Tagan Turbo-jet 1100W Review

[Super Nade]

Wolf posted this on the Front Page and I'm transposing it here for discussion.

http://www.hardwaresecrets.com/article/403/9

Superbly done review! If only he had an experimental setup to present some hard data.

To summerize:

It is a garbage unit in more ways than one. Note that the fancy box can be interpreted as a fancy coffin when it blows up. ;)

Unfortunately we don’t have the necessary equipment to make a true power supply review; we would need to create a real 1,100 W load with a load tester to check if this power supply could deliver its labeled power or not. As mentioned, our friends at Planet3Dnow.de have one and their Tagan TG1100-U95 burned when they tried to pull more than 768 W @ 25º C on the +12 V lines.

[burebista]

OK, I'm not a big fan of Tagan neither, but this it's not true:
Also, the idea of using fancy connectors is good, however you won’t be able to use a GeForce 8800 GTX with this power supply, because this video card needs two auxiliary PCI Express power connectors and the size of the connectors used by this power supply prevent two of them to be installed together on this VGA. What is the point of having a “1,100 W” power supply with four auxiliary PCI Express power cables if you cannot use it with a GeForce 8800 GTX?
This picture (http://www.crazypc.ro/forum/attachment.php?attachmentid=17175&d=1163755950) is from our tests with

eVGA 680SLI
intel E6600
TEAM Xtreem M1800HC5DC
2X EVGA 8800GTX
Tagan 1100W

Everything was smooth and flawless.

[Oklahoma Wolf]

Yeah, it's an awful design for something claimed to be this powerful. Not much else I can say - Topower's staying off my recommended list for the forseeable future.

[jonnyGURU]

I'm not sure why Gabe said what he did about the PCI-e connectors. I hadn't heard of any problems myself.

But yeah... that's pretty whack that they make an "1100W" that can't even put out more than 800W... by design.

Chris from Planet 3D Now tells me he's working on the English translation of their review. I can't wait! :) I tried to read it earlier, but it just wasn't working for me. ;)

[burebista]

LOL! Now I see the prices in my country. Tagan 1100 $583 :crazy: and Galaxy 1000 $461 (yeah I know, a small country in far East Europe... :()
Pretty hard to choose between two, huh? :lol:

[jonnyGURU]

Even the $349 they ask for it in the states is absurd.

It's about the same price as a Galaxy 1kW.

The Coolmax CTG-1000 comes in at under $300.

And when the SilverStone Olympia comes to town, forget about it! That should be under $300.

[Bbq]

Another reason to stay from tagan.

I'll never understand how people keep recommending tagan for everything, stating how magazines testing on a pc that draws no more than 300w, is superior to many better units, and they don't even measure voltage.

[madmat]

They're fulla shit, I've put 840W on my tagan's 12V rails and it didn't blow up.

[jonnyGURU]

They're fulla shit, I've put 840W on my tagan's 12V rails and it didn't blow up.

You writing that review for TechGage, right?

[madmat]

Yeah...put 1kw on the 1100 with 840W on the 12V rails. It got loud as hell and put out 80mv of ripple on the 12V lines but it was in spec and the voltages didn't sag much at all.

[Super Nade]

Interesting madmat. Did Topower quietly release another version?

[madmat]

Mine looks identical to his, he didn't put a heavy load on it but Planet3Dnow.de had one go to hell at 800 some-odd watts so they naturally call shens and say it won't do 1100W. I call shens on them and say they just got a defunct unit. It was flawed, yes but that doesn't mean none will do over 800W. I read another review of the same unit and they pushed it to 1100W with no problems.

[nat_raven]

i dunno how you guys hate topower so much?

sometimes they produce gold but sometimes sh*t! but that doesn't mean that they will never produce gold anymore right?

i'm not a supporter of tagan nor topower but from what i read on this forum, with all my respect to jonny, it's kinda bias to some brands.

i don't say that tagan win my respect but those review didn't win me at all.... :lol:

it's not a review imho.... sure you know better electronics than me but a review of a psu w/o any load? c'mon man u must be joking! you always state that "our friends at planet.de test like this... our friend at planet.de say like this...

sh*t happens man... that's why madmat's perform ok while those planet.de folks scream for help. i smell something is not right, agree?

