2013 Mac Pro - My variation

[bamboreta]

Good news or almost. The thing is improving: What a relief! Well, I was a little disappointed, but at last I achieved to start the OS X's installation (only with Intel 4600 HD graphics ) and terminate and install the necessary adjustments with Multibeast and after the restart I had a kernel panic, I guess because I forgot removing the flash drive from the USB port, then remove it off and started normally. All works fine, but without wifi or BT as I knew, so I checked the Atheros ethernet network worked well and managed to open Safari and I could browse and download the kexts of Toleda (bcm4352.kext and toledaARPT.kext) put them on the desktop and I downloaded, installed and run Kextbeast and then rebooted and I can to set up the wifi (I guess it's the 2.4 GHz) and connect to my Airport's network as seen on the picture but still don't works the BT 4.0. I done something wrong or I forgot something. I don't know.
Anyway, I'm more pleased and encouraged by what I am going to try to put the 760 GPU to see what happens. Again fingers crossed...

 

[JuanLobo]

Did you replace the wifi card?

I only use bcm4352.kext, and not the Toleda kext with AW-CE123H card. If that is the card you have, they may be stepping on each other.

5Ghz can be enabled, with a patch.

I am really glad to hear that the motherboard is working. That is a big step.

 

[klausi13]

@JuanLobo:

Did you find over the last weeks a good spot for the wifi antennas ? As i also wrote in the low power Mac Pro Mod thread i've tried around 1-2 weeks different wifi cards and antenna combinations and the speed still was very bad. So i've switched over to powerlan adapters and are now happy with a constant 100 mbit network speed which not slows down my internet connection.

But if there is hope to speed up the wifi i will give it another try again

 

[JuanLobo]

@JuanLobo:

Did you find over the last weeks a good spot for the wifi antennas ? As i also wrote in the low power Mac Pro Mod thread i've tried around 1-2 weeks different wifi cards and antenna combinations and the speed still was very bad. So i've switched over to powerlan adapters and are now happy with a constant 100 mbit network speed which not slows down my internet connection.

But if there is hope to speed up the wifi i will give it another try again

Klausi, I ended up putting the antennas external mounted to the base. Also, added some RF shielding inside the case. The last time I looked, it was connected at 513Mbit to my router in another room (802.11ac). I haven't tested the actual throughput, I'll do that when I get everything closed up again.




Edit: Test with case off using iperf going wifi from this hack to a wired Mac Pro through Netgear R7000 running DD-WRT:

------------------------------------------------------------
Client connecting to 192.168.1.10, TCP port 5001
TCP window size: 129 KByte (default)
------------------------------------------------------------
[ 4] local 192.168.1.147 port 49454 connected with 192.168.1.10 port 5001
[ ID] Interval Transfer Bandwidth
[ 4] 0.0-10.0 sec 474 MBytes 397 Mbits/sec

I'm happy with that! Works out to about 49MBytes/sec.

 

[bamboreta]

Did you replace the wifi card?

I only use bcm4352.kext, and not the Toleda kext with AW-CE123H card. If that is the card you have, they may be stepping on each other.

5Ghz can be enabled, with a patch.

I am really glad to hear that the motherboard is working. That is a big step.

Yes, John. That's was one of the first things I did because it was clear that it wouldn't work.
Well, I partially dismantled the setup and I tried the GPU directly on the PCIe slot and it works fine so then I have returned to rebuild with 3M extender included and I have returned to test the computer and Voilá! works well. But where I am stuck and didn't advance is the amount of testing I've done installing and deleting kexts and patches for that works best for the wifi and I have not even managed to change the settings of country or channels (only FCC and US has changed and the amount of channels), nor takes the patch effect for the 5 GHz and of course the BT nothing and that it is simply not recognized although I have also updated (BTFirmwareUploader.kext), just says Azurewave in the USB section on system profile, but bluetooth nothing in system profile and on system preferences does not appear of course. I tried with a usb dongle and boom! BT recognized at the first time and I was able to pair my mouse and keyboard without any problems, then I removed the dongle and I've rebooted and nothing ... again without BT . I have also tried to change some box codes with Hex Fiend (the first time I'm use it ), but don't let me save the replace it . I don't know, probably I don't do it well or just don't quite understand by the language issue. Anyway ... I put some images.

