More IT (scrap NAS)

More IT going on at my home. Some time ago I experimented with proxmox virtualization (déja vue!) and virtualization clusters. At the time, I built a second AMD E-350 rig, which at some stage was converted into a scrap-NAS for data migration purposes. To do this, I grabbed a 2TB-USB2 storage device which was on sale, salvaged the HDD and dumped the rest. The actual HDD would have been more expensive at the time. I figure the USB3 marketing helped to get a cheap HDD (WD green series).
At some stage, I decided to buy a second 2TB-USB2 storage device, in order to salvage the HDD.
Both disks are now living happily next to one another in the same mITX case, alongside with the E-350 board.
This setup is now my new 4TB scrap-NAS, running FreeNAS w/ZFS (stripe) for increased performance.
In the course of experimentation, I played with noise, energy and compression levels provided by FreeNAS. My preliminary decision was to run at maximum energy saving, minimum noise level and lzb-compression. Although very appealing, zfs-dedup remained dis-engaged, remember, it is a scrap-NAS.

Overclocking the Hack

By now I am building Hackintosh setups for years. In the past, I was sticking to what the average user would be doing, happy to see my hackintosh doing what I want it to do.

With my present "production" hackintosh, yes, I do own the licenses, i.e. bought the DVD, thing are different. This particular build (i5-3570k) opens the possibility to overclock... and this is exactly what I am describing in this post.

The Hackintosh I am using is built in a BitFenix prodigy (white), I is equipped with an Antec Kühler 60 closed water cooling system and 1.6GHz DDR3 RAM. Speaking of the cooling system, the fan provided by Antec was too noisy for my taste, hence, I used the back fan provided by BitFenix for the radiator.

This blog is about overclocking... now that you know the baseline, let's talk about overclocking!

After some experimentation, I came up with the following parameters:

Base clock frequency: 109MHz
CPU clock ratio: 46 - resulting in a CPU clock of 5.01GHz
RAM clock at 1.744GHz (the RAM I am using being rated at 1.6GHZ)
Cinebench (on OS-X) reports a CPU score of 5.84 pts


CPU temperatures, running SETI@home at full cpu power, stayed below 50C.

Mind you, I am running a water cooled system! Overclocking from 3.80GHz to 5GHz is not nothing...

improve your web experience

Howdy!
Yet another off-topic post. Lately I am not doing a lot of radio. However, I think that the occasional experience with IT can be at least as much fun as good ole ham radio.
One of the last experiments was virtualization using proxmox on an AMD E-350 low power dual core processor. This particular system is now equipped with an additional ethernet card (100Mbps) and serves as a firewall w/ caching web-proxy using Smoothwall Express 3.1 RC1. The hardware I use is a total overkill: 8GB RAM and 250GB HDD space. As I said, it was used for experiments with virtualization.
The surfing experience has improved a lot since static features of webpages are now locally stored in the caching proxy.
Using smoothwall with squid is highly recommended!
Cheers!

Delidding the i5-3570K CPU

Yet another off-topic (non-radio that is) blog entry.
Having built the Core i5-3570K and the Core i3-3225 Hackintosh, the next field of experimentation was to understand the temperature differences of the core, as displayed by some software products.
To run the CPU cores at full load, I used SETI@home.
Doing that, I observed a 10 degrees centigrade difference between the hottest and the coldest core. I Believed that this would cause a lot of thermal strain on the CPU chip.
Hence, I decided to do what others did before, delidd the CPU and regrease the cooling.

I took some pictures during the process, there are not dissimilar to the ones which could be found on the www, i.e. a knife being stuck between the aluminum heat spreader and the CPU's PCB, etc. on demand, I will share my pictures...

My system is cooled by an ANTEC Kühler 60 and ran up to 52 degrees centigrade for the hottest core. Using MX2 between the heat spreader and the chip, as well as the heat spreader and the cooling pump, the hottest core shows 50 degrees centigrade. There is still a 10 degrees difference between the hottest and the coolest core.

Was it worth the trouble delidding a CPU, potentially damaging it? NO certainly not. However, it was fun to do it, and hence, I would do it again.

Should you intend to to this exercise yourself, here a tip: run your Ivy Bridge CPU hot before trying to cut the glue. The glue being soft helps a great deal reducing the risk of damaging the PCB due to larger forces being applied.

Mind you, delidding your (expensive) CPU is at your own risk!

Update: I noticed that the temperatures of the cores are much more linked as they were before. One core running at 100% load, whilst the others are at idle will increase the temperature of all core essentially equally. Before the delidding, this single core would got hot, while the others remained cool. Therefore, I would recommend the mod, which seems to remove thermal stress on the die of your cpu.

Update 2: Decided to nevertheless de-lid the Core i3-3225. In the process, I cut myself with the stupid hobby-knife. Probably, due to the lower price of the i3-3225, I was not taking as much care, and hence, the knife went straight into my thumb. Same procedure as before, cleaned the die, applied MX2 to the inside also to the outside, i.e. between the heat spreader and the cooling fan.

inactive ...

