Mac Mini version of a CAPS music server - Step by Step

[Superdad]

Sandyk,

Thanks for confirming that reconnecting the tacho line should be ok- now all I need to do is find which wire it is!

I agree that it is important to keep the feedback if you don’t know what customers will do with the product. I cloned my SSD to an SDHC, and there were some alarming spikes to 60-65C on the CPUs every hour or so. They settle back fairly quickly, but I did wonder if I was going to have to stop the copying. I can live with an uncontrolled fan, but I know what I’ve done….I have not yet tried video streaming, but that might need a higher fan speed.

 

Alex,

Nice idea, but no way will the fan wires reach! It’s awkward even to disconnect while holding the loose fan. I think you would need a very thin (8-way?) ribbon cable terminating in the right plug and socket, and this would have to do a very sharp twist at the fan end.

 

Bob:

Thanks for the additional info. I realized we will need to have a wire/connector pair running back over to the motherboard anyway, so one I figure out the series of connectors Apple uses I'll try to find an in-line mating jack for the end coming off the fan and just make that a close part of the wiring harness that runs to the motherboard.

 

If indeed your are running your 3.5V into the 12V line of the fan (and not into the PWM line), then John has a really great idea for how to hack the 4-wire PWM system and go beyond just filtering. With some kind of follower transistor (plus a small RC network) he said he could allow the PWM and tach lines to work (and at a MUCH lower current than present) while providing clean and quiet power to the fan--keeping all pulses out of the computer!

If this circuit works then after the Mac mini we may have to look at doing a board for PCs--to compete with the SOtM…

 

If you can PM me with a clear drawing/description of which pins are the Ground/+DC/PWM/Tach lines for the mini, that would be a huge help. I'd rather have John send me a pro to board to try than to have to send him my mini to experiment with. If I was not spending so much money on production parts right now I'd by another mini for all this work. May have to anyway.

 

Ciao,

ALEX

[sandyk]

some kind of follower transistor (plus a small RC network) he said he could allow the PWM and tach lines to work (and at a MUCH lower current than present) while providing clean and quiet power to the fan--keeping all pulses out of the computer!

If this circuit works then after the Mac mini we may have to look at doing a board for PCs--to compete with the SOtM…

 

Alex C

That's great news. The Mac Mini especially, is very compact, and correct cooling is paramount. It's an added bonus if you can do this without additional RF/EMI radiated by the fan leads as at present.

 

Regards

Alex

[AudioPhil]

Hello Folks

 

Don't forget about the right OS. Windows Server 2012 R2 and AudiophileOptimizer run on a mac mini like a charm! -> http://www.computeraudiophile.com/f10-music-servers/whos-used-audiophile-optimizer-windows-2012-server-or-windows-8-a-17897/

 

Best,

Phil

[tranz]

Hello Folks

 

Don't forget about the right OS. Windows Server 2012 R2 and AudiophileOptimizer run on a mac mini like a charm! -> http://www.computeraudiophile.com/f10-music-servers/whos-used-audiophile-optimizer-windows-2012-server-or-windows-8-a-17897/

 

Best,

Phil

 

Hi Phil,

 

Thanks for the info. If I ever build a CAPS your OS tweaks are a no brainer. I used to build DAWs on Windows and had to do a lot of manual OS tweaking.

 

For Mac many of us are using the CAD scripts and manual tweaks to shut down unnecessary processes, as it makes a big audible difference.

 

But the only reason I even started using Mac was for the audio software Amarra, and now Audivarna which are Mac only.

 

Cheers

[tranz]

Externally powering, clocking and filtering USB output and powering memory is still something I have not been able to do in a Mac.

 

I wholeheartedly think that an SoTM PCIe USB card is a key differentiator of a CAPS.

 

Has anyone found a Mac alternative?

 

Has anyone ever tried externally powering the memory sticks in a Mac?

 

Cheers

[Superdad]

I wholeheartedly think that an SoTM PCIe USB card is a key differentiator of a CAPS.

Has anyone found a Mac alternative?

