Mac OS 9 Lives

Mac OS 9 Discussion => Hardware => Topic started by: urdvurk on September 05, 2016, 04:30:17 AM

Title: MDD silencing effort
Post by: urdvurk on September 05, 2016, 04:30:17 AM
Having recently gotten my hands on a nice dual 1 GHz MDD I decided that that could do with some quieting too (after my previous effort of silencing a Quicksilver G4 (http://macos9lives.com/smforum/index.php/topic,3391.0.html)). I knew these were loud, but hearing it in person is something else again! I used to have an old Proliant server that was similarly loud, but at least that didn't try to pass itself off as something you might want to put on your desk. To me that's what it sounds like the most: it turns the room into a server room. I honestly can't imagine how people can live with this thing on a daily basis.

But enough complaining, I did manage to get it down to manageable levels, and there's more to come. Here's how I did it.

I started out with the CPU fan. The PSU fans are loud too, but not as easy to get at. Right away you notice how the fan isn't actually pointing at the heatsink all that much, and moreover it is not pointing at the part of the heatsink where the CPUs are. So I simply tried swapping it with a 120mm Papst that I mounted right on the heatsink:

Simply put rubber mounting things on the fan:

(https://c7.staticflickr.com/9/8515/28845649494_11a19f3552_b.jpg) (https://flic.kr/p/KWZuCo)

...and push down on the heatsink. The rubber things will wedge themselves between the cooling fins, holding the fan in place.

(https://c2.staticflickr.com/9/8285/28848378193_0c1b0d8ed1_b.jpg) (https://flic.kr/p/KXetLV)

I only have two drives, so I fitted them in the front and removed the rear drive cage to make room. I ran the fan off the logic board connector, with a kludged fan splitter to make it fit. I figured that since the original fan, while 150 cfm, was only blowing over the heatsink for less than half of its surface, surely I could replace it with a 55 cfm fan that was placed better, right? Wrong. Temperatures went up straight away, and the fan was running at full speed all the time.

So then, having seen what this guy did (http://www.xlr8yourmac.com/systems/mdd_cooling_mods/MDD_cooling_mods.html), I put back the original fan and kept the additional fan for extra capacity. That worked: originally the system ran at slightly over 59 degrees, and this pushed it down around half a degree. That may not sound like much, but it does matter: the fan starts ramping up at around 57 degrees, and every little bit you can take off the idle temperature will have the fan running at a lower speed, resulting in less damage to your hearing.

I also tried raising the fan by about 40 mm to point straight at all of the heatsink (leaving out the DVD drive cage for the moment).

(https://c2.staticflickr.com/9/8555/29470159945_c58a1eb996_b.jpg) (https://flic.kr/p/LUbgPX)

That had pretty much the same effect. Temperatures would be lower, but more importantly, the cooling system was more effective. It would idle at around 58,5 degrees, and when the fan started ramping up when temps went up, it would actually push temperatures to below what they were at idle. Temperature under load would be lower than at idle.

(https://c2.staticflickr.com/9/8047/28849095633_643eb029a2_b.jpg) (https://flic.kr/p/KXia3z)

The red line is the CPU board, the bumps are temperatures rising under load, and then being pushed back down by the fan spinning up.

The green line is one of the hard disks, and in the second bit of graph there I had swapped the two cpu fans: the original on top of the heatsink, and the new one in the position of the original. As you can see this didn't do much except raise HD temperatures, so I changed them back, and HD temps were back to normal.

The biggest difference however came from replacing the PSU fans. I read somewhere that the official Apple replacement fans were only 25 cfm, down from the original 38 cfm, so I thought it would be OK to fit a single 55 cfm fan. I took out the PSU, removed them, and fitted a single 120 mm fan in their place:

(https://c7.staticflickr.com/9/8128/28845644174_7ae5a9d9e4_b.jpg) (https://flic.kr/p/KWZt3E)

Unfortunately I can't show you better pictures, I had to sort of tape it into place as the PSU went back in, so now it's stuck there and I can't remove the PSU without cutting it all off again. However I can tell you generally how to do it and what to watch out for. I started by cutting the basic shape of the fan surround, and making preparations for sticking the fan in. I connected it to one of the fan terminals inside the PSU, by cutting off the connector off one of the original fans, stripping the cut wires and filling them with soldering tin. These I then stuck into the new fan's connector.

First, I taped part of the fan surround over the slot on the back of the PSU that allows for the fans to be removed:

(https://c3.staticflickr.com/9/8398/29362525602_375b03d66c.jpg) (https://flic.kr/p/LJEBTq)

I used thin cardboard, and made sure it fit flush with the PSU housing, so there would be no trouble fitting the PSU back in the case. I found that the fan surround would only fit into place if it would fold flat against the PSU as it is moved into position, so at first I only taped it to the back of the PSU, then sort of folded and taped it into place as the PSU went back into position, fitting the fan inside along the way. I was going to maybe drill some holes in the PSU housing later to improve airflow, but seeing as how that would involve taking it all out again, I don't think I'll bother. I did cut out the fan grilles in the PSU housing. The new fan now runs at 9-10 V, and compared to the original fans it is whisper quiet.

