Jump to content

ROC-RK3399-PC (Renegade Elite)


tkaiser

Recommended Posts

3 hours ago, Da Xue said:

This really takes the cake since it has the SoC on the bottom where you can't even put a heatsink.

 

Same with your ROC-RK3399-PC (Renegade Elite): https://libre.computer/products/boards/roc-rk3399-pc/

 

How do you plan to solve heat dissipation there? And do I get it right that the USB-C in the upper left corner is meant as a generic 12V input?

Link to comment
Share on other sites

On 6/24/2018 at 2:43 AM, tkaiser said:

 

Same with your ROC-RK3399-PC (Renegade Elite): https://libre.computer/products/boards/roc-rk3399-pc/

 

How do you plan to solve heat dissipation there? And do I get it right that the USB-C in the upper left corner is meant as a generic 12V input?

The SoC side is the top side and we have a massive heatsink spanning the size of the board there. USB-C is for power delivery. 12V headers are for PoE.

Link to comment
Share on other sites

56 minutes ago, Da Xue said:

The SoC side is the top side and we have a massive heatsink spanning the size of the board there.

 

To be fair that is effectively the same thing, but yes, that should do nicely.  Will the production units be extruded or will they also be machined as pictured here?  It would be nice if more of these SBC's came with proper enclosures, <speculation> I suspect if this much effort is being put into that you hope to keep this form factor for a while?  </speculation>

Link to comment
Share on other sites

1 minute ago, TonyMac32 said:

 

To be fair that is effectively the same thing, but yes, that should do nicely.  Will the production units be extruded or will they also be machined as pictured here?  It would be nice if more of these SBC's came with proper enclosures, <speculation> I suspect if this much effort is being put into that you hope to keep this form factor for a while?  </speculation>

That basic design will be used on all 5W-20W power envelop solutions including the S905D2/X2 and S922. It is CNC milled but designed with mass production in mind. 

Link to comment
Share on other sites

1 hour ago, Da Xue said:

Still being designed but photos attached.

Thanks for sharing. Do you sell the ground plane only? :P There are (I guess) some MIPI CSI/DSI interfaces on the left and the right which aren't accessible with the case (maybe just cut out a small gap, might be not what most of the people want). Also, as soon as you use the GPIO header for some 'hats/caps' the case might be 'problematic'. Would be nice if people can use your nice big heatsink even when they don't use the whole case. :) 

 

20 minutes ago, TonyMac32 said:

Will the production units be extruded or will they also be machined as pictured here?

Did someone ever compared head dissipation of extruded vs. milled blocks? I think if you want a price that people are willing to pay, a case milled from a block will probably be a bit to expensive. 

Edited by chwe
Splitted from M4
Link to comment
Share on other sites

3 minutes ago, chwe said:

Did someone ever compared head dissipation of extruded vs. milled blocks?

I think in real terms it would be negligible.  Aluminum isn't like steel where the processing method radically changes the grain structure.  From an aesthetic point of view milled is always "nicer"

Link to comment
Share on other sites

2 minutes ago, chwe said:

Thanks for sharing. Do you sell the ground plane only? :P There are (I guess) some MIPI CSI/DSI interfaces on the left and the right which aren't accessible with the case (maybe just cut out a small gap, might be not what most of the people want). Also, as soon as you use the GPIO header for some 'hats/caps' the case might be 'problematic'. Would be nice if people can use your nice big heatsink even when they don't use the whole case. :) 

 

Did someone ever compared head dissipation of extruded vs. milled blocks? I think if you want a price that people are willing to pay, a case milled from a block will probably be a bit to expensive. 

The idea is to make it just slightly bigger so you can hook up an LCD with FPC cable. It doesn't interfere with the mezzanine since the entire design built to be stacked. If you were developing on the GPIO and have an LCD, you would have to have the LCD detached.
For all intents and purposes, I don't think extruded and milled matters for heat dissipation at this power density. We are not doing 50W+. 

Link to comment
Share on other sites

5 hours ago, Da Xue said:

Still being designed but photos attached. Big enough for you? @tkaiser 

 

Yep, this looks nice and sufficient. :) 

 

I still don't get how the board is powered (is this USB-C compliant powering with circuitry that can cope with both 5V and 12V -- same with PoE and 12V) but doesn't matter that much yet. I assume we'll get info when available.

 

BTW: For other readers IMO important your comment about PCIe Mezzanine board and PCIe attached SATA (not 'native' ;) ): https://www.cnx-software.com/2018/06/24/odroid-n1-canceled-due-to-ram-supply-issues-odroid-n2-coming-later-this-year/#comment-554498

Link to comment
Share on other sites

10 hours ago, tkaiser said:

 

Yep, this looks nice and sufficient. :) 

 

I still don't get how the board is powered (is this USB-C compliant powering with circuitry that can cope with both 5V and 12V -- same with PoE and 12V) but doesn't matter that much yet. I assume we'll get info when available.

