The BR-549 Sampler Jr. project

Makalu

well it was about a month and a half ago that I cobbled together a 700W PSU load tester and hot box and used it enough to verify that my design was sound but I couldn't really do anything with it besides tear it apart. It was just cobbled together with a lot of baling wire and duct tape and hot glue and took up about 5'x2' of space...no pictures were taken of it...too embarrassing. It had no switches, everything was just direct wired or some type of disconnect or jumper etc. The original hot box was however the very best that Newegg sells

it wasn't gonna hold up long...but anyway I used it to test an Antec EA500D which went ok except for some problems with some rails and regulating the hot box temps and other little snafus but enough to come up with some data...

http://www.fileden.com/files/2006/11...0D%20Chart.jpg

So anyway it needed an enclosure and switches and stuff. It'll put the smackdown on anything less than a 1600W unit. The major rail load specs are 3.3V/30A in .5A increments, 5V/30A in 1A increments and 12V/110A in 1A increments...voltages and temps are logged on a PC. Dunno what else to say about it so I'll just share some pics and comments...

minor rails and my bulk load precision is covered by ridiculously underrated and oversinked Dale aluminum wirewounds. Vishay rates these at 100%, the military rates them at 70% and I rate them at 50% because when I was in the military they said that there's a right way, a wrong way and the Army way

hmmm my dead grandfathers jigsaw...takes a licking and keeps on ticking...it and a palm sander, variable speed drill and a dremel tool are the only power tools I have for this project

guess I was admiring my work cutting switch holes...lot more left ugh that was a boring day!

2 banks of 4 1ohm grid resisters + 8 Evercool high speed blowers = ~1150W of bulk 12V rail sucking powah

erm casework?

finally the last joint

patches? yes we need some stinking patches...

Krylon rattle can crap finish...shoulda gone and bought some real paint like an isocyanate or some pigmented conversion varnish and fired up a sprayer but uhmm I didn't...dumb move...oh well this'll work

some local color

omg I have to laugh...and have to get back to work on this and get it done so I can start work on the BR-549 Sampler Sr. It'll have a front porch and rocking chair...

maxfly · 06-01-2009, 11:58 | posts: 6,299 | Location: Glass City | User is Offline

nice, cant say i understand what most of the components are but i do know what they do when its finished lol. keep up the good work

avivoni · 06-01-2009, 12:33 | posts: 1,071 | Location: PR | User is Offline

i odnt know what that is but it definitly looks interesting. anyone care to explain?

JohnMaclane · 06-01-2009, 12:44 | posts: 3,919 | Location: Malta | User is Offline

I think its a load bank for PSU testing?

How are the tolerances on the resistances? Can the system measure noise? I know anandtech uses a croma load bank system which is extremely expensive accurate tool.

Pretty cool home system.

Makalu · 06-01-2009, 17:14 | posts: 3,984 | Location: Eugene, Oregon | User is Offline

yep it's resistive load banks...very simple really...just switches and resistors, each on it's own circuit. The rated tolerance of the Vishay Dales is 1%...the grid resistors were surplus and not marked by a maker so no info but the eight of them I have are within ~3% of each other. I'm also using some Xicon cement resistors which are 5% tolerance. Anyway my incremental adjustments are all handled by the Vishay's so if for some reason the other's aren't giving me what I expect I can fine tune it there.

Nope no noise metering...although I could switch off the loads and/or fans for a moment and get a reading...but at the moment it's just set up to monitor fan voltages at the different load/temp levels and that combined with the manufacturers data should give people an idea of what to expect noisewise.

The Chroma is for factory quality control mass production type of thing...crazy four decimal place accuracy yeah but Anandtech publishes data in a graphic format that's not visually that accurate. The metering accuracy of my system is not really something tied to what I am building except for the temp metering is handled by a T-Balancer on board and that's ±2°C which is really disappointing if you ask me but at least it's on par with the K-type thermocouples that most use. The volts and amperage accuracy is from a selection of low cost data logging multimeters (hell everything here is low cost)...voltages are read by back probing the connectors above the switch panel (that's as close to the PSU connector end as possible which is what we want) and amperages are read with a clamp meter around the exposed wire there also. erm so anyway my accuracies with my current metering equipment is like ±.5% and three decimal places for voltages and ±2% and 1 decimal place for amperages. So I think that's better than necessary...I round the numbers off same as Anandtech does. My system is more flexible than a Chroma because it's entirely connector based meaning it will work just fine with any number of 12V rails whereas the Chroma and other ATE equipment you're locked into not only the number of rails that they designed it with but also which connectors are on which rail which more often than not is different from the unit being tested. Anyway that's enough about that.