With the PSU now completely in the background, it turned out the original CPU fan gives off a sort of whining whirring sound on contact with the case that is surprisingly annoying. Because of this I fitted the fan in its raised position with thick double-sided tape, which eliminated the vibration completely. The result so far:

(https://c7.staticflickr.com/9/8314/29435889206_ec585aeb41_b.jpg) (https://flic.kr/p/LR9Ck9)

Note the bit of cardboard in the back to guide air over the heatsink rather than straight out the back of the case. Hardly radical modifications, but the MDD is now producing an acceptable level of noise, rather than the server-level racket it produced before; around 9 dB less. I was actually considering replacing my Quicksilver with it, until I booted into OS 9: has anyone ever found a solution to the fans running that much faster in OS 9? I do have the firmware upgrade installed that was supposed to help.

So, next is some supposedly high-end thermal compound; if that will get the temperature down by even half a degree that will help keeping the insan-O-fan running at a lower speed, which would be a good thing. Speaking of which: I noticed that the DVD drive cage is just over 140 mm wide:

(https://c2.staticflickr.com/9/8705/29470175545_6360a9f31f_b.jpg) (https://flic.kr/p/LUbmsV)

...and the fan in its raised position will still fit vertically inside the bottom drive position:

(https://c3.staticflickr.com/9/8474/28845654914_9308eba253_b.jpg) (https://flic.kr/p/KWZweQ)

...so cutting out a bit of the drive cage floor will allow me to fit both the DVD drive and the fan in its raised position. I have ordered a 140 mm fan to replace the original CPU fan, a Fractal Design HF-14. It's still 118 cfm, which should be fine, but at 26.5 dB it's a LOT quieter than the stock fan.

To be continued!

Title: Re: MDD silencing effort
Post by: devils_advisor on September 05, 2016, 05:37:19 AM
good job, looks like a winter project i might  take on.
Title: Re: MDD silencing effort
Post by: widdly on September 05, 2016, 06:11:11 PM
Does OS9 support the fan temperature sensor hardware? 

On my emac, the fans have temperature sensors but they run full speed all the time in OS9.  I ended up wiring a potentiometer in.
Title: Re: MDD silencing effort
Post by: urdvurk on September 06, 2016, 01:36:12 AM
Fan speed does vary with processor load, so I guess so. It seems to use a much more aggressive profile though; when I rebooted from OS 9 into Tiger, CPU temperature was 56 rather than the usual 59 degrees.

Wiring a potmeter in is not a bad idea at all, I will probably give that a go.
Title: Re: MDD silencing effort
Post by: GaryN on September 06, 2016, 05:53:04 PM
Does OS9 support the fan temperature sensor hardware? 

No.
Title: Re: MDD silencing effort
Post by: miracman on September 06, 2016, 06:50:25 PM
Nice job there hehe :)

I'm growing tired of my noisy MDD, and I found 2 Gigabit Ethernet DP 500 models for peanuts really, so.. not sure what to do here...  They're so quieter (plus you can fit that old school mini jack mac microphone, in the pre MDD native soundcard-not sure what was in there before the Texas Instruments one).  :)

Anyways, there's one detail worth mentioning in that setup (and I hope to be corrected if I'm wrong), and maybe it's just a silly detail after all as well.

Your model has one ATA-100 plug for two HDDs (which would most probably be the rear bay you've condemned), and one ATA-66 (which would be the one you're using).

It might have no impact in speed whatsoever, I'm no expert here, and I keep reading in the posts about bottlenecks I wouldn't have thought about and real techy stuff and so on, so I don't know, but still.. possibly maybe, probably, your hard drives are hooked on the ATA-66 thing. 

Maybe it's easy to just switch the inputs on the board if that has any real speed impact at all.

Anyways, keep us informed :)

Title: Re: MDD silencing effort
Post by: DieHard on September 06, 2016, 09:21:43 PM
From Miracman
Quote
Your model has one ATA-100 plug for two HDDs (which would most probably be the rear bay you've condemned), and one ATA-66 (which would be the one you're using).

Yes, unfortunately, there will be a noticeable speed decrease booting and using the slower PATA Controller, But simply buy an SSD with an SATA/PATA adapter and Velcro it in place; load the OS and all apps (60GB OWC SSD on ebay for $25);

You can also Mod it by checking out this (works under OS9):
http://macos9lives.com/smforum/index.php/topic,687.msg2114.html#msg2114
Title: Re: MDD silencing effort
Post by: geforceg4 on September 06, 2016, 09:25:45 PM
good job i didnt read the entire post but i plan to ;)
Title: Re: MDD silencing effort
Post by: urdvurk on September 07, 2016, 03:03:24 PM
Right, I've got an update: I picked up the new thermal compound (Gelid GC Extreme, which is supposed to be very good) today and have just applied it. And it's a result!

I was looking for a small decrease in CPU temperature, and that's exactly what I got, and a little more: it now idles at 56 degrees C, and goes up to 57 under load - right at the beginning of the fan ramp and just enough for the first little step up in fan speed. The higher fan speed then pushes it straight back down again.

Here is the before and after graph:

(https://c8.staticflickr.com/8/7549/29448114271_47816697ee_b.jpg) (https://flic.kr/p/LSehq8)

You can see right away that it takes a lot longer to get up to temperature, showing that the processors can get rid of their heat much more easily, and that it ends up at a lower idle temperature. The little bumps are Geekbench runs starting, and the subsequent step-up of the fan. Except this time, I had to turn down the music to hear if it had actually stepped up. Big difference from before.

So anyway, in Tiger the MDD is now completely usable, it simply never gets loud. I can't wait for the new fan to arrive, that should really take it down to whisper-quiet level.