 

Hi Tranz: I'd love to consider pursuing/developing one for us Mac enthusiasts, but the only obvious connection in/out (for an external ultra-USB box) would be via the Thunderbolt port. While there are some Thnderbolt>PCIe card adapters available, they themselves are probably not optimum (power source, regulators, extraneous noisy circuitry?), and since the effort would still involve development of a USB card, I think the whole thing would be best as a dedicated, from-scratch endeavor. A potentially expensive development project--especially since s/w drivers would be required.

On the other hand, Thunderbolt via OS X already supports Ethernet, video, Firewire, audio/mic, etc.--as seen on those $200 hubs--so doing one dedicated to USB (probably 2.0 rather than 3.0) should not be a massive project or we would not see such devices popping up (though it too a while for them to do so). Does anyone know if those hubs all come with their own proprietary drivers, or does Apple do its usual good job of building support into the OS?

 

Just thinking out loud--way too late at night--after a long session of tracking down and ordering parts for forthcoming PS production (my big electrolytics are out of stock most everywhere, argh!).

Ciao,

Alex

[tranz]

Hi Tranz: I'd love to consider pursuing/developing one for us Mac enthusiasts, but the only obvious connection in/out (for an external ultra-USB box) would be via the Thunderbolt port. While there are some Thnderbolt>PCIe card adapters available, they themselves are probably not optimum (power source, regulators, extraneous noisy circuitry?), and since the effort would still involve development of a USB card, I think the whole thing would be best as a dedicated, from-scratch endeavor. A potentially expensive development project--especially since s/w drivers would be required.

On the other hand, Thunderbolt via OS X already supports Ethernet, video, Firewire, audio/mic, etc.--as seen on those $200 hubs--so doing one dedicated to USB (probably 2.0 rather than 3.0) should not be a massive project or we would not see such devices popping up (though it too a while for them to do so). Does anyone know if those hubs all come with their own proprietary drivers, or does Apple do its usual good job of building support into the OS?

 

Just thinking out loud--way too late at night--after a long session of tracking down and ordering parts for forthcoming PS production (my big electrolytics are out of stock most everywhere, argh!).

Ciao,

Alex

 

Thanks Alex.

 

It is hard to know, without spending the considerable extra cash and time. whether daisy chaining Tbolt devices (sonnettech, magma) will do more harm than good. Or using a Mac mobo (like Lechu) that has a PCIe connector with the unicorn USB board for Mac.

 

But designing, and profitably bringing a Mac USB Audio card to market seems almost impossible. If the boys at SoTM can't...

 

It is interesting that companies that specialized in PCIe audio cards like RME and Lynx do not have a pro-audio solution for USB out for Mac either.

 

By the way, does your PSU design have a separate rails and method to power the RAM slots? Have you found a way to bypass where the 12V input is stepped down in the various voltages needed downstream on the mobo in order to feed them the voltages using multiple rails? Sorry if you have already explained it in one of your threads.

 

Cheers

[Superdad]

It is hard to know, without spending the considerable extra cash and time. whether daisy chaining Tbolt devices (sonnettech, magma) will do more harm than good. Or using a Mac mobo (like Lechu) that has a PCIe connector with the unicorn USB board for Mac.

 

After looking at it some more, it is easy to see that Thunderbolt involves anther layer of controller chip between it and the PCIe bus--just so that it can multiplex together all the other interface functions (display, USB, Firewire, Ethernet, etc.). I think it is is the wrong direction.

 

But designing, and profitably bringing a Mac USB Audio card to market seems almost impossible. If the boys at SoTM can't...

 

Well, not impossible, but there would be some s/w headaches...

 

By the way, does your PSU design have a separate rails and method to power the RAM slots? Have you found a way to bypass where the 12V input is stepped down in the various voltages needed downstream on the mobo in order to feed them the voltages using multiple rails?

 

The JS-2 has two rails, and their output voltages are set with separate switches (see back panel pic). But while I am offering a Mac mini DC conversion board (hopefully with linear fan controller with the first version), our supply is meant to be usable for everything from PCs to drives, to small DACs and headphone amps. So we are not getting into highly specialized (and motherboard soldering required) extreme tweaks such as separate RAM powering lines and removal of any DC-DC converters that are on motherboards. Of course someone could do that and use the second rail of our supply to run it. (Output labeled DC-2--the one with the "sense" switch and SMA jack near it--can be set of any voltage down to 1.25V with the right resistors at the other end of the SMA-jack-connected coax line; so 3.3V is no problem.)

