[oo.viper.oo]

Hi guys, regarding topic about compatibility of memory modules for ASRock J4205 ITX motherboard I report Crucial Ballistix Sport 2x4GB SO-DIMM DDR3L 1866MHz CL10 (BLS2C4G3N18AES4CEU) works only on frequency 1600MHz in dual-channel mode. It must be set manually in BIOS. Only one module is working when it is set AUTO or 1866MHz.in BIOS.

Hi! Same problem here with that Ballistix RAM modules, i am in contact with Asrock support and i hope they provide a fix for this.

I think the problem is more complex.

I have the same memory modules, pair of BLS2C4G3N18AES4CEU. They're certified for this board according to Crucial's web page and they do work flawlessly @ 1866Mhz in dual-channel mode. I run extensive memory tests of them with my ASRock J4205 ITX, >24 hours of Microsoft memory test embedded in Windows and >24 hours of Memtest86. My ASRock, being also used as NAS, has run with >1 month up-time without any issues.

But

Since the very beginning I had problem with this board to actually power up from cold stop (with power cord disconnected). Sometimes it got stuck in POST and required one or more resets to come up properly. But again, once it went up, it was working flawlessly (as stated in previous paragraph) and I could power it down and up again and all was good until I disconnected the power cord. I made following observations while troubleshooting it:
- I replaced lithium cell just to rule out problem with BIOS settings backup even though the former cell was ok according to voltmeter, this had no impact on the issue
- Using only single memory module as opposed to two modules did not have any impact
- Replacing picoPSU with big ATX PSU (to rule out issue with PSU) fixed this cold start issue and made the board powering up normally. But the picoPSU is ok, was used in my previous build for years, it was reliable and it is working reliable in different machine. And as stated in previous paragraph, once powered up, it's running fine. It's 120W picoPSU btw, so plenty of power for 10W TDP ASRock :-).

So I thought to myself "screw you", I can power it up after one or two resets and I keep it up all the time so it's not an issue for me. Well, there was short blackout in my location because of some construction in the neighbourhood, after 1 month of machine's up-time and then I couldn't power it up at all. Not with one module, not with big ATX PSU, not with anything I tried. So I had to RMA the board and now waiting for replacement.

And while I am waiting for replacement, I google around and I stumbled upon numerous reports of similar "cold power up" problems associated with ASRock boards in general, not just with J4205 ITX. There was one guy explaining that ASRock is not following ATX power specifications in their boards when it comes to powering them up from cold stop. It has something to do with right order of powering individual components like CPU, Memory, etc. They have to be powered up in specific order and with specific delays in between, otherwise system ends up with corrupted memory and cannot get past POST. ASRock is apparently using low quality capacitors which control these "delays in between" so it can happen that something is powered up faster than it should and it causes this kind of problems after cold start. If you're lucky, your capacitors have exactly right capacity and you have no problems. If you're unlucky, your capacitors have slightly different capacity and are occasionally causing this race condition. And here's where different types of PSUs come into play as well. Because picoPSU and similar DC/DC PSUs are generally very "fast" and they just immediately start supplying current at all the voltages after cold start. Big ATX PSUs may have bigger delays between individual voltages ramp-up and PSU delays may either compensate for the wrong delay from low-quality capacitor, or the other way round. It's just a matter of chance. This explains why earlier my board had problems starting with picoPSU, but it was starting just fine with big ATX PSU. Then after 1 month of up-time low-quality capacitor changed it's capacity ever so slightly, but enough so that I could no more power up the board with any type of PSU.

All the reports of similar ASRock cold power up phenomenon I stumbled upon on the Internet were successfully solved by board replacement.

Sorry for long post, it's not exactly related to Kodi or Apollo Lake in general, but as these ASRock Apollo Lake boards are so popular, chance is that people with similar problems would look into this thread and find this post and they'd know they're not alone :-)

Edit: Guess memory has to be powered up first and then CPU so that memory is live once powered-up CPU starts initialising it. Memory is powered by 3V rail, CPU by 12V rail, DC/DC PSU powers both at the same time, big ATX PSU may have slight delay on 12V rail...and you get the idea, lack of proper delay on board can be compensated by delay in PSU.