Nato.dbnz · 06-01-2009, 17:40 | posts: 2,476 | Location: Auckland NZ | User is Offline

You've always been to go-to-guy for PSU advice on this forum, but not once did I imagine you were quite this hardcore

Good work!

JohnMaclane · 06-02-2009, 00:27 | posts: 3,919 | Location: Malta | User is Offline

Quote:

Originally Posted by Makalu

yep it's resistive load banks...very simple really...just switches and resistors, each on it's own circuit. The rated tolerance of the Vishay Dales is 1%...the grid resistors were surplus and not marked by a maker so no info but the eight of them I have are within ~3% of each other. I'm also using some Xicon cement resistors which are 5% tolerance. Anyway my incremental adjustments are all handled by the Vishay's so if for some reason the other's aren't giving me what I expect I can fine tune it there.

Nope no noise metering...although I could switch off the loads and/or fans for a moment and get a reading...but at the moment it's just set up to monitor fan voltages at the different load/temp levels and that combined with the manufacturers data should give people an idea of what to expect noisewise.

The Chroma is for factory quality control mass production type of thing...crazy four decimal place accuracy yeah but Anandtech publishes data in a graphic format that's not visually that accurate. The metering accuracy of my system is not really something tied to what I am building except for the temp metering is handled by a T-Balancer on board and that's ±2°C which is really disappointing if you ask me but at least it's on par with the K-type thermocouples that most use. The volts and amperage accuracy is from a selection of low cost data logging multimeters (hell everything here is low cost)...voltages are read by back probing the connectors above the switch panel (that's as close to the PSU connector end as possible which is what we want) and amperages are read with a clamp meter around the exposed wire there also. erm so anyway my accuracies with my current metering equipment is like ±.5% and three decimal places for voltages and ±2% and 1 decimal place for amperages. So I think that's better than necessary...I round the numbers off same as Anandtech does. My system is more flexible than a Chroma because it's entirely connector based meaning it will work just fine with any number of 12V rails whereas the Chroma and other ATE equipment you're locked into not only the number of rails that they designed it with but also which connectors are on which rail which more often than not is different from the unit being tested. Anyway that's enough about that.

Very interesting! The time and effort you put in this thing deserves its own front page article IMO.

When I meant noise I was referring to electrical noise on the lines, is there a way to measure that? Also have you considered a micro controller to process the readings? so that it could poll them and just print the whole lot of readings?

Damn interesting project though dude, like a billion thumbs up.

Makalu · *Last edited by Makalu; 06-02-2009 at 03:39.*

well I've done some tinkering in areas that interest me...I only had to buy one heatsink. One of the four external ones there...it's ever so slightly different from the ones I bought several years ago...plus there's three more internal ones for the cement resistors. It worked out perfect for the amount of space I had and the amount I needed and used up all of the heatsinks I had left over from my PPP (Passion for Peltiers Phase)...that's not called hardcore though, it's called serendipity

Aluminum plates would have worked btw but I didn't have that...I had these.

oops AC ripple & noise...you need an oscilloscope to measure those. I've checked into it and it'd be a PC one...USB Instruments Stingray...I don't think there's room to mount one inside without blocking airflow for no good reason nope so it'd be external nearby. The cost with a BNC probe and a filtering harness would be about $250. I need one yeah...we'll see how it goes.

The T-balancer logs temp data on the PC in a database file...and the two data logging multimeters I have also will log in a file that I can export into erm Open Office spreadsheet is what I use. They also all have their own software that displays data onscreen in real time of course...and adjustable polling rate.

I think maybe the Senior model could make the cover of Better Homes & Boxcars

Makalu · *Last edited by Makalu; 06-03-2009 at 05:52.*

some progress...

