Topic: Official FutureBit Apollo BTC Software/Image and Support thread - page 39. (Read 47056 times)

I have free electricity at the moment and want to stack as many freshly minted Sats as I can. Without modifying my house cabling and without wearing earplugs 

And I don't have much of a clue, yet, about how to tweak a miner, therefore I use the Turbo setting because I figure it must be reasonably safe, having been pre-set by the designer.

And I am adding external, quiet fans to reduce the noise, but the stock fan will not spin any slower than 1800prm, even when the ASIC temp is as low as 52degC due to my external cooling. So it remains fairly noisy.

Why do you need a fan that can go silent, even though that would mean temps are going to skyrocket? Just to run it in combination with an auxiliary fan at reasonable noise levels or to run it in an underclocked setup?
Because in my experience, it was already a relatively inefficient hashboard at Turbo setting (compared to current-gen chips), even though the 3TH/s Turbo setting is the most efficient in terms of hash per joule.

Those specs look good, except maybe for the temperature range, if you throttle the fan down you can get higher than 70C air going through the fan. But that should be avoided anyway.

But that looks like an unapproved spec sheet, is that fan actually for sale somewhere?

I have to add :Actually available for purchase to the wish list!  



Few days ago I found this one but I missed the fact that it is much thicker than 25mm. However, I have not searched that manufacturer for a 92x92x25mm fan, they might have one.
https://www.farnell.com/datasheets/2243384.pdf

Could this be because you are using Tor? I vaguely remember reading that 10 connections is the max when using Tor, but might be wrong about that.

I haven't been able to find discussion of issues with the Eero wireless router.  This doesn't allow you to add/change port forwarding the typical way; you do it through the Eero mobile app.  I reached out to Eero support to learn about this, and followed their instructions to the best of my limited abilities.  I now have port forwarding enabled for 8333 on my Eero, but the ApolloBTC node dashboard continues to only show 10 connections.  Can anyone help please?

As for your wishlist, did you check out this Delta fan yet? https://www.delta-fan.com/Download/Spec/QFR0912GJ-00P0.pdf
That one seems to be able to spin as low as ~ 1100 RPM, while blowing at 5600 RPM max speed. Didn't find anything better suited so far, being both powerful and also able to turn that low.

Thanks for your tests at lower frequencies. With reasonable temp range I meant close to what the stock fan does, so not more than max. ~ 10°C above when compared to stock. But unfortunately, the Noctua seems to be really sub par as a replacement fan for this purpose...

What is a reasonable range? I honestly don't know.

I assume lower temps are better, but is that actually true, and if so, then why?

And it makes me sad to see Polli being lazy when she could be running in Turbo mode and be just as quiet as with the Noctua NF-A9 in Eco mode, but with a lower ASIC temperature and an additional (partially) redundant cooling system.

Gotta make that curve go up again, soon.... 

Running a Standard Apollo in balanced mode:
Air temperature: 24-25degC
No auxiliary fans, stock fan replaced with a Noctua NF-A9 PWM fan

ASIC temp settles at 73degC and Noctua 92mm fan runs at 1555rpm
Noise is about 50% due to the power supply fan


Running a Standard Apollo in Eco mode:
Air temperature: 24-25degC
No auxiliary fans, stock fan replaced with a Noctua NF-A9 PWM fan

ASIC temp settles at 68degC and Noctua 92mm fan runs at 1340rpm
Noise is mostly PSU noise.


So, here is my updated wishlist for a ASIC fan replacement:
- 92mm x 92mm x 25mm PWM 12V
- minimum speed 1000rpm or less
- maximum at least 4000 -5000rpm
- I think the noise will sort itself when the rpm are lower.

My impression is that the Noctua fans are not that much quieter than other fans that I could detect it with my ears, in a noisy bush setting with all that wildlife doing what it does best: Make noise!
In a quiet lab, yeah, probably easy to show that Noctua's are superior, but the main factor for real world application is the fan speed / rpm. You need a fan that can go very slow, but also very fast, and which fits into the space available.

The search continues ....

I'm not sure the Noctua NF-A9 PWM fan will be much quieter than the stock fan when you use it by itself.

It will make a massive difference only if an additional external fan does the work (quietly) so that the 92mm fan is throttled to a lower speed than the 1800rpm minimum of the stock fan.

And you take the risk that it gets forgotten that an insufficiently powerful fan is in the modded Apollo and then someone sets it to Turbo mode or something goes wrong and it overheats.

My advice is NOT to use the Noctua NF-A9 to replace the stock fan.

But I'll run some tests with Balanced and Eco mode and let you know how that goes.

Correct:
I have not yet tried other setting with this fan, only installed it yesterday.

What happened since then is this:

-The piggy-back connection I put on the 12V output of the PSU failed and the 200mm Noctua fan and the additional 80mm fan on the PSU stopped running.
- the ASIC temp went back up to 79degC with the Noctua 92mm fan running at 1919rpm and the PSU temp (IR thermometer on silver side of housing) had increased to 35degC (usually is 30-31degC).
- the 200mm 3D-printed PLA fan shroud had warped/melted a little bit due to the slow but hot air coming from the ASICs.
- I then connected the 20mm Noctua fan with the modified Y-cable to the miner board fan power connector. I cut the third wire in the cable so that the 200mm fan is not throttled by PWM signal, but the 92mm Noctua fan is throttled down.
- I set the trimpot for the (previously) stock PSU fan to zero
- now ASIC temp is 63degC, 92mm fan is running at 1130rpm, PSU outside IR temp is 30degC
- the PSU fan is now clearly the loudest part of the total noise from the miner. I have spliced a 1k ohm trimpot into it's red cable so I can control it's speed.

