Reversing the Air Flow

Black Hole

May contain traces of nut
There has been some discussion about turning the fan over so that it blows into the box instead of sucking out of the box (but I can't immediately find it). IIRC it has been reported that the RE model comes out of the factory like that, and there have been reports of the temperature reducing as a result.

I've been thinking about it, and I advise caution for the following reasons:
  1. The temperature monitoring (standard or custom) is taken from the SMART stats for the HDD, so the observed cooling is for the HDD and says nothing about the temperature the SoC is running at. An overheating SoC is suspected as one cause of crashes.

  2. In "blow" mode, the inlet air is directed onto the HDD before it has a chance to cool the SoC. In "suck" mode, outside air drifts over the SoC without first passing over the HDD (preferred).

  3. When in "suck" mode, warm air from all around the box is drawn out. Outside air drifts in through the vents to replace it, without much momentum to carry dust with it (although obviously some will hang in the air and be drawn in). In "blow" mode, there is a concentrated inlet airflow which can carry more dust.
Discussion welcome.
 
I must admit, I adapted all my HDRs to what you call the blow mode.
The fan label faces inwards, towards the hard drive.
I recall doing some tests during the summer and notice it helps keep the drive cool (and I have to increase the fan speed to 75-90% during the days it gets to 30c outside temp).
I think it all depends on the siting of the HDRs, ambient temp, air flow etc.
It helps keep the drive cool for the occasions I run fixdisk, which seems to create a toastie HDR.
I also place some extra raspberry pi type heatsinks on 2-3 of the chips as a precautionary measure.
These machines are getting old so I'm trying to keep mine going!
Oddly enough I consider this the suck mode, as it sucks air into the unit.
 
Oddly enough I consider this the suck mode, as it sucks air into the unit.
Yes, I was in two minds – but the clincher was what the fan acts on.

It helps keep the drive cool for the occasions I run fixdisk, which seems to create a toastie HDR.
HDDs are specced to run much hotter than SMART complains about, and as I noted in the OP I would be more concerned about the SoC. The best way to resolve this is with some long-term monitoring of two units, with thermocouples on the SoC heatsinks and opposite-running fans. At some point the fan directions would need to be swapped so that individual differences are cancelled out.
 
I recall why I did my amateur mods now.
I have only purchase used Humax HDRs, so the history of them can vary.
All of my HDRs have been the 1st gen ones, horizontal tuner connections.
The only thing I can say for sure is that one or two of them still had the tamper sticker intact, so I guess the HDD smart Power_On_Hours and HDD temp may give a good indication of how well they were kept.
I found my first HDR was functioning great and the only modification I made to it was to introduce a larger hard drive after cleaning out and and blowing air through the board etc to rid it of any dust.
I left this first HDR with the fan drawing air away from the drive - so in suck mode in your terms.
This HDR eventually had the HDMI green screen problem that I couldn't resolve so I bought another used HDR.
So what I have done since then is to set the fan to blow air into the drive - so blow mode in your terms.
Additionally I place stick on heat sinks on the following five SoCs - HDMI Xmit, HDMI reciever, the 2 DVB-T2 and the 32MB MLC flash
It is a bit of a scatter gun approach, but the hardware is getting old and I am just trying to make it last as long as possible by reducing any possible heat stress.
I also increase the fan speed during summer months to keep the drive temp less than 40c.
I prefer to keep the drive temp relative low as 'overflow' air will get blown over the remainder of the Humax HDR.
There is no noticeable increase in dust on the units since reversing the fan.
Just my 2 cents.
 
Last edited:
I've had relatively little trouble with my units, which don't get heat cycled due to being turned on and off.
That's a good point. We have slightly different ways of using the units which may affect their reliability.
I've just check my HDD temp history and it seems to be less than 38c for the most of the last year, with occasional ventures to 39-40c - when I then set the fan to max speed and noise!
For hard drive I usually go by something like this https://www.buildcomputers.net/hdd-temperature.html.
So my initial task was to keep the drive cool, and then try to reduce heat in other areas by introducing the extra heat sinks as I had changed the original airflow.
One thing I have noticed is that Seagate drives often have a historic highest temp attribute as well as a current temp attribute, whereas my WD drive doesn't record the historic highest temp.
 