There'll be a bottom panel here between the hot box and the blowers just above the wires that cross over...not airtight...just slots for the wires and maybe just resting on the wires since they're 12 gauge and really stiff. The heater and fan will be on the right...92mm exhaust fan on the left. The four 8-pin PCIe connectors across the right side...erm well they still piss me off. They were sold as being 8-pin PCIe extension cables but they aren't...they're just 8-pin EPS connectors and they just put their colored wires on the opposite side of a normal EPS connector which the PCIe does but it's still not quite right since the 8-pin PCIe has 3 hots and 5 grounds...not four of each. Anyway the electricity doesn't care what color the insulation is but the keying on the connector is wrong. It'd work with some PCIe connectors if you force it but some have a bridge that will prevent it from going into an EPS connector period. So I had to cut out the wall there where the bridge would fit and I also beveled the outer corner of the D connector so that the square pegs will fit without force.

hmm got most of the Dales and their switches wired up here I think except some waiting for the Xicons. The resistance thru the switches and wiring is...well it's worse than I had hoped for but not as bad as planned for. I didn't expect a perfect 10A out of them but was hoping for 9.5+ and the remainder can be attributed to metering accuracy and rounded up. But 9.4A total off the 12V so nope. Anyway it's ok because I only need 5 or 6A to fill the gaps in the grid resistors. They had three electrodes so I wired (and switched) them to give ~3.5A, ~8.5A and ~12A each...8.5A (100W) is the most useful.

So it remains to be seen how much of a gap the 3.3V and 5V Dales have to fill from the "bulk" Xicon loads which I haven't tested at all yet.

These are the Xicon cement resistors...they handle the -12V rail and the 20A each on 3.3V and 5V....not quite sure how I'm gonna mount these yet but they'll be in the airflow under the switches and across from the Dales.

yumm spaghetti...that's a 4 channel rheobus on the left (not mounted down yet). It'll control the 8 blowers and the T-Balancer mounted on the other side of the wall behind it will control the other fans. These will all be powered by a jumper ATX "bench supply" btw. That way I won't lose fan speed as the test units 12V rail sags under load and also I won't lose my cooling if the test unit shuts off or blows or my AC circuit breaker trips. Also I want to be able to cool things off for awhile after a heavy load run probably.

The 12V switching section is connected with bullet connectors so that I can move it onto whichever PSU connector I want. That'll be useful for determining if a PSU labeled as multi-rail really is and if it is then I want to know what the OCP limits are and this'll enable that.

yes yes I'm a lousy photographer...

there you get a better idea of how it'll look and how the current clamp will go around the loops of wire...there will be enough slack to pull out and measure groups of connectors at once (like say for instance all of the 12V3 connectors). High speed Yate Loons there...there'll be fans on the sinks too.

JohnMaclane · 06-03-2009, 06:34 | posts: 3,919 | Location: Malta | User is Offline

Nice work, looking good.

Those blowers however look loud.

Makalu · *Last edited by Makalu; 06-03-2009 at 07:26.*

yeah they're Evercool 75mm X 30mm 3400RPM and rated 36db...pros of cons of using the grid resisters are they don't need to be sunk but they need to be force convection cooled. They're a lattice of strips of a resistive alloy...kinda like an electrified radiator and it takes high pressure to force the air thru the lattice and regular axial fans just won't do...these are rated 12.19mmH20 pressure.

Anyway yeah they're loud but they won't have to be run full bore to cool a resister in the 3.5A or 8.5A mode and I'm just not sure about 12A mode yet. With the "prototype" they were just sitting out in like 20°C ambient and it'll be higher now especially if all 8 resisters are going and the hot box above is at °50C. They could be ran at 10.5-11V before which makes a pretty big noise reduction...we'll see.

There is some room for some sound absorbing material on the wall behind them...hmm

hallryu · 06-03-2009, 08:42 | posts: 6,282 | Location: England | User is Offline

I'm not sure what that is but I'm glad your on our side! lol

Makalu · 06-06-2009, 12:17 | posts: 3,984 | Location: Eugene, Oregon | User is Offline

welp nothing real dramatic photo wise to report since I've been mostly working on the wiring. It's getting there...

although I'm not totally happy with some of my decisions but they can be changed later if it doesn't work out. So anyway all the 12V loads and 10A each of 3.3V and 5V are hooked up and I've just been test running thru those...works real good so far. I guess the 5Vsb is all hooked up now too but I haven't messed with it yet...it'll give up to 5A in .5A increments btw. Kinda waiting on the Xicon cement resistors because those will cramp the amount of work space. Ack and I can't test all 8 grid resistors at once because one of the channels on the rheobus shorted out and I'm working with a 3 channel backup till I get a new 4 channel one. So lessee...oh I got the fans on the sinks as you see.

and the action shot

ignore the caulking and touchup...I think maybe it looks slightly better than the gaps there.