[GalvanizedYankee]

I will follow further testing of this unit with genuine interest. I doubt I'll ever need 1kw of comp power but as a learning experience, it is fun to watch titans battle.

Four gimmicks that come with this unit made me smile :)
1-The gold plated mains power plug.......OK.
2-The ferrite coil on the mains power cord. Err, on a VGA cable yes, but on the power cable?? Yes I have seen ferrites on PCI-e power cables, a nice touch but not really needed, imo.
3-A 12AWG grounding wire for the PSU. What are all the black wires for in the harness?
4-The SATA to 4 pin Molex adapters. Kinda cool if one is going to be replacing HDDs down the line but they do add to overall depth and they are just waiting to be bumped into disconnect.
I'd rather use a wired adapter. *shruggs*

None of the above will cause any direct harm, I just find them gimmicks :cool:

[Oklahoma Wolf]

sometimes they produce gold but sometimes sh*t! but that doesn't mean that they will never produce gold anymore right?

That's one of the issues I have with them right there - for the price these things cost, I want to see a design with build quality that rivals Zippy, Etasis, and PC P&C Turbocool. I don't see that in this unit. Instead I see what I've come to expect from Topower - some die before they get to full load, some don't.

[jonnyGURU]

i'm not a supporter of tagan nor topower but from what i read on this forum, with all my respect to jonny, it's kinda bias to some brands.


"With respect?" Sorry.. but it sounds pretty disrespectful to me if you think that I'm going to base an opinion on something without reason. Unlike most people, if you ask me my opinion of Kwung Poo Jung PSU and I've never heard of them, I'm going to withhold comment. Others would jump all over them just because they never heard of the brand and then recommend an Antec because 10 years ago they had good luck with them. Poppycock!

Let's see... Just from my own reviews.....

BFG built by Topower had poor voltage regulation, dropped voltages in the hot box and mediocre efficiency: http://www.jonnyguru.com/review_details.php?id=21

Mushkin built by Topower had poor voltage regulation, failed crossload tests and had lots of ripple: http://jonnyguru.com/review_details.php?id=31

Tagan EasyCon built by Topower had poor voltage regulation, failed crossload tests, mediocre efficiency and ran hot as a mofo: http://www.jonnyguru.com/review_details.php?id=18

And I have a 1kW Power Train sitting here that has 200mV of ripple.

Biased is what most people have when it comes to power supplies, blindly suggesting for and against products that they know little about, and the whole reason this site exists.

You give me a good reason, outside of the OCZ 520ADJSLI (not even the OCZ 520ADJ) NOT to have "bias" against Topower and maybe I'll consider a change of heart.

[nat_raven]

hey

i'm not in any way defending topower nor tagan since they don't give me any money to talk sh*t to every brand against them here.

what i'm feel that once mr. jonny tell that this brand s*cks then everybody is become a true follower and junk those brand to sh*t! i don't even think that those "followers" and jonny lovers even see those stuff.

also i'm very irritating to read one review that dissambles psu and call that a total review while not doing any load test at all! just saying that a friend at german website said so and that's it! as i said once again, i'm no electronics expert like him but pleaseeeeee.... no load on psu review and give a conclusion that those unit can't deliver 1000W by design and by some german friends? you should do better than that... ;)

rather than dis-regard some brand that "might" want to correct what err they might made, why don't you.. i believe already become one of the most readed psu review website just asked to tagan to give what their best psu to be tortured and judge by ourself?

sh*t....... i already sound like topower/tagan spoke person! :D

[Oklahoma Wolf]

what i'm feel that once mr. jonny tell that this brand s*cks then everybody is become a true follower and junk those brand to sh*t! i don't even think that those "followers" and jonny lovers even see those stuff.

