• Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!

GPU Now this could be big: R9 Nano!

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
Hm, very interesting. It seems to be a good card, and it's the best you can get for mini ITX by a good margin, but right now the price is a bit tough to swallow. I understand why it's that expensive and I'm pretty sure that we'll never get an mITX 980, even if it's power envelope would allow for that, but paying the price of two 970s to get a 10-30% increase in performance is hard to justify.
 

Vittra

Airflow Optimizer
May 11, 2015
359
90
HWC noted their unit had horrible coil whine. At the price they want, this isn't acceptable, especially for it's intended usage scenarios.

Edit - looks like Bit-Tech had the same issue.
 
Last edited:

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
Yeah but the coil whine only appeared at really high framerates above 120Hz, right? I don't really see that as an issue, you can use VSync to alleviate that. And it's pretty hard to avoid coil whine as well, I don't think there even is a general way of counteracting it.
 

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,948
4,953
Most of my cards had coil whine at very high framerates. Not that it's not a real issue in the review units, but before developers capped their menu fps, I could always hear the "stress".
 

veryrarium

Cable-Tie Ninja
Jun 6, 2015
144
44
I was surprised after reading Tom's Hardware review how little power it drew through the motherboard socket, it almost looked as if AMD designed the distribution to make sure using Nano on a PCIe x8/x4 slot is perfectly fine.
 

Vittra

Airflow Optimizer
May 11, 2015
359
90
Yeah but the coil whine only appeared at really high framerates above 120Hz, right? I don't really see that as an issue, you can use VSync to alleviate that. And it's pretty hard to avoid coil whine as well, I don't think there even is a general way of counteracting it.

No, it was perceivable during their gameplay of Hitman over the fans. However, I would actually be concerned with older games running at higher framerates that could benefit having coil whine at this point - we're moving towards higher refresh rate screens at this point - 144hz is common, and the PG279Q will be 165hz.

Yeah but the coil whine only appeared at really high framerates above 120Hz, right? I don't really see that as an issue, you can use VSync to alleviate that. And it's pretty hard to avoid coil whine as well, I don't think there even is a general way of counteracting it.

Covering the chokes (Asus calls this "Hardening" apparently) or the less elegant glue method can mitigate the vibration of the coils that cause this "whine" quite effectively. The idea isn't to eliminate the noise as that is a fools errand, but to reduce the noise such that it isn't perceivable in virtually all usage scenarios.
 

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,948
4,953
These seem to be the covered or "hardened" coils found on older motherboards and GPUs:



This is a glued coil, mainly done with PSUs:



And these are solid-state coils, used in today's motherboards and GPUs:

 
  • Like
Reactions: Vittra

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
So thiese coils are all pre-prepared to emit as little coil whine as possible, right? How would someone go about designing a circuit that counteracts coilwhine? Can you really do anything against it?
 

EdZ

Virtual Realist
May 11, 2015
1,578
2,107
The 'solid state' coils are essentially regular encapsulated coils, but with a case around them.

'Coil While' is due to current passing through coils causing them to heat up, and the heated coil expands. Current then reduces, the coil cools and contracts. Do this many times per second, and you have a tiny speaker. The high cycle rate (frequency) means the coils emit a high pitches tone: the 'whine'.
The entire reason coils are on the board is part of the power delivery circuitry. They are fed AC at high frequencies, and are part of the system that converts that to DC for delivery to the GPU*.
The only ways to eliminate coil whine, no matter what encapsulation is used, is to either eliminate the expansion/contraction, or to change the frequency at which it occurs to move it out of the audible range. Changing the frequency is trickier than it sounds because the frequency varies (as GPU load varies, which is why coil whine a a weird varying screech rather than a pure tone), as frequencies get higher you start to have even worse RF leakage issues, and even at high vibration frequencies you can still get audible harmonics.
You could make the coils out of Inconel, or another superalloy that has very little thermal expansion. However, these materials are not cheap, and may not even be suitable for construction of inductors (I can't find any suppliers of SMT Inconel inductors).


* long answer: 12VDC is received by the card, MOSFETs chop that up at several kHz into 12V pulses with around a 12:1 duty cycle, coils shape those pulses into a sine-like waveform, capacitors buffer that waveform into a smooth output. Because of that 12:1 duty cycle, the final DC voltage is 1/12th of the input voltage, so you have a DC conversion of 12V in to ~1V out, perfect for feeding the GPU. This is a super-simplified description for how a switched-mode DC converter works. While there are other DC-DC converter designs, switched-mode converts are best at handling the high powers required efficiently and in a compact form factor.
 

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,948
4,953
Post of the Week award goes to EdZ !

Very informative and easy-to-read piece, I didn't know they vibrated because of heat, expanding and compressing. But those solid-state coils seem to generally be smaller than the encapsulated coils, so I thought these were different physically.
 

EdZ

Virtual Realist
May 11, 2015
1,578
2,107
SMT inductors are not constructed any differently from regular inductors, they just have a different packaging. Some may not even be encapsulated in epoxy, but just placed inside a plastic box.



 
  • Like
Reactions: Phuncz

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,948
4,953
And there goes the dream that solid-state coils were actually something different, like solid-state capacitors.
*shudder* Or is that also just an electrolytic cap inside a plastic case ? :eek:
 

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,948
4,953
I've read multiple times that coil whine can't be completely avoided (on all hardware combinations), just that it can be minimized.