However in OS 9 it is still loud of course. The new fan should help there too, but think I'll get one of those separate temperature control boards and try to get that to work. If it does, that will give me OS-independent temperature control, and the OS 9 noise problem should be gone.

After paying proper attention I noticed that fan speed does vary in OS 9, but not with processor load: it starts high, then slowly goes down a couple of steps, then holds. Does anybody know how or why?
Title: Re: MDD silencing effort
Post by: urdvurk on September 07, 2016, 04:05:30 PM
Your model has one ATA-100 plug for two HDDs (which would most probably be the rear bay you've condemned), and one ATA-66 (which would be the one you're using).

It might have no impact in speed whatsoever, I'm no expert here, and I keep reading in the posts about bottlenecks I wouldn't have thought about and real techy stuff and so on, so I don't know, but still.. possibly maybe, probably, your hard drives are hooked on the ATA-66 thing. 



Thanks for pointing that out, I thought I remembered seeing markings on the logic board for ATA-66 and ATA-100 next to the front connectors. Wishful remembering, clearly, because you are quite correct: only the back one is ATA-100.

Fixed it, for now at least, with a longer IDE cable.
Title: Re: MDD silencing effort
Post by: urdvurk on September 11, 2016, 03:10:47 AM
Another update: the new 140 mm fan has arrived, and I have been experimenting with it for a while. It's a Fractal Design HF-14, rated at 118 cfm, which I figured would be enough since the original fan now would only run at low speed, even under load. It's got removable rubber corners, which is fortunate because it turns out it may need some careful positioning to avoid the bits of the CPU board that stick out from under the heatsink. Being able to remove a corner piece of the fan will definitely give me some extra room, so that was lucky.

So I put it in, and while it was able to keep temps in check, it had to run faster than I liked. I figured it might be a better idea to sit another fan on top of the heatsink, with both fans running at low speed, blowing towards the heatsink. This worked quite well, I ran both fans off a PCI slot fan speed controller, and was able to run the 140 mm fan at 8 V, with the 120 mm fan at 9 V. Temperatures would be 57 degrees when idle, 58-59 under load.

Incidentally, this is also how I solved the OS 9 noise problem: I ran the computer at full load in Tiger, found a fixed fan speed setting that would keep temperatures in check but was still quiet, and left it there. Booting into OS 9 would then have the fans running at the same low speed. Problem solved.

By the way, this is where it starts to get a little more involved. If you are simply looking for a quick and easy way to get the MDD's noise down, I would simply replace both the PSU and CPU fans, and be done with it. I would probably use a CPU fan with a higher CFM rating, rather than the one I used, which is geared more towards silence. This would allow the fan to be run at very low speed, and net you a system that, while not completely quiet, is still perfectly acceptable to work with.

However, I can never leave well enough alone, so I kept fiddling. I also ran into a problem: over time, around an hour or two, temperatures would creep up to over 59 degrees when idle. Also, when I opened the case, then closed it again, temperatures would spike temporarily but then drop to a more healthy 56-57 degrees, only to then creep up again over time. Clearly it wasn't getting rid of hot air properly, and heat was being trapped in the case. I reversed the direction of the fan on top of the heatsink, to try to blow hot air out of the case, but this didn't solve the problem.

So I set to work: by now some temperature sensors I had ordered had arrived, and I installed them: on in the front of the case near the bottom, one where the bottom DVD slot would be, one in front of the heatsink, one behind the heatsink, one next to that in the back of the case where hot air is supposed to exit, one on the video card and one in front of the PSU fan. I removed the side of the case, and stuck the LCD readouts more or less where the sensors are located:

(https://c7.staticflickr.com/9/8156/29491821582_cabf6848dc_b.jpg) (https://flic.kr/p/LW6i5s)

This is an 'after' picture, with the 'CPU in' temperature at a comfortably low 32-ish degrees. Before, 'PSU in', 'out', 'out 2' and, most importantly, 'CPU in' would all creep up to around 40 degrees, hindering cooling. To fix this I compartmentalised the case, making sure that the CPU fan would only get air from the front of the case, and that air from the heatsink fan would be guided out. Temperatures are now down all around, 'CPU in' is now clearly separated from the other sensors, and the 'out 2' sensor, which is located here:

(https://c3.staticflickr.com/8/7495/29311482650_3ece50dfdc_b.jpg) (https://flic.kr/p/LEa1Aw)

is now around 2 degrees higher than the rest of the case, showing that heat is in fact being removed there. The inside now looks like this:

(https://c7.staticflickr.com/9/8097/28976317734_d9f57072d5_b.jpg) (https://flic.kr/p/L9xcK5)

It doesn't look like much, but it does work: CPU temps are now around 53 degrees idle, 54 degrees under load, with the fans running at around 8 V, or quiet enough to have it sitting right next to me on the desk. Weirdly, I seem to have stumbled upon some kind of fan speed sweet spot: running them at lower speed will raise temperatures, but running them at higher speed will also raise temperatures. It's almost as quiet as my Quicksilver, the loudest sound is now the hum from the hard drives.

