I'm not sure with the RAM height requirement, but Thermalright XP-90, ID-Cooling IS-60 or IS-50 can be considered as well.
It's really weird, at least with BIOS v1.6 I have similar temp readings. No idea what's up with the WiFi - when I install Windows I can see all networks around, but as soon as installation is finished and I'm logged in, I can only see the network I'm connected to. Same happened on both motherboards.Weird that you've had so many issues - I've got that board in my HTPC, and it works perfectly. No weird thermal readings (I've got the system's only fan set to switch off below 55°C for fully passive cooling, and it tends to hover below that pretty steadily with desktop and video usage - spikes would set the fan going, so I'd notice), and while I'm using Ethernet on it, I've tested the WiFi and found it ... okay. Nothing spectacular, but then the spot it's in is a worst-case scenario for router coverage.
I'm using a kit of Crucial Ballistix Sport LT 16GB (2x8GB, specifically part no. BLS2K8G4D32AESBK) in the B550M-ITX/ac HTPC. They're specced at 3200MT/s CL16, but they're Micron "E-die", so they clock very high very easily (all Ballistix with 'AES' in the part number should be E-die). They're standard/"low" priofile (similar to Corsair LPX) and worked well in my build. I'm currently running them at 3800 CL16 @1.38V using numbers from 1usmus' Dram Calculator for Ryzen. Only other change required was setting vSOC to ... I think 1.15V? Can't quite remember. 100% stable and incredibly easy, with IF at 1:1 on my Ryzen 5 Pro 4650G. If you don't need more than 16GB, I would definitely recommend these kits - if you can find them, that is. They're getting quite old, and Micron's "D-die" has largely replaced E-die. D-die should clock even better, but I have no idea which kits to look for to find those specific memory chips.It's really weird, at least with BIOS v1.6 I have similar temp readings. No idea what's up with the WiFi - when I install Windows I can see all networks around, but as soon as installation is finished and I'm logged in, I can only see the network I'm connected to. Same happened on both motherboards.
I thought antennae issues were because of drivers - all worked fine all day after a fresh Windows install, but the issue started again later in the evening. All worked fine until I set XMP in BIOS and started tweaking memory - maybe just coincidence, but could it be RAM?
What RAM do you have, @Valantar ?
Would anyone have any recommendations on what to use?
I'm tempted to return my RAM and buy a different one - if I can't reach advertised speeds I'd consider it as faulty.
With XMP enabled, I had USB randomly disconnecting, computer started stuttering and WiFi antennae issues started. When I disabled XMP all seemed to return to normal (but with crappy RAM speeds), then I tweaked memory manually as per XMP profile - issues came back and didn't go away even after CMOS reset. I had to reinstall Windows again and now all works fine - that's why it feels like XMP/RAM tweaks are a coincidence, but unstable RAM can cause OS issues so not sure.
I'm only installing latest Nvidia drivers and chipset drivers from ASRock's website - nothing else so really not sure what may cause problems.
Thanks @ignsvn and @SiKiaTriK
Weird that you've had so many issues - I've got that board in my HTPC, and it works perfectly. No weird thermal readings (I've got the system's only fan set to switch off below 55°C for fully passive cooling, and it tends to hover below that pretty steadily with desktop and video usage - spikes would set the fan going, so I'd notice), and while I'm using Ethernet on it, I've tested the WiFi and found it ... okay. Nothing spectacular, but then the spot it's in is a worst-case scenario for router coverage.
Thanks for the detailed info, your posts made me dig a bit more. Finally found the bugger - it's the PCIe extender causing interference. WTF...I'm using a kit of Crucial Ballistix Sport LT 16GB (2x8GB, specifically part no. BLS2K8G4D32AESBK) in the B550M-ITX/ac HTPC. They're specced at 3200MT/s CL16, but they're Micron "E-die", so they clock very high very easily (all Ballistix with 'AES' in the part number should be E-die). They're standard/"low" priofile (similar to Corsair LPX) and worked well in my build. I'm currently running them at 3800 CL16 @1.38V using numbers from 1usmus' Dram Calculator for Ryzen. Only other change required was setting vSOC to ... I think 1.15V? Can't quite remember. 100% stable and incredibly easy, with IF at 1:1 on my Ryzen 5 Pro 4650G. If you don't need more than 16GB, I would definitely recommend these kits - if you can find them, that is. They're getting quite old, and Micron's "D-die" has largely replaced E-die. D-die should clock even better, but I have no idea which kits to look for to find those specific memory chips.
