The EPA and you.

Some of you may have read about the EPA's proposed new ratings system for automobile fuel efficiency and emissions:


Well, I decided that I'd add to things by coming up with my own ratings system that consumers would more easily be able to relate to the environment, as opposed to misinterpreting the EPA's letter grade as a judgment of the car as a whole. Then again, any car that'll save you -$6900 over the course of five years and is likened to an atom bomb detonating by the EPA must be a really awesome ride. Let me know what you think ;-p


(click to enlarge)

Perfectly normal thought process...

What's the first thing you think of when you're jarred awake by a sudden, deafening thunderclap shortly followed by rain, the noise from which can only be described as a roar, in the middle of a summer afternoon... and you notice that the air conditioner is running?

Well, if you answered "Hmm, if it's still 103°F outside, then the increase in humidity that'll result from this rain will even further reduce the efficiency of our central air conditioning system," then you and I are a lot alike because this is exactly what occurred about two days ago. What's particularly fascinating to me, in retrospect, is the chain of thoughts which resulted from this very simple starting point. They went something like this:

"I wonder if the evaporative cooling effect caused by rainwater being sucked through the condenser would offset the efficiency loss caused by increased ambient humidity? How much of an overall efficiency gain could be had by setting up a simple condenser sprayer/mister system? What about building a water jacket for the condenser or putting together a refrigerant-to-water heat exchanger and plumbing it to a large evaporative cooler (Engineering folks: think a small cooling tower. Computer water cooling folks: think a really big bong cooler)? Taking the evaporative cooler approach, you'd be able to improve heat transfer by attaining a larger temperature delta between the condenser and the fluid cooling it than would otherwise be possible with direct-to-air cooling."

To complete what ended up turning into a mini engineering thought exercise I, of course, had to also go through some of the disadvantages associated with such experimentation which boiled down to:

- Cost to construct a sufficient evaporative cooling system vs. possible electricity savings from improved system efficiency (experimentation would have to start with a mister system and small scale condenser immersion testing with a cheap wall-mount air conditioning unit to determine project feasibility).
- Increase in household water usage during the months the evaporative cooling system is in use.
- Increased cooling system complexity (and a corresponding drop in likely reliability).
- Increased cooling system maintenance (lines and fittings would need to be checked for leaks, circulation pump intake filter(s) would require cleaning, the cooling system would have to be manually topped-up regularly (if an automatic coolant reservoir refill system isn't constructed), fan intake filters (if used) would require cleaning, coolant would have to be treated to mitigate biological growth, mineral deposits, and possible corrosion (if incompatible metals are used in the absence of a sacrificial anode), etc.).

I realize that, to most, this entire idea would probably come off as being rather ridiculous but, regardless of the merits of the original idea, it's the process that's important here. I constantly launch into this sort of thinking and it can be sparked by practically anything... after all, how else are we supposed to keep our minds sharp?

Phobos is soon to go...

Well, that's that... Phobos has been listed on Craigslist. With any luck, someone will be interested in it and I won't have to drive too far. While the computer certainly doesn't represent my best work (I have a different set of aesthetic standards for things I toss under my desk that nobody gets to see), I'm still going to miss it.

The listing, along with plenty of photos, can be found here: http://losangeles.craigslist.org/ant/sys/1825803081.html.

Four cores on the cheap... all while consuming less than 100W

A couple of months ago, the guys in the XtremeSystems World Community Grid forum reminded me of how wonderfully inexpensive old Intel server hardware has become... so, without further ado, I bring you a $200 quad-core system (just add a hard drive, E-ATX compatible case [optional], and video card [optional]).