So, next is to put the new fans in properly and permanently. For this I conveniently have two Quicksilver fan surrounds lying around, and what do you know:

(https://c8.staticflickr.com/9/8003/28978707343_b1e79c43d9_b.jpg) (https://flic.kr/p/L9Ks6e)

It's a straight fit! Well, almost. However, note how the hole at the top exactly coincides with the screw hole that fixes the MDD drive cage in place. I'll need to make a big hole in the side of the case of course, like there is in the Quicksilver, so no drives are ever going to live here again. The other fan surround I'll combine with the DVD cage, which I will also make airtight to make sure air only comes in through the front, and to improve airflow I'll remove bits of the case here and there. I may even take out the PSU again and open up the casing a bit to improve airflow, but I am still trying to figure out a way to mount the fan in place that will allow removal and refitting of the PSU.
Title: Re: MDD silencing effort
Post by: MacTron on September 11, 2016, 08:56:50 AM
... years before, I had faced the same problems and I had tried most of your solutions, but not into noise reduction effort but a extreme cooling effort.
With this goal, I had found that evacuating hot air from the MDD case is very important, at some point.

After paying proper attention I noticed that fan speed does vary in OS 9, but not with processor load: it starts high, then slowly goes down a couple of steps, then holds. Does anybody know how or why?

The CPU fans speed vary upon the CPU module temperature. You can check this with a half an hour video encoding at different room temperature, by example.

Does OS9 support the fan temperature sensor hardware? 

Of course it does. But depends on the Mac model. The temperature sensors was changed more times than the third part software to show it :)  but the apple software was always aware about the temperature sensors measures.
You should keep in mind that CPU temperature in Mac Os 9 isn't as smooth as in Os X - especially in dual CPUs configurations- because Os X can distribute task between CPUs more efficiently .

The CPU fan vary its speed as its required by temperature measures. I strongly recommend against replacing this with a fixed speed fan -as some are saying- , because at some point you could end with fried CPU module.

In this picture we can see the original CPU fan placed in the way that it's more efficient in cooling the CPU. As a consequence of this, it works at lower speeds and it is more silence. ( I should had removed some dust before take the picture ) :-\

(http://macos9lives.com/smforum/index.php?action=dlattach;topic=927.0;attach=242;image)

Here we can see the CD-ROM bay support that I had to make to let room for the new fan placement.

(http://macos9lives.com/smforum/index.php?action=dlattach;topic=927.0;attach=244;image)
Title: Re: MDD silencing effort
Post by: DieHard on September 11, 2016, 10:07:55 PM
It should also be noted, as Mactron and I have discussed in the past, the "Copper" MDD G4 Cooler/Heat-sink is a huge plus when it comes to keeping the CPU cool, probably get at least 10 degrees with a Copper one and "Arctic Silver" or similar paste like the one you mentioned.

Remember, at lower temps, the fans spins slower... and quieter.  Since my all my G4 MDDs use Single 1.33 Ghz. CPUs, I run the stock Aluminum (non-heat pipe) heat-sink and it does fine.

Apple Part Number 076-0983 is for the MDD G4 1.42 Dual but can obviously be used with all MDD G4s single or Dual
(http://ep.yimg.com/ca/I/lovemacs_2433_1470884813)

Here is the Aluminum and the Copper side by side
(http://www.xlr8yourmac.com/systems/MDD_copper_heatsink_swap/image004.jpg)

Lastly, note that Mactron... Owner of "M.A.R.L" uses the copper in the photos above :)
Title: Re: MDD silencing effort
Post by: urdvurk on September 12, 2016, 03:00:25 PM
... years before, I had faced the same problems and I had tried most of your solutions, but not into noise reduction effort but a extreme cooling effort.
With this goal, I had found that evacuating hot air from the MDD case is very important, at some point.

Definitely, but especially if the main CPU fan gets its air from the general environment of the case. I reckon if you added some separation between the intake side of the main fan and the rest of the case, temperatures would be down.

The CPU fan vary its speed as its required by temperature measures. I strongly recommend against replacing this with a fixed speed fan -as some are saying- , because at some point you could end with fried CPU module.

This is why I ran the processors at 100% for extended periods of time while I was mucking about with fan speeds. The fan speeds are now such that they will keep temperatures down to below where the stock fan would start to ramp up, basically indefinitely, while the computer is running under full load. Thankfully, they are still much quieter than the stock fan running at its lowest speed. I have high hopes that the finished product will perform even better, but it turns out it's quite a lot of work. Aluminium and steel is harder to work on than cardboard, who would have guessed?

Apple Part Number 076-0983 is for the MDD G4 1.42 Dual but can obviously be used with all MDD G4s single or Dual
(http://ep.yimg.com/ca/I/lovemacs_2433_1470884813)

Here is the Aluminum and the Copper side by side
(http://www.xlr8yourmac.com/systems/MDD_copper_heatsink_swap/image004.jpg)

Yes please.
Title: Re: MDD silencing effort
Post by: mrhappy on September 12, 2016, 09:32:17 PM
The copper sink even LOOKS cool!! ;D
Title: Re: MDD silencing effort
Post by: urdvurk on September 13, 2016, 12:32:22 PM
Does the dual 1.25 FW800 have the copper heatsink too? There's one up for sale here and I might pick it up. The processors should be usable too, right?
Title: Re: MDD silencing effort
Post by: joetramp on October 06, 2016, 12:58:06 PM
Hi,

I've tried to silence my MDD couple minutes ago. I bought a SilenX iXtrema Pro Quiet to replace the big 120mm fan. Unfortunately the SilenX fan is broken but the stock fan isn't the main problem with the noise so not a deal breaker here. I also replaced the power supply fans with 2 EVERFLOW F126025BU but they seem to be just as loud as the original ones... so no success so far.

