Yes, I agree the centrifugal fans are naturally noisier than axial fans. By how much I don’t know.
Cool. That’s what I was curious to see. So it does indeed have a radial fan. That could mean anything in terms of noise a guess. Could be fairly silent if run slowly or loud if running fast. It can however offer high pressure at even low operation speed if low flow rate suffices.
There were a few things I picked up from the Notebookcheck review. First was the comment that the HP was the noisiest mini they had ever tested. The second was, and this is me reading between the lines, that tests are done to expose best performance of the machine being tested. I understand that they choose the highest performance option available to provide as relative a comparison as possible i.e. show them in their ‘best light’
I looked at quite a few HP reviews. One, and I can’t recall which, suggested a high 40db in quiet mode but none that I saw compared performance or noise at various modes … so, unknowln at this point
As for ‘perfect’ comparisons; totally agree that this is a pointless quest
If they were quiet I’d buy a HP Z2 mini G1a tomorrow …
Edit(2): … because I’d get a shipping date at the time I place my order as it’s been available in Europe for months now
Since we’re looking at ~7W idle draw there isn’t much to cool anyways. Yes, Notebookcheck states that’s “Idle Min” but they also measured 10W “Idle Min” for the desktop.
The fact that they describe the fan algorithm as a “hardware circuit” means that if there is an issue seen inconsistently in the field, it is very probably either a manufacturing defect or logistical mix-up (such as a manufacturer sending an incorrect configuration to Framework). Assuming that “hardware circuit” is accurate, this would be an issue impossible to fully fix without opening the machine.
EDIT: Everything following this is semi-invalidated. It has been confirmed that the delay on the 128 GB desktops is due to a manufacturing shortage somewhere in the process for the high-memory mainboards.
The delay could then be explained by (a) Framework internally deciding how to handle the issue for units experiencing the issue and by (b) Framework taking steps to ensure units not-yet shipped will not have this issue.
If the issue on the 128 GB shipping pause is not chip availability, my money would be on the PSU manufacturer sending Framework some units not exactly matching the agreed custom configuration. The question of liability is hellish, and if that’s the issue, my heart goes out to whoever at Framework is dealing with it.
Are the 64 and 32 GB configs still shipping out fine? They use the same PSU, so that could help tell if this issue is the one causing that delay.
Same issue here, the PSU fan (which I thought was the CPU fan at first) kicks in every ~3 minutes when idle with the ambient room temp 20C. Batch 8, 230V.
I guess there’s no point in overloading support and asking for a replacement unit now since it’s not guaranteed to help, so I’m just going to wait for a couple of months to see if there are any improvements. Worst case scenario, I’ll just put the mainboard into a mini-ITX case.
I also wonder if sticking an additional Noctua 40mm fan (running constantly at the lowest rpm) either to the exhaust or the intake of the PSU is going to help, to keep it cool enough at least when idling.
I have done this with a 60mm fan, it’s helping with idle noise currently but isn’t 100% quiet and isn’t helping under sustained load. Still better than it was before with the irritating noise every few minutes. Here’s a thread I started about my temporary solution with some pictures. I will be receiving a replacement PSU but it hasn’t been sent yet.
That is a very interesting teardown of the HP
Got my replacement PSU & cooler today. Cooler swap was quite nerve wrecking and swapping both is a complete teardown of the whole system. ![]()
Sadly it did not change much. PSU fan already kicking in without any ramp-up and every few minutes.
Don’t have much hope that there is a fix for this first PSU iteration.
Hope there is a silent one in the future, because right now the constant cyclic noise is really annoying having the machine sit on the table.
A few more measures / observations:
I’m running Ubuntu 25.10 and the system initally was idling around 10 - 12W. After appplying cpufreq the system is idling around 6-8W.
Source + Inspiration: This excellent Strix Halo Homelab guide
I use glances to monitor temperatures. Only “Sensor 2” (1) and “r8169_0_bf00:00 0” (2) stand out. They are usually bit elevated compared to the rest, but not by much.
This is my system idling with two browser windows open (CPU fan set to always on at 25% in the bios - at that setting the Noctua fan is barely audible and it usually keeps temperates in general a bit down):
Temperatures really don’t seem that high overall. This screenshot was taken right as the PSU fan kicked in. So the PSU itself still seems to get hot compared to the rest of the system.
I received my replacement PSU today and installed it. Took about 10m while watching YouTube videos in the background.
I’ll monitor for noise. I’m not hopeful though based on the replies from others.
My replacement PSU experiences the same noise behavior. Different resonance/tone but the same behavior. I’ll update my Framework Support ticket and see if they are considering alternatives but the replacement route is a bust.
Got a replacement PSU today. Set it up. Exact same issue.
I’m thinking of getting an 80 mm front panel fan. Then I’ll get rid of the PSU’s side duct, and craft some custom duct that funnels the front panel fan flow right into the PSU’s intake. What do you think of my genius idea? Also any recommendation on which fan should I get (Noctua seems to have multiple varieties and I have no idea which one to pick)? And how to craft the duct (I’‘m thinking duct tape)?
Good luck!
After installing my replacement I wrote support a long summary and additional infos. Then I shared a screenshot of the temperature sensor readings on the forums and added that as follow-up email too.
Support seems to have only seen my follow-up email and told me:
Thank you for the information your shared. Please continue to monitor your Desktop.
Please let us know if there is anything else we can help you with. We’re here to help.
Thank you for reaching Framework support. Have a great day!
Completely ignoring that this issue is not resolved, my questions if they are actively investigating this issue, what next steps are, etc. ![]()
Got my Desktop last week, same issue with the PSU. The fan comes on every few minutes and stays at full speed until suddenly switching off.
I already reached out to support, so far no answer.
In the meantime I installed a 80mm front fan, maybe that way the PSU will stay cooler.
I got my Framework Desktop machine on Oct 13th. Its the 64 gb version and running Fedora 42. Just like the others, it was quiet initially and then started this whining noise. I first thought it was Firefox and its not offloading video the GPU properly, but then it started happening even with just regular browser tabs open and reading stuff. No video etc.
Ill be starting a return
I was batch 14 but have now cancelled my FW Desktop as this obvious to owners PSU problem does not seem to be an obvious problem to FW. As such it shows no signs of being addressed
I was Batch 13, and I cancelled for mostly the same reason. I was going to use the FW Desktop to get into LLM work for the first time, and so I just put that on the back burner, as I have other projects I need to work on anyway. But whether Framework solves the PSU problem will determine whether I eventually buy a desktop.