 

[JuanLobo]

Odd that you are having so many issues with BT. I only use Toleda's bcm4352.kext. I don't use the ToledaARPT.kext or the firmware uploader. Nothing patched except for 5Ghz.


So glad to hear the extender is working too.

 

[JuanLobo]

Finished up the fan controller. Works well so far... but I am missing one blasted 3mm nut needed to mount the circuit board to the rail inside. Anyone have small nuts? ;-)



Used 123Circuits to draw out the circuit. Its free, but can be very buggy. Frequently doesn't erase lines that it should, I had to refresh the browser frequently to get rid of them. Also has the nasty habbit of placing lines on top of each other for no good reason



I used these neat little breadboards: http://www.amazon.com/gp/product/B0040Z6OK6 But, if you did want to have circuit boards printed, they can be ordered through 123Circuits... 3 for $15 didn't seem too bad.



The code is very simple. It just checks the voltage provided on the motherboard case fan connector and converts it to a matching PWM signal, speeding it up if the temp sensor detects the GPU getting hot. For this motherboard, the case fan is voltage controlled, not PWM, even though it is a 4-pin connector on the motherboard. When I was first testing on an HP desktop computer, the motherboard fan controller would lose its mind if the fan didn't obey. That would have required me to provide a fake tachometer signal to the motherboard so I could speed the fan up faster than it was requesting. Luckily, the Gigabyte motherboard doesn't care. It just says "Here is 5-12v volts, do what you want with it."

        `*******************************************************
        `PICAXE 3-wire to 4-wire fan controller with
        `auxiliary temperature sensor speed boost
        `v1.0 - Jaun Lobo
        `
        `Inputs:
        `  C.1 = Pin 2 of motherboard fan via 7.5k/4.7k divider
        `  C.4 = TMP36 temp sensor (mounted to GPU heatsink)
        `
        `Outputs:
        `  C.2 = 25KHz PWM signal to pin 4 of PWM fan
        `
        `Motherboard Fan Input:
        `  1 - Ground
        `  2 - Fan voltage
        `  3 - Tach input
        `  4 - Not Connected (5v)
        `
        `Fan Controller Output:
        `  1 - Ground
        `  2 - 12 Volts
        `  3 - Tach output
        `  4 - PWM input
        `
        `Tach signal is passed through from fan to motherboard
        `*******************************************************
        #picaxe 08m2
        #terminal 4800
        #define testmode
        
        symbol maxduty = 160
        symbol minduty = 48
        symbol boostsetpoint = 165


        setfreq m4
        input C.1,C.4
        output C.2
        
        `Set 25Khz/100% duty cycle
        pwmout 2, 39, maxduty
        `1 second at full blast
        pause 1000
        
        `Set starting average voltage
        w2 = 960
        `Set starting average temp
        w4 = 150
        
        `Sample current voltage
main:    readadc10 1, w1
        `Weighted average of 4 samples
        w2 = w2 * 3 + w1 / 4
        
        `Sample current temp
        readadc10 4, w3
        `Weighted average of 4 samples
        w4 = w4 * 3 + w3 / 4
        
        `Boost fan speed if temp over setpoint
        if w4 < boostsetpoint then
            w5 = 0
        else
            w5 = w4 - boostsetpoint * 3 max maxduty
        end if
        
        `Duty should be 100% if adc >= 960 (maxduty * 6)
        w6 = w2 / 6 + w5 min minduty max maxduty
        pwmduty C.2, w6
        
        #ifdef testmode
        sertxd("Voltage:", #w2, " Temp:", #w4, " Boost:", #w5, " PWM:(", #w6,"/", #maxduty, ")", 13, 10)
        #endif
        
        pause 500
        goto main

The result is when the GPU get hot, the top fan gets a boost. Here is Valley running while monitoring the debug info from the Picaxe

 

[JuanLobo]

Here is the solution I cooked up for front panel audio jacks that should work with OSX. I have the parts on order, but I am pretty sure my theory should pan out.

http://www.tonymacx86.com/hardware-...ront-panel-audio-without-hard-find-jacks.html

 

[JuanLobo]

Bamboreta, something I forgot about.... Instead of mounting the fan grill to the case, you may want to mount it to the fan itself. I am thinking of doing that, just to see if I can get a bit of the Mac Pro recessed lip effect of the real mac pro.

 

[JuanLobo]

Finally got front panel jacks wired up and working in OSX. Detects when headphone/mic is plugged in, and sound control panel switches from Speaker to Headphones.