I'm really happy with the cooling they give btw...can run the full load on them and keep the fans at just a whisper and it's very hard to feel any warmth anywhere.

kinda sorta mounted the PSU bracket scavenged from my Stacker TC01 and the door fits just snugly...

I know you guys like photos of smokin' hot hardware so I took a couple

Can't think of anything else happening with it...

Tat3 · **Memory**: 3*2Gb 8-8-8-20 @ 1640MHz

Nice work !

One thing, Hilbert doesnt use any load tester (other than actual hardware) in his PSU reviews, do you think this could help testing PSU's to max (say like 1,5kw one) ?

Makalu · 06-06-2009, 15:22 | posts: 3,984 | Location: Eugene, Oregon | User is Offline

Thanks Tat3 (and all)...well without going into detail of the variety of tests this is a PSU load tester and that's what it's for yep lol...there are a variety of ways to load a PSU and it all pretty much depends on just what sort of tests one wants to do.

Makalu · *Last edited by Makalu; 06-22-2009 at 11:40.*

Resistors-On-A-Stick

These are great but the deep fried ones at the County Fair are even better.

Ok, last two shots of the "bottom end"...all wired up here except my transient surge circuit which is proprietary so no pix...oh and the thermal sensors are dangling there (gray wires)...not in place yet.

Load testing the load tester...

I don't have a 1kw+ PSU so I used a Ultra X-Finity 600, Thermaltake Toughpower 700 and Antec NeoHE 380...there's more wattage to be had there from both the PSU's and the load tester but uhm it's just not set for multiple PSU's...although it did take those to fill the 6 PCIe connectors anyway but all the 3.3V and 5V loads are on one connector (24-pin) and so only available to one PSU but none of them can do 28A on both the 3.3V and 5V rails at the same time and also the 12V adjustment is only available to one PSU so there's some rails that I could normally pad out without tripping their OCP which I couldn't here...anyway it's there...final totals are 3.3V/26A, 5V/28A and 12V/108A.

Got the turbine mounted for the high altitude drop test...

and landing lights on the empennage.

kiddin'. It's a 120V fan than was part of a 1500W space heater I dissected. After my prototype trials and errors I realized that trying to heat a box to 40°C+ by using the heat from the load tester just doesn't work at sub200ish watt levels and it's impossible to regulate across the full load range and different units radiate more heat than others and so I toyed with the idea of a separate load bank for the bench PSU to power but it's not very energy efficient and wasn't wild about the idea of the wear and tear on a $50 PSU. So I just bought a 120V AC space heater for $11. The pretty radical looking fan there wasn't really what I had in mind but selection of the heaters is limited this time of year...erm and the bolts weren't part of the heater assembly, I just got a little carried away. Anyway it looks better than the square plastic heating housing and it makes me laugh.

Some kinda Compunurse type device there I installed since I had it...it's not for test measurements...just so I can see the relative box temp without looking at the T-Balancer on the computer. Plus the thermisters have faster response time than the more accurate diodes. Knob for the thermostat and the switch gives fan only with no heat.

Lessee...ah the rheostat lights...tried clouding them with lacquer thinner but it wouldn't faze that plastic so I just sanded them which at least got rid of the eye-piercing pin-point beam. To the far left is the molex plug for powering the T-Balancer, rheostat and Compunurse thingie off the bench supply. And the AC cord for the heater exits below that. T-Ban USB data cable to the right goes to the PC...

Uhm oh yeah I changed the heater so it's just 350W now...here's the inside:

TT strapped into the electric chair with some big thumbscrews...

I DID have the lexan side windows cut right until I ground the opening out fixing some wows and I went too far...oh well...I can make some new ones if it gets to bugging me but it'd be $8.

That's about wraps it up guys...

MM10X · **Memory**: 4GB ocz 1000Mhz 5-4-4-15

looks awesome, its for testing the load capacity of a power supply, right?

Makalu · 06-22-2009, 12:33 | posts: 3,984 | Location: Eugene, Oregon | User is Offline

It's for testing the quality and quantity of the outputs under controlled loads in a controlled environment...similar to these:

http://www.driverheaven.net/reviews....d=549&pageid=3

http://www.extremeoverclocking.com/a...Testing_3.html

http://forum.xcpus.com/mods/5438-des...ad-tester.html

http://www.velocityreviews.com/forum...ad-tester.html

http://www.techpowerup.com/reviews/A...x_1000W/3.html

techpowerup uses big adjustable wire wound resistors