 

[pooger]

Thanks Alex.

 

It is interesting that companies that specialized in PCIe audio cards like RME and Lynx do not have a pro-audio solution for USB out for Mac either.

 

Cheers

 

I think it's more a matter of only Mac Pros still having slots in them, and the fact that pro audio hasn't used USB in quite a long time. So, if you want a pro sound card, you either have a PCIe card interface (in your older Mac Pro, or externally in an expansion chassis for the new paint can Mac Pros) that ties into a separate box (like the Echo set up I used to use) with the converters and pre-amps all on their nicer power supplies, or you get a separate box that just connects via firewire. In ways, it's more interesting to me that home audio is still using USB rather than firewire.

[Audio_ELF]

Actually both RME and Lynx have released USB devices more recently whereas their earlier interfaces were FireWire (or PCI/PCIe). In addition, both RME and Apogee have also started to create Thunderbolt interfaces.

 

Even in studios using USB interfaces with Windows based computers; I doubt you'd find any SotM or PPA USB card though...

 

Eloise

 

PS I'm not suggesting this shows USB is better than FireWire just more ubiquitous.

I think it's more a matter of only Mac Pros still having slots in them, and the fact that pro audio hasn't used USB in quite a long time. So, if you want a pro sound card, you either have a PCIe card interface (in your older Mac Pro, or externally in an expansion chassis for the new paint can Mac Pros) that ties into a separate box (like the Echo set up I used to use) with the converters and pre-amps all on their nicer power supplies, or you get a separate box that just connects via firewire. In ways, it's more interesting to me that home audio is still using USB rather than firewire.

[sandyk]

Even in studios using USB interfaces with Windows based computers; I doubt you'd find any SotM or PPA USB card though...

 

Eloise

 

I would be surprised if they didn't at least use some kind of USB isolation though.

Then again, with the loudness wars , they probably wouldn't notice much benefit anyway.

 

In ways, it's more interesting to me that home audio is still using USB rather than firewire.

Perhaps partly because not too many computers have Firewire ports on the motherboard these days, and onboard Optical Out is pretty lacklustre ?

[miguelito]

Holy molley!

[miguelito]

Update 6 - DO NOT HAVE A FAN...PERIOD

 

Thanks to lechu's recommendations I also purchased a Streacom case and was able to get the Mac Mini setup to be fanless.

 

Even though I had the fan setup to be powered externally it is still shocking how much of a difference it made removing it all together!

 

This had me curious as to why it would make such a difference when I was not taxing the mobo's power.

 

The pictures below will tell the story, and I cannot recommend enough that a music server should NOT have a fan! A more than doubling in electrical EMI field and a more than 10-fold increase in magnetic EMI field with the fan on.

 

The last picture is the nastiest one with a 22mG field, which when placed next to the CPU causes listening fatigue/increased noise floor.

 

Baseline EMI field with Fan Off

 

Electric

[ATTACH=CONFIG]11503[/ATTACH]

Magnetic

[ATTACH=CONFIG]11504[/ATTACH]

 

EMI field with Fan ON

 

Electric

[ATTACH=CONFIG]11505[/ATTACH]

 

Magnetic

 

[ATTACH=CONFIG]11506[/ATTACH]

 

How about mumetal under the fan? prob an easy mod in the factory case.

[Superdad]

Miguelito: You can't put mu-metal (or any other EMI/RFI absorbing/blocking material) on the bottom or top of the fan--it needs to be open on both sides for air flow!

 

 

Tranz:

I am not at all convinced that it is the magnetic field (generated by the fan bing a little motor with coils) which accounts for the fatigue.

We have been looking at and measuring the 25KHz PWM signal pulses going between the fan and the motherboard (this is how 3 and 4-wire computer fan speed controllers work, as it would take too much space and generate too much heat to have a variable DC supply on the motherboard to adjust fan speed). We think the current flowing on the fan wires and the noise of those pulses is the real culprit. (That's why we are working on a special linear controller circuit for the mini's fan--and one of its chips has to have a good size heatsink on it! But I am not allowed to talk about products or development here; an update e-mail will finally be going out soon to those of you on my "list.")

