Hardware Insights Forum

Heads up: This is a long post.

Macron MPT-xxx series (xxx can be any 3-digit number; all have the same basic design) are the PSUs I've personally seen most often. Most of the parts look to be good for what the labels say, but capacitor failure is typical. At least they don't use custom sizes that aren't replaceable with standard components from the good brands. Capacitor complement includes:
Primary - 2 x 680µF 200V 25mm
+3.3V, +5V, +5VSB - each 2 x 2200µF 10V 10mm
+12V - 1 x 3300µF 16V 12.5mm
-5V and -12V - each 1 x 220µF 16V 6.3mm (doesn't sound like much but considering the topology, I don't actually think it will be a major problem)
1 x 470µF 25V 8mm in the power circuit for the PWM controller
4 x 10µF 50V 5mm (2 in the base driver and 2 in other areas)
4 x 1µF 50V 5mm (1 of these is in the fan controller or at least in the general area and the original is usually a 4mm even though the silkscreen supports 5mm)

I've seen them with 12KJ, CEC-ECAP, Fuhjyyu, G-Luxon, and GoldLink.
It probably doesn't help that they rarely use actual low-ESR capacitors (if they do it's G-Luxon LZ and I don't remember seeing more than 2 or 3 of those in any individual unit). At least their occasional use of low-ESR caps gives an idea of the specifications.

G-Luxon LZ 2200µF 10V 10x25: 1450mA/39mΩ
G-Luxon LZ 3300µF 16V 13x40: 2200mA/26mΩ
G-Luxon LZ 220µF 10V or 150µF 16V 6.3x11 (the 220µF 16V listed is 8mm): 300mA/220mΩ
G-Luxon LZ 470µF 25V 8x20: 1000mA/67mΩ

Based on this information, the closest replacements would be Rubycon YXJ for the 220µF 16V and Chemi-con KY/Nichicon HE for the others (note: the 2200µF should be 6.3V to match exactly).

My observations:

• The 2200µF 10V and 3300µF 16V caps fail the most.
• I haven't seen the 220µF 16V or 470µF 25V caps fail yet, but that's probably just because they aren't under as much stress.
• I haven't noticed a substantial difference in failure rates between the brands of general-purpose caps used.
• I haven't seen many G-Luxons in these PSUs (whether low-ESR LZ or general-purpose SM), so I can't conclusively prove that low-ESR is making those last longer, but so far I haven't seen them failed in these power supplies.
• I've probably seen an overall failure rate of over 50%.

If there's anything about the design that truly makes me concerned about ripple, it's that the +5V and +12V have no pi coils. (+5V has its two caps directly in parallel with each other. They could have designed it with a single 4700µF 12.5mm cap and saved a bit of space.) (I've even seen a PSU, Seventeam ST-250BLV, with two pi coils on +12V.)

The topology in MPT-xxx is a pretty standard half-bridge affair. Main switchers are 13009L and 5VSB switcher is KSC5027. Main transformer is an ERL-35.

There are versions with 80mm and 120mm fans, however the model number doesn't differentiate the two. The 120mm-fan version may actually perform worse because the heatsinks are not correctly designed for the airflow path (they have fins extending horizontally - basically the heatsinks are just vertically shrunk from in the 80mm-fan version, so the top row of fins blocks air from reaching those below!!!).

MPT-xxx2 (I presume the 2 is supposed to mean second-generation) is a completely different design. I've only seen one of those, an MPT-4012 with a 120mm fan. It's a forward converter, don't quite remember but I think it was single-switch. The platform goes up to 500W according to the PCB markings, the heatsink design is much improved, and the main transformer is an ERL-39. The primary capacitors are now 1000µF, and that's as much as I can be bothered writing about it...

Anyway, I'd at least consider reviews of these units. Even if in the next el-cheapo round-up.

I know it's not related to those units specifically but I have an idea for another polymod experiment - why not try polymodding just the +5VSB of a PSU using only as big a cap as necessary to get low enough ESR??? (I somehow doubt it's actually necessary to have, say, 2000µF worth on the 5VSB...)