Due to circumstances, I have been inactive for quite a while. In a way, I regret this, on the other hand, this provided me with some fresh ideas and also with the opportunity to cut off some old stuff.

Lately, I did some hackintoshing, which at some stage, I will document. Radio had a rest, however, I feel an urge to return to the wireless stuff.

Sooner or later, I will pick up things that are in the pipeline for long... and finally publish some here.

Cheers, 73, Joachim

Samsung Chromebook 303C

This is very off-topic, compared to what I usually would publish.

Some week ago, I got myself a Samsung Chromebook 303C. The device is best described as a 10" ARM-powered netbook-like gadget.


Some details in short

The device has got

• a really nice full-sized keyboard,
• multi-touch touchpad,
• a built-in webcam,
• a built-in microphone,
• built-in stereo speakers,
• 2GB RAM,
• 16GB SSD storage,
• a integrated WiFi interface,
• 1 USB 3.0 port,
• 1 USB 2.0 port,
• an HDMI interface,
• an SDHC card reader,
• a single jack headset connector,
• and a power jack.

I may have forgotten one or the other thing, more specific information can be found on Samsung's webpage.

First of all, I bought the thing for having something lightweight, inexpensive to carry about daily. One of the most important points for my was a decent keyboard... and actually, I am very very happy with this one!

Some observations

Things I like:

• the keyboard is really smooth and precise
• the sound of the little speakers is impressive
• the touchpad is very responsive
• the display is crisp and has an excellent brightness range and is matte
• the lower power device does not generate a lot of heat and no noise at all
• very low battery drain during sleep
• the start-up time from cold boot is amazing!
• the device needs 12V, making it ideal for field-day operations

Things that could be better:

• the white power-LED on the right side of the keyboard is somewhat irritating
• a replaceable battery would be a benefit for longer journeys
• individual sound-in and sound-out connectors
• the headset connector is not really smooth
• WiFi occasionally stops transfers, although the connection did not drop

Things I don't like, but can understand / live with

• the display hinge projects quite a bit, I figure, this way it still is sturdy
• there is an access port in the back, which is for a SIM card of more pricey models, with a very flimsy lid
• one needs 2 hands to open the display lid
• the plastic feels cheap, but than again, it is a cheap device

Things I really don't like at all

• there are no left / right mouse buttons, a right-click is a strange two-finger gesture
• there is not obvious way to quickly disengage the touchpad, which would be practical for writing longer texts
• an SDHC-card projects a whopping 7mm out, rendering the card reader useless a storage extension, the card reader has no "spring action", a card to be easily accidentally pulled out... what were they thinking?!
• the power supply is really out of date and weight...

Conclusions and Thoughts

All in all, this is a cloud device. Being offline means that many things can't be done. There are some applications which can be used offline, hence basic functions as text-processing, using a calendar or a basic spread-sheet are still available.

ChomeOS, which runs on the device, is a very down scaled Linux, which in essence uses the Chrome browser for running HTML5 applications.

There are presently first attempts to create full Linux distributions, e.g. Ubuntu by Canonical. Over time there should be some stable distributions available for a full offline experience.

I am happy with the device, knowing its' limitations and the intended use.

Acer Aspire H340 & NAS4Free

Once again an off-topic post in the blog. Somehow the strange desire to share computing / IT topics occasionally took over.

Presently, I am rebuilding my IT, trying to catch up with recent technologies.

Years ago, I purchased an Acer Aspire easyStore H340. The neat little device came with 3 1TB HDDs and MS Windows Home Server. The latter, although doing its job, sucked. Finally, I decided to look for alternatives, in particular some with RAID redundancy and modern file-systems, such as ZFS.

The first alternative I found was "freeNAS". This product, along side some others, is supposed to run from flash drives or SSDs. This would helps to speed up boot and also preserves valuable HDD slots for volume data devices.
And here it comes, the H340 carries an onboard 256MB flash memory device, which is used for MS WHS recovery.
Early versions of FreeNAS were small enough to fit on this device. However, FreeNAS has evolved and grew somewhat larger.

The good new is, that there is some other product, which originates in FreeNAS and is still small enough... check out NAS4Free.

To install NAS4Free, I figure, there is only one option: equip the headless H340 with a head, i.e. a keyboard and a screen. I choose to obtain a PCIe-1x graphics card and use a USB keyboard for input.
Additionally, JP3 needs to be installed!!! Do not remove the jumper at any later stage... at least my H340 would not boot NAS4Free w/o it. Contrary to JP3, the graphics card can be removed, e.g. to reduce power consumption.

Here comes the tricky bit, the H340 needs some strange tweaks to get it to boot from USB sticks, USB CD drives etc. The CMOS setup is not that straight forward, but, it will get you there. The F12-key will help to select the boot device, if it has been recognized by the system.