Way back in 2006, the tech industry was clamoring for a cool and efficient alternative to Intel's lineup of 95W+ TDP NetBurst-based Xeon processors. Intel, having just released single and dual-core Yonah variants (based on the Dothan Pentium M CPUs) for the desktop/mobile market, responded by adapting the dual-core Yonah for multi-socket applications and adding support for 36-bit memory addressing. Thus, the 31W (Xeon LV) and 15W (Xeon ULV) dual-core Sossaman was born. The tidbit that I find most interesting about the Sossaman/Yonah/Dothan CPUs is that they're, essentially, an evolution of the old Pentium Pro's P6 microarchitecture. Anyway, getting back on track, the Sossaman and Yonah CPUs turned out to be very, very short-lived products for Intel, as they were used as stepping stones to reach what we now know as the Intel Core microarchitecture. The greatest shortcoming of the Sossaman/Yonah CPUs is their lack of x86-64 support and it was this key shortcoming which contributed to their short, dead-end lifecycle.

So, what does all of this mean to enthusiasts? Well, since the Sossaman is a 32-bit only dead-end product (there is no upgrade path), it's not something which the IT industry has a demand for. What happens to old computer tech that is no longer in demand (come on, everyone should know this one by now)? It becomes dirt cheap!

Here are the basics that you'll need:

- (2x) Intel Xeon LV SL8WT (2.0GHz Dual-Core Sossaman): here (note: this eBay auction has ended, it's provided as an example of about what you should be paying per chip [$25-$35])

- (1x) Intel SE7520BB2 dual-socket Baker Bay server motherboard: here ($9.99)

- 2GB-4GB 1Rx4 PC2-3200 ECC registered memory: here ($20-$60)

- (2x) CPU heatsinks [$10 ea.] (This is something you'll have to improvise unless OEM XS gets more of the Intel "ABB" 'sinks back in stock [Intel p/n: ABBHEATSINKS] or you can find them elsewhere. Keep in mind that the Intel "ABB" style heatsink has a 3mm step in the base to allow for proper contact with bare-die CPUs)

- (1x) Xeon backplate: here [$2.95] (only needed if you're either running the system caseless or you're using a case which does not have standoff locations for Xeon heatsink mounting)

- (1x) Quality power supply [$50] (doesn't need to be rated for more than 350-450W... but it needs to be a quality unit with a 24-pin ATX connector and an 8-pin EPS-12V connector): something like this, this, or this would be fine

Simply add a spare HDD and you're good to go since the motherboard is equipped with onboard video (VGA). Granted, the onboard GPU is an old ATI Rage XL... so don't expect much from it. The board does have a PCI-e v1.x 8x slot, though, which can be modified to accommodate a PCI-e 16x video card--all you have to do is carefully hack off a portion of the motherboard's PCI-e 8x connector. Now, since it's only a v1.x 8x slot, throughput will be limited and this is something which will be noticeable with cards much faster than an HD4850 or GTX260 (throughput is so limited, in fact, that overclocking said cards results in almost no appreciable performance gain in 3DMark). The above system should be able to score about 9600 points in 3DMark '06 when running with an HD4850. If you're one of those folks who enjoys having their computer make sounds, then you'll have to invest in a used sound card (I managed to score an Audigy 2 ZS from an XS forum member for $30) or pick up a random cheap one.

It's about now that I should probably mention that the hardest part of this build is tracking down those Xeon SL8WT processors at a reasonable price. There may not be a significant industry demand for these parts anymore, but there's no shortage of scalpers on eBay hoping to gouge some poor IT guy in need of a replacement part. They do pop up in bulk at sane prices from time to time, but you'll have to be patient and vigilant.


There's SammyOne during its caseless testing phase. In case you're wondering about the name, yes, I have two of these things now (one on Vista, one on Ubuntu) and have even replaced my gaming rig with SammyOne (Phobos will be hitting Craigslist shortly). Using World Community Grid output as a benchmark, this Sossaman system isn't that much less powerful than Phobos' E8400 running at 3.6GHz (~10,213 WCG PPD vs. ~9,000 WCG PPD), it uses substantially less power, produces far less heat, and has been handling daily use multitasking quite well (maybe even a bit better than Phobos when BOINC is running).