My opinion of Topower predates JG's first review, based solely on observed circuit design - usually, I find I'm not that far off when the good reviews come up :p

[jonnyGURU]

nat_raven: I even know people in the industry who have used Topower as an OEM in the past only to eventually conclude that the partnership was a mistake.

For example, why isn't OCZ still using Topower?

And I've never seen any kind of "dog pile bashing" here. In fact, I try to prevent it. Take Hiper, for example. I pointed out flaws in their design and personally I wouldn't recommend them. But I find no fault with their M series.

If you think the members of this forum don't have enough of a head on their shoulders to make up their own minds, you'r mistaken.

As for Gabriel, he probably knows more about electronics than most of the folks here. I know he knows more than me. If he can look at a design and say, "this can not effectively deliver X amount of power" I'm going to take his word on that.

[nat_raven]

ok wolf

at least you let me know that you're not just a "usual" follower...

but jonny, there's many troll (like you called me) like me that only bcome some follower and s*cker! and that's really irritating

for example: like u said in one of your post, ocz might one hella good psu on its class on his time but loosing his teeth now to many newcomers.... that's make sense! but you can't see maybe the "comeback" of some brand that might awfully lousy once if you didn't try to lower your voice.

ok, those power train is lousy but how do you know this topower based 1100w is the same or different since u don't have any in your hand? call tagan, ask them to send you one! then we can call it a day rather than debating here

to be honest, this is the spirit i'm looking for!

[Super Nade]

Gabe Torres did rip apart the unit and actually point out what he thought were component level deficiencies.

For example:
Instead of using only one electrolytic capacitor on its active PFC circuit, this power uses two 1,200 µF x 200 V connected in series, which equals to a 600 µF x 400 V capacitor.

Although this isn't very critical on the Primary side, the ESR's of those caps are gonna add up, right? I would also consider Tosin an iffy company (mostly used in crap Jetway boards).

This power supply uses two GBU1006 rectifying bridges on its primary stage, which can deliver up to 10 A each (rated at 100º C), so the total current the rectifying section of this power supply can handle is of 20 A. Just for a comparison, Enermax Galaxy 1000 W uses two 20 A bridges, meaning it can handle the double: 40 A.

Ofcourse, bigger is not always better as it really depends on the design, but from the block-schematic posted, the bridges do looked under-specced.

It isn't a fake/crap review by any stretch of the imagination. Just do a search for OCZ Powerstream 600W and DFI, you will find tons of posts with folks complaining of units that blew up on them. Also, look for "dead Powerstream" to guage the quantum of this problem.

[jonnyGURU]

for example: like u said in one of your post, ocz might one hella good psu on its class on his time but loosing his teeth now to many newcomers....

Maybe it's the language barrier, but I never said that.

My opinion of OCZ is that they had one good model (the 520ADJSLI, made by Topower) switched to FSP and used their high-ripple (upwards of 200mV) Epsilon unit and then finally made a modular from a quality 3Y unit, that's priced a little bit out of reason.

That doesn't echo what you claim I said at all. In fact, that kind of sounds like the OPPOSITE! I don't judge PSU companies or re-branders over all. I do judge them on a model by model basis. And currently, I do not see anything from Topower that I like.

ok, those power train is lousy but how do you know this topower based 1100w is the same or different since u don't have any in your hand? call tagan, ask them to send you one! then we can call it a day rather than debating here

to be honest, this is the spirit i'm looking for!

I do admit that the Power Train is not the same platform as the Tagan units. In fact Tagan has an exclusive on design that Topower is not allowed to use for anyone else.

If Madmat has a unit to test for Techgage then I trust it will be thoroughly tested. He has my old load tester and he has built his own "hot box." He's also borrowing my USB Stingray for a couple weeks so he can measure ripple. So I'll withhold any additional opinion until I see that review, but that's not to say that I don't take Gabe or 3D Now's experiences with a grain of salt.

[Oklahoma Wolf]

The two caps in series on the primary isn't too big a deal - I've seen that done before. What I take issue with is the boneheaded insistance on using all four switching transistors for the whole PSU, when they could have just as easily done better with an independant design.

But, they just had to cut down the parts used in filtering the 12v output, didn't they?

[nat_raven]

So I'll withhold any additional opinion until I see that review, but that's not to say that I don't take Gabe or 3D Now's experiences with a grain of salt.

yes sir!

hopefully with many load test data ;) that's i find easier for me to digest

u won't hear anything from me until the day come

[Spectre]

Topower is decent enough and if someone buys one they are in all likelyhood never going ot see any problems with it..... but if there is something better at the price point......... But then again I like dumpster diving.....*cough*SuperFlower*cough*

[krampak]

The two caps in series on the primary isn't too big a deal - I've seen that done before.
I have 12 units from 28 right here which uses 2 capacitors in serie on the primary side, so I would say that is common. They are built by SevenTeam, AcBeL, Elwin, Topower and 2 unknown.

[Micutzu]

Hello everyone !

I registered here just to be able to chat with knowledgeable persons about the hardwaresecrets.com review, as i think there are some flawed points in the author's argumentation. I've had my fair share of experience with Tagan/Topower PSU's working at two of the local Tagan distributors, and also being quite involved into overclocking ... you can see some of my results on Futuremark's ORB. I don't have an artificial load tester or even oscilloscope to monitor line noise/ripple, so all my observations are based on practical testing with real-life loading and regular multimeter monitoring.

As my friend burebista pointed out here (http://www.jonnyguru.com/forums/showpost.php?p=5353&postcount=2) the connectors work perfectly with 8800GTX, as you can also see in the picture. That shot was taken when i was testing 8800GTX's for the first time and i've encountered no problem whatsoever installing or using the PSU with the GTX's. The guys at hardwaresecrets didn't even have a 8800GTX to test with.

Two rectifier bridges that can take 10A each will allow an average power input of well over 2000W in the worst-case scenario, wich is a decent margin in my opinion. The Galaxy has stronger bridges, good boy - have a cookie, but that is definetely not a thing to be considered a design flaw.

3-transistor PFC ... i don't know if the 4th is necessary or not, but as most PSU's have 2 or even 1 transitor in the PFC stage, i'd say 3 is OK; just a personal opinion.

Just one +12V filtering circuit is logical, the +12V lines are anyway joined and wether they met just after the rectifiers or after the filtering, it doesn't really make any difference. I agree that a filter projected to handle a certain amount of current and ripple won't handle 2 times more safely, but also it's phisical size depends on the switching frequency ... a circuit with 2 times higher switching frequency will need 2 times smaller output filter. Now i don't know what's the switching frequency for the Topower and it seems neither does hardwaresecrets.com (or at least don't say anything about it), and an observational opinion like that without knowing ANY facts at all or without backing it up with any kind of testing is simply worthless.

As for the lack of modular connectors, it's more a matter of personal preference. They add wire resistance and they're harder to work with - yes, you need at least 10cm more in front of the PSU as you can't bend the wires because of the connectors. I don't like them, i don't need them, i don't care about them.

The Tagan U95 pdf is up there on their site, here's a link (http://www.tagan.de/pages/products/turboS/Actionpaper_tagan_turbojet_1100_DE_mini.pdf), but it seems it hasn't been observed by the reviewer. Come on, you write a 10 page review and you can't type www.tagan.de in the browser and make exactly 2 clicks to get to the pdf ? On top of that, the most obvious thing, that kept repeating all over, is the feeling that the title of the review is wrong ... yeap. I think it should be called "Enermax Galaxy 1000W Power Supply, part II".

jonnyGURU, i see you've had some experience with Topower built PSU's, but except the PowerTrain every one of those has used the same old electronics, that i agree can be improved. But, the Tagan U95 models are based on Topower P9 electronics and are something else. I really want to see it tested by you, if i could i would've sent you one :).

SuperNade, indeed the ESR adds up for the two capacitors, but you should also consider that higher capacity generally means lower ESR. Not that ESR matters too much in that particular aplication.

[Super Nade]

SuperNade, indeed the ESR adds up for the two capacitors, but you should also consider that higher capacity generally means lower ESR. Not that ESR matters too much in that particular aplication.


Still digesting the design, so, no comment at this point on that account. However, I did mention that the caps in series was not critical on the primary side. :)

A hint regarding the switching frequency can be obtained by looking at the RF damping circutry (at least the number of components involved). I'f I'm not mstaken, this is the trade-off between using a higher switching frequency and maintaining EMC.

[jonnyGURU]

Mine looks identical to his, he didn't put a heavy load on it but Planet3Dnow.de had one go to hell at 800 some-odd watts so they naturally call shens and say it won't do 1100W. I call shens on them and say they just got a defunct unit. It was flawed, yes but that doesn't mean none will do over 800W. I read another review of the same unit and they pushed it to 1100W with no problems.

I just heard that there were different revisions of this PSU and that the early revision had a lot of issues.

If I had one here, I'd look for a way to differentiate the revisions.

Matt, anything on that PSU that looks like a revision number?

EDIT: Apparently, we can tell from the S/N. There's a date code built into it somewhere.

[Micutzu]

My SN is TD60 461001, if that helps.
A special thing i've noticed about these PSU's (both the 1100W and the U25 700W wich is based on the same electronics) is that the output voltage raises a little bit under load.

[madmat]

TD60 461297 is mine, no clue how to decipher it though.

BTW, Yankee...I think you're unclear on the adapters, they're not molex to SATA they're inverse and adapt from SATA to molex. The TG 1100 has like 10 SATA cons and 3 molex cons, if you need more molexes you're hosed unless you use the adapters.

[jonnyGURU]

Wow! Those two S/N's are close!

Let me see if I can figure out how to decipher the S/N.

[Micutzu]

I don't think they made too many 1100W units to be honest :).
I know for a fact that there has been a faulty batch but they were quickly recalled, mine is the fixed version. Here's the rail splitting for you testers:

12V1: 12V 8pin;
12V2: 12V 4pin, 12V 8pin, 20+4pin, HDD, S-ATA;
12V3: PCI-e 1 & 3;
12V4: PCI-e 2 & 4.

[Iantikas]

As my friend burebista pointed out here the connectors work perfectly with 8800GTX, as you can also see in the picture. That shot was taken when i was testing 8800GTX's for the first time and i've encountered no problem whatsoever installing or using the PSU with the GTX's. The guys at hardwaresecrets didn't even have a 8800GTX to test with.

http://img245.imageshack.us/img245/7975/pcietopcx3.th.jpg (http://img245.imageshack.us/my.php?image=pcietopcx3.jpg)

coff coff :rolleyes: ...bye

[madmat]

http://img245.imageshack.us/img245/7975/pcietopcx3.th.jpg (http://img245.imageshack.us/my.php?image=pcietopcx3.jpg)

coff coff :rolleyes: ...bye

Umm, that clearly says "Hardwaresecrets.de" on that pic so your point is?

As burebista illustrated in his pic the connectors fit the card he tested with or are you trying to call that a fabrication? I'm sure that there's some variances going from card to card especially as the card makers start to stray from the reference designs. I know that the 7600GT shows that beautifully, there are 7600GTs built on 7600GS PCBs.

[jonnyGURU]

Actually is esay PC-Experience.de. :D

But I see the point. As you say, Matt; cards are going to vary. Not even from fudging the reference, but from just soldering the connectors down a mere millimeters apart!

Of course, one could always FORCE the connector. ;)

I don't so much blame Topower for the connector fit. I don't like the connector on the card. Put me on your design team. :) They should have connected to the "back" of the card with a 45 degree angle.

In nVidia's defense, the choice to go with two 6-pins was last minute. They were waiting on PCI-SIG to decide on the 8-pin PEG connector, it never happened, and nVidia needed to launch the card.

[madmat]

My bad, I meant that but Hardwaresecrets was on my mind due to the whole topic and them saying it'd never work...that and the pic clearly not being theirs that is.

Edit: I'd like to see the 8 pin design be adopted by the 8800GTX now that those PSUs are coming out. I know it'll take some time but I'm hopeful it'll happen.