I'm in the process of upgrading my main rig, and for that I've bought a 32GB (2x16GB) kit of G.Skill FlareX 3200c14 (F4-3200C14D-32GFX) - they're Samsung B-die, so they should also clock quite high. So far I've barely made an attempt, but I'm getting CPU interconnect errors in HWInfo at 3800c16 1:1, so I probably have to push my vSOC higher. Still, the RAM seems to handle it fine, and timings and latencies are significantly better than the Crucial kit. It even booted fine into windows at 3800 1.35V as I forgot to adjust DRAM voltage on my first attempt - rather impressive, at least superficially, as Samsung B-die is known to need a bit of voltage to scale well (Dram Calc suggested 1.42V for that speed). This is on the Phantom Gaming B550 board though - I'm going all ASRock this generation, apparently!
Thanks for the detailed info, your posts made me dig a bit more. Finally found the bugger - it's the PCIe extender causing interference. WTF...
I noticed today that my keyboard+trackpack start to miss input when I place it on the aluminium plate which has the mobo underneath. There were problems with the keyboard only if I placed it on the half closer to me - if I lifted the keyboard or pushed it further along the aluminium plate everything worked fine. I started downloading GTA 5 and poked everything - download speed was dropping if something was wrong.
Can't believe it - in the end it's not RAM, BIOS, drivers or some other black magic, just good old cables just the right size in the right spot... Thanks everyone for info on this thread, really helped debugging.
Black Ridge came - it comes with a fan, my bad there
Looks brilliant Thanks for the photos and suggestion - the cooler is great. Against the aluminium plate, it works like with a duct, just need to drill some extra holes in the bottom aluminium plate - BR is much wider.If this can help you, a couple of photos of the BR on the Gigabyte Z390i
Although there is no collision on any component, as you can see, the space is veeeeery tight (I love it so much )
Cheers man!
Thanks. ADT-Link replied, their cables already have EMI shielding and asked about details of the setup and hardware. Really good customer service.Glad you figured it out.
That news article is hilariousElectromagnetic Interference can cause all kind of wired issues, better check with your neighbors if their Wi-Fi is still working...
An old TV crashed entire town's broadband every day for more than a year
Electrical interference from the appliance knocked out internet services in a small Welsh town.www.cnet.com
Thanks. Really neat setup inside!Yes, the gpu blows air out the bottom left and the cpu blows out the top right. Both have the fan orientation swapped to pull case air through the heatsink and out of the case.
At some point I wanted both to exhaust out the top but
the pcie cable was too short, the gpu was too wide and the top left vents are blocked by the keyboard anyways.
The cpu heats the top cover + u-shaped aluminium while the gpu heats the bottom panel.
They don't seem to get too hot except the top right part of the aluminium but that might be solved by some vent holes or slots.
Thanks. I tried to find more info, but there's a lot of conflicting posts.Given the use of shielded risers, it's strange that you're having interference problems. Is your setup properly grounded? I.e. are your motherboard/GPU/riser mounting points all grounded?
Thanks. I tried to find more info, but there's a lot of conflicting posts.
Do I get it right? In a nornal case, PSU is grounded (earthed).
Mobo is grounded by the 24pin connector.
Case is grounded to the PSU.
Is the GPU grounded in any way apart from the PCIe connector? I/O shield to case?
I found that some manufacturers ground standoff holes, so the mobo would be grounded by 24pin and via standoffs to the case which would separtely be grounded to the PSU? I found that ground loops can interfere with wireless signals, but not sure how I would create a ground loop in my rig.
My motherboard is grounded by the GxR-ONE's 24pin connector (correct me if I'm wrong, @REVOCCASES).
GPU - not sure in my case. Would it even need a separate ground?
