You can stress test with your ROCK Pi 4 to see if the hardware is stable or not.
There is no problem with the size of the heatsink, but the installation process needs to be very careful.
If your ROCK Pi 4 hardware does have problems and the stress test failing. Of course it can be replaced well.
Please remove the rib by filling it.
Why it has that rib is there is a pure mystery as it serves no purpose apart from drastically increasing the chance of damage.
That way the spacer size can be increased and also gain more normal and easily available sizes.
I also have the same and because there is movement on fitting usb…
Most who have never removed the heatsink will prob find they are the same, if they do.
I actually put a dab of epoxy back on what I think is an inductor, but you may have a short or at least resistance to ground.
There is a huge problem with the size of the heat sink rib (pointless) as the board mounts have movement that chip away at the ‘inductor’.
The damaged element is a 2.2microH choke, according to RK808-D datasheet.
Before replacing, just check if it’s soldered correctly.
Before I tear apart my rockpi4b’s, Is only for the big black heatsink
or does it also apply to the allnet case with the integrated heat sink ?
The integrated heat sink does have all 4 corners cut off. Still, care has to be taken when assembling the Rock Pi 4. You can find the assembly instructions on video on the bottom of the product description.
Can you please also make sure that the bundled screws are of the right length?
In the current package they are way too long and they do not fit in the holes of shorter standoffs.
The idea of filling the rib is also worth consideration IMHO. There would be no need for copper pads then.
I think so the rib has zero thermal effect have never seen its like on any cpu package.
But yeah if filled we no longer need copper shim just better more standardised pillars that do have screw depth.
Also there will be more air gap from the cpu heat sink to other components.
The corners are shaved but m2.5 pillars sit in m3 mounting holes and when fitting usb devices its near impossible to stop movement.
The spacing is less that the overall tolerance of movement due to the rib and the choke @Zoran mentioned.
So is the small chip off the corner of the choke likely to be the cause of my instability, or is it likely that there is another problem with the board?
I posted it in another thread already but just wanted to share again.
I use this small heatsink to cool down my Rocky Pi which I use for media center and it turned out really great.
I do have the big heatsink but haven’t tested it due to these problems you guys show.
Forgot to put the link…
The downside there, is that you’ve got moving parts. For something like this, that is extremely undesirable.
Regarding the “problems”, as long as you are careful with assembly, you’ll surely be fine. I think a lot of people buy things like this with a picture of the end result, but coming from a consumer products mentality, where everything is already put together in a plastic box that protects it even if your dog grabs it and buries it in the yard for a few days. I’ve seen far too many threads about people complaining about things that they eff’d up themselves, like the one where someone was complaining about it not coming with spare parts to replace the ones that he carelessly dropped into the rug. Not only a danger for carelessly losing parts, but rugs are great for helping you build up static charges.
I had the same issue.
It’s possible that you just ripped of the choke from the board, i.e. the choke lost contact with the board. Before replacing, try to solder it, very carefuly, or ask a pro if you are not experienced, it’s easy to do more damage else.
Tried reflowing the solder, no change still crashes all the time.
I found the suggestion of resoldering the inductor to be somewhat dubious. Its not really the kind of thing that would lead to an intermittent crashiness, because it would mean that you’ve lost electrical connection, thus whatever is running through it wouldn’t power on at all.
A chipped corner on that inductor is not likely to cause instability. The reality is that all that part is, is a coil of wire around a ferrite core.
If you’re having stability problems, the first thing you should try is a different/better power supply. I was having a really serious stability problem with one of my rock960c’s (same as a rockpi4, but with the vastly superior 96boards footprint), and it went completely away when I switched to a better power supply than the one it came with.
Thanks for your reply.
I have tried 4 different power supplies.
Motorola turbo charger
Nintendo switch power supply
Blitzwolf qc3 charger
Raspberry pi 5v power supply.
All have same problem.
One thing that does change the problem,
The board seems slightly more stable without the heatsink attached.
If anything, that is about the opposite from what I would expect. It should be MORE stable at lower temperatures. Suggests to me that something is being shorted out.
I don’t know the details of most of those chargers, except that most or all of them appear to contain some form of “smart” to them, which ultimately can do nothing but muck things up. Further, they are all likely to start at 5v output until they are instructed to step up the voltage.
The schematic for this board calls for 12 volt power input. 12V input runs to a pair of 5 volt buck regulators. While the datasheet for the 5 volt buck regulators isn’t explicit in this, at every instance they show an output set to 5 volt, they are showing a minimum input of 6 volt. So if you’re feeding the board with a 5 volt power supply, the output of those buck converters is probably somewhere in the range of 4-4.5 volt, which is too low. While the instruction manual may indicate that USB PD/QC/etc. adapters are compatible, the startup state in ALL of them is to output 5V. This is so as not to cause damage to devices not capable of accepting the higher voltages.
Now the board does have an LDR6015T PD controller;
That means that it is supposed to signal the power adapter to switch to a higher voltage.
HOWEVER, the functionality of this part is subject to a whole bunch of other conditions being met, none the least of which is the fact that your WIRE needs to be implemented correctly.
If it were me, what I would do, is rig up a true 12 volt 3 amp adapter, and wire it straight up to the VCC12V_DCIN contact point beside the power measurement resistor (R2).
I have an old USB C power cable that my dog chewed, I salvaged the plug on the end, can I apply 12v directly to the black and red wire in this cable?
Or does the LDR6015T need to request a higher voltage before it can receive a higher voltage?
I have an intermittent instability as well. Most of the time it works fine, but sometimes the whole system just halts (a hardware stop, not a kernel panic or a segfault). I have no idea what is causing it, though it seems to happen when the system is under load. I have the AllNet case with the built-in heatsink.
I am the “someone [who] was complaining about it not coming with spare parts” (@lbdroidman above) so I may have damaged something in my multiple attempts to build my system. I’ve been used to getting a RPi board and just screwing on a case. It seems that the Rock Pi requires mechanical engineering skills to assemble.
No. It just needs you to use something a little more sensitive than a sledge hammer to assemble it.
The only thing that the LDR6015T does is tells the power plug that it is OK to send a higher voltage.
Your dog chewed wire should be fine, HOWEVER, I suggest that you confirm that the red and black wires are, in fact, the positive and negative power wires. Its a bad idea to make any kind of assumption that the manufacturer used predictable color coding. I’ve seen USB wires with all kinds of random colors.
Kids under 10 can assemble it
Amazing where the world goes…, I just wonder you use words “mechanical engineering skills to assemble” but are unable to assemble
