HDD running for timeshift. Peak startup was 22W. Cheapo Jaycar power meter.

[raymondjpg]

The U4's printed specs say 65/40/< 1W for peak/normal/shutdown power. The power supply brick is rated 12V/3.5A. Measured consumption is 20W with the 2GB 2.5" HDD running for timeshift. Peak startup was 22W. Cheapo Jaycar power meter.

Those power specs suggest to me like some fan cooling would be necessary in normal situations, definitely under peak load, and possibly not with HDD running for timeshift. That latter one might be mitigated by use of a SSD, but it is normal operation that I would be looking at and in that case an SSD might not be of much benefit vis a vis heat output.

I guess I would need to be hearing from anyone who has tried to run this unit fanless. If there is provision to set the fans at 0% I would have thought that someone would have tried it.

[Grumpy_Geoff]

...I guess I would need to be hearing from anyone who has tried to run this unit fanless. If there is provision to set the fans at 0% I would have thought that someone would have tried it.

That feature was implemented only about three weeks ago. I think not many users will even be aware of it.

[MrQuade]

I personally wouldn't recommend it at all.
Fanless devices are generally purpose built to directly couple their heat generating components to the chassis to maximise radiation and convection areas. Even the standard NUCs have a fan, you need the special versions if you want to go fanless.

Whether in standby or fully-on, the U4 CPU will be generating enough heat that it will need to be forced out of the case one way or another.

[raymondjpg]

That feature was implemented only about three weeks ago. I think not many users will even be aware of it.

Maybe not, but I would think someone will try it when they become aware of it.

I personally wouldn't recommend it at all.
Fanless devices are generally purpose built to directly couple their heat generating components to the chassis to maximise radiation and convection areas. Even the standard NUCs have a fan, you need the special versions if you want to go fanless.

Whether in standby or fully-on, the U4 CPU will be generating enough heat that it will need to be forced out of the case one way or another.

In principle I'd agree with that, but if I were in possession of a unit I'd probably try it at least. If, for example, fan operation were really only necessary for playback of HEVC material, I'd gladly skip that in favour of fanless operation. Also, Id be looking to try lowest possible fan operation which didn't result in the casing feeling too hot.

Again, probably waiting on some intrepid user to try it.

Thanks for the comments.

[Paul_oz53]

It would be brave to operate the U4 fanless without changing the heatsink. Personally, I doubt it will be possible.

Been quite a while since I last designed a cooling system but the size of current heatsink looks unlikely to handle more than a few watts static cooling.

[peteru]

I might not even go as far as calling the aluminium block a heat sink. Maybe not even a heat spreader. It's more like a fan mounting block.

I've had the unit running without cover in an air-conditioned room with ambient temp set to 21C without the fan for several hours. The metal block on the CPU felt uncomfortably hot as did one of the tuner cans. The CPU did not get hot enough for the thermal protection circuit on the chip to trigger a thermal shutdown.

If you are going to experiment, you want to look at:

Code: Select all

root@devu4:~# cat /sys/class/thermal/thermal_zone0/temp
38946

That translates to 38.946C or about 39C. This is on an idle U4 with fan speed set to 2/3 of max, without case and in a room air-conditioned to 21C with some minimal air flow over the board.

The SoC max operating temperature is 75C. In reality, that means you don't want to see anything higher than about 70C coming from that temperature probe. If you want to do thermal management properly, you really need a FLIR camera to understand where the hot spots are.

[raymondjpg]

I've had the unit running without cover in an air-conditioned room with ambient temp set to 21C without the fan for several hours. The metal block on the CPU felt uncomfortably hot as did one of the tuner cans. The CPU did not get hot enough for the thermal protection circuit on the chip to trigger a thermal shutdown.

That would suggest to me need for at least some degree of fan operation i.e. not fanless.

That translates to 38.946C or about 39C. This is on an idle U4 with fan speed set to 2/3 of max, without case and in a room air-conditioned to 21C with some minimal air flow over the board.

Could maybe read much the same with a lower fan speed?

Thanks for these results.