Like you said, those MPT units are very well built. They will do their labelled wattage and I even think the ripple would be in spec with replaced low ESR caps. I have recapped one and it's been great...seems to run cool for an old half bridge unit.

If you're pulling <200W from the 12V, 3300uF should be plenty to keep the ripple in spec.

Are the primary caps branded by G-Luxon? My MPT-301 had 680uF G-Luxon primaries (Rarely see primaries made by them)

Alright then, I'll see if I can source a Casing Macron for the next el-cheapo PSU roundup. I see a few MPT-400s on eBay sometimes. It still won't get any awards, though, if it uses Fuhjyyu caps. 3 points usually come off for that.

I've also seen "Auriga" which I think is a sort of low-end brand for Macron PSUs. Some of them (models starting with MPT) are the same as the "real" Macrons, while other models are gutless wonders (and yes, those are made by Macron).

Do you have an extra penalty for using general-purpose caps on the outputs???

I would love to see a Macron PSU review. The one I had (branded Ever) used 100% GoldLink caps besides the primaries. They're good PSU's considering their old topology and price. Power Logic fans seem to be reliable too which I've seen in every Macron

LongRunner wrote:Do you have an extra penalty for using general-purpose caps on the outputs???

I would assume so as they create more ripple and it exponentially reduces the already low life span of GP caps

It depends on the effects. If it causes borderline ripple suppression (which is what will usually happen) then extra points will come off for that. But If it still has very low ripple due to other aspects of the design (ie, The PSU is designed well so as not to stress them), then no.

All but one of the caps in my MPT-4012 are Fuhjyyu and most of them are TN series. So far they haven't popped but I don't know whether that's due to better cooling, lower stress, lack of use, or luck.

Primary - 2 x 1000µF 200V 25mm LP
+3.3V and +5V - each 1 x 3300µF 10V 10mm (but the board is silkscreened for 12.5mm) TM and 1 x 2200µF 10V 10mm TN
+12V - 1 x 3300µF 16V 12.5mm and 1 x 2200µF 16V 10mm, both TN
Unregulated negative rail (this PSU uses a 7912 for the -12V) - 1 x 220µF 25V 8mm TN
-12V from 7912 (low ESR unnecessary) - 1 x 100µF 25V 6.3mm TN
+5VSB - 2 x 1000µF 10V 8mm TN
Other - 1 x 47µF 25V, 2 x 22µF 50V, 1 x 10µF 50V, 1 x 0.47µF 50V, 1 x 4.7µF 50V, 1 x 22µF 25V, all Fuhjyyu TN except for the 0.47µF which is GoldLink RK. The 47/25 is on the fan output and the 22/25 in parallel with the thermistor.

I've failed to find a difference between MPT-xx0 and MPT-xx1 though I have seen units with the same model number and different cables, and I even have one MPT-301 with 2 x 12A rectifiers on +12V (like the -40x units I've seen) while the other -301s I've seen have 1 x 16A there!!!

While most of the units I've seen have Fuhjyyu primaries I have seen a few with G-Luxon primaries. Oddly enough I've seen a unit with G-Luxon primaries rated at 105°C (HW series). (I have also seen 85°C G-Luxon primaries (TW series) and I don't know what they used the 105°C G-Luxons for, and while Fuhjyyu does have 105°C primary caps (LG series) I've never seen those in a Macron PSU.)

(Some of Fuhjyyu's datasheets contain the statement, I'm not kidding, "general purples". It also appears in Jun Fu datasheets. There must be some sort of merger similar to Teapo/G-Luxon as Jun Fu used to use the cross-shaped vent.)

I had to upload the GoldLink logo here as capacitor.web.fc2.com has hotlink protection.

There is some sort of IC for the standby supply in the MPT-xxx2 but with a separate MOSFET. The old MPT-xxx had a self-oscillating 5VSB but at least without an electrolytic "critical cap". MPT-xxx usually have a 30A Schottky rectifier on +3.3V and two on +5V, and a 6A fast recovery or 10A Schottky (why not just stick with one or the other???) on +5VSB.

