• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FRONTPAGE Rosewill TACHYON 1000 Watt 80+ Platinum Power Supply Review

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Overclockers.com

Member
Joined
Nov 1, 1998
Rosewill has decided to get into the High Wattage High Efficiency power supply game, their TACHYON series runs from 550 W up through 1000 W. Naturally enough, I went for the 1000 W flavor. It boasts 80+ Platinum efficiency, modular cables, as well as general awesomeness. As a side note, a "Tachyon" is a theoretical particle that always moves faster than light. Unless this PSU is playing serious games with time, it isn't that sort of tachyon. It's a bit big for a single particle anyway.
... Return to article to continue reading.
 
Was waiting for these Tachyon units to land Finland, ended up going with the Super Flower Golden King instead. Looks like this would've been a good choice too!
 
nice review, I have had an LED that's leads where not completely covered slip out of a reservoir and trip the protection (*edit* because the lead hit the 5.25" drive cage) Not sure what would have happened if it did not (melted LED lead?) but that is a bit scary to see it did not trip.

How does great ripple affect the computer compared to good or below average ripple?
 
How does great ripple affect the computer compared to good or below average ripple?
I'm not completely sure is it because of small ripple, transient spikes or smaller "cross-talk" between the different rails, but I've found my Golden King which has very little ripple to not have the buzzing/whirring/chirping on sensitive speakers, depending on CPU and GPU loads.

Modern motherboards have lots of multi phase power filtering so I don't think any of the PSUs which clear the most current ATX revision should have issues even if the ripple was relatively large, but still within specs, but analog sound circuitry seems to benefit from a high-quality PSU, the ancient one I used for a week was NOISY on the speakers.
 
I'm not completely sure is it because of small ripple, transient spikes or smaller "cross-talk" between the different rails, but I've found my Golden King which has very little ripple to not have the buzzing/whirring/chirping on sensitive speakers, depending on CPU and GPU loads.

Modern motherboards have lots of multi phase power filtering so I don't think any of the PSUs which clear the most current ATX revision should have issues even if the ripple was relatively large, but still within specs, but analog sound circuitry seems to benefit from a high-quality PSU, the ancient one I used for a week was NOISY on the speakers.

That is interesting I could understand choke whine being less of an issue on a better PSU wether it was because of the ripple or not and it may very well have been a factor to your audio phenomena but there are other unexplanables that added to your issue with your speakers. I was wondering though if the lower ripple effected overclock stability or stability in general and again comparing good to great not poor or out of spec to good or great.
 
Realistically speaking, it depends on the ripple.

The long waveform ripple that comes from the feedback loop in the regulator is almost always audio frequency, having that flavor nice and low will reduce the chances of inductor whine and possibly help with stability on the absolute ragged edge OCing. More on that in a bit.
This is a good example of long waveform ripple:
long duration.jpg
Note that you can see the medium and short duration ripple imposed on the long duration.

Medium duration ripple is an artifact from switching the MOSFETs on and off, when they're switched on the voltage rises (way above 12v/5v/3v3, if you leave them on too long), when they're switched off the voltage drops. All SMPS PSUs have some of this ripple, it's unavoidable.
You won't hear this ripple directly as typical switching speeds are ~70kHz. You can hear the 3rd and 4th harmonic though, if the inductors aren't well secured and there is a lot of this flavor. Here's an example:
medium-switching.jpg
This flavor is less detrimental to OCs than the long duration, enough that Kingpin and TiN use NEX1500 units that have more ripple than some others in that price range. The extra ripple's very slight chance of causing issues is outweighed by being able to advertise the product. (Alternatively, they may simply have been told what PSU they're going to use, I don't know)

Last is short duration ripple, also known as transients. These are extremely short spikes that happen when a MOSFET switches. Typical duration is 500ns or less. They can cause issues with an extreme OC, but are unlikely to. Massive transients are a potential hazard to hardware that doesn't have proper filtering to remove them, as they can pass straight through VRM sections fairly unchanged. Your HDD motor doesn't care if there is a 400ns spike of 500mV. Your CPU would appreciate it less, though.
This picture is a wonderful example of both medium and short duration ripple:
short-switching.jpg
Short duration ripple doesn't have to originate in the rail it comes out in, either. As the MOSFETs charge and discharge the inductors (APFC) and transformers (5VSB, 12V, 5V, 3V3) there is an external magnetic field that changes violantly, that's the only reason this stuff works in the first place. However, changing magnetic fields induce current in conductive materials they are around. Welcome to EMI! The field of the main transformer switching for 1500w is quite intense, while the electrical output goes through proper filtering to smooth it out the field can induce transients in the modular PCB or in the output wires. The PSU case acts as a faraday cage to keep this EMI hell internal, and good PSUs are set up so that the wires and PCB traces are the wrong length to receive the EMI signal strongly (just like an antenna, the wrong length means terrible reception).





On OCing:
Normal 24/7 OCs can take a lot more ripple than extreme OCs can, and even at the extreme level the difference between 20mV on 12V and 80mV on 12V is going to be extremely difficult to find.
Audio stuff generally has very good input filtering, but if that is overpowered by a messy PSU you'll know it. With good input filtering (be it on a card or onboard) any PSU within the ATX spec should be fine in general.

A general note:
The last units I had with wild, but not especially dangerous, ripple were the InWin Glacier units. They had transients in the 190-200mV area on all three rails at full load. I sold both of them (with large disclaimers and warnings of course, and a rather low price) and both as still in service with no issues at all as far as I know. As you can see in the ripple pictures, the MOSFETs are creating nice spikes when they switch. Internal EMI or poor filtering? Hard to say. I've had other units where pushing the cable pack into the PSU 1/8" or pulling it out 1/8" could raise or lower transients by >100mV.

On inductor whine:
All inductors will ring at some frequency(s). They also ring magnetically at some frequencies, sometimes the same, other times totally different. Whether we want them ringing magnetically really depends on what we're doing.
Unfortunately we have a lot of harmonics to worry about.
Example: An inductor (anything, really) switched at 100kHz will have potential to ring (both ways) at 100kHz, 50kHz, 25kHz, 12.5kHz, 6.25kHz, 3.125kHz, etc. Each harmonic is weaker than the one before, so a good (for audible noise, anyway) is one that does all its ringing at 100kHz and 50kHz, and doesn't have enough energy left over at 25kHz and 12.5kHz to overpower the physical inertial of the thing and the silicone glue good manufacturers put on it.
A lot more detail could be gone into here, but I have reviews to write and cookies to make, so I'm going to cut it off here.


In closing:
If it's within ATX spec it is very unlikely to cause any issues with anything. That's the point of the ATX spec.
Devices with lousy input filtering or extremely sensitive devices that don't have staggeringly overkill input filters may be different.
 
Thanks for the post, was looking for some good insightful knowledge on how can a PSU affect sound if the circuitry has less than optimal power supply rejection ratio, really informative stuff!
 
awww you just got into the good stuff too!

wow @ that post, very informative I must say it helps my understanding of the situation in there
 
Thank you for the explanations, both in the main piece and in your long response. I think it is a clear explanation of issues in PSU 102, and needs to be posted in a citable place somewhere. Great job.
 
Back