I have made some enquiries about fire retardant filament for printing, but I'm not sure I want to tackle printing these parts from polycarbonate, because I don't have a heated print chamber. Maybe the 200mm fan shroud will print OK with PC, it has a large footprint and the round shape might distribute the warping forces evenly and prevent disaster.

I will set fire to some part of the Noctua 200mm fan and the stock fan, if they burn well, then I don't need to worry about getting fire retardent printing filament, because it will not make a difference.

Thanks for that valuable information! So just to confirm: When your Apollo hit 79°C with the Noctua A9 PWM, you were running "turbo" mode (brd_ocp=75 and osc=50)?

Did you also check temps and fan RPM at lower frequencies back then, what they were like at 60/40 and 48/30 (which relates to "balanced" and "eco" mode)?
Because if the temps were within a reasonable range, the Noctua would be still fine at least for my purposes and probably others who are looking for a silent setup.

Lesson learned is this:

The specs for the required fan are something like this:
- 92mm x 25mm thick
- PWM lowest speed below 1800 (ideally zero or at least totally silent)
- max rpm at least 3000, but that might not be enough, better 5000-6000 or higher.
- basically need a powerful, noisy fan that is happy to go to sleep when told to do so, but makes a heck of a lot of wind and noise when needed.

I have not been able to find one, yet, any hints would be much appreciated!

Same contraption shown in infra-red. Very handy to find hot-spots.



The ASIC temperature appears stable at 79degC with only the Noctua NF-A9 running (but I have turned the added PSU fan back on low to keep the PSU happy).

This is what happens when I also turn on the Noctua 200mm fan:

- It gets quieter while the internal fan slows down and the ASIC temp eventually settles at around 61degC;
- the Noctua NF-A9 slows down to 1150rpm or less
- the noise level at arm's length (0.5m) is clearly lower than the noise from a practically unmodified Full Package Apollo operating about 3.5m away from me (with an ASIC temp of 59degC and the stock fan running at 2905rpm).
- the wildlife, birds, crickets etc outside are seriously interfering with my ability to hear the modded Polli at arms length.

 

But the problem remains that the Noctua NF-A9 does not have the Oooompf to cope with a heat wave, and I want Polli to stay at least as safe as it was when I started to deconstruct it's warranty! 

Working on it!

But it is a 92mm fan that is required.

The contraption shown in the photo has a Noctua 200mm fan on top, a Noctua NF-A9 PWM where the stock fan used to be, a Duratech Precision 80mm Silent Case Fan 12V 0.08A 1600rpm YX-2570 on top of the power supply, and the stock power supply fan is throttled down with a 500ohm trimpot until I cannot hear it any more.

It's whisper-quiet when I turn on the added fans, but the Noctua NF-A9 PWM lacks power when I run it by itself.

In 24-25degC air temperature, with 'Auto cooling' equivalent settings for the standard Apollo, running in turbo mode, the ASICs temp gets to 79degC while the Noctua fan runs at 1924rpm.

That means the fan is almost maxed out (max speed is 2000rpm) and it will not cope with much hotter ambient air temperature once summer comes around properly.

That said (and stressing again that I do NOT recommend replacing the stock Apollo fan with the Noctua NF-A9 PWM), it is much quieter than the stock fan when operating in conjunction with the auxiliary fans.
It is also much quieter than the stock fan when it is running by itself (auxiliary fans turned off), but the ASICs are a lot hotter, 79C vs 60C. That cannot be good for their life expectancy.

And the Noctua NF-A9 PWM will not cope by itself if ambient temperatures get into the toasty range.

The cooler fan in my batch 1 standard unit also never goes below 1800 RPM. Has anybody made good experience with replacing the stock fan with a Noctua NF-A8 PWM or a NF-R8 redux-1800 PWM yet on the Apollos? These should be able to turn at least as low as 325 RPM.

Sounds like the IP address of the Apollo is in a different range to the one the new router is using, maybe the Apollo had been given a fixed IP address rather than using DHCP to pickup the address.

NEW Router and Apollo isn't showing up as "connected".  Everything looks like it's working...red light blinking, ethernet port on router green, etc. However, in the router dashboard the Apollo isn't showing up. Spectrum says that I need to call you guys, because the router is recognizing every other device that I have.  My Apollo was working fine until this router upgrade. Any suggestions?  I'd be grateful.

Thanks in advance to all.

Most likely you have you BTC directly connected to the Internet, maybe in a DMZ. In this case it is directly exposed and most likely compromised, like in about 5 seconds. It should sit behind a firewall/router with port forwarding only.

I started a new thread without direct mentioning of the Apollo hardware.

So if anyone is interested, follow me here: Worst case ASIC heat scenario

https://bitcointalk.org/index.php?topic=5424883.new#new

i will, thanks