Can anyone recommend fans for the power supply that are actually quiet?
Title: Re: MDD silencing effort
Post by: geforceg4 on October 06, 2016, 02:07:08 PM
Can anyone recommend fans for the power supply that are actually quiet?


i used ones from fractal Design http://www.fractal-design.com/home/product/case-fans/silent-series-r2-60mm
Title: Re: MDD silencing effort
Post by: joetramp on October 07, 2016, 01:41:25 AM
Thank you geforceg4! I'm going to try these.
Title: Re: MDD silencing effort
Post by: urdvurk on October 18, 2016, 02:32:06 AM
Right, I've finally managed to put the MDD back together again. Photo of work in progress:

(https://c1.staticflickr.com/6/5751/30105480360_4f94c8935c_b.jpg) (https://flic.kr/p/MSjssd)

In the end I wouldn't necessarily recommend using the QS fan surrounds, they took a lot of mucking about to fit into place properly. The final result is slightly different again from this picture. That said, they're in and they fit very nicely. You may also notice the large hole in the back of the case, where you can see through. This is the QS-like exhaust hole to get hot air out. I've made it so the heatsink will fit straight onto the exhaust fan, with some foam to separate it from the CPU fan airflow.

So, I put all the bits back in, and started it up. It worked extremely well: CPU temp under full load was 47.5 degrees idle, 48.5 degrees under load. The exhaust fan was working very well, taking heat out from the top of the heatsink and straight out of the case:

(https://c3.staticflickr.com/6/5338/29772194274_dc79a21c3a_b.jpg) (https://flic.kr/p/MmSh7o)

(That would be the OUT 2 reading, if you hadn't guessed.) I say 'worked', because in the meantime I had gotten my hands on a dual 1.42 FW800 MDD, with the copper heatsink, so the processors of that went in, with the result looking something like this:

(https://c5.staticflickr.com/9/8132/30366541076_071b99491d_b.jpg) (https://flic.kr/p/NgosH3)

(https://c6.staticflickr.com/6/5553/30402722525_544d8af5a7_b.jpg) (https://flic.kr/p/NjzUc8)

The copper heatsink is smaller than the aluminium one, so I had to fit more foam to prevent air from the CPU fan being sucked straight out again. Also its top is flat and closed, as opposed to the fins of the aluminium one, so airflow is not as nice. I would have liked to fit the aluminium one, but a couple of the coil-like things on the processor board fouled the heatsink:

(https://c5.staticflickr.com/6/5498/30366541156_e13670aa47_b.jpg) (https://flic.kr/p/NgosJq)

...so if I'm going to do that I'll have to take some material off. I think I will though, the extra foam also obscures almost half of the CPU fan:

(https://c3.staticflickr.com/8/7772/30286051802_70a03bc427_b.jpg) (https://flic.kr/p/N9gW3W)

It still worked quite well though: there is some heat buildup in the case, but if I leave it running at full load overnight it ends up at 53-ish degrees, with about 1 degree difference between idle and full load. Not bad at all, given that the 1.42 processors will also be producing more heat, but still not as nice as it could be.

By the way, this is how I fitted the PSU fan:

(https://c3.staticflickr.com/6/5342/30286052002_3049c5ede7_b.jpg) (https://flic.kr/p/N9gW7o)

Not visible in the picture, but I cut out the facing corner of the PSU housing, and out of aluminium sheet made the bit that holds the fan:

(https://c3.staticflickr.com/6/5611/29772191714_3424d2226e_b.jpg) (https://flic.kr/p/MmSgmf)

It slots straight into the holes of the old fans, and basically clamps itself into place quite well. I made it out of a few small bit of sheet because that is what I had lying around, but if I were to do it again I would make it out of a single larger piece of sheet. The fan is still connected to the PSU fan connector.

So, with regard to the noise it makes: I can run the CPU and exhaust fans at their lowest speed, but temps will be up then, and the PSU fan will be the loudest by far. If I tune them to be just quieter than the PSU fan, I end up with the temperatures I mentioned. At that point the HD was making an annoying hum, so I bent open the drive cage and suspended the HD from simple elastic bands. It is not quite as silent as the QS, about 1.5 dB up from that, but it is whisper quiet and perfectly fine to have sit on the desk right next to me. Temperatures are much lower than stock as well, almost to the point where you have to wonder why Apple couldn't be bothered to think up something better themselves to begin with.

Up next is to fit the aluminium heatsink, and then to take it all apart again to properly fit all the cables and do some final finishing here and there.
Title: Re: MDD silencing effort
Post by: urdvurk on October 19, 2016, 04:13:38 PM
Continuing: to fit the aluminium heatsink I took a wood router to it. I'd read somewhere that wood routers can also be used for aluminium, so I decided to simply give it a go.

Before:

(https://c2.staticflickr.com/6/5832/30316608321_bf1cb8f6aa_b.jpg) (https://flic.kr/p/NbYxrt)

and after:

(https://c7.staticflickr.com/6/5801/29805559534_6c5020096b_b.jpg) (https://flic.kr/p/MpPhru)

Slightly wobbly surfaces and an error in the top edge, but otherwise quite acceptable I think. So in it went, and out came most of the foam:

(https://c1.staticflickr.com/6/5731/29805561224_02014736c5_b.jpg) (https://flic.kr/p/MpPhWC)

It turns out I did have to add another layer of foam to get a snug fit onto the heatsink, but with that it fit perfectly. This is looking in through the new hole in the side of the case, and through the exhaust fan, at the heatsink:

(https://c3.staticflickr.com/9/8670/29805564074_ecb4939f60_b.jpg) (https://flic.kr/p/MpPiML)

So, did it work as well as the copper heatsink? Well, no.

