The LaST Upgrade Part 33


Last updated April 27, 2017


As many know already, I am not a fan of the "Pico PSU" simply because when I purchased 4 of them a couple of years ago, from various sellers, none of them worked correctly or simply did not work at all. I am not saying I am against these PSU's, many people use them without problem. Though I have also heard stories over the past year that PICO's might have been the cause for some Falcons being killed off.

In light of such issues, and myself personally wanting to re-vist the PICO technology, I ordered 3 Pico PSU's as illustrated below. One looks to be a genuine PSU, the other 2 seem to be a cheaper clone. I brought these off evilbay (and no I am not linking evil bay links into my site pages!) and I already see that the clones do not even give any sort of amps rating per supply rail. Likely I will test in 1 amp increments until something dies a death.

Testing PSU's and reviewing parts can be time consuming. I am not going to spend huge amounts of time on these, but will do some basic tests. In order to explain what is "good and bad" I need to explain some basics about Inductors and Mosfets. I'm not going to explain everything in detail as it would make a huge article in itself. So I will try and keep things simple to understand. If you are not intetested in some small tech info, then skip the following text.


INDUCTORS - I have wrote about this somewhere else, probably on my own designed PSU. Though here is a crash course.

Inductors are used to store energy for switchmode PSU's. Good inductors will have low resistance, a good core material (A bad core can lose 60% of the energy) and must be able to hold its inductance value within a good tolerance.

RESISTANCE - Generally the larger the inductor, the larger the wire used and the lower the resistance. Higher resistance will cause heat build up and lose efficiency.

INDUCTANCE - Must maintain its value at rated current to approximately 10%. Difficult to find quality inductors.

CURRENT RATING - Must be a genuine figure. The problem is, a lot of manufactures either don't list the current, or they list the "fuse current" of the wire, not the amps the inductor is typically capable of. This is a huge trap for people who assume a 10uH 10amp inductor can delivery that, as normally they can't. The reality is, after just 1amp, the inductance can drop by 10% or more. In fact 10amps 10uH inductor, might just act like a straight piece of wire at 10amps or lose 90% of its value.

This boils down to saturation current, Which the only way to tell is to look at the graphs, assuming the manufactures list them. If they do not, then avoid that manufacture totally as they are likely going to be poor quality inductors.

As the switchmode controller is trying to deliver a voltage and maintain it under heavy loads, the controller will push more and more current into the inductor when more amps are needed. If the inductor starts to saturate, the Inductance will drop anywhere between 10% to 90%. If the inductor loses 50% of its value, then the switchmode controller will likely try and push 50% more current into it to maintain the output current. This then ends up with a huge drop in efficiency where things start to heat up to the point they melt. Of course its possible to melt a so called 10amp inductor with just 1 amp of current.

Overall, what I calls as a good inductors is one which maintains its value within 10% and has a genuine rated current. While I can test resistance to get a idea on quality, I am not going to test saturation currents as it would be very time consuming. Even so, I will be testing efficiency of each PSU, so if efficiency starts to drop down, then we can assume it has hit the saturation current of the inductors. Looking at the tiny sizes of them and considering they are switching 8amps on the 3.3V rail, it will be interesting to see how they hold up.


MOSFETS - Total nightmare to give even a simple summary, huge nightmare to find a mosfet suitable for a application as there are so many conflicting specs.

RDSon - On state resistance. Of course lower resistance is always better. But actually its not that simple. One of the PSU's quotes 4mR resistance (0.004ohms) which is great, but there are other issues at work here as well. Generally low resistance mosfets have a huge die. As a result, they are slower and need more peek amps to turn them on and off. Resistance is when the mosfet is fully turned on (this isn't the same as switching resistance which is a complex topic in itself)

CISS - Input capacitance. Large die is large input capacitance. Typically in the order of 3,000pF. (3nF) which is actually a lot. This capacitance has to be charged and discharged at a rapid rate. The faster you want the mosfets to switch, the harder it becomes. Don't forget a capacitor acts as a short circuit when at 0 volts. So the mosfet driver circuit needs to be able to deliver some serious peek amps to get things moving. Also this is a problem as several peek amps can be needed, this means the pico's PSU rail itself, needs to be stable and be able to deliver the peek amps also. This can result in a lot of noise being generated in the PSU's own circuit, which is bad.