[Tazz]

Well if the PSU didn't use the cosmetic covers on the PCI-E connectors it would probably have connected without any problems.

[jonnyGURU]

They come off easily enough.

Throw the cables on the ground and stomp on them and the plastic cracks open. Used to have to do that all of the time with the Ultra X-Connect cables.

Unfortunately, you're left with some exposed wires (individually insulated) so have som heatshrink handy to slide over the end and cover up the gap.

[Tazz]

Yup, I prefer cutting them with a knife rather than stomping on them though. :)

[burebista]

As burebista illustrated in his pic the connectors fit the card he tested with or are you trying to call that a fabrication?
Oh, I wish I've had that setup. :D
Micutzu is the man, you can trust him like I trust jonny, OW, Super Nade, madmat & others.

[nat_raven]

Quote from Tweak Town

Link: http://www.tweaktown.com/news/6746/overklokk_sets_a_new_3dmark_03_record_86_198_point s/index.html

Overklokk sets a new 3DMark 03 record: 86,198 points

Following their success at this year's DreamHack event, Kinc & crotale have another challenge which lies ahead now that Overklokk over at the XtremeSystems Forums has just managed to pull a new 3DMark03 world record of 86,198 points. This was made possible with extreme cascade cooling on a pair of 8800GTX GPUs and Core 2 Extreme CPU.

For full details on his achievement head over to Nordic Hardware (http://www.nordichardware.com/news,5206.html) or the XtremeSystems forums (http://www.xtremesystems.org/forums/showthread.php?t=125525).


From the link, overklokk also using dual 8800GTX and have no probs with PCI-E connectors. He's also using Tagan Turbojet but "only" 900W instead of 1100W

@jonny: you can't stomp the PCI-E connector and hope it to break, it's made of some kinda rubber and inside there's some caps for filtering RFI & EMI. It can be opened only by slide it down. It seems kind same with my newly buy dual engine PCI-E con.

[Tazz]

@jonny: you can't stomp the PCI-E connector and hope it to break, it's made of some kinda rubber and inside there's some caps for filtering RFI & EMI. It can be opened only by slide it down. It seems kind same with my newly buy dual engine PCI-E con.

http://tazz.ndtechs.net/Tazz/PCI-E-Cover.jpg

Jon is talking about stomping on the blue connector at point A. This would effectively work, but could also cause the main part of the connector to break in someway (just my opinion). The filtering should be handled at point B. I don't have the specific unit here or I would show an example with the covers removed.

[jonnyGURU]

You're right about where I was taking about stomping, Tazz. But there are small caps underneath "A" tat could potentially break under stomping. :D

[Tazz]

Thanks for the correction, as stated I don't have a PSU here out of the many I have that utilize the caps so I am not sure. Which is another reason I prefer cutting them at the seem :).

[nat_raven]

http://tazz.ndtechs.net/Tazz/PCI-E-Cover.jpg

.....The filtering should be handled at point B. ......

The inside of Point B is Ferrite Core and yes it's also for filtering, but the caps at point A also for filtering.

[kimandsally]

Just to add my bit which some might think worthless but nevertheless it means something to me.
I went through a phase of building PC's for part of my income a while back and used quite a few Tagens the U22 and U15 series, after having several fail and some were below ATX spec out of the box I decided enough, as someone who earns money from building PC's it didn't make economical sense to continue buying them, then shortly afterwards the supplier stopped selling them telling me that there was far too many returns.

Just because of the inconvienience and money I lost I wouldn't try another Tagen until it was widely reported that they had improved, I cannot afford to buy products that are unreliable compared to the normal average.

[GalvanizedYankee]

The inside of Point B is Ferrite Core and yes it's also for filtering, but the caps at point A also for filtering.

Why is this filtering necessary on the PCI-e cables? The PSU has output caps that filter, the VGA card has input caps that filter. I find this to be a marketing redundancy.
Please tell me I'm wrong and that the newer graphics cards need this.