(https://c5.staticflickr.com/6/5574/30139627180_c40d3f880f_b.jpg) (https://flic.kr/p/MVkt7u)

Left line is the copper heatsink, right line the aluminium one. Spikes are stress runs and/or me opening the case to poke around, the dip at the end is enabling nap mode. In the end, what made the biggest difference was removing the side panel, which enabled the case exhaust fan to work that much better. It's weird, on the Quicksilver it seems quite able to blow out hot air that way, even with the panel in place, but on the MDD temps drop by 5 degrees when I remove the panel. Needless to say it stays off for now, and I'm on the hook for one of these (http://www.s155158671.websitehome.co.uk/Resources/Side-Panel-on.jpg).

I have since experimented some more with various configurations of heatsink and fans, and to the great surprise of probably no-one at all it turns out that both MacTron and DieHard were completely correct: you need to get hot air out of the case, and the copper heatsink works better than the aluminium one.

I've ended up with running the CPU fan off the original logic board connector. If you can keep temperatures below 57 degrees the fan will always run at its lowest speed, and you get the benefit of temperature control. Even in OS 9 it isn't that bad; the fan will typically run at 8-9 V, which is around the voltage at which I would be running it anyway. The case fan now runs at a very low speed, maybe 6 V. Idle temperature is around 54 degrees, and under load it will rise to slightly over 56. This is with the aluminium heatsink, because of the neater interior, the copper heatsink will take off around 2 degrees. I'm quite happy with it like this, it is now closer again to being as quiet as the QS. The loudest is now the PSU fan, so I'll try to get that to spin lower. I've thought of maybe thermally connecting the heatsinks inside the PSU to its casing, so heat can flow away that way, which in turn should make the fan spin slower.

The copper heatsink will probably find its way into the FW800, along with the dual 1 GHz processors, to try a minimum-effort quieting mod: replacement PSU and CPU fans, nothing else.
Title: Re: MDD silencing effort
Post by: geforceg4 on October 20, 2016, 05:19:24 PM
i havent had time to read this entire thread..
does anyone want to give a quick synopsis here? whats been accomplished? lower heat? lower noise? all of the above?
im interested to see the 120mm fan in place of the duals by the power supply..behind the speaker..
i guess i will have to read the thread to find out exactly how this was accomplished!
Title: Re: MDD silencing effort
Post by: urdvurk on November 08, 2016, 02:57:25 AM
I'll do a summary next, but the short version is that both temperatures and noise are down compared to stock.

First however there is the side panel that needed some holes to facilitate the removal of hot air from the case. Going by this example (http://www.s155158671.websitehome.co.uk/Resources/Side-Panel-on.jpg) I thought basically any holes at all would go a long way towards solving the problem of heat buildup in the case. I came up with a design that I thought would work:

(https://c7.staticflickr.com/6/5654/30819462766_ef8c520739_b.jpg) (https://flic.kr/p/NXpNxG)

To test, I took an old scuffed side panel and started to drill some holes in it. I tried it, but straight away it became clear that just a few holes were not going to cut it: temperatures went up to almost what they were with a closed side panel. So, I drilled some more holes in my test panel:

(https://c3.staticflickr.com/6/5766/30554889330_054386dc1f_b.jpg) (https://flic.kr/p/Ny2N5J)

It doesn't look like much, but the total surface area of the holes is about the same as in my design, so it should be good to test. As a side note, the panel is very easy to work on, as can be seen from the tiny threads of material left between the holes on the second row from the bottom. Polycarbonate is quite elastic, and does not break or crack easily.

There was a clear improvement, with temperatures only going up 3-4 degrees compared to with the side panel off:

(https://c4.staticflickr.com/6/5539/30220501883_23dcc75a7a_b.jpg) (https://flic.kr/p/N3tYkV)

To clarify: I started up the system and did some stress tests, during which I put on the side panel. Temperatures go up straight away as you can see. Then, after leaving it on overnight, I did some more stress tests, during which I took the side panel off again, resulting in temperatures dropping again.

Clearly, more holes are better, so first I enlarged the hole I cut in the side of the case:

(https://c3.staticflickr.com/6/5333/30554890570_592835185a_b.jpg) (https://flic.kr/p/Ny2Ns7)

...and also stuck the wifi antenna out of the way. Then, I added another row of holes to the design, and increased the diameter of all of them by 1 mm:

(https://c1.staticflickr.com/6/5453/30819462536_1b4fc60404_b.jpg) (https://flic.kr/p/NXpNtJ)

This increased the total surface area by 39%, which I hoped would be enough. After a few hours of careful drilling, I ended up with this:

(https://c6.staticflickr.com/6/5654/30220478453_45a31452c6_b.jpg) (https://flic.kr/p/N3tRnX)

Given that most of it was done with an electric hand drill, I'm not too unhappy with the result:

(https://c2.staticflickr.com/6/5798/30767996161_16135a2d48_b.jpg) (https://flic.kr/p/NSS2jr)

One more, with the panel in place:

(https://c6.staticflickr.com/6/5778/30855922965_f6b30592f0_b.jpg) (https://flic.kr/p/P1CETP)

So did it work? Thankfully, yes it did. After some more fiddling with cable routing, temperatures are now between 53 and 55 degrees, so not as low as without the side panel, but low enough for me. I expect they will drop some more over time as the thermal compound sets or cures or whatever.
Title: Re: MDD silencing effort
Post by: urdvurk on November 08, 2016, 03:47:33 AM
To summarize: my MDD is now whisper-quiet:

(https://c3.staticflickr.com/6/5744/30554891610_40faa451b4_b.jpg) (https://flic.kr/p/Ny2NL3)

This is taken at ear position, with the MDD sitting on the desk next to the keyboard, just visible in the background. I had to take that reading late at night, as during the day background noise would get in the way.