There is also other capacitance's at work such as "reverse transfer capacitance". Where you also have a capacitor between the mosfet gate and the load you are switching. For example, if you are turning your mosfet on and off with 12 volts, and are switching a 50volt rail, then you will have a small capacitors (well several hundred pF probably) between the 12volt and 50volt parts internally in the mosfet. This means the mosfet driver also has to have good control over the mosfet gate, else it could see this 50volt rail and this could kill the driver circuit. Of course all this is a very complex topic in itself.

Td(on/off) Total delay on and off. Large mosfet dies take longer to turn on and off. The delays for turning and mixed in with the actual rise and fall times of the mosfet can really add up. Generally Mosfets can switch anywhere between a few nanoseconds to 100ns or more. The faster they can switch the better.

ID ( AMPS) - The more amps you can switch the better. Well not really. Again large dies have lower resistance, higher current ratings, but take longer to turn on and off. Great if you don't need speed. Though if your are using them in a switchmode PSU, then generally you need speed, and a large die isn't going to work.

Confused ? it gets better. What if we want a high amps mofets to switch fast ? Well, basically you can't. Higher amps mosfets need a larger die and the larger the die, the slower they switch. You can't have it both ways. If we want low resistance then we need a larger die again. A smaller die will switch faster but will have a larger resistance, which we don't want either. To add more confusion, there isn't only on state resistance (which is the resistance when the mosfet is fully turned on) but also you get switching losses as well. The time it takes the mosfet to turn on or off is again depending on the die size. You can have a mosfet which can switch 100amps and have near zero on state resistance, but may take 500ns to turn on and off. This is a huge loss (called switching losses) where the mosfet will heat up a fair amount and will get hotter the longer it takes to turn on or off. So while higher amps and lower resistance is what at first appears a good thing, it depends on the application. If you are needing to switch 5amps, you don't use a 100amps mosfet. That would be a very dumb thing to do.

Now take into account the switching speed of the PSU controller. It can vary from a few KHz to even low end MHz range. If the total switching delays of the mosfet are 50ns ON and 50ns off (which is pretty typical) and the actual ON time is 100ns, then the total cycle length is 50+50+100 = 200ns. In fact This means the mosfet will spend 50% of its time in switching losses. Which is bad. To give a idea, assume you lose 50% efficiency of the mosfet (again I am not going into this in depth) and spend 50% of time fully turn on. You could have a zero resistance mosfet (RDSon) and the mosfet can catch fire due to the heat build up in switching losses.

ironically , I have seen 200mR mosfets (0.200ohms) switch a few amps and remain stone cold. Their switching time was below 5ns so the switching losses were very small, even though the on resistance is very high. Such mosfets were running around 200KHz so they spend a lot of time turning on and off, so fast switching is a must.

You could have a higher amps mosfet with zero resistance, but forget trying to turn it on and off fast. Try to drive such a device in the upper KHz ranges and the mosfet can spend 100% of its time in switching losses and in that case, you should probably find a easier hobby :)

Mosfets are a conflicting nightmare of chaos. Its enough to give any professional design engineer a headache, nevermind people just venturing out in mosfet technology. I will look at the spec sheets for the mosfets used to see what ratings they are, but without knowing exactly how the circuit is working, it probably wouldn't be fair to say any mosfet is good or bad as they are very specific on the application. Of course if we have a 1amp mosfet switching a 10amp rail, then I will probably frown upon it somewhat :)

Overall, I just wanted to make the point that just because a manufacture claims lower resistance mosfets, or high current mosfets, not to assume that is the good thing to look for as it is not so simple.


CAPACITORS - We all know what they are. Generally tantalum, ceramic, electrolytic. Problem is, SMT capacitors are generally small, and have poor ESR ratings. Larger capacitors have higher ESR ratings. What is ESR ? Basically to been seen as a resistance in series with a capacitor. For example, if you had a capacitor in a supply rail, if you run that capacitor via a 10ohm resistance, then you would think, why bother having the capacitor there in the first place ? You would be right to question exactly that!

Ceramic capacitors are my type of capacitor, You can't really go wrong with them. Though once you get into the 100's of uF, then you are pretty much forced to look at electrolytic types. Though that in itself isn't exactly true. Generally larger uF values are used to improve stability. Though as a side effect of higher uF, is physically larger capacitors, which have lower resistance. So its not fair to simply say more uF is better. For example, you could have a 1,000uF electrolytic capacitor which may give 10mR resistance. Or a 100uF ceramic which will have near zero resistance. Again, depending on application, I prefer to use ceramic where ever possible. Larger values and voltages of ceramics are on the market more lately than ever. Why bother using a electrolytic when there are ceramics available. Well, maybe cost. But if a ceramic is available in the right voltage and value, then use it! That is one thing of "quality" that I will be keeping a eye on.