[Micutzu]

You should consider that we have (1) line noise from the PSU (ripple), (2) line noise produced by high transients from other consumers (CPU, other VGA, HDD motors etc) and (3) RFI picked up by the wire from the surrounding enviroment.
The VGA's only have few capacitors to filter the supply line, but without an inductor efficient filtering is not possible. By using an inductor a PI-type filter will be formed together with the VGA on board capacitors and the capacitors inside the PSU. It works both ways, ripple/noise attenuation at the VGA's input will be much better and high transients from the VGA won't escape back to the other components. I don't know how much this helps in practice but it is something extra and i rather have it if possible.

[Oklahoma Wolf]

By using an inductor a PI-type filter will be formed together with the VGA on board capacitors and the capacitors inside the PSU.

Only if the inductor's actually in the circuit, else it's not really a pi filter ;)

[Makalu]

hmmm comparing the labels on the TG-900 and the TG-1100, I'd say there's about 100w unaccounted for on the latter. Something fishy there if the 900 specs are to be believed. I mean, supposedly the extra 200w all comes off the 12v's if the 3.3v and 5v combined are the same which is what the labeling indicates. If so then the 12v should have a max combined rating of like 88a and we know that ain't happenin'. Wouldn't surprise me one bit if it's only 72a (same as the 900) or maybe 75a in reality.

[Micutzu]

The "max combined power" for +3.3V, +5V and +12V has a 200W difference for the two models, wich can mean there is a limitation from one of the common circuits for all the 3 voltages or from the cooling limitation for the weaker model. The difference in max current per output is small indeed, but not that gives the overall power capacity.

Only if the inductor's actually in the circuit, else it's not really a pi filter ;)

Well, it is in the circuit, even if not close by ...

[Oklahoma Wolf]

Well, it is in the circuit, even if not close by ...

I mean it needs to be in the +12v between the two capacitors to be a proper pi filter... if it's just surrounding the wires without being part of them it's likely just there to filter EMI. I'm nitpicking I know ;)

[nat_raven]

micutzu:

i hope u are the same micutzu on XS ;)

since i read u use many tagan to do WR, how is personally u thinking about tagan? honestly, i prefer to believe some "hardcore" user that torturing pc everyday as living rather than some reviews or people that never hold one in hand but always critize... :)

[Makalu]

The "max combined power" for +3.3V, +5V and +12V has a 200W difference for the two models, wich can mean there is a limitation from one of the common circuits for all the 3 voltages or from the cooling limitation for the weaker model. The difference in max current per output is small indeed, but not that gives the overall power capacity.


Well, the heatsinks are the same size and design and the fans and airflow are the same so I kind of doubt that they improved the cooling to the tune of 100W without changing those. It could be that they've done something to unleash 100W on a common circuit even though the unit has some pretty independent circuitry for all three voltages. I suppose the Right Reverend Sun Moon will tell us what it can output soon enough :)

[jonnyGURU]

micutzu:

i hope u are the same micutzu on XS ;)

since i read u use many tagan to do WR, how is personally u thinking about tagan? honestly, i prefer to believe some "hardcore" user that torturing pc everyday as living rather than some reviews or people that never hold one in hand but always critize... :)

Trolling again, I see. :D

[nat_raven]

:o

[madmat]

Well, I just put 840W on the 12V rails again and still no catastrophic failure, no smoke, no flash, no bang...not even an electronic sausage. Overall there was 997W on the entire PSU and it coped just fine. Sadly 1kw is the limit for me so that's as far as I can push things.

Now I did see some alarming ripple at first glance but I think it's a byproduct of my summing all four 12V rails then splitting the load out to the 2 inputs on the SM268. The first time I ran it I did things a bit differently and only saw 80mV under the same load so I'm thinking the readings I'm seeing now (roughly double that) are a byproduct of the rails being out of electrical phase with each other. I'm going to isolate the rails and do some testing that way and see what the rails do at 17.5A load individually.

[Edit:] Yep, sure enough the rails individually only hit about 40mV of ripple under a 17.5A load. Combining a couple of rails increases that reading even though the load is halved per rail. I'd have to say that it's not as nasty as I first took it to be.