I'll update this post tomorrow with a photo of SammyOne in its long-term home: a Stacker 832. Eventually, I hope to cram both SammyOne and SammyTwo into the same Stacker 832 case through modifying the hinged fan carrier to accommodate a ghetto-fab acrylic E-ATX motherboard tray.

Update: Here are those photos I promised (TwitPic thumbnails, click to view the whole photo).

 

A few photos from Willow Springs

For those of you who don't keep track of my Twitter account, I added a few of my photos from last weekend's trip to the Willow Springs International Raceway to TwitPic.  The gallery of thumbnails can be viewed here (just click the thumbnails to view the whole image and enlarge).

What's with all of this stuff?

Some of you have likely noticed the mini-explosion of content on the blog over the past week or two and are wondering what's up... especially since it's likely that you've already seen the content elsewhere.  Well, simply put, I'm taking some time to collect selections of my work that have been posted in various tech forums (like XtremeSystems) and centralize it on my blog as an archive.  Why, you ask?  There are several very good reasons for doing this, but the most important of which is that, for my own sake, I need to put together a "portfolio" of work demonstrating my expertise, experience, some of my many skills.  Having owned and operated a business for five years is great and all, but it leaves me scratching my head when it comes to providing references for prospective employers (especially pertaining to technical capabilities, since many of our customers weren't even aware of the stuff I did behind the scenes).

With any luck, this should help open some doors... and, even if it doesn't, it still makes my some of my past work easier for you guys to track down and reference.

[Forum Archive '09] Project: CES '09 LMM (Last Minute Mod)

At the very end of 2008, I was approached by Super Talent Technology (a producer of DRAM and flash memory products for the PC and consumer electronics market) and was asked to put together a display system for their booth at the upcoming Consumer Electronics Show (CES) in Las Vegas, Nevada. Not everything about the arrangement was particularly rosy, though, as I ended up shelling out somewhere in the ballpark of $300-$400 for cooling parts, modding supplies, and custom appliqués in exchange for permission to put the store's URL on the case (and, no, I didn't get to keep the system). To top it all off, I had maybe two weeks to order all of the parts necessary, build the system, and haul it off to Las Vegas while still running the store and dealing with all of the personal stuff that was going on--to do this sort of project "right" I'd usually ask for at least six to eight weeks. Total forum thread views: 11,083 (as of 5/05/2010) and the original thread can be found here. As a side note, all images have been re-sized via HTML, rather than my usual two image (preview & full image via link) approach, to speed along the archiving process. You can still right-click on the images and view them at full-size in a new tab/window. Also note that all vBulletin emoticons and their associated code have been removed from the archive entry.

!--Start of Archive--!

[Entry 1]
I usually don't post worklogs... mostly because I'm too busy actually modding but I figure that I probably should for this one (even though it isn't really a "major" mod at all).

Anyway, last week (or perhaps the week before... I really can't remember) I was contacted by Super Talent Technologies (they do memory) to put together a display machine for their CES booth. Frankly, it's a little late in the year to be asking for custom display machines but, whatever, I'm game.

They dropped off the goodies on some Friday (was either last Friday or the Friday prior) and I've spent time collecting other parts, getting custom appliqués on order, and prepping stuff. Now, the catch with this build is that I get to foot the bill for all modding-related stuffs, water cooling, and appliqués. In exchange, I can put our logo and stuff on the case.

What Super Talent dropped off:


Aside from the Antec 900, not too shabby.

The two main items that I need to focus on 'showcasing' are the solid state drive and the 3GB DDR3 kit. I'm thinking that I'll work up some sort of bracket to mount the SSD up near a case window or something. Aside from that, I've mostly just been thinking about cooling (then again, when am I not?) and mucking about in Sketchup to see what'll fit and what won't.