 

BTW: For other readers IMO important your comment about PCIe Mezzanine board and PCIe attached SATA (not 'native' ;) ): https://www.cnx-software.com/2018/06/24/odroid-n1-canceled-due-to-ram-supply-issues-odroid-n2-coming-later-this-year/#comment-554498

Native *speed*. Thanks for the catch. 5V and 12V works. Still trying to get 20V working.

Link to comment
Share on other sites

20 hours ago, Da Xue said:

The SoC side is the top side and we have a massive heatsink spanning the size of the board there.

This radiator will be inefficient. The main location of the Board \ radiator is horizontal and without a fan, with natural air circulation. Therefore, grooves, such depth and shape, along the long side, on the surface of the radiator is not needed. Such grooves will work only in the vertical position of the radiator (that the air would move along the grooves). But then it will be difficult to connect the cables to the connectors. If you make the radiator half the thickness (reduced by the height of the grooves) and with a smooth surface it will work better than this option and the cost will be half (less metal consumption and do not need to make slots in production). Or change the direction of the slots to 90 degrees (parallel to the short side). And do not varnish or paint the radiator surface.

Link to comment
Share on other sites

21 minutes ago, balbes150 said:

This radiator will be inefficient. The main location of the Board \ radiator is horizontal and without a fan, with natural air circulation. Therefore, grooves, such depth and shape, along the long side, on the surface of the radiator is not needed. Such grooves will work only in the vertical position of the radiator (that the air would move along the grooves). But then it will be difficult to connect the cables to the connectors. If you make the radiator half the thickness (reduced by the height of the grooves) and with a smooth surface it will work better than this option and the cost will be half (less metal consumption and do not need to make slots in production). Or change the direction of the slots to 90 degrees (parallel to the short side). And do not varnish or paint the radiator surface.

The design is not final yet. The heatsink is sufficiently big to act as a buffer so the groves are not meant to serve as headspreading fins. They are there to reduce contact surface area should you put your hand on it and lift it when it is hot. Otherwise with higher power chips, you can burn your skin.

Link to comment
Share on other sites

19 minutes ago, Da Xue said:

The design is not final yet. The heatsink is sufficiently big to act as a buffer so the groves are not meant to serve as headspreading fins. They are there to reduce contact surface area should you put your hand on it and lift it when it is hot. Otherwise with higher power chips, you can burn your skin.

There is no difference what kind of burn you get, from a solid surface or from a "striped". :)

If the calculations can surface temperature (of any size) more than 50-60 Degrees, in such cases, use a simple protection device - a grid (fence) with a large cell (by analogy, as using protection on the fans from falling on the blades).

Link to comment
Share on other sites

1 hour ago, Da Xue said:

The heatsink is sufficiently big to act as a buffer

reminds me about the test @TonyMac32 did with a heat sink without fins.

It is useless, once the heatsink reached its capacity it is just a hot brick - that will not get away its heat. That said, I fully support @balbes150 statement.

 

@Da Xue, make sure you are not just wasting time on that piece. By the way, the Lamobo R1 housing had a 'foot' for natural air convection.

 

Link to comment
Share on other sites

46 minutes ago, Tido said:

It is useless, once the heatsink reached its capacity it is just a hot brick - that will not get away its heat.

 

Funny sharing of assumptions and opinions here. This stuff can be tested. And (not so) surprisingly it has already been tested:

 

 

 

Link to comment
Share on other sites

1 hour ago, tkaiser said:

Funny sharing of assumptions and opinions here. This stuff can be tested. And (not so) surprisingly it has already been tested:

1. Turning the slots (if left) is useful for the option if you put the Board vertically. The Board orientation option is vertically limited by the placement of the connectors on the short sides -> to put vertically (with support on one of the short sides) is unlikely.

2. In Odroid, the position of the fins is designed for the forced air flow (fan) option regardless of the position H1. Put H1 vertically (something to support the table was on the short side with connectors) with the movement of air along the ribs and the second option vertically, with support on the long side (vertical movement of heated air across the ribs). I don't know (it's hard to estimate the real size on the photo) how big these edges are, but according to the laws of physics, the higher the height of the edges, the more effect. By the way, I advise you to look at the cooling radiators for powerful sound amplifiers of good manufacturers, how they have ribs and what size they are on the back side of the amplifier. :)

Link to comment
Share on other sites

4 minutes ago, balbes150 said:

advise you to look at the cooling radiators for powerful sound amplifiers of good manufacturers, how they have ribs and what size they are on the back side of the amplifier.

 

I know these things well but as already said: With RK3399 it's simply not necessary. This huge pile of metal will be sufficient to prevent throttling if cpuminer on an ODROID N1 without any heatsink and active cooling at all can generate 6.6 khash/s and still stays below 85° C.

