Refurbishing an old Macron MPT-460

[Pentium4]

Some of you may remember me posting about this a long time ago. My last week at my last tech company I ordered this on eBay and has been sitting in a box ever since. I recently thought about this and remembered how much of a beast it is!

It's for the most part a standard Macron PSU but at the same time I've never seen one like it. First of all, it's got a dual layer PCB. Cool! It's also got short but very hefty heatsinks. Look at the chunk on that primary one!

The thing has a pretty good input filter. It's got a coil, X cap, 2 Y caps on a small board not pictured. Then on the board is another coil, X cap, 2 Y caps, and 3 MOVs. The bridge rectifier is 8A. Bulk caps are 1200uF! The switchers are strong 2SC3320's with a solid 15A rating.

The secondary uses two 30A schottky for 5A, two 16A ultra fast for 12V, and a 40A for 3.3V. The supervisor chip is an SG6105D.

I'm sure you all noticed that caps All of the GL caps look bulged, except for the 470uF 25V one on the 5VSB. There are only three Fuhjyyu caps, and they are on the 5VSB as well so naturally they are bulged! The other is on the 3.3V rail before the coil. I'm guessing these are all GP caps. There is no way with the capctiances listed that they are also Low ESR. Most of the little caps are G-Luxon. I will test all of them. I will also test the Gincon caps and see if they are actually 1200uF. If you look closely on the solder side, that big GL cap bulged so hard on the bottom that it started to desolder itself!

Soldering is mostly good, but some sloppy hand soldering and some barely filled joints. What I found interesting is the date codes. The transformer date codes are all from very early 2005, the Superred fans were made in 2004 (Same model number, 7 months apart!) but the unit itself was made in January, 2008.

Viotek isn't fooling any (nerds) with their bullshit label. 700W continuous? No way! You can even see the "ATX460 MACRON POWER CO. LTD" on the top of the PCB and "MPT-460" on the bottom. Also, "Customizable cable system" ?? It's not even semi-modular

Originally I was thinking about upping the 12V capability of this thing, but I was thinking I would actually use it in my dual PIII system I am working on because it has a really strong 5V rail, and all molex connectors. The main motherboard plugin is all 16AWG and is 24 pin (Which is actually what the PIII board has)

When I tested to see if this thing worked in March, 2015 it powered up but made an insanely high pitched squeal. Louder than any Antec Stupid Power even. I wonder what is causing it?

[Wester547]

Re: Refurbishing an old Macron MPT-460

I think the Gincon primaries are G-Luxons rebranded (TW85˚C is a G-Luxon series). Here is another thread showing Gincon primaries, once again another G-Luxon 85˚C series (GR), with the sleeves bearing (no pun intended...) the same color profile and date / factory code style. It's at least possible they are distributing caps from the same OEM. So I'd guess they are really 1200uF unless they are out of spec.

I'm wondering if they (Viotek) simply decided to take an old Macron PSU made in 2005 and house it in a 2008 chassis. I also noticed that there are two chokes regulating the unrectified +5V transformer tap voltage down to +3.3V (on the AC side of the heatsink, attenuating AC and bringing down the voltage). I wonder if they are both for that (in parallel) or if the second choke is for something else.

Yeah, the GL (Goldlink) cap blew its bottom rubber bung out and desoldered itself. It outgassed so quickly, there must have been more than enough internal pressure. KZGs, YECs, Sacons, and Lelons were known to do that as well, back in the day. I wonder how many of those GL and Fuhjyyu caps bulged sitting in storage if any. And as far as I know, PSUs that use ultrafast recoveries on the +12V rail may not take too kindly to being swapped for schottky rectifiers because of the transformer secondary design (I believe you once encountered the same issue).

[kaboom]

Re: Refurbishing an old Macron MPT-460

I also noticed that there are two chokes regulating the unrectified +5V transformer tap voltage down to +3.3V (on the AC side of the heatsink, attenuating AC and bringing down the voltage). I wonder if they are both for that (in parallel) or if the second choke is for something else.

Magamp chokes for regulating +3.3 and/or +5.

Two chokes can mean:

One on each AC leg to the +3.3 rectifier (two total)

-or-

One on an anode of the +5 rect, and one on an anode of the +3.3 rectifier; the other sides of each rectifier being directly connected to its respective transformer tap. This would allow both +5 and +3.3 to be seperately regulated, via their own magamp driver ckts.

The main loop would control +12, and +5/+3.3 would be rough-regulated by feedback resistor network weighting, and fine-regulated with their magamps.

If they're regulating all three...

Otherwise, it's typical group-reg, with +5/+12/-5/-12 using only the coupled inductor and feedback weighting, with +3.3 using a magamp.

We'd have to trace the traces and study the unit.

[PeteS in CA]

Re: Refurbishing an old Macron MPT-460

The two magamp inductors might be connected in series. The wire might be too heavy to get enough turns for the right inductance. Or one core might not be able to handle the necessary flux density, so it is split by splitting the turns across two cores (I think that's how it might work). Or both.

Is the inverter full bridge? I think it may be. One little rule-of-thumb tip regarding power transistors is to look at the collector current vs. beta chart in the datasheet and note at what current the beta falls to 10. That is its useful rating. In this case the 2SC3320 "15A" transistor is useful to about 10A.

I noticed the Y-caps are safety agency approved. I also noticed that the O/P inductor cores are Micrometals -52 material rather than the older -26 material. The inverter switches being transistors rather than MOSFETs, -26 would have been fine, and less expensive. So it's a little impressive that the mfr spent a little extra to get somewhat lower core loss.

The substantial heatsinks, hmmmm. I'd be impressed if each was one solid extrusion. But what they are is a thick piece of sheet metal, with extrusions bolted on with heatsink compound to fill air gaps. Even if properly applied (there is such a thing as too much thermal grease!), those composite heatsinks aren't going to be as good as a solid extrusion of similar fin pattern.

The 460W silkscreen on the PCB simply reflects when and for what the PCB was originally designed. As long as parts fit and aren't overheating, uprated parts could yield higher output power.

I can't see "standard" 4-digit datecodes on any of the 'lytics, but there's a 3-digit marking on the 85C (!) I/P 'lytics that suggests they were 4YO when that PSU was built. Not great, though I suppose they got away with it. Was this a case of one company buying out another? This (and the 3YO magnetics) has the feel of miscellaneous older parts from different PSUs being taken off the shelf and thrown together to create a higher power PSU than was originally made.

[momaka]

Re: Refurbishing an old Macron MPT-460

Yes! Good to see you are reviving this beast-of-a-Macron.

Those huge capacitors nearly made me piss my pants! This thing must have had 0.1 mV ripple on the outputs when it was new.

I'm guessing these are all GP caps. There is no way with the capctiances listed that they are also Low ESR.

Yes.
I too would venture to say those caps are GP. Or at best, they may be at the very far end of the low-ESR spec. Give than this is an old half-bridge design with relatively low switching frequency, the output capacitors actually don't even need to have that low of an ESR anyways.

And I absolutely love how much space Macron left for the diameter of these caps. This is how PSUs should be designed! Not with some skinny custom-order caps that are unobtanium once they pop and need replacing.

I guess it would be super-easy to find replacement caps for this PSU. Awesome!

The two magamp inductors might be connected in series. The wire might be too heavy to get enough turns for the right inductance. Or one core might not be able to handle the necessary flux density, so it is split by splitting the turns across two cores (I think that's how it might work). Or both.

I thought that might be the case too, but then I traced the coils' traces on the pictures that Pentium4 uploaded, and it seems to be that each coil goes to each side of the 3.3V rectifier, like Kaboom suggested above as the first possibility.

Out of curiosity, I checked the pictures of both my Macron MPT-301 and MPT-401, and sure enough, they are wired like that too.

So each coil goes to each leg on the 3.3V rail rectifier, and the 3.3V rail is sort of "dual" mag-amp -regulated. I'm guessing Macron did this for better regulation and power handling on the 3.3V rail.

I also noticed that the O/P inductor cores are Micrometals -52 material rather than the older -26 material. The inverter switches being transistors rather than MOSFETs, -26 would have been fine, and less expensive. So it's a little impressive that the mfr spent a little extra to get somewhat lower core loss.

Maybe that's part of the reason why these Macron PSUs were getting around 80% efficiency (or at least according to the review of the MPT-301 on hardwareinsights).

It's funny that they used a somewhat more modern (for Macron) 2-layer PCB, but the design is the same good ol' half-bridge found in just about every other Macron PSU. Even the 5VSB is the same 2-transistor design as on the older units (but unlike other PSUs with a 2-transistor 5VSB design, I find that Macron's is very stable and is unlikely to go over-voltage with age.)

So I guess it goes to show that a good design matters quite a bit. Now I have an Aspire/Apevia PSU that I got recently, and that thing is 180 degrees opposite compared to this Macron. But I'll post it another day. (It's going to be a long post, as there is just too much humorous nonsense and penny-pinching to describe.)

Even if properly applied (there is such a thing as too much thermal grease!), those composite heatsinks aren't going to be as good as a solid extrusion of similar fin pattern.

You're right.
But in this application, I don't think that will make much difference, if at all. Reason why is because the contact surface area between the heatsink parts is quite large, so the difference in temperature between the "core" heatsink and the bolted pieces will be minimal.

Viotek isn't fooling any (nerds) with their bullshit label. 700W continuous? No way! You can even see the "ATX460 MACRON POWER CO. LTD" on the top of the PCB and "MPT-460" on the bottom.

Technically speaking, Viotek might have actually rated this PSU correctly. After all, they specified 700 Watts continuous, but they didn't specify which numbering system they were using .

I'm sure they meant 700 Watts in the Octal numbering system though, because that comes out to... wait for it... 448 Watts in our well-acquainted Decimal numbering system.

See! They were right the whole time and you hastily accused them for nothing.

Also, "Customizable cable system" ?? It's not even semi-modular

Maybe Viotek included a pair of cheap wire cutters with the PSU originally, in order to "trim" the unnecessary wires. But it just so happened that you didn't get them, since you bought this PSU used.

I think Viotek is a great! company. You computer noobs just don't know anything.

Originally I was thinking about upping the 12V capability of this thing, but I was thinking I would actually use it in my dual PIII system I am working on because it has a really strong 5V rail, and all molex connectors. The main motherboard plugin is all 16AWG and is 24 pin (Which is actually what the PIII board has)

You can probably power two dual P3 systems at once with this PSU. I don't think there is any old 5V-heavy PC that will draw more than 140 Watts from the 3.3V and 5V rails together. Maybe if you added a Radeon 9800 Pro/XT, you could push that 3.3V/5V consumption a bit further, but I doubt this PSU will be breaking a sweat at all powering your PC.

Anyways, I can't wait to see the recap / finished job.

[goodpsusearch]

Re: Refurbishing an old Macron MPT-460

3 MOVs! Wow!

Cheap UPSs have just 1 MOV, sometimes smaller too!

This thing could do 500W continuous, probably 550W peak, depending on the frequency of the switching transistors.

Did you ever recap it?