The main difference I see is lack of ground for the case - I tried connecting a wire between the case and ground of the spare 6pin PCIe, but didn't help. I'll try grounding the case and GPU properly later this evening
Mostly everything in a standard PC is grounded through the case - motherboard (standoffs), PSU (screws and casing if not painted), GPU (I/O bracket), etc. Given that your GPU is mounted with metal screws I would expect those to provide grounding - at least if the mounting holes on the PCB are plated (it's difficult to tell from the pics, but it looks like they are). Of course there should also be grounding through the DC cabling for each component, but redundancy in grounding is always a good thing. Still, it looks like you should be in good order there.Thanks. I tried to find more info, but there's a lot of conflicting posts.
Do I get it right? In a nornal case, PSU is grounded (earthed).
Mobo is grounded by the 24pin connector.
Case is grounded to the PSU.
Is the GPU grounded in any way apart from the PCIe connector? I/O shield to case?
I found that some manufacturers ground standoff holes, so the mobo would be grounded by 24pin and via standoffs to the case which would separtely be grounded to the PSU? I found that ground loops can interfere with wireless signals, but not sure how I would create a ground loop in my rig.
My motherboard is grounded by the GxR-ONE's 24pin connector (correct me if I'm wrong, @REVOCCASES).
GPU - not sure in my case. Would it even need a separate ground?
The main difference I see is lack of ground for the case - I tried connecting a wire between the case and ground of the spare 6pin PCIe, but didn't help. I'll try grounding the case and GPU properly later this evening
Proper grounding (earthing) of all components could help but it could also be an EMI issue / improper shielding of your setup. Sometimes when I test something on an open bench I have similar issues like you. E.g. Wi-Fi drops or my wireless keyboard not working. I mean there is a reason why "normal" PCs and Laptops from HP, Dell, etc are tested for FCC / EMI compliance.
You could also try to use different (e.g. shorter) antennas or a USB Wi-Fi dongle and see if that works better.
Thanks for your messages. I'm trying to find some concrete information on grounding. I sent an email to ASRock support asking if the mobo is grounded via standoffs at all - may come in handy when building custom cases.Mostly everything in a standard PC is grounded through the case - motherboard (standoffs), PSU (screws and casing if not painted), GPU (I/O bracket), etc. Given that your GPU is mounted with metal screws I would expect those to provide grounding - at least if the mounting holes on the PCB are plated (it's difficult to tell from the pics, but it looks like they are). Of course there should also be grounding through the DC cabling for each component, but redundancy in grounding is always a good thing. Still, it looks like you should be in good order there.
You could always try covering the riser cable with RF shielding tape, just make sure it's not shorting out anything.
Also, remember that with your current design, with the motherboard I/O recessed into the case, there's zero shielding between whatever is connected there (WiFi antennas, USB receivers, etc.) and the internals of the PC. Potentially making matters worse, it's all sandwiched between two layers of aluminium, which might cause any RF noise generated to bounce around between these layers, increasing interference. Of course you also have a solid aluminium plate interfering with any external communications (WiFi, BT, USB receivers). It might be worthwhile to try adding some sort of backplate or holder for the motherboard's I/O shield to protect the rear I/O. You might also improve your WiFi by getting some laptop-style antennas and mounting them in the monitor casing or some other location not blocked by metal plating. These typically perform worse than traditional antennas in a vacuum, but they are eminently flexible in how they can be mounted, potentially improving things notably.
Motherboard mounting holes are always grounded. AFAIK it's a part of the ATX spec. If you have a multimeter you could easily measure this, just do a continuity test from a ground pin in the 24-pin, EPS connector, any rear I/O connector housing, or any other grounded point to any mounting hole. You could also do a continuity test between two mounting holes - given that they are on opposite sides of the board, the only reasonable reason for them to be in continuity is if they are grounded (they're definitely not live! ).Thanks for your messages. I'm trying to find some concrete information on grounding. I sent an email to ASRock support asking if the mobo is grounded via standoffs at all - may come in handy when building custom cases.
The WiFi drop is a symptom, I was wondering what was the cause. FINALLY had the time to properly troubleshoot the bastard and found the issue. I also found some info on GPU frequencies and harmonics overlapping WiFi frequencies on specific channels - really interesting, but not the problem I had.