 

I also tried this with a Banana Pro (SoC on the right side!) screwing it into a simple small Aluminium box with a thermal pad between SoC and metal enclosure. 8°C less in idle and the whole box got somewhat warm. And the thickness of the enclosure is just 0.8mm. The thermal mass of @Da Xue's metal brick is way higher.

Link to comment
Share on other sites

I don't understand, if you write that rk3399 under full load, without additional cooling system, can work for a long time within 85 degrees, why do need such a massive (and relatively expensive) radiator ? In this case, @Da Xue writes about possible burns from the radiator (not from the processor) -> the temperature of the processor is significant. Am I missing something or am I misreading it ?

Link to comment
Share on other sites

@Tido remember the heat sink I tested was 20 mm x 20 mm, so it simply didn't have sufficient surface area to be useful.  In the case of this aluminum plate, it has both a large thermal mass and a large surface area when compared to the SoC alone.  Also, as compared to a lot of heat sinks, the notches run longways, providing the thickest cross section to move heat from the SoC.  (The other Libre computers do it the opposing way, reducing effectiveness somewhat, although I agree that at 5-10 watts it's not critical) 

For the fin orientation, remember on pi-factor boards there is no air path through the header if a hat is plugged in.

 

2 hours ago, tkaiser said:

like RK3288 on Tinkerfurnace or the Exynos on Furnace XU4)

 

The RK3288 is ok, the included heat sink is not.  With a 25x50mm heat sink with fins running longways it is a much cooler operator.  The XU4 though, even liquid cooled it gets hot.


 

Link to comment
Share on other sites

31 minutes ago, balbes150 said:

if you write that rk3399 under full load, without additional cooling system, can work for a long time within 85 degrees, why do need such a massive (and relatively expensive) radiator ?

 

The 85°C were obviously a throttling treshold (simple software/DT setting). The more important part was that RK3399 without any cooling improvements (just the standard heatspreader) at this temperature can generate 6.6 khash/s with cpuminer and that big cores throttled down to ~900 MHz (another setting). This was just utilizing CPU cores (with demanding NEON code though) and RK3399 also has a capable GPU inside.

 

With that large metal thing @Da Xue will prevent throttling for sure in normal environmental conditions. And since he said this is a design suitable for other boards to follow maybe there's also some safety headroom (no idea how much power for example an RK3399 Pro with CPU, GPU and NPU stuff will waste and how much heat it will generate)

Link to comment
Share on other sites

IMHO it's just thoughts, I'm not trying to convince anyone :) . Such a massive radiator with additional processing for the grooves will have a significant cost in production with no guaranteed result. All passive heat removal systems have a serious disadvantage, they are highly dependent on the environment and the position in the room. Therefore, it may be better to use a small radiator with a fan (with automatic temperature control) + a simple (properly designed) universal housing (which can be easily used on all models). This ensures that the entire system operates in a wide range of temperatures / loads and the device is positioned in space (eliminating unintended user errors).

Link to comment
Share on other sites

10 hours ago, Da Xue said:

Still trying to get 20V working

 

Just curious: why this? I always thought 20V are only needed when PD profiles 4 (60W) or 5 (100W) are needed. For profile 5 more expensive USB-C cables are needed and I already question the need for 60W 'somewhat'. Wouldn't be profile 3 with 36W be enough? Or will you also provide a mezzanine for full sized PCIe cards?

Link to comment
Share on other sites

On 6/26/2018 at 10:02 AM, tkaiser said:

 

Just curious: why this? I always thought 20V are only needed when PD profiles 4 (60W) or 5 (100W) are needed. For profile 5 more expensive USB-C cables are needed and I already question the need for 60W 'somewhat'. Wouldn't be profile 3 with 36W be enough? Or will you also provide a mezzanine for full sized PCIe cards?

There's no plans for full size PCI-E cards because that is like Pandora's box in terms of customer issues. If 20V works, we could potentially design a mezzanine to drive PoE devices and act as a terminator for PoE camera system for computer vision applications.

Link to comment
Share on other sites

10 hours ago, Igor_K said:

 

The way both demo videos are produced it could be fakes (statically overlaying different display contents). @Da Xue just as a suggestion (from someone who had to learn to spot faked SBC demo videos): IMO it would be better if there's a little bit of human interaction overlapping the display contents at least for a short time.

Link to comment
Share on other sites

8 hours ago, tkaiser said:

 

The way both demo videos are produced it could be fakes (statically overlaying different display contents). @Da Xue just as a suggestion (from someone who had to learn to spot faked SBC demo videos): IMO it would be better if there's a little bit of human interaction overlapping the display contents at least for a short time.

I can assure you that it would take more time to fake that video than it would be to do it for real. I'm not too happy with that video because the lighting all wrong. It would kind of destroy the reputation of everyone involved if we published a fake video. The campaign was launched before all the materials were ready because of scheduling deadlines with product delivery.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines