Corsair RM750x PSU Review

A Look Inside And Component Analysis

Parts Description

Before proceeding with this page, we strongly encourage you to a look at this article, which will provide you with valuable information about PSUs and their operation, allowing you to better understand the information given below. Our main tools for disassembling PSUs are a Thermaltronics soldering and rework station and a Hakko 808 desoldering gun.

Primary Side
Transient Filter6x Y caps, 2x X caps, 2x CM chokes, 1x MOV
Inrush ProtectionNTC Thermistor
Bridge Rectifier(s)1x GBJ25L06 (600V, 25A @ 113°C)
APFC MOSFETS2x Vishay SiHF22N60E (650V, 13A @ 100°C, 0.18 Ohm )
APFC Boost Diode1x Vishay 8S2TH061 (600V, 8A @ 120°C)
Hold-up Cap(s)2x Nippon Chemi-Con (400V, 390uF & 470uF or 860uF combined, 2000h @ 105°C, KMR)
Main Switchers2x Vishay SiHG20N50C
(560V, 11A @ 100°C, 0.27 Ohm )
APFC ControllerInfineon ICE3PCS01G - CM03X
Switching ControllerInfineon ICE2HS01G
TopologyPrimary side: Half-Bridge & LLC Resonant Converter
Secondary side: Synchronous Rectification & DC-DC converters
Secondary Side
+12V MOSFETS6x Sinopower SM4021NAKP (40V, 100A @ 100°C, 2.7 mOhm ) @ VGS=6V)
5V & 3.3VDC-DC Converters: 4x M3006D & 2x M3004D fets
PWM Controller: APW7159
Filtering CapacitorsElectrolytics: Chemi-Con (105°C, KZE & KZH series)
Polymers: Nippon Chemi-Con
Supervisor ICWeltrend WT7502 & Weltrend WT7518
Fan ModelNR135L (12 V, 0.22 A, Rifle Bearing)
5VSB Circuit
RectifierPFR20V45CT (45V, 20A, VF: 0.42V max @ 125°C)

Basically, this is exactly the same platform as the RM750i. The only exception is the absence of the digital interface, which allows the monitor to partially control the PSU through the Corsair Link software. In both cases (the RM750i and RM750x), analog circuits control the PSU's operation in the APFC converter and the primary/secondary sides, meaning that the digital interface acts only as a bridge between the PSU and the system. Hence, it has nothing to offer, performance wise.

All RMi and RMx units are manufactured by CWT, Corsair's favorite OEM. In the primary side we meet a half-bridge topology along with an LLC resonant converter. In the secondary side a synchronous design is used, as usual, along with a couple of DC-DC converters for the rectification of the minor rails. The use of only Japanese capacitors promises increased reliability and offers a stable performance over time. Finally, as CWT used to do, this platform doesn't use any proper heat sinks in the secondary side. This doesn't seem to affect this unit's reliability; however, if it did, the company wouldn't offer such a lengthy warranty.

The AC receptacle hosts two Y caps. The other parts of the EMI filter are located on the main PCB, including four Y and two X caps, two CM chokes and a small MOV.

The single GBJ25L06 bridge rectifier is bolted on a dedicated heat sink. It can handle up to 25A of current, easily meeting the demands of this PSU.

Two Vishay SiHG22N60E FETs are used along with a single Vishay 8S2TH061 boost diode by the APFC converter. The pair of bulk caps is provided by Chemi-Con (400V, 390uF & 470uF or 860uF combined, 2000h @ 105 Celsius, KMR) and their combined capacity is more than enough for this specific platform.

The APFC controller is hosted on a vertical daughterboard. The APFC controller is an Infineon ICE3PCS01G that is backed up by a CM03X Green PFC controller. On the same board we also found the LLC resonant controller, an Infineon ICE2HS01G.

Arranged into a half-bridge topology, two Vishay SiHG20N50C FETs are used as main switchers.

As we already stated, the secondary-side CWT didn't use any proper heat sinks, even though this PSU features a semipassive mode. To minimize noise output, the fan isn't spinning at all when the PSU is in this mode. The FETs that regulate the +12V rail — four Sinopower SM4021NAKP — are installed on two vertical PCBs. In addition to their power-transfer duties, a small number of bus bars also work to cool the PCBs. Four Chemi-Con electrolytic caps (rated at 105 C) are installed between the PCBs that hold the +12V FETs. These caps are used for ripple-filtering purposes.

A large daughterboard houses the couple of DC-DC converters that generate the minor rails. The common PWM controller is an Anpec APW7159, and each VRM uses a couple of M3006D FETs, along with a single M3004D FET.

Housekeeping is done by a Weltrend WT7502 IC, which is installed on the main PCB.

A PFR20V45CT Schottky barrier diode (SBR) regulates the 5VSB rail.

The MCU PCB in the RM750x looks empty, since the PIC32MX that is used on the same card in the RM750i is missing. In addition, instead of a four-pin fan header, a two-pin fan header is used in this case because the PWM doesn't control it.

At the front side of the modular PCB, quite a large number of Chemi-Con polymer caps filter the rails. At the rear side of the same board, several thick cables transfer power from the regulators to the modular sockets.

Soldering quality is good overall, and we didn't spot any long component leads, which we hate to see, since they can cause trouble.

To reduce production costs, Corsair used a rifle-bearing fan instead of an FDB fan. The NR135L fan (135mm, 12V, 0.22A) is a low-speed unit with very low noise output even at higher speeds. The semipassive mode and the highly relaxed fan profile combine to help this fan offer a very silent operation for the RM750x, under all conditions.

This thread is closed for comments
19 comments
    Your comment
  • Blueberries
    I've been promoting these for a while, it's nice to see Tom's do an article on them. The only reason these are Gold rated is because they just miss the Platinum rating at 20%.

    They have a 650x as well that's a little cheaper.
  • Amdlova
    High Rate Faliure can be the the name of this series of corsair. for Now EVGA or SEASONIC power supplies. don't spend a penny on corsair products...
  • Aris_Mp
    Please keep in mind that some Corsair PSUs are also made by Seasonic. Also this series is very new to have a high rate of failures. Unless you have some solid facts to share on the older RM line which is out for quite some time now.
  • jonnyguru
    1335368 said:
    High Rate Faliure can be the the name of this series of corsair. for Now EVGA or SEASONIC power supplies. don't spend a penny on corsair products...


    Wrong on so many levels.

    YOU do not have the failure rate for this or any other Corsair PSU.

    This PSU is the RMx, not the RM, so even if you did have a failure rate, it would only be about two weeks of data.

    If you were talking about the RM and not the RMx, and you actually had failure rate data, you would see that the failure rate on the RM wasn't really high at all.
  • PureBlackFire
    Quote:
    High Rate Faliure can be the the name of this series of corsair. for Now EVGA or SEASONIC power supplies. don't spend a penny on corsair products...


    *sigh* nonsense comment of the day.
  • chalabam
    Tomshardware:

    I don't know if this is a problem of your site, or my PC/IP, but frequently the charts do not load, even when the page "ends" loading.



    Sometimes, if I "reload", then the charts also load.

    I open all the article pages simultaneously, on different tabs, so I don't need to wait for each one to load.
  • Rookie_MIB
    The older RM units weren't 'bad' really, they were ok, but Corsair has been stepping up its game with quality parts and build on some of these newer units which is nice to see. You can never have too many solidly designed units to choose from - competition toughens the breed.
  • Blueberries
    Quote:
    High Rate Faliure can be the the name of this series of corsair. for Now EVGA or SEASONIC power supplies. don't spend a penny on corsair products...


    Did you even LOOK at the article? The only problem with these are the Sinopowers on the secondary side, and that's not even a "bad" thing. Oh and btw, some of the best power supplies in the world are Superflower OEMs produced by Corsair.
  • mavikt
    With these in depth coverage of PSU's I propose introducing a tier'ing table thing at the and of each review (or perhaps a best buy PSU of the month) equivalent to what's done for CPU's and GPU's, ranking PSU models (perhaps too the makers). I saw a comment here on toms on another PSU news flash in the comment section referring to such thing in the forum but now I can't find it.
    Permanent'ing such thing from the editorial side would be great!
  • anort3
    Quote:
    Quote:
    High Rate Faliure can be the the name of this series of corsair. for Now EVGA or SEASONIC power supplies. don't spend a penny on corsair products...
    Did you even LOOK at the article? The only problem with these are the Sinopowers on the secondary side, and that's not even a "bad" thing. Oh and btw, some of the best power supplies in the world are Superflower OEMs produced by Corsair.


    Corsair doesn't use SuperFlower anywhere in its lineup. CWT, Great Wall, Flextronics, Seasonic and Chicony which has since been dropped are all the OEMs Corsair uses or has used.
  • JackNaylorPE
    Tier lists and OEM brand are by no means an indicator of quality. Just because a certain OEM is used is no guarantee of quality. Just like Corsair sells Great, Good, moderate and crappy PSU's, most OEMs also produce a wide range of quality.

    Tier lists will include an entire product line in a category based upon a single model review. The Corsair HX 750/850 were great, the 1000 / 1050 were dogs but all were placed in the same tier. Below we see that CWT made their lowest tier as well as some of the best products Corsair ever sold.

    CS was made by Great Wall
    CX and GS by CWT
    VS by CWT
    RM 450-650 were made by CWT,
    RM 750 and 850 by Chcony
    RM 750v2 -850v2 and 1000 were made by CWT
    RMi / RMx series are made by CWT
    HX 650 was made by Seasonic
    HX 750, 850 and 1050 were made by CWT (2/3 were excellent)
    HXi series was made by CWT
    AX series was made by Seasonic
    AXi was made by Flextronics
  • Blueberries
    218893 said:
    Quote:
    Quote:
    High Rate Faliure can be the the name of this series of corsair. for Now EVGA or SEASONIC power supplies. don't spend a penny on corsair products...
    Did you even LOOK at the article? The only problem with these are the Sinopowers on the secondary side, and that's not even a "bad" thing. Oh and btw, some of the best power supplies in the world are Superflower OEMs produced by Corsair.
    Corsair doesn't use SuperFlower anywhere in its lineup. CWT, Great Wall, Flextronics, Seasonic and Chicony which has since been dropped are all the OEMs Corsair uses or has used.


    Wrong. The AXi is a SuperFlower Leadex OEM. Which is sometimes produced by Corsair, Seasonic, and EVGA.
  • anort3
    The AXi OEM is Flextronics..... Unless you know something all the review sites don't?

    http://www.realhardtechx.com/index_archivos/Page447.htm

    http://www.orionpsudb.com/corsair

    http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story6&reid=317

    http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story5&reid=378


    And why would Seasonic make a SuperFlower platform? They don't. The 2 companies are direct competitors. You are correct about EVGA using SuperFlower though.
  • Blueberries
    I'll shut up now. I thought the AXi was SuperFlower which is the stem of my confusion
  • jonnyguru
    35894 said:
    The Corsair HX 750/850 were great, the 1000 / 1050 were dogs...


    They were?!?!?! Says who???
  • JackNaylorPE
    Quote:
    They were?!?!?! Says who???


    The guy who gave it a 8.5 performance * and 7.0 Build Quality rating :)
    http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story5&reid=245

    The 850 got a 10 in performance but back then no BQ ratings
    http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story5&reid=153

    * 1000 wasn't that bad with a 9.0 performance rating. I used it 3 times as I recall, had problems with 2..... one would hiccup under SLI loads, a user build had fan noise issues.
  • jonnyguru
    35894 said:
    Quote:
    They were?!?!?! Says who???
    The guy who gave it a 8.5 performance * and 7.0 Build Quality rating :) http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story5&reid=245


    Right. And if you read the fine print:

    "More loose heatsink screws to deal with. But at least there were only two of them in there this time that could be considered a potential problem. Two points off for that. It'll probably never be a problem for 99% of these sold, but I'm an electronics tech. I get to be picky with stuff like that."

    This doesn't make a product "a dog".

    AFAIK, it really wasn't a problem for 99% of the folks out there and since it was pointed out in the reviews, CWT started using nylon washers or thread lock on all screws.
  • tacgnol06
    RM750i is currently slightly cheaper than this model on Newegg, which leads me to wonder why they'd release this version to begin with, unless they're phasing it out to replace it with something cheaper to make for around the same retail price... oh.
  • JackNaylorPE
    112719 said:
    Right. And if you read the fine print: "More loose heatsink screws to deal with. But at least there were only two of them in there this time that could be considered a potential problem. Two points off for that. It'll probably never be a problem for 99% of these sold, but I'm an electronics tech. I get to be picky with stuff like that." This doesn't make a product "a dog".


    I did read the fine print, but I am speaking "in context" ... As you said ... 2 points off... that only brings it to a 9. In the general context of things, I normally would call a 9.0 build quality rating a dog.... perfectly acceptable, perfectly suitable for an office or budget restricted gaming box.

    But when one is looking at, and prepared to pay for what the HX series bought to the table for a presumed moderate to higher end build for overclocking or gaming, the mindset is "top end".

    I expect a 10.0 here. When you are paying 10.0 quality a prices, a 9.0 BQ / Perf. rating is gonna get a "dog" label, at least in my eyes. If I wanted 9.0's, I'd look at cheaper model lines. As a TX, and without the careless assembly, the moniker would not have been used.