Dual MCR220's, modded drive cages, some flat black interior paint, white LED fans, a custom cut window, black tubing, compressions, and stainless radgrilles on the side panels acting as exhausts for the radiators is the current plan... with the two main issues being mounting the radiators and getting ahold of an extra right-side (blank) panel for the 900. I can't get particularly wild with this build not only because I don't have the time to do so, but also because they want the case easily transportable/shippable (so, no external craziness).

Cutting should commence tomorrow.

Would anyone in the bay area happen to have a spare Antec 900 side panel?
[/Entry 1]
[Entry 2]
Sorry for the lack of updates, but I've been busy juggling this project, the store, and personal stuff.

Anyway, I leave for CES in two days... and, boy, do I have my work cut out for me.

Here's a quick summary of where I'm at and how I got there:



Started by stripping the case (pulling all the cages, covers, and other crap out).


A little blue painter's tape on the PSU so as not to scratch it up while test fitting stuff.


Marking up the case to act as a guide for the primary cuts. I'm very much a hands-on, throw stuff in, figure out where to cut, go crazy with Sharpie kind of modder who's projects tend to change as I move through them--it's fun!


...and the carnage begins! Before anyone asks, yes, I'm wearing both safety glasses and earplugs (otherwise, I would have become blind and deaf long ago). What the picture can't show you, though, are my poor Logitech Z-2300's blasting away in the background (a mix of Muse, Led Zeppelin, and Audioslave, in case anyone is curious).


By this point, I'm starting to drill the rivets out of the case to take it apart so that I can finish most of the cutting.


Whadda 'ya know... it's back together again. After pulling the case entirely apart and cutting it up some more, I partially reassembled it so that I could test fit some of the components and check for any additional modifications that it needed. All that excess Sharpie really makes the cuts look a lot worse than they actually are...


Looking familiar?


I'm also using the opportunity to mark spots where mounting holes will be needed and figure out the SSD mount (not shown).


Turns out that a few more mm had to be trimmed from the motherboard tray so that I could mount fans up to the radiator on the right... Eh, at least I was close on the first go.

Which brings us to today...


Here the case has been disassembled again, each panel deburred, filed, wet sanded, thoroughly cleaned, and hung for priming.


It started getting quite cold, so out came the dish heater to keep the panels warm (paint was stored in the front office, so I didn't have to worry about warming it).


Finished priming and allowing things to dry for an hour before rushing into a color coat. I really don't like rushing paint because it's just one of those things that, if you take the time to do it right, can make the difference between an amazing mod and a mediocre mod... but, with so little time to work with, I obviously can't spend weeks on paint (which is why I'm not even considering external paint).


Applying color coats to the panels... in the dark.... with only the dish to keep the parts warm and cure the paint (probably about 45°F out by this point).


After things cured enough to be carefully handled, I moved all of the panels into the office and programmed the thermostat to hold the temperature at 75°F all night to help the paint cure (by the time I left, the temperature in the warehouse was lingering around 40°F... not a great environment for paint to cure quickly). The sad part is that it takes a full week for this paint to completely cure, so it won't actually be done until after CES is over... but that won't stop me from finishing the case anyway!

So, my 'to do' list for tomorrow (Monday) consists of the following:

- Cut the side panels (window for the left panel and one 'radgrill' hole for each side).
- Deburr, wet sand, clean, prime, and paint the insides of the side panels.
- Pop rivet the chassis back together.
- Start assembling computer.
- Pray all of the parts work (I haven't had time to test any of the components that Super Talent dropped off).

Tuesday's list should go like this:

- Finish assembling computer.
- Fill water cooling system.
- Install Windows.
- Apply the custom Appliqués which should arrive on Monday (despite a few setbacks, Terry over at Shatteredweb has been great to work with, as always).
- Box the system up, stuff it into the 328i (since there's no way it's fitting in the roadster), and start the first leg of my drive to Vegas (breaking it into two days because I have to drop by SLO to deliver a TV).