 

I confess that up until last night, I had been trusting (based on reports from Tranz, BobL, and others here) that getting rid of the fan's electrical nasties would be a good thing sonically--perhaps on par with other supply and OS tweaks--but I had done comparisons in my own system. Well yesterday, before I completely tore apart my mini again for more physical fitment measurements, I listened, then pull the fan entirely, then listened again (to beginnings of my usual test tracks, and just for 7 minutes total as I did not want to overheat anything with no fan at all).

What a shock! Wow, without the fan--or should I say without all the PWM nasties--it sounded fantastic (ooh, an unintentional pun!). Quite noticeable, with great ease and more acoustic and detail.

But for the Mac mini at least, you have to have a fan, and the computer should be able to adjust its speed based on thermal needs.

 

Cheers,

Alex C.

[BobL]

How about mumetal under the fan? prob an easy mod in the factory case.

 

Well, you could, and I spent some time thinking about exactly how to do this before I got the results in my post no. 142 above (04-27-2014). However, at least for my system, I have not bothered because I get no difference between the Mini fan with external power, and with zero power to the fan. Similarly Alfe prefers his external fan, with its magnetic field, to a no-fan setup. The discussion above by others indicates fairly strongly that the SQ culprit is probably the pulses from the internally powered fan, not the magnetic field.

 

If you do want to try shielding, it would be much better to try to 'enclose' the magnetic field, not just put some under the fan. The mumetal should be on both sides, and so far as possible these should be magnetically connected, but you still have to allow air to flow. Incidentally, if anybody wants a mumetal sheet, I have one which is now surplus to requirements.

[miguelito]

Ah, ok. My gut tells me the nasties come through the powering of the fan, not EMI (but of course not having actually experimented I wouldn't know for sure). Seems you've confirmed this. Is your device essentially a filter (there's a fan filter at SOtM Soul of the Music but it's for PC like fan connections). It seems to me the best solution is heatsinking heavily and remove the fan altogether.

 

BTW... Is there a way to "clock down" a mac so that you'd get less power consumption?

 

Finally: Can I get on your special list?

[Superdad]

Ah, ok. My gut tells me the nasties come through the powering of the fan, not EMI (but of course not having actually experimented I wouldn't know for sure). Seems you've confirmed this. Is your device essentially a filter (there's a fan filter at SOtM Soul of the Music but it's for PC like fan connections). It seems to me the best solution is heatsinking heavily and remove the fan altogether.

 

BTW... Is there a way to "clock down" a mac so that you'd get less power consumption?

 

Finally: Can I get on your special list?

 

I have to be careful here, but the SoTM is a filter for the fan lines, but the pulses remain. The Mac mini fan has a high-current PWM circuit in it and lends itself to a more comprehensive approach. Our device takes the signal coming out of the motherboard, runs it into a power opamp (designed for servos) and along with a cap and some other circuitry delivers a smooth, variable (based on the Mac and OS X reading the tach line and temp sensors) DC voltage back to the fan. I am not explaining it well, but I'll get it straight when we release it.

 

No fan would be nice, but you can't do that with the Mac mini's form factor, and the mini has other advantages which are hard to replicate in other ways unless someone gets extreme.

 

As for under-clocking, there is a utility called CoolBook (CoolBook ; meant for extending battery time and keeping heat down on Apple laptops), and while it supposedly does not work on anything past Snow Leopard and they say that i5/i7 processors are not supported, I have some links and study to do on hacks that some people have made to allow CoolBook to run under Mavericks (not sure about i5/i7, but my music server is Core 2 Duo).

 

Lastly, yes miguelito, you are on the "list." You have not missed anything as I have been too busy with the project to compose the bulletins (Got to cut down forum posting time, but conversing with you guys is my social life! Wow, that sounds pathetic. Well what can I say, I live way out in the country with 3 kids and a wife that travels a lot on business; This place is more interesting than my chats with the butcher and produce people at the market…)

[miguelito]

[Superdad]

Well, you could, and I spent some time thinking about exactly how to do this before I got the results in my post no. 142 above (04-27-2014). However, at least for my system, I have not bothered because I get no difference between the Mini fan with external power, and with zero power to the fan.

 

I have previously told BobL that those results are what inspired us to dive into doing the linear fan controller. He will be rewarded. And after hearing my system last night without the fan, I am even more excited about all this.