I put the mobo and GPU on a large cardboard box, started downloading 80GB file and moved the motherboard around, put aluminium plates at different angles, moved the mobo around, upside down, etc. It's definitely the PCIe riser and whatever it emits, but under very special circumstances:
Basically, if the aluminium plate is angled like a mirror would be, "pointing" at the motherboard and "reflecting" the riser, WiFi drops.
- if the riser goes underneath the motherboard and I place an aluminium plate under the motherboard closer than around 5cm - WiFi drops,
- if the riser goes underneath the motherboard but I put the aluminium plate between the motherboard and the raiser - no issues with WiFi,
- if the riser goes underneath the motherboard, no aluminium plates, and I lift the motherboard at a diagonal towards the GPU, if I get the angle just right - WiFi drops,
- if I lift the mobo as above, find the "sweet spot" where WiFi drops and put an aluminium plate between the motherboard and the riser - no issues with WiFi. If I remove the plate - WiFi drops,
- if I lift the motherboard and WiFi is OK, when I put an aluminium plate next to the motherboard, facing the riser (where it bends and goes under the mobo) - WiFi drops,
- I turned the mobo around (PCIe facing the GPU) and folded the riser 4 times (90 degrees left, 90 right, 90 right, 90 left - basically twisted it 180 degrees), if the raiser is folded like that and the aluminium plate is closer than around 5cm - WiFi drops.
I sent ADT-Link an enquiry for a "crossed" PCIe extension (straight connector, but one end is rotated by 180 degrees). If soldering something like that at 5-10cm without any losses is possible, I would still be able to mount my GPU and mobo to the same plate (keep all vent holes on bottom plate). In the meantime the only option is to mount the GPU to the bottom plate and use a 5-10cm riser with no folds (like @ruleh's build).
Hopefully Zen3 APUs will bring brilliant 1080p gaming performance and I'll be able to get rid of the GPU from the case and neaten everything up
Good point, may try before I reassemble the rig. Wonder what ASRock will reply too, last time I had a PC motherboards were mounted on plastic trays or with nylon washers on standoffsMotherboard mounting holes are always grounded. AFAIK it's a part of the ATX spec. If you have a multimeter you could easily measure this, just do a continuity test from a ground pin in the 24-pin, EPS connector, any rear I/O connector housing, or any other grounded point to any mounting hole. You could also do a continuity test between two mounting holes - given that they are on opposite sides of the board, the only reasonable reason for them to be in continuity is if they are grounded (they're definitely not live! ).
It was an odd issue - a lot of potential causes and a lot of speculation. I actually had WiFi without any antennae, only with antennae WiFi was dropping. I tried below mobo, above, even with a 0.9mm aluminium backplate from my laptop - WiFi dropped. I'm glad issues were there without any aluminum plates - really points at the riser.What you're describing sounds pretty much exactly like what I was thinking: that the aluminium plates of the case effectively trap RF noise from the riser inside of the case, reflecting it back and forth and creating too much interference for the WiFi to handle. Isolating the rear I/O with a vertical plate might solve it, though that depends how good the shielding on the WiFi adapter is. RF shielding tape like I linked might also help you - I would try wrapping the riser in that, then putting one of the riser mounting screws through the tape to ground it.
Looks good! Btw, AFAIK RF shielding tape isn't superficially conductive - I would assume the shielding mesh/foil is insulated by plastic layers on both sides (that's why I suggested sticking a screw through it for grounding). But your solution looks like it worksHigh tech solutions for high tech problems
I wrapped the riser in aluminium foil and put some fish paper on top to prevent shorting the mobo from underneath, then folded the riser and mounted the motherboard as before.
I once found a spot where the wireless keyboard cut off, but can't find it again - must be just the right spot and angle, otherwise everything works like a charm. Thanks for brainstorming @REVOCCASES and @Valantar
I'm really tempted with clear polycarbonate sheets for the proper build - 2mm for case and 3mm behind the screen, it could look great, especially with the glow from the back of the screen Should be lighter, nicer to handle and strong enough, they are as cheap as aluminium. Heat transfer would definitely be worse, but I could go back to the Noctua cooler or Thermalright AXP-90 full copper with a Noctua A9-25 fan as exhaust.
I'll use the prototype at least until I have the battery working and until I'm sure I don't have to drill any more holes for anything