Internal Temperatures (SoC)

[Ezra Pound]

The point I was making is that if the 'working' temperature is exceeded and the unit starts to fail, there is a very good chance that the unit will not have suffered any permanent damage after cooling, giving the user thr chance to fix the problem

 

[Stummery]

I did not like the temperature the SoC heatsink was getting to so as a precaution and possibly extend the life of my Humax boxes I have fitted both with an additional fan, glued in place, to cool the SoC heatsink. I have connected it across the HDD power supply to achieve on/off. See photo



I have used a multimeter with data logging to measure/monitor the temperature of the SoC during operation. See graph



Point A is in full standby B is when powered on to make 2 simultaneous recordings C is when 1 recording finished
D is when the 2nd recording ended and the unit powered down E is when I opened the lid and removed the probe
Time of recording from B to C is 30 minutes
Fan speed of rear fan is set to 55%

I presume the spike at D is showing the point were both the HDD and SoC fan powered off but the SoC was still transferring heat to the heatsink.

 

[Black Hole]

I did not like the temperature the SoC heatsink was getting to

I agree with you. Clearly there's not a lot wrong with Humax's engineering decisions considering how long my units have been running, but the key point is that mine run continuously. Units turned on and off all the time will be cycling between ~20°C and ~70+°C and suffering far more mechanical stress in the process. Even without thermal cycling, I would like my silicon to have a much lower maximum temperature.

From an engineering point of view I find plenty wrong with your rough-and-ready solution, but it does the job and gets it done with far less expenditure of time and effort than being "fancy" (it might not survive a trip through the post though, and I would be worried about it potentially ripping off the spreader with it). I'm just concerned about the power to the fan, which could conceivably go negative (7V derived from the difference between 5V and 12V). You could protect the fan with a series diode, run it from 12V through a zenner, resistor, or regulator, or just run it from 5V (or even 12V), or simply use a 5V fan.

I think the best idea would be to run the CPU fan from the HDD fan output, then the fans can be throttled through the fan package. The drawback is that we have no idea what the current specification for that output is. A Rolls Royce solution would be to take the PWM control signal from the existing fan output and use that to provide a beefed-up drive to both fans.

It would also be very nice to provide CPU temperature monitoring from CF. Softies (eg @af123, @xyz321, @prpr, @/df): if a temperature reading were made available as a USB device, would you be able to interrogate it and log it through sysmon?

 

[prpr]

if a temperature reading were made available as a USB device, would you be able to interrogate it and log it through sysmon?

I don't know much about USB, but (if a driver is needed, and I expect it is) I would expect finding the source and compiling the driver for whatever device would be much more difficult than getting any raw data produced in to and out of sysmon (which is 'just' Jim and Javascript).
Did you have a device in mind?

 

[Stummery]

I only wanted a quick and simple solution with the parts I had to hand (12v fans).

The HDR-Fox T2 is excellent for what I need, especially with the custom firmware installed, and will never be sold so no worries about transit.

I initially tested the fans with a reverse 5v supply and they survived a 6 hour test, but for peace of mind a series 1N4148 diode is fitted to one connector, the fan was still faster than I wanted so there is a 22R series resistor fitted to the other connector.

I did consider using the PWM supply from the main fan but discounted it due to likely problems of balancing the speed requirement of the SoC fan.

The PWM drive to one of the fans failed about 3 years ago requiring the P-Channel MOSFET to be replaced, its Drain specification is for a steady 2A at 70°C so unlikely to be a problem for anyone wanting to take that route for controlling a SoC fan.

 

[everthewatcher]

I only wanted a quick and simple solution with the parts I had to hand (12v fans).

FWIW on a long gone PVR-9200 I used the small temp-controlled fan PCB from a scrap PC PSU to control the main fan, with a bit of fine-tuning of the base speed.

 

[bottletop]

As an alternative to fitting an extra fan that will reduce the SoC by 15C or so, I plan to change my fan to
blow over the hard drive

1. replace fan with a faster one (more CFM/airflow)
2. add a guide to channel a bit of airflow to the SoC (see attached pic)
3. enlarge the opening for the HDD caddy to SoC

The image here a green tool where I plan to place the guide.
It won't reduce SoC temp by much - maybe 1-5C at a guess

 

[Black Hole]

replace fan with a faster one (more CFM/airflow)

More airflow inevitably means more noise. You can get more airflow simply by cranking the fan setting to 100%.

 

[Stummery]

As an alternative to fitting an extra fan that will reduce the SoC by 15C or so

Comparing my graph to the readings taken by BH I have managed a reduction of over 30°C
A dedicated fan is the easiest way to obtain the lower temperature without a noticable increase in noise, my SoC fan is barely audible even with the lid removed.
Unlike the BH modification my SoC fan is positioned at an angle blowing across the heatsink towards the HDD enclosure and rear fan giving a worthwhile improvement and even helps slightly with the airflow within the case/main PCB.

 

[Rick88]

I've had an HDR-Fox T2 for several years now. When I bought it on eBay it had hardly been used. I changed and moved the hard drive to a 2.5" unit and it is now on the right hand side. The hard drive casing has been removed and a new fan/heat sink has been mounted on the original heatsink and is running at 40%. Today with recording and playback at the same time the temperature at the heat sink is 41C with an ambient temp of 19C. My modification has been working for at least two years with no problems.

 

[Black Hole]

So you've abandoned forced air case ventilation? How is the new heatsink secured to the old one?

I venture to suggest the temperature of the old heatsink is much higher than recorded by the thermocouple on the new one, simply because of the thermal resistance at the interface between the old and the new (see my primer in post 1).

I am also concerned that there is now no regulation of the cooling. The only means the firmware/software has to monitor internal temperature is via the HDD SMART data, so if the HDD gets hot the firmware will trigger 100% fan but all that does is move (internal) air across the SoC. You have what control engineers would call an open loop!

 

[Rick88]

I knew someone would pick holes in this. The HDD just doesn’t get hot. Heat sink to spreader, thermal paste and touch of superglue on the edge to hold. It works for me…

 

[Black Hole]

It works for me…

I'm not saying it doesn't, but it can't be recommended.

Have you measured the temperature of the SoC (or the SoC heatsink) rather than just the extra heatsink? It looks to me like there is a gap between the two heatsinks caused by the second heatsink being larger than the flat area of the first, so there is a stand-off and you are relying on the thermal paste... which actually isn't very good and only meant for filling any slight voids.

Without CF running the fan (and bear in mind that the fan package has been known to go off-line), your reported fact that the HDD doesn't get hot would mean the SoC never got any cooling. The HDD isn't so much used for sensing the temperature of the HDD as the temperature within the whole cabinet. And then, if the internal temperature does go up so the HDD triggers the fan, all the fan is doing is stirring the air within the cabinet and not drawing in external air.

I don't understand why you didn't keep the HDD caddy and fan as is, and add a fan to the SoC in parallel. If that heatsink fitted properly, you wouldn't need a fan on it. And adding a fan to the original heatsink wouldn't need an additional heatsink.

I knew someone would pick holes in this.

Picking holes? No, just submitting your "design" to normal engineering review. Having spent my working life as an electronics design engineer, I think I'm qualified!