Time for sleep....
[/Entry 2]
[Entry 3]
Well, I worked late into the night last night (like every night for nearly the past week) and am almost done... should be able to finish everything up today. Sadly, there's no way I'll be able to make it to Vegas for the opening day of CES (because I'm driving, the best I'll be able to do is Thursday evening). I've had to take a lot of shortcuts and make a lot of compromises to get this thing done and I'm not very happy about that, but it'll still look pretty cool (just not up to my usual standards).

Pictures will be posted if I have time... otherwise you'll either have to go to Sands Convention Center booth #72421 or wait 'till Sunday to see it.

As a side note... Dave, your pump brackets are too damn strong ;-)
[/Entry 3]
[Entry 4]
Okay, okay... I just finished sifting through photos and resizing stuff last night, so here we go!


After getting the chassis riveted back together...



Checking radiator and reservoir fitment...





A quick spray and cook of the PCI bracket covers... That dish heater really comes in handy (even if the color and finish are a little different as a result.


A little cutting, filing, sanding, tapping and... voila! Compression fitting equipped, G1/4 tapped D4.


Prior to the fitting of tubing.











Bleeding the loop...


Installing Windows

Now some shots from the Super Talent booth at CES... as you can tell, I kept having problems with sagging window molding through the show because I just didn't have time to tack the molding down before leaving for Vegas. Heck, I was still applying appliqués on the show floor.





...and this one's just for you, Dave:


Oliver ( Fr3ak ) has some more photos from the CES floor.
[/Entry 4]
[Entry 5]
"Very nice and clean look. Only 1 thing i don't like :P Paint that Dvd-rom inside of case :P" -Borys

Oh, there are a ton of little things that bother me about this build... the unpainted DVD-ROM, unpainted internals of the 5.25" bay covers, didn't have time to paint the insides of the two side panels (not that you really see them), the window molding (and the fact that I had to use it), not having time to paint the exterior (deep, gloss black with lots of clear would have been nice), wiring could have been nicer, didn't have time to sleeve stuff, the cuts could have been better and could have used more clean-up... the list goes on and on... I'm my own worst critic.

Still looks good, though :-)

"How are the actual temps on the system? I am thinking of doing this on my TT armor with a 3x120 on one side and 2x120 in the top." -Camaroz06

Ya know, I can't really recall... Honestly, I put it together, checked the temps to make sure that everything was working alright, and moved on to more assembly. As for what you're planning, a triple and a dual fan radiator should be more than enough to handle an i7 system with a couple video cards.

"Excellent build. Sorry I did not make it to the Sands and sorry I did not see you at the XS party." -DarthBeavis

Thanks :-)

I wasn't able to get away from everyone at CES early enough to meet up with many at the XS party (I think that I got there at like 8:30 or 9:00pm... but was there until like 3:00am). There's always next year!

"Very nice! It'd be nice to get to build rigs for free." -atomic ferret

Free? Hah... I ended up putting over $100 in fittings alone into that case (not to mention all of the cooling gear, paint, appliqués, and modding supplies)... and I didn't even get to keep it :-(

"In a more productive note petra do is there a version of the Antec 900 without a case window because all the ones I'm seeing have a case window already built in that funky shape." -silverphoenix

Nope but the side panels are interchangeable and Antec sells the blank panel on their website. I ended up stealing a panel from some guy's home system (has a 900, works for Super Talent, offered it up so that I could finish in time for the show).
[/Entry 5]
!--End of Archive--!

[PT Archive '09] A Quick Camera Mount Test

This is an external/internal camera mount test that I put together in 2009 for a couple suction mounts that I purcahsed. The amusing detail is that the video is composed of four separate trips around a local neighborhood block (well, local when I lived in Saratoga) that were painstakingly assembled in iMovie (would have been a lot easier in Premiere or Final Cut). There's really only one point in the video where it's very obvious... stupid turn signals.