It also runs cooler than stock, below 55 degrees even under full load. This means that the central CPU fan is always spinning at 5 V. At the same time, it has gone from dual 1 GHz to dual 1.42 GHz processors, so it is much faster now as well.

To get here, I replaced the dual PSU fans by one 120 mm fan. I cut part of the PSU housing:

(https://c8.staticflickr.com/6/5617/30768003551_a42a3dd9ce_b.jpg) (https://flic.kr/p/NSS4vR)[

...and made a thing out of aluminium sheet that holds the fan:

(https://c4.staticflickr.com/6/5802/30220484643_3247e81c0f_b.jpg) (https://flic.kr/p/N3tTdF)

...which slides over the PSU housing and slots into the old fan mounting holes:

(https://c2.staticflickr.com/6/5783/30220487153_ac35fdc8ca_b.jpg) (https://flic.kr/p/N3tTXX)

In the center of the case, I replaced the big CPU fan with a 140 mm fan that blows air directly at the heatsink. This gets air from the front of the case, which I have separated as much as possible from the rest of the case, to prevent hot air being circulated. In addition, I have added a 120 mm case fan to the side of the case that sucks air straight from the heatsink and out. To do this I made a fan holder, a large hole in the side of the case, and a fan grille in the side panel.

The inside of the case now looks like this:

(https://c7.staticflickr.com/6/5324/30819418846_dd43dcc953_b.jpg) (https://flic.kr/p/NXpzus)

(https://c2.staticflickr.com/6/5449/30221884553_94a004b041_b.jpg) (https://flic.kr/p/N3B4n4)

Everything slots straight in and is held in place by the DVD cage and the black processor board support. The heatsink fits exactly onto the foam lining, so the case fan only gets hot air from the heatsink.

As a final note, it turns out that airflow below the heatsink is important as well. After rerouting a few fan wires below the heatsink bracket, right up to the ports in the back of the case, I noticed that processor temperatures were up by a degree or two. I couldn't figure it out, the heatsink was in its proper place, the wires were not even completely blocking airflow there. Still, rerouting the wires to elsewhere lowered temperatures again. So: airflow, any airflow, around the heatsink is important. Big surprise there.

This is what it looks like from the outside:

(https://c6.staticflickr.com/6/5730/30768002501_458b9bc9a3_b.jpg) (https://flic.kr/p/NSS4cK)

(https://c8.staticflickr.com/6/5458/30855921855_847b1aabc2_b.jpg) (https://flic.kr/p/P1CEyF)

...except it is now under the desk, and has cables coming out the back so it actually works. Still, as MDDs go, I'm quite happy with it.
Title: Re: MDD silencing effort
Post by: dr bu on November 08, 2016, 04:22:17 AM
Looks beatiful.✌
Title: Re: MDD silencing effort
Post by: DieHard on November 08, 2016, 07:50:06 AM
Love the outside pic :)
Title: Re: MDD silencing effort
Post by: ELN on June 12, 2017, 12:27:56 AM
urdvurk, could you please post the design for your power supply fan mount, for the record?
Title: Re: MDD silencing effort
Post by: FdB on June 12, 2017, 11:00:40 AM
Urdvurk, looks great!
Title: Re: MDD silencing effort
Post by: 1@n on June 20, 2017, 02:00:49 PM
A few years ago, after reading some sound experiments by a person named Charles Altman, I made custom wooden (mostly pine) boxes for some guitar amps, pa speakers and a bass amp.  One unexpected benefit from the reboxing was that they were all way quieter than before.  The wood absorbs and nullifies a lot of vibration.  A lot of the sound was coming off of the plastic covering the speakers had before.  Sound only comes off a final surface, it doesn't come from the middle of something.  And whatever that final surface is made of will piggyback the waves coming off a surface, like surfing.  I don't doubt that there would be a big benefit to reboxing a Mac, just don't varathane it when you're finished.  The Mac case is acting like a big, ugly resonator - plastic has horrible harmonics.  Even sitting the Mac on some pine might help.  It's important to test the pine when you buy it by hitting it and getting pieces that have pleasing harmonics.
Title: Re: MDD silencing effort
Post by: torvan on June 29, 2017, 10:43:34 PM
Love the holes in the side of the case. I would love to have a template that I could apply to the side and drill it. I am not exactly handy with a lot of modifications (I refuse to use the popular term "hack") because I am a bit clumsy, but I have a couple test side panels from a "way beyond dead" Quicksilver.

I have recently added Noctua fans to it in order to cool and silence it more. I replaced the main fan with a same sized Noctua using 4 pin to Molex connectors they provide. I used two more smaller Noctuas (same size as original ones) with the same connector for the power supply.  The fans are fairly quiet at full speed, but I have no temp monitors to measure the resulting heat in the G4 MDD Dual 1.25. I am only running 9.2.2 at the moment, but will install Tiger here soon.