General purples

Hahaha. On the LZ series. never noticed that until now

Jun Fu caps are just re-sleeved fuhjyyus.

LongRunner wrote:so I can't conclusively prove that low-ESR is making those last longer, but so far I haven't seen them failed in these power supplies.

Low ESR capacitors do not last longer than higher ESR capacitors unless the application of said capacitors absolutely requires that low ESR capacitors are used. Lower ESR capacitors must pass more ripple to "ground" (more ripple has to pass through them) so as to maintain a lower ripple voltage than higher ESR capacitors do. It will prolong the life of the occupied computer (low ripple going to the computer) but not the PSU. This is why you see low ESR capacitors from junk brands fail more than higher ESR ones, and why the capacitor plague has more than anything involved low ESR touted capacitors. And it's also why areas of the motherboard that require low ESR capacitors fail more than more general purpose "fielding" sections of the board (or coupling/decoupling/bypassing otherwise) - lower ESR rating means there's less "resistance" to passing ripple, and the only place for that ripple to go is through the capacitor - IE more internal heating. Lower ESR capacitors have to work harder (despite the higher ripple rating which also helps).

That might be a crudely simple explanation to a more complicated subject, but that's how I understand it. Regarding PI filters, you can get away without them, I think, if an equivalent inductor of sorts would be used, or even just capacitors with a rather low ESR rating (like three of them in parallel with an excellent ripple rating), but not too low for SMPS unless we're talking +5VSB or linear regulated +3.3V rails (those rails do benefit from extremely low ESR capacitors). Of course, without PI filters, capacitors will have to work much harder to keep the ripple low, so that's why I don't recommend skimping out on them ever. As far as the MPT PSUs go, they are alright power supplies... not the best but certainly not the worst and better with a recap than a lot of the truly horrifying cheapos out there.

(they have fins extending horizontally - basically the heatsinks are just vertically shrunk from in the 80mm-fan version, so the top row of fins blocks air from reaching those below!!!).

Not sure if that would have the same effect in the application of a PC since most of the time, PSUs are mounted upside down in a standard ATX case.

Wester547 wrote:Low ESR capacitors do not last longer than higher ESR capacitors unless the application of said capacitors absolutely requires that low ESR capacitors are used.

I think these PSUs do require low ESR caps.

Lower ESR capacitors must pass more ripple to "ground" (more ripple has to pass through them) so as to maintain a lower ripple voltage than higher ESR capacitors do. It will prolong the life of the occupied computer (low ripple going to the computer) but not the PSU.

That might be what PCBONEZ said but I don't think it's applicable to the real world except when working with multiple caps in parallel.

PCBONEZ's explanation (half ESR = twice the ripple current) assumed that the ripple voltage remains the same.

If you replace (say) one cap out of eight in parallel in a VRM with one of lower ESR, you will indeed get more ripple current through that individual cap, because they are at the same voltage and lower resistance across the same voltage does indeed mean more current, but if you replace all of the caps with lower ESR units the ripple voltage will drop, and the current through the group will (hypothetically) remain the same. (In practice it's not that simple, but I'm sure it's closer to the truth than what PCBONEZ said.)

And it's also why areas of the motherboard that require low ESR capacitors fail more than more general purpose "fielding" sections of the board (or coupling/decoupling/bypassing otherwise) - lower ESR rating means there's less "resistance" to passing ripple, and the only place for that ripple to go is through the capacitor - IE more internal heating.

As explained above, that has nothing to do with the caps having lower ESR. It's just because general-purpose caps, used as they should be, aren't put under as much stress. It's true that bad caps fail quicker but stress has the same effect on them as it does on good caps.

linear regulated +3.3V rails

Linear regulators can be functionally explained as a dynamic resistor that adjusts on-the-fly to keep the output voltage constant with varying input voltage and load. Therefore they don't produce high-frequency noise to begin with.