[prl]

...
If you are going to experiment, you want to look at:

Code: Select all

root@devu4:~# cat /sys/class/thermal/thermal_zone0/temp
38946

That translates to 38.946C or about 39C. This is on an idle U4 with fan speed set to 2/3 of max, without case and in a room air-conditioned to 21C with some minimal air flow over the board.
...

Should it be reported as System temperature or Processor temperature in the About screen?

[peteru]

Should it be reported as System temperature or Processor temperature in the About screen?

It will be when I push the latest firmware. I added that functionality last night.

[prl]

Should it be reported as System temperature or Processor temperature in the About screen?

It will be when I push the latest firmware. I added that functionality last night.

[prl]

Of course, if you want silent running in a Beyonwiz PVR, there's always the T2, which is fanless and can be fitted with an SSD.

[raymondjpg]

Of course, if you want silent running in a Beyonwiz PVR, there's always the T2, which is fanless and can be fitted with an SSD.

Sure, I have that option already. Still tossing up whether to take a punt on a U4 with 1TB HDD to start off with, because of the extra tuner and gigabit networking. I'll probably end up with one if the current round of software updates deal successfully with restart issues reported in the EPGRefresh and U4 threads. I'd also like to be reassured that the U4 will work trouble-free with IceTV interactive. So holding off for now.

[Paul_oz53]

Of course, if you want silent running in a Beyonwiz PVR, there's always the T2, which is fanless and can be fitted with an SSD.

Sure, I have that option already. Still tossing up whether to take a punt on a U4 with 1TB HDD to start off with, because of the extra tuner and gigabit networking. I'll probably end up with one if the current round of software updates deal successfully with restart issues reported in the EPGRefresh and U4 threads. I'd also like to be reassured that the U4 will work trouble-free with IceTV interactive. So holding off for now.

I'm getting normal service from IceTV on the U4 with no restart issues (touch wood!). I do customise a fair bit. Mine has the 1TB drive and I can hear it ticking on timeshift recording more than the fan when the TV is off.

prl is also an IceTV user and doesn't appear to have any issues.

Don't blame you for being cautious though. Gully does appear to have a correlation between at least some restarts and a timer operating. Not sure what type of timers though, or if this is just coincidence. I think we would also like to know the cause - and the cure!

[peteru]

As far as I know, no-one has reported any IceTV issues that would only affect the U4.

[prl]

...
prl is also an IceTV user and doesn't appear to have any issues.
...

True, but our in-use device is a T4. I have IceTV accounts enabled on a test T2, T3, T4 and U4, but they don't get regular use.

All models use the same code for IceTV, and there's only a tiny bit of model-dependent code in it - used to set up the initial default device type when you create an account, and various pieces of text that are generated with the device's brand and model.

That's not to say that it's impossible for the IceTV plugin to behave differently on different models due to its interactions with other parts of the UI code.

[raymondjpg]

As far as I know, no-one has reported any IceTV issues that would only affect the U4.

Me neither, but when I asked for comments on performance of the U4, including IceTV, on the U4 thread there was no response, good or bad. I guess I could ask the question in the IceTV forum.

[MrQuade]

Me neither, but when I asked for comments on performance of the U4, including IceTV, on the U4 thread there was no response, good or bad. I guess I could ask the question in the IceTV forum.

I think the only reason you got no response was that there was nothing to say. It really does *just work*. No need to ask on the Ice forums since I am pretty sure you are not going to find any U4 owners there who don't also post here.

[raymondjpg]

I think the only reason you got no response was that there was nothing to say.

I'm not that sanguine. I'd be more reassured if users piped up and said it works well.

Chacun a son gout.

[dRdoS7]

Of course, if you want silent running in a Beyonwiz PVR, there's always the T2, which is fanless and can be fitted with an SSD.

Sure, I have that option already. Still tossing up whether to take a punt on a U4 with 1TB HDD to start off with, because of the extra tuner and gigabit networking. I'll probably end up with one if the current round of software updates deal successfully with restart issues reported in the EPGRefresh and U4 threads. I'd also like to be reassured that the U4 will work trouble-free with IceTV interactive. So holding off for now.

Well, hold off no longer!!