Anyway, let me know what you think... and, remember, it's a camera test so I'm not really supposed to be amusing or talking about anything meaningful. I did throw in a couple shots from a weekend outing at the end, including some sightings (Viper, F-430 Spyder, GT-R). The full-size version can be seen on YouTube.

[Forum Archive '08] Fan Controller Analysis

Sometimes a bit of investigation must be done before a question can be answered...this is one of those times. In late 2008, a controversy erupted in XtremeSystems over whether or not certain fan controllers utilized some form of PWM (a debate stemming from the acoustic behavior of certain fans when used with different controllers). Below are a few of my posts from the thread documenting my look into the Sunbeam Rheobus fan controller. As a side note, all images have been re-sized via HTML, rather than my usual two image (preview & full image via link) approach, to speed along the archiving process. You can still right-click on the images and view them at full-size in a new tab/window. Also note that all vBulletin emoticons and their associated code have been removed from the archive entry.

!--Start of Archive--!
[Entry 1]
Well, the four channel Rheobus has a couple quad op-amps, along with some caps and resistors, smack dab in the center of the PCB and the only reason I can think to have those there would be to act as an oscillator (timing circuit) for PWM regulation. If you measured the output with a multimeter, it would indicate that the actual output voltage was changing when the knob is turned but all it's showing you is apparent voltage (since the DMM really doesn't refresh fast enough to know what's going on... you'd need an oscilloscope to see that).

I'll see if I can borrow an o-scope from someone either this weekend or next week and put together a little video showing the output of various fan controllers.
[/Entry 1]
[Entry 2]
Erm, as a bit of an update... I think that I'm going to have to re-evaluate my statements regarding the Sunbeam Rheobus...

Margaret came up to visit this weekend and we managed to pull an ancient oscilloscope outta her parent's attic to test a few fan controllers with. What we found was interesting but kinda confusing.

As I mentioned previously, I had assumed that the Sunbeam Rheobus was using some form of PWM due to the components which are on the PCB; however, when we hooked the controller up to the o-scope, we got something that looked like this:



Hmmm, thought us... that doesn't look like any sort of PWM that we've seen before. Then we remembered that the fan which was hooked up to the channel we were testing would act as an inductive load and, therefore, what we were seeing was feedback from said fan. Hooking a pot up to the controller (resistive load) resulted in the output looking like a flat line (straight DC output).

So, what does this mean? Well, it means that, assuming the old o-scope wasn't acting funky, the output from the Rheobus is not a PWM signal, it's just straight DC voltage. However, that still doesn't explain why we were able to start fans at lower voltages with my lab PSU (or a computer PSU and some resistors to give the desired voltage drop) than the Rheobus, it doesn't explain why they are using two op-aps per channel in the Rheobus (because, as Margaret demonstrated on my white board, you can create a PWM fan control circuit with just two op-amps), and it doesn't explain why it looked like they were using one of the op-amps as an integrator (which would imply some kind of PWM). Due to the unanswered questions, Margaret is taking a couple controllers back down to Cal Poly with her, where she has access to better equipment and will more easily be able to put together a circuit diagram for the controller.


As a side note, we got the same output from the Scythe Kaze Master.
[/Entry 2]
[Entry 3]
Sorry for the double post... but some of the lingering questions have been answered...

They're only using one op-amp per channel (which leaves 4 of the 8 op-amps unused), unlike what we had originally thought. The one op-amp that's being used per channel, which initially looked to be used as an integrator (causing confusion), turned out to be serving as a comparator with a capacitor shorting any high frequency noise. In the end, Sunbeam is using (or, as I would put it, wasting) the op-amps to trigger the LED's color change from blue to red at about 7V and below... and that's it. The mystery of the Sunbeam Rheobus has been solved.

I guess I forgot about those LEDs in my initial assessment of the circuit...
[/Entry 3]
!--End of Archive--!