thermaltake tr2 430w

[tom66]

Re: thermaltake tr2 430w

So then... having low ESR/impedance/high ripple rating capacitors on the secondary guarantees stabler power and matters far more than having the highest uF rating?

Both factors are important. An upgrade from 3300uF to 4700uF might be less noticeable than a corresponding 30% decrease in ESR. However, I believe changing ESR and capacitance can affect compensation which can make ripple -worse- due to lots of complex mathematical things I don't understand.

And I always thought Antec rebranded PSUs, so whether their PSUs were good or not would depend upon who they're branding (whether it be Seasonic, Delta, FSP, etc...).

Yes - plus the "make it cheaper" fudge factor you have to include when Antec talk to OEMs.

[Phaihn]

Re: thermaltake tr2 430w

I attached a pic with my suggestions. The cap circled is 85 Celsius cap. This is unacceptable for a power supply. You should replace it with 105C cap. I am not sure if this cap is for -12V or 5vsb. If it is responsible for 5vsb filtering along with the other 470uF I would advise you to put a 1000uF there too.



You better stick to the original voltage values. 6.3V can be replaced with 6.3V or higher Volts (10V, 16V).

10V -> 10V or 16V

16V ->16V

is it safe to go from 1000uf to 470uf?

[hkqq2010]

Re: thermaltake tr2 430w

i just put them back the way they were

Good. It should take out the hot air out of PSU and PC.

[Phaihn]

Re: thermaltake tr2 430w

Good. It should take out the hot air out of PSU and PC.

[Phaihn]

Re: thermaltake tr2 430w

had to use 1200uF caps didnt have any 1000uF 16v caps.

[c_hegge]

Re: thermaltake tr2 430w

^
You recapped with OST?

[goodpsusearch]

Re: thermaltake tr2 430w

Have you even checked the series and the esr of the old and the new caps?

[LDSisHere]

Re: thermaltake tr2 430w

^
You recapped with OST?

The tall green one is a Teapo, I am surprised you did not mention it. Unless OST is so much worse than Teapo that they seem good in comparison.

I would not use these either (Thanks to this site educating me.) unless it was something I needed right now or it was just for temporary use. If it was something that I wanted to use long term than I would do it using good caps. If getting caps is a problem then you just have to do the best you can with what you have and hope it does not go .

[momaka]

Re: thermaltake tr2 430w

Several things that need to be addressed here...

1) It's nice to post the stash of caps you have, but without listing what series they are or what dimensions they have, then we can't really recommend which ones would be appropriate replacements. Same goes for when you listed the caps in the PSU. Ideally you want to put it in this format:
voltage rail: -- cap brand -- cap series -- cap voltage -- capacitance -- cap diameter (in mm).

2) The stash of caps you have all seem of inferior brand or inappropriate series (or both: the United Chemicon KZG, for example). In this case, I would suggest not to replace the caps in the PSU except those 85C rated ones that goodpsusearch pointed out, along with any other 85C rated caps (except for the big high voltage Teapo caps on the primary side - those can stay). The small caps can stay too

3) You might want to get rid of this loose wire. Looks like a manufacturing defect/sloppy work.

If momaka saw this, I am sure he would warn you about that resistor near the 5vsb capacitor. Try to bend the resistor away of the capacitor if you don't want it to cook the cap.

It's like you can read my mind
All good, though. Saves me from having to write the same stuff over and over .

[goodpsusearch]

Re: thermaltake tr2 430w

Indeed. Replacing TEAPO with OST is not an upgrade, it's a downgrade.

[momaka]

Re: thermaltake tr2 430w

In general at high frequencies capacitance is less important: it's ESR which is the killer. Basically you want to dump all that high frequency to ground and get flat DC out. Lower ESR, less impedance for the AC. That's a basic view of the problem.

In theory, you are correct. But there's more to those PSU filter circuits than that. Unfortunately, my knowledge is a little rusty now since most of my theory comes from university physics courses I took several years ago (and I actually dropped out of uni last spring, so don't ask me for formulas or too many details please - it's all gone from my head ). Therefore, the below information may or may not be entirely accurate (and please correct me if it isn't).
...
So basically, what kind of caps you can use in a PSU has a lot to do with the arrangement of the circuit - mainly, the filter circuit of PC PSUs has both caps and inductors. They are arranged in what is called an LC (i.e. inductor-capacitor) series circuit. An LC circuit is like a mass hanging from a spring. If you raised the mass and dropped it, the mass would begin to oscillate up and down. In an ideal (frictionless) world, this oscillation would continue forever. In the real world, however, this is not the case since there are resistive and frictional forces, so eventually the mass would stop oscillating.

Therefore, in reality, an LC series circuit actually is an LCR (inductor-capacitor-resistor) series circuit. The resistance of this circuit is present in the copper traces, the wire in the inductor, and most importantly the ESR of the capacitor. LCR circuits are special - they are called tuned circuits because they can filter out a range of frequencies, depending on the components' parameters.