[tranz]

Wireless Connection Update

 

Since I had ripped out the wireless antenna from the Mac Mini, but still require an iPad connection for song access, as well as for internet radio, I had my Ethernet cable plugged in. This still did not sit well with me since I also have TV connection, modem, routers, switches and all other computers hardwired to the same Ethernet switches. And yes, even though Ethernet helps in lifting ground and has small transformers to help block noise, it still was not enough. I have found a few helpful bits of kit to still have a wireless connection to cut any potential noise from the other networks, but not have it powered by the motherboard and allow for physical distance.

 

1. NETGEAR Universal N300 Wi-Fi to Ethernet Adapter (WNCE2001) for $47

 

This is a handy small wireless receiver with one LAN out connection. This is tied, still through the Acoustic Revive filter to the Mac Mini.

 

2. HY1503C Variable Linear DC Power Supply. Analog. $70

 

To remove the switching supply wall wart, I use a bench LPSU to provide 1A, 5V

 

3. A connector plug for DC power. 1.3mm x 3.4mm DC connector. 171-3224-EX from Mouser.com

 

Overkill? Probably, but it is a cheap way to add separation.

[Superdad]

Tranz: Please remind me how you are getting music tracks to you mini. In other words, where are you storing them and what connection to the computer are you using for retrieval of tracks at play time? IIRC, you are using a SATA connected SSD or HD, is that correct? I assume that you are not streaming the files via your new Netgear adapter--rather just using that to make iPad remote control possible.

 

How do you manage your music library? Is your music Mac mini headless or do you Screen Share/VNC to it from another computer?

 

Best,

 

ALEX

[tranz]

Hi Alex,

 

The wireless access is purely for using Apple Remote and VNC apps on the iPad for a headless, mouseless, keyboardless setup.

 

All my music files are stored on a WD thunderbolt external HD powered by a linear PSU.

 

Amarra in cache mode loads the songs into memory and I can see the WD stop streaming before playback starts.

 

Your question does make me wonder if storing the files on a NAS is possible again, and it will help reduce electrical noise by no longer having the thunderbolt connected. Amarra streaming is not possible, but purely in Cache and Playlist mode it might work.

 

Cheers

[EuroDriver]

I was working on moving my DH61DL mobo from a Streamcon 8 enclosure to a Fractal Node 304 and had a 140 mm Noctura fan powered off the mobo 5 volt supply as a backup when the temperatures got too high. DAC was an Exasound E20, amp a Cambridge Audio 650A and a pair of Canton 720.2 DC speakers. Turning the fan on produced an immediate SQ deterioration.

 

So same upstream rig moved from Southern France to Germany, but hooked up balanced to Hypex NCore and B&W 805D and when I turn the fan on, I can hardly hear any SQ degradation ! What's going on ?

 

my speculation

 

Perceived SQ degradation is due to a rise in the noise floor caused by electrical and magnetic field of the operating fan.

 

- the AC mains is cleaner in Germany due to no other consumers, fridge TV etc

 

- the Hypex NCore has much lower noise floor the the CA 650A

 

- the effect of an increase in noise floor is not that speaker sensitive

 

so the additional fan caused noise on top of a significantly lower noise floor is barely perceptible

 

could the balanced connection from the DAC to the Amp be reducing the effect of the fan noise ?

 

Can some one enlighten me on how noise levels add up and a good text to read on the subject of noise summation ?

[fmak]

[ATTACH=CONFIG]9857[/ATTACH][ATTACH=CONFIG]9857[/ATTACH]

 

There is another solution external fan 6,7 db max temp 35°C

 

How have you (or anyone else ) measured this? In an anechoic chamber?

 

It is hard to fine ambients with at 30 dB linear or less.

[fmak]

Yes I did check temp for SSD,Ram etc.. nothing over 40°c the fan 120mm 43,7 CFM

 

NF-S12A ULN Ultra Low Noise 120mm Cooling Fan, 600/800 RPM

 

a sales pitch for the unknowing. The A weighting rolls of sharply at low and high frequencies. dBC or dBlinear tells

you how noisy a fan is.

 

There is also the issue of piezoelectric effects from vibrations, as well as noise from the control loop.