Last edited:
I have left the fans in the original expel air direction but on my 2 units I have fitted 12v 50mm fans blowing down at an angle on to the SOC heatsink and toward the HDD enclosure then let the rear fan expel the air. I find that doing this and monitoring via Sysmon gives me a reasonable idea of the internal temperature.

I am using the Fan Speed package and I control the SOC fan by using the HDD SATA power, it is connected across the 12v and 5v connections so runs on approx 7v for a reduced speed/noise.

Edit: I have a resistor fitted in series with the SOC fan to protect the 12v and 5v supplies just in case the fan should develop a fault.
 
Last edited:
it is connected across the 12v and 5v connections so runs on approx 7v for a reduced speed/noise.
Novel!

My concern with that would be if the 12V rail was turned off but not the 5V rail, -5V gets applied to the fan.
 
Last edited:
There has been some discussion about turning the fan over so that it blows into the box instead of sucking out of the box (but I can't immediately find it...
Here's my post of HDD and SoC temps for airflow in and out: click

The later V1 V2 (REV if you must) and the DTR-T1000 blow air in.
 
Last edited:
Here's my post of HDD and SoC temps for airflow in and out: click
having the fan blow air in results an 8C cooler HDD as tested and has no effect on the SoC.
Perhaps that is a good mod if combined with some kind of filtering (even a coarse-weave filter should cut the dust down). It should also be fairly easy to devise a clip to add a CPU fan to the SoC heatsink, the worst bit being having to tap power off somewhere.

Okay, I know the heatsink is no more than stuck on, but CPU fans are very light and the HDR motherboard horizontal (and not subject to being moved around much). Maybe a little downward force could be applied by the top cover.


REV if you must
It was designated "RE" by Humax. I don't like it much either, but…
 
.. It should also be fairly easy to devise a clip to add a CPU fan to the SoC heatsink, the worst bit being having to tap power off somewhere.
I don't use the common interface, so what I do is to remove the plastic cover for it (on the rear) to help with air flow a little, as I've changed my fan to blow over the drive. The small rectangular hole may act is a mini exhaust port. Also this gives the opportunity to do something like routing a usb fan & cable via this hole from the rear usb to the main SoC heatsink. (May need to add a grommet.)
 
I have one unit with a stock fan and one turned around. I noticed no difference in the amount of crashing before and after. The new unit gives the same temp readings as the old. I needed to turn up the fan on both to get a decent reported temp. No idea what the SOC is reading on both units. FWIW and this might be purely a placebo effect but since dimming the already dim display trying to preserve what's left prior to removing the filter - I have not had a crash since.
 
I've been wondering about the design of the HDD caddy. There's an opening in it near the SoC heatsink, with not apparent function other than it could direct air over the heatsink if the fan were in blow mode.

There's a flaw though: there are so many other exits for the blown air (including the ventilation in the underside of the chassis) that only a small proportion would be directed across the SoC. Nonetheless, I wonder if the fan ended up in suck mode by mistake?!
 
Curiously, although my "new" 1TB unit is still of the V1 (not RE) layout, the caddy is slightly different in that it only has two screw holes for the fan. Weird. I can understand economising by only fitting two screws (four is overkill), but why bother changing the molding?
 
Surely it was a question of which type of caddies was in the warehouse? Hence the usual disclaimer that manufacturers use about reserving the right to "improve" the product compared with what's documented in the product specification and user manual.

As the heat-sink doesn't appear to be critical to the SoC operation, I suppose that the Humax designers turned down the opportunity to increase its ventilation
 
Surely it was a question of which type of caddies was in the warehouse?
Why should there be two types of caddy at all?

As the heat-sink doesn't appear to be critical to the SoC operation
I don't know how you come to that conclusion. Heatsinks do not only rely on airflow, the increased area also provides more dissipation by radiation and convection. The heatsink fitted to the HDR-FOX has (I estimate) 20x the surface area of the naked SoC.
 
Heatsinks do not only rely on airflow, the increased area also provides more dissipation by radiation and convection.
So what is convection if not airflow?

While here I'll point out that finned heatsinks with a fin pitch of less than say 10mm need a forced airflow as convection drops with fin pitch.
 
Back
Top