The LCR circuit can be classified in 3 states: under-damped, critically-damped, and over-damped.
- An under-damped LCR circuit is one where the resistance is so low that the circuit behaves more like an LC circuit (i.e. lots of oscillations).
- A critically-damped circuit is one where the resistance is just right and there are a few oscillations that die out very quickly. If you're trying to build a good filter, you want to get your LCR circuit in this state.
- Finally, there's the over-damped circuit where the resistance is too high and there are no oscillations.

Now you're probably wondering what all of these oscillations have to do with the PSU filtering. Well, it has to do a lot with the power transfer of the ripple. The better the power transfer of the circuit, the more the ripple current will be "shunted" to ground and the less of it there will be present on the outputs of the PSU (i.e. you get a cleaner power output).

In an under-damped LCR circuit, the circuit may begin to oscillate at its own tuned frequency. This is highly undesirable in a PSU because while the circuit may filter out the ripple, it will introduce it's own characteristic ripple on the outputs. This is called ringing and that's the sole reason you DO NOT want to use polymer or very low ESR caps in a PC PSU. An exception to this is if the inductors in the PSU are tiny, such as is the case with most LCD monitor power supplies (but that opens a whole new can of worms... you have to change the compensation on the feedback circuits and the lack of inductors also puts much more stress on the schottky diodes). Also, the ringing can mess with the compensation of the feedback circuit, so that's another reason to stay away from poly and ultra-low ESR caps when there are large inductors in the circuit.

In an over-damped LCR circuit (let's say from general purpose caps or out-of-spec high ESR caps), the circuit just won't do as much to the ripple current and so the majority of the ripple current may continue forward to the outputs of the PSU. This is also undesirable because you get "dirty" power.

In a critically-damped LCR circuit, most (ideally all) of the ripple current is filtered out by the circuit and the ringing is very small. This is how you want the circuit to run.

As far as reducing/increasing the capacitance in a PSU LCR circuit - well, if you do it, do so reasonably. Generally, you should not decrease the capacitance because this raises the filtering frequency so you get more ripple from the higher frequencies as well. This adds up with the ripple from the lower frequencies and you get more ripple on the output.
On the other hand, increasing the capacitance can help reduce ripple. However, if you go too high you can cause problems again.

Based on what I have seen in various 200-300W PSUs (both good and crappy ones), this is generally what I think should be the minimum configuration for each voltage rail:
2x 2200uF caps on the 3.3V rail
2x 2200uF caps on the 5V rail
1x 2200uF or 2x 1000uF on the 12V rail.

Now, depending on whether you'll be putting a heavier load on the 5V rail or the 12V rail, you can probably make small adjustments to those values. For a 12V system, I recommend to up the capacitance on the 12V rail to at least 3000uF or so. You may also be able to lower the capacitance on the 5V rail (if you have to due to not having the proper caps or similar), but if the PSU is from an older design with a heavy 5V rail, then it's best not to lower the capacitance.

[Phaihn]

Re: thermaltake tr2 430w

^
You recapped with OST?

that was all i had otherwise i would of used rubycon or pansonic.

[Phaihn]

Re: thermaltake tr2 430w

Have you even checked the series and the esr of the old and the new caps?

how do you do that?

[Phaihn]

Re: thermaltake tr2 430w

i didnt replace teapo with ost i replaced YEC with ost.

[Phaihn]

Re: thermaltake tr2 430w

Several things that need to be addressed here...

1) It's nice to post the stash of caps you have, but without listing what series they are or what dimensions they have, then we can't really recommend which ones would be appropriate replacements. Same goes for when you listed the caps in the PSU. Ideally you want to put it in this format:
voltage rail: -- cap brand -- cap series -- cap voltage -- capacitance -- cap diameter (in mm).

2) The stash of caps you have all seem of inferior brand or inappropriate series (or both: the United Chemicon KZG, for example). In this case, I would suggest not to replace the caps in the PSU except those 85C rated ones that goodpsusearch pointed out, along with any other 85C rated caps (except for the big high voltage Teapo caps on the primary side - those can stay). The small caps can stay too

3) You might want to get rid of this loose wire. Looks like a manufacturing defect/sloppy work.


It's like you can read my mind
All good, though. Saves me from having to write the same stuff over and over .

i only changed out 3 and they were 85c ones.

i cut that off yesterday that was my own handy work i removed cables that were no longer of use. mostly extra old style molex plugs that were never used anymore.

[Wester547]

Re: thermaltake tr2 430w

Based on what I have seen in various 200-300W PSUs (both good and crappy ones), this is generally what I think should be the minimum configuration for each voltage rail:
2x 2200uF caps on the 3.3V rail
2x 2200uF caps on the 5V rail
1x 2200uF or 2x 1000uF on the 12V rail.