To get up my system, I choose to boot the NAS4Free live CD with a USB CD-drive.
The option to install an embedded system w/o swap will install the OS on the H340's internal flash drive, just about... no room to spare, all done with a screen/kb attached.

Here comes the more fancy bit.
In such a setup, you would like to go for the most senior option of storage, which presently seems to be ZFS.
The configuration of ZFS is actually not very well documented, neither at SUN, nor at NAS4Free. So, here's what I did to get ZFS up and running on an H340.

1. install NAS4Free from a CD using a screen and keyboard attached
2. reboot after installation has finished, the console should offer a possibility to use DHCP now
3. note the IP-address given to the H340
4. using a remote computer, connect to the H340 using a webbrowser
5. go to the "disks" menu and "import" all disks
6. go to the "disks format" menu and format all HDDs with ZFS
7. go to the "disks zfs pools" menu and create a virtual device (I used single parity raid)
8. go to the "disks zfs pools management" menu and create a "pool"
9. go to the "disks zfs datasets" menu and create a dataset using your pool(s)
10. go to the "disks zfs volumes" menu and create a volume using your dataset(s)

The volume(s) should now be ready to use, i.e. assign to services. Under the "services" menu, one can activate various services such as NFS, CIFS/SMB, AFP etc. Assign those services to a mount point in your volume(s).

Update: It appears not to be necessary to use zfs volumes. It is advisable to mount the dataset. Additionally, it seems a good idea to use the "dedup" feature, although caution should be taken when removing a dataset, cf. advice given by nas4free.

You'll by now be running a rather robust ZFS NAS made of relatively cheap WHS hardware.
I figure it is pretty cool that NAS4Free still fits on the onboard flash drive of the H340.

PSK Transceiver Kit

Very unfortunately, but understandably, Dave (Small Wonder Labs) has discontinued his marvelous PSK-series transceivers.

Lucky for us, there is another kit vendor, who's kits have the potential to replace the ones of Dave.
Have a look at the KN-Q7A kits.

The 20m KN-Q7A operates with an i.f. of 4.194MHz, using a ladder filter made of standard xtals. The transceiver makes us of a moderately pulled VXO of 18.432MHz. All parts a relatively standard and well known, apart from a couple of inductors. The design makes use of subtractive mixing, which increases stability.

Here's what we can learn from Dave's PSK-series, the combination of standard crystals to result in an operating frequency near enough to the PSK bands.

• 30m: 4.000MHz + 6.144MHz
• 20m: 5.0688MHz + 9.000MHz

The only difference to the KN-Q7A is that additive mixing is used. 

The 40m KN-Q7A involves two different i.f., dependent on the frequency range ordered, either 8.467MHz or 8.192MHz. The l.o. will make use of any of those frequencies: 15.360MHz, 15.418MHz, 15.500MHz, 15.536MHz or 15.570MHz. I have not yet figured out a combination to reach 7.040MHz, however, I am sure that one can be found.

Further, I am convinced that the design can easily be adapted to the 80m band. Think of 10.000-6.400 for starters.

As to QRSS/WSPR:

• 30m already cover the mod above
• 15m 4.000MHz i.f. and 25.000MHz l.o.
• 15m WSPR:  4.096MHz i.f. and 25.000MHz l.o.
• 20m could be reached by 4.000MHz i.f. and 18.000Mhz l.o.
• 20m WSPR: 4.096MHz i.f. and 10.000MHz l.o.
• 40m best option would be 4.000MHz i.f. and 11.000Mhz l.o. (alternatively 11.059Mhz)

Easy 472kHz Superhet

Just a thought, a 4.000 MHz (industrial xtal) signal mixed with a 3.530 MHz (80m xtal) signal would result in 470kHz, somewhat shy of our band.

The 4MHz would make a nice intermediate frequency, with a cheap ladder filter. Tweaking QRG is hence restricted to the 3.5x MHz frequency

A couple of options

• pull 80m xtal down
• pen the 80m xtal down (super VXO)

A thought outside the box could be to get dividers into the equation. Douple 3.530 and you will get 7.060. As stated above, the 3.53 are just a bit too high.
Expanded Spectrum Systems sells 40m crystals, which, divided by 2, would result in in-band frequencies.

• 7.042 MHz => 479kHz
• 7.050 MHz => 475kHz
• 7.055 MHz => 472.5kHz

Also here, a super VXO would be an option.

For a WSPR transceiver, the 7.050 MHz option seems ideal. A small downward pull of some hundred Hz should be easily doable without compromising stability. Mind you, WSPR need a "USB dial frequency" of 474.2kHz.

600m Octaplumb update II

A while ago, I built the Octaplumb octagonal RX loop, made from heavy gauge copper wire and PVC plumbing parts (see earlier posts). The loop was tuned to 504kHz, since that was what we had at the time.

Very recently, we know got a slightly different range. Hence, the center QRG of the Octaplumb had to be changed. Some experimentation showed that adding 82pF to the 680pF which are in parallel to the butterfly configed poly-vary-con.