According to OWI, the T2 has Gigabit:

Maybe someone could add a "0" to the Link Speed so we can use the extra?



dRdoS7

[peteru]

OWIF is correct. The SoC has GigE Ethernet interface that has a choice of external 10/100/1000 PHY via RMII or the built in 10/100 PHY.

The T2 uses the built-in PHY so you get a GigE Ethernet interface that supports up to 100Mbit link speed.

The T4 uses an external PHY to give you up to 1000Mbit link speed. The external RMII PHY on the T4 is the reason why WOL can not enter low power mode, because doing so would turn off power to the PHY.

Those are all design compromises. T2 was designed to be low power, quiet device whereas the T4 was designed to pack maximum set of features.

U4 is from a new generation of hardware, so the goal posts and constraints have changed somewhat since the T-series.

[raymondjpg]

Temptation got the better of me so I ordered a U4 with 1 TB HDD.

Before installing I'd like to check on what firmware to apply. I'd prefer not to go on the beta feed, in order to keep things less confusing (for me).

1. My understanding is that v20180216 will fix any issue with autotimers in the menu, but not high pitched fan noise.
2. If I encounter high pitched fan noise then I should apply v20180326.

Does v20180326 apply a high pitched fan noise fix that was first introduced in the online version v20171230, and which resulted in the fan being operated at 100% all of the time? If so, does v20180326 just fix the noise problem by setting its operation at 100%, but also introduce settings to progressively enable reduction of that fan operation to a point where it still operates all the time but hopefully with less noise and without a high pitched whine?

I see from what peteru said in the 20171236 thread that there had been no comments about the fan operation fix. My comment would be that I would prefer the fan to only operate at a speed necessary for cooling at whatever load is experienced by the processor at the time. That was my understanding of the original software control of the fan operation i.e. variable speed, depending on processor load. However, if that results in high pitched noise whine that is noticeable and annoying then application of firmware v20180326 would probably be the lesser of the two evils for me.

One final question. With application of firmware v20180326, is there any option to override the fan setting options i.e. to restore the fan operation to what I understood to be its original variable speed operation?

[prl]

I'd recommend simply going to the latest 17.5 beta USB. At the moment, that's 20180408. That will definitely give you fan speed control (separately for standby and running).

...
One final question. With application of firmware v20180326, is there any option to override the fan setting options i.e. to restore the fan operation to what I understood to be its original variable speed operation?

I'd have to revert back to it to find out, but 20180408 doesn't, so I'd assume that 20180326 doesn't, either.

[raymondjpg]

I'd have to revert back to it to find out, but 20180408 doesn't, so I'd assume that 20180326 doesn't, either.

Please don't do that. I'll take it that if I apply fan control firmware that it is non-reversible.

Does this mean that all future firmware releases for the U4 will implement fan control? If so, then it looks like it will not be optional.

[prl]

Peteru is probably the only one who could say what might coming up with U4 fan control.

'It's tough to make predictions, especially about the future.' Yogi Berra (attrib)

[MrQuade]

It is unlikely that a load dependent Auto fan nice is going to be of any use in the U4.

Best if just finding a fan speed that annots you the least and just leaving it at that.

Just install the latest USB image, set the fan, and don't overthink it

[raymondjpg]

It is unlikely that a load dependent Auto fan nice is going to be of any use in the U4.

I was thinking that was the default option on leaving the factory. If not, then what is the default option?

Best if just finding a fan speed that annots you the least and just leaving it at that.

Just install the latest USB image, set the fan, and don't overthink it

Probably what I'll end up doing, and would really be the only course of action if fan speed control is the only option in the firmware.

Regardless of what the default factory setting might be, ideally I'd prefer to have the option to retain it. For instance an OFF option against the fan setting, which I guess is now somewhere in the SETUP.

[prl]

It is unlikely that a load dependent Auto fan nice is going to be of any use in the U4.

... what is the default option?

In the current 20180410 firmware, the U4 fans have 14 accessible speed steps 2..15. The default "running" setting is 12, and the default "standby" setting is 4.

[peteru]

You are definitely over-thinking and over-complicating the fan operation.