I also replaced the drives with a SATA controller and two 500GB SATA 3.5" drives. I would pull in the SSD drives I have, but these two were sitting there in the tech closet looking lonely..... So I never hear any real HDD sound anymore, and they are resting comfortably in the space where the second CD/DVD would normally sit. Plus side of that location--I can change them out without opening the case (in theory).

But I would love to add another fan to the machine, just to get some of that hot air out of it given the G4 heat issue and the poor cooling this design has.



Title: Re: MDD silencing effort
Post by: DieHard on July 01, 2017, 10:21:04 AM
This is still... by far... the coolest (pardon the pun) MDD Modification ever done :)
Title: Re: MDD silencing effort
Post by: IIO on July 03, 2017, 07:50:08 AM
i have one of those in the dual 1,25, works okay but you can not really file that under "silencing".

it originally was some USD 200+ so you wouldnt buy one new today.

i have yet to move one of my G4s into a 6 HE 19" rack mount case with lots of space to put the components far from each other, eventually with the PSU outside the case, and only one big fan over the processors.

Title: Re: MDD silencing effort
Post by: urdvurk on November 06, 2017, 02:35:15 AM
Hi all,

I have been meaning to reply here but had not yet gotten around to it, sorry. Thanks for the compliments!

urdvurk, could you please post the design for your power supply fan mount, for the record?

My original squiggles would not be very useful I suspect, I mostly just made stuff up as I went along. However I have made a quick drawing of the end result, with all the relevant measurements:

(https://farm5.staticflickr.com/4539/26433862159_6651f2b228_b.jpg) (https://flic.kr/p/GgSsh8)

Left is the top and bottom part, to the right the connecting part. You could cut the whole thing out of a larger piece of sheet, but this way the top and bottom parts can be made exactly the same so it fits better. If I were to do it again I wouldn't bother with the sliding pins (https://flic.kr/p/N3tTXX) and just fix it into place some other way. This means that the indicated room to slide the thing into place is optional. From memory it is around 5 mm, so the step in the left edge of the part would move down by that much (the 18 mm dimension would become 23 mm, 20 mm would become 15).

I would love to have a template that I could apply to the side and drill it. I am not exactly handy with a lot of modifications (I refuse to use the popular term "hack") because I am a bit clumsy, but I have a couple test side panels from a "way beyond dead" Quicksilver.

The pattern with measurements is a couple of posts up, but I could make a pdf if you want it. You get the best results using a pillar drill and one of these:

(https://farm5.staticflickr.com/4564/38177832092_da81700dc7_c.jpg) (https://flic.kr/p/21aDnom)

...which will allow you to drill lines and grids of holes much more precisely. I had to use a hand drill on many holes, and it's much more difficult and inaccurate.

Anyway, hope this helps.
Title: Re: MDD silencing effort
Post by: mrhappy on November 06, 2017, 09:48:05 AM
Thanks for the info Urd!! I'm sure some people will make use out of this! ;D ;D
Title: Re: MDD silencing effort
Post by: DieHard on November 07, 2017, 08:20:31 AM
OK urdvurk.... tell the truth, did you used to work for Apple ?  Or do you just design bridges for the government, this thread gets "cooler" every time I read it
Title: Re: MDD silencing effort
Post by: torvan on November 07, 2017, 11:50:04 AM

I would love to have a template that I could apply to the side and drill it. I am not exactly handy with a lot of modifications (I refuse to use the popular term "hack") because I am a bit clumsy, but I have a couple test side panels from a "way beyond dead" Quicksilver.

The pattern with measurements is a couple of posts up, but I could make a pdf if you want it. You get the best results using a pillar drill and one of these:

(https://farm5.staticflickr.com/4564/38177832092_da81700dc7_c.jpg) (https://flic.kr/p/21aDnom)

...which will allow you to drill lines and grids of holes much more precisely. I had to use a hand drill on many holes, and it's much more difficult and inaccurate.

Anyway, hope this helps.
[/quote]

I was referring to a file that was the exact size so if I printed it, I could fasten it to the side and fire away at it.  But I also realize what I should really do is create a design of my own since I couldn't really demand a precise template from someone else without compensation for said art.
Title: Re: MDD silencing effort
Post by: urdvurk on November 08, 2017, 01:42:07 AM
That's why I suggested a pdf, I did the drawing in Illustrator on a 1:1 scale, so if you print a pdf made from that at 100% all of the dimensions should be accurate. Anyway, here you go.

A movable vise or cross table will be more accurate, but I'd be interested to see how you get on with a hand drill. If I remember correctly, I did all of the holes at a small diameter first. After drilling everything up to their final diameter I finished the edges with a countersink drill bit, but only on the outside, because the inside of the panel has a layer of silver paint. Countersinking on that side would make the holes look bigger.

You're completely welcome to it by the way - but we do expect pictures of the result.  :D
Title: Re: MDD silencing effort
Post by: torvan on November 08, 2017, 10:52:48 AM
Yeah, photos once I get it done. It will take a while though, so do not hold your breath or you will pass out!
Title: Re: MDD silencing effort
Post by: smilesdavis on August 26, 2022, 10:22:28 PM
i got one original marathon coming that i will run some current upgrades on

-remove meachanical noise sources like hdds to sdcards, old fans with noctuas and decouple parts with shock absorbers and finally sound insulation
i might even go as far as water cooling . this bad boy is totally silent  by the end of the year