Very informative! Thanks for taking the time to write that out. And well, an old Dell Newton Power 200W I have only had 2x 1500uF/10V (Rubycon YXG) on the 5V rail and 1x 1500uF/10V and 1x 2200uF/10V (Rubycon YXG) on the 3.3V rail. Sounds a bit low but note that this PSU is spec'd for ATX, not ATX12V, so I suppose for a 200W it's acceptable. There are toroidal coils on every output which helps with the filtering. It's single transistor forward and on the 5V and 3.3V rails there's a TO-247 30A part from General Semiconductor (with a 45V reverse block voltage - MBR3045PT).

Now, depending on whether you'll be putting a heavier load on the 5V rail or the 12V rail, you can probably make small adjustments to those values. For a 12V system, I recommend to up the capacitance on the 12V rail to at least 3000uF or so. You may also be able to lower the capacitance on the 5V rail (if you have to due to not having the proper caps or similar), but if the PSU is from an older design with a heavy 5V rail, then it's best not to lower the capacitance.

An old Huntkey 250W I had only had 1x2200uF/16V FCon GL on the 12V rail and powered an AMD Athlon 64 3800+ (85W TDP) and GeForce 8600GT/9600GT LP (50W/60W TDP) for years without a problem... and that was with a low speed, 80mm temperature controlled fan from Yate Loon. I dunno how it pulled that off. It had two 10A ultrafast recovery transistors from NXP (two BYQ28E-200s - a 200V block reverse voltage on both) on the 12V rail and it was half-bridge, though that still doesn't sound like a lot to me... it did have a pair of TO-220 13009s, though, as its switching transistors. And it also had toroidal coils on every output on the secondary.

OST is definitely an upgrade from YEC, but it's best to use Japanese capacitors.

[Phaihn]

Re: thermaltake tr2 430w

ill use japanese caps when i have the money to buy them the $8 shipping cost doesnt help.

[Wester547]

Re: thermaltake tr2 430w

Also, does the 20-+% tolerance in the LCR circuit in terms of capacitors somehow carry over to primary capacitors as well or is that totally different?

[hkqq2010]

Re: thermaltake tr2 430w

... an LC series circuit actually is an LCR (inductor-capacitor-resistor) series circuit. The resistance of this circuit is present in the copper traces, the wire in the inductor, and most importantly the ESR of the capacitor. LCR circuits are special - they are called tuned circuits because they can filter out a range of frequencies, depending on the components' parameters.

The LCR circuit can be classified in 3 states: under-damped, critically-damped, and over-damped.
- An under-damped LCR circuit is one where the resistance is so low that the circuit behaves more like an LC circuit (i.e. lots of oscillations).
- A critically-damped circuit is one where the resistance is just right and there are a few oscillations that die out very quickly. If you're trying to build a good filter, you want to get your LCR circuit in this state.
- Finally, there's the over-damped circuit where the resistance is too high and there are no oscillations ...

Vow ... it is really a good info ...
Althounght I do not fully understand all of it, I had copy it down. I will re-read ... re-read and re-read it later.

Thanks for your explanation and teaching.
Many thanks.

[momaka]

Re: thermaltake tr2 430w

An old Huntkey 250W I had only had 1x2200uF/16V FCon GL on the 12V rail and powered an AMD Athlon 64 3800+ (85W TDP) and GeForce 8600GT/9600GT LP (50W/60W TDP) for years without a problem... and that was with a low speed, 80mm temperature controlled fan from Yate Loon. I dunno how it pulled that off. It had two 10A ultrafast recovery transistors from NXP (two BYQ28E-200s - a 200V block reverse voltage on both) on the 12V rail and it was half-bridge, though that still doesn't sound like a lot to me...

Well, the output filter caps don't really dictate how much power the PSU can output. Technically, it probably is possible to design a SMPS PSU capable of doing 100A @ 12V with just a single 2200uF cap. The problem it might have, however, is with ripple and noise on the outputs. And computers don't really like noise and ripple.

That Huntkey seems to have a decent power capability on the 12V (2x 10A rectifiers in a half-bridge gives a maximum of 20A capacity). However, the power probably wasn't very clean due to the lack of filtering. Most likely it was just enough to be in spec and/or make the computer not crash.
Generally, decent board are designed to be able to work even if the power they receive has out-of-spec noise and ripple. This puts a heavier toll on the board's capacitors, but it still works.

Also, does the 20-+% tolerance in the LCR circuit in terms of capacitors somehow carry over to primary capacitors as well or is that totally different?

Not sure what you mean here.
Most electrolytic capacitors specify a +/-20% tolerance for their capacitance. It doesn't matter whether you're designing an LC filter circuit or not - you still have to take into account that if you're using electrolytic capacitors, their capacitance may vary up to +/-20%.