Old driver code programmed the hardware PWM at a frequency that caused resonance and as a result the circuit would produce an audible squeal. The fan was initialised to a fixed speed that was somewhere around 66-75% (can't quite remember the exact figure). The driver never adjusted the speed of the fan after the initial setting took place. The user space software did not provide any facility to adjust the fan speed either.

When I discovered the cause of the squealing problem, I released a temporary workaround that forced the fan speed to 100%. This removed the squeal and improved the cooling. The fix was done purely in user space software.

After a period of time, the Koreans provided updated drivers to "fix" the squealing. The temporary workaround described above was removed when the updated drivers were added to the firmware. This effectively took the user space software to square one - no facility to adjust fan speed at all. Unfortunately the "fix" that came from Korea was as simple as defaulting the fan speed to 95%. As part of the change a new bug was introduced in the driver that would completely disable the fan if the user space manipulated the fan speed in a particular way. The squealing was still present if you tried to change the fan speed to anything less than 95%.

I was not happy with the state of the driver or the lack of control from user space. It didn't look like the Koreans would come up with a "proper" fix, so I decided to fix both parts of the problem. I fixed bugs in the driver and changed the hardware PWM circuit programming to avoid resonance at all fan speeds. I also implemented fan speed control in user space, so that you can set separate fan speeds for normal operation and standby mode.

I have no plans to make any changes to the fan functionality in the driver, unless some show-stopper bugs come to light. It works.

As far as user space is concerned, the ability to have a choice of separate fan speeds for normal operation and standby mode is already an improvement. It's probably sufficient. At this stage I have no plans to add temperature sensitive fan control. The kernel API has enough capabilities to do this, but it really should not be done from enigma2. It needs to be done at the system level (either using the standard kernel thermal management features) or using a separate daemon. It'll need proper hysteresis handling at a minimum. The last thing you want is to cook your CPU because enigma2 had a spinner while updating autotimers or trying to read a large file over flaky WiFi network and as a result could not turn the fan on.

tl;dr - The implementation in the most recent image (17.5 20180408 as of now) is the best choice. All previous implementations were inferior in some way.

[raymondjpg]

Old driver code programmed the hardware PWM at a frequency that caused resonance and as a result the circuit would produce an audible squeal. The fan was initialised to a fixed speed that was somewhere around 66-75% (can't quite remember the exact figure).

Then nothing is lost and everything gained by implementing the fix.

I have no plans to make any changes to the fan functionality in the driver, unless some show-stopper bugs come to light. It works.

Understood. It aint broke.

tl;dr - The implementation in the most recent image (17.5 20180408 as of now) is the best choice. All previous implementations were inferior in some way.

Noted.

Thank You.

[sub3R]

You have control over the fan speed on the U4. There are two settings - one for normal operation and one for standby. You can adjust the speed of the fan for each mode independently. The speed can be varied in a number of steps between 0% and 100%.

Should it be reported as System temperature or Processor temperature in the About screen?

It will be when I push the latest firmware. I added that functionality last night.

Both very nice features. Thanks.

I did a few quick tests on our U4 which sits on its own shelf in an open front & rear cabinet with plenty of space all round.

The following results were with the ambient temperature fairly consistent around the 20ºC mark & the fan speed settings at the default settings:

• With the U4 switched to deep standby for 5 hours the CPU temp indicated 37ºC when first switched back on.
• With the U4 then switched to standby for 4 1/2 hours with nothing being recorded the CPU temp indicated 53ºC when first switched back on.
• With the U4 then switched on playing back a recording from its HDD & nothing being recorded the CPU temp indicated 49ºC after 2 hours.
• With the U4 then switched on playing back a recording from its HDD & recording services from four different channels (recording on all 4 tuners) the CPU temp indicated 53ºC after 2 hours.

I think the default settings are pretty good but I may increase the speed up a notch for standby.

Edit: Tests over 4 days showed by moving the standby speed setting up one notch from the default in standby gave 52ºC, up another notch gave either 50ºC or 51ºC, & up another notch gave either 50ºC or 49ºC. Left at 3 above the default setting (5 above the lowest setting) to give 49ºC to 50ºC in standby.