[jonnyGURU]

Cool.

And there is two versions of this PSU out there, where the early version was problematic, but I haven't been able to find any info that helps me decipher the serial number into something that tells me which one's good and which one's not.

[Micutzu]

Oklahoma Wolf > actually i haven't cracked the inductor's casing open to see if the wires are winded around the ferrite core or they just pass trough it, wich would only help for some RFI protection.

Makalu > maybe the 900W is just artificially limited, it's simpler for them to sell the same product under two different ratings. This would make it a very interesting PSU in that range ... unfortunately i don't have neither the 900W or the 1100W models in stock anymore for a closer inspection.

nat_raven > thanks mate, along the time i've used alot of different PSU's for benching ... Hiper, Antec, OCZ, Seasonic, PCP&C, Silverstone, Tagan etc. Rule of thumb is "use the strongest PSU I have available" :).

madmat > your testing confirms my expectations ... how good was the +12V regulation at max load ? I don't see why the ripple was out of phase on the rails though, internal splitting is very simple.

jonnyGURU > i don't think there are early versions of the PSU out there anymore, they were few and the problems were corrected a good couple of months ago.

[madmat]

The TJ has 2 (not one) transformers for the 12V rails, that and the resultant circuitry (2 virtual rails per transformer) can mean that you can have phase variances. that's pretty obviously the issue here.

The 12V rails dropped from 12.28V at 2A per rail to 12.10V at 17.5A per rail. none too shabby.

[Makalu]

Makalu > maybe the 900W is just artificially limited, it's simpler for them to sell the same product under two different ratings. This would make it a very interesting PSU in that range ... unfortunately i don't have neither the 900W or the 1100W models in stock anymore for a closer inspection.


Yeah that's possible, and it would be simpler than producing a unit with lesser components. It would be kind of strange though since the cheaper unit wouldn't cost any less to produce and it'd be simpler still to just produce the TG1100. Stranger things have happened though :lol:

[jonnyGURU]

jonnyGURU > i don't think there are early versions of the PSU out there anymore, they were few and the problems were corrected a good couple of months ago.

Then there IS an early version. If there were a few (an older version) and the problems were corrected a good couple months ago (a new version) than that's two versions. ;)

[Micutzu]

Haha, agreed, i was just saying there are virtually no chances to find the old version now, everything out there is the new stuff.

The TJ has 2 (not one) transformers for the 12V rails, that and the resultant circuitry (2 virtual rails per transformer) can mean that you can have phase variances. that's pretty obviously the issue here.

The 12V rails dropped from 12.28V at 2A per rail to 12.10V at 17.5A per rail. none too shabby.

The two transformer outputs are joined before the filtering stage, the two out-of-phase ripple signals being overlapped here. The "rail splitting" is actually made of different current sensors for each "rail" and nothing more, ripple and even voltage (not considering wire losses) should be identical on all +12V outputs.

[madmat]

Well, removing the rails from the outputs going into the ATE lowered the ripple at the same level of load. IE, the rails were loaded as a whole at X load and the ripple was at Y level. Leaving the load at X and simply pulling rail 1 lowered the ripple to Z while the ATE remained at X load. The only conclusion I can come up with is that it's a phase issue. If you've got a better idea I'd love to hear it but I know that before I modified the interface to marry all four rails together into one rail and instead had the rails running in pairs into the inputs on the ATE the ripple was roughly half of what I saw this way.

I had to do it as one large rail due to the SM268 only being able to do a max of 32A on rail 2 and 38A on rail one. I wanted to load all four rails evenly rather than two at 16A and two at 19A.

I'm not saying that I'm happy that there's a phase issue as the Duro didn't react the same way, in fact it was quite happy with all the rails lumped together. I wish that I had the ability to do more rails so that I could isolate the rails to the order of 12V1 on it's own channel, 12V2 on it's own channel and tie 12V3 and 12V4 together to see what the ripple values look like for those three outputs as users that are using the TJ 1100 on 8800GTX's are likely to have the rails mixed in that manner.