AND THE REST - There as so many aspects of a PSU that it would be almost impossible to go though it all. PCB copper thickness for example. I myself use thicker copper as its improves efficiency and regulation.

There are many types of switchmode designs also. We have not mentioned P&N type mosfets. Generally P-channel mosfets sux, nobody uses them (well ok some do) but I personally just like using N-channel only. There isn't any need for P-channel unless your looking for a cheap circuit design. In my view, if you want to use 2 mosfet switching design, then use a driver capable of driving 2 N-channel mosfets!! While there are always "pro's and con's" for any design, I'm not going to go into it all here. There are many guides on the Internet about switching technologies already.

Tests are only for guidance only. The 3.3V rail was tested using a rheostat resistor where I cannot set 100% correctly.Unfortunatly I don't have the equipment to accurately set resistance values. Also I made the assumption that the 3.3V rail was actually 3.3V. So tests will have some small percent error. I think overall tests give a really good indication of operation anyway.

I am only quickly testing each PSU out. Testing methods are not ideal, though to test things out better would need a great deal of time and better testing equipment. Both things I do not have! My rheostat is a variable 20ohm wire wound rated at 200watts. Changing its value is a nightmare of trial and error. Its value was measured with a proper low resistance meter and pretty close to the values I needed. Though the 3.3V amps testing are not totally accurate but as close as I could get. Overall, the "error" factor in testing is across all PSU's and I was more interested in comparison of these PSU's rather than rock solid diagnostics of each one. Even so, I think my results are pretty close and good enough for the series of tests I wanted to do.

Pico PSU. M3-ATX 125W

So what we have is a claim of better than 94% efficiency with 50% load on all rails. 5V rail we have 6 Amps and 6 Amps also on the 3.3V rail. Not to shabby!

125 Watt total output power.

5V X 6 = 30 Watts
3.3V X 6 = 19.8 Watts
5VUSB X 1.5 = 7.5 Watts
-12V X 0.15 = 1.8 Watts
12V X 4 = 48 Watts

TOTAL = 107.1 Watts.

As the label quotes 125watts, then it would appear to add up at least. This could equate to about 15% efficiency loss at full load (85% efficiency). Of course I used 4amps for the 12V, Though it does appear to get a bit more in depth with ratings which I will not cover.

I can't find much information about this module other than ..

The M3-ATX is an intelligent, high power, automotive DC-DC ATX power supply designed for car pc applications. Intelligent auto on/off controller makes this power supply a necessity for automotive use. The M3-ATX operates on a wide-range of input voltages (6 to 24V) providing safe and consistent power to any application. The M3-ATX is even design to safely withstand engine cranks. The M3-ATX is based on a similar design as the M2-ATX but has been dramatically reduced in size resulting in a plug-in style unit.

By using Patent Pending HyperWatt[TM] technologies, the M3-ATX packs an impressive amount of power relative with its very small footprint. The M3-ATX has several key advantages over traditional power supplies:

As to what "HyperWatt" is I have no idea. I've seen such terms used by manufactures of mosfets. They all invent their own "cool names" for stuff.

The patent number 7,539,023, seems to have been filed in 2005!



Arrived in a small box, anti-static bag with some bubble wrap.

The board looks like a solid construction. Its actually on 2 PCB's with parts stuffed on both sides. I'm not goint break apart to see whats used on the rear of the boards (well maybe later ;) as I need to test the thing before I kill it :)

One thing which caught my eye right away is they appear to have used flat wound wire inductors! Flat wound are something I really like. They offer the lowest resistances and generally indicate good quality. It also makes me think right away that someone has done some proper research into inductors to go ahead and use them, as they are not generally that common on the market yet. Looking how there isn't any part numbers on the larger inductors, I can only assume they are custom manufactured. It would be interesting to do some saturation tests on them to see what the core material is like, though it is not something I have time to do currently. I will assume if they are custom run and they used flat wound wire, that they are not going to bodge it with a cheap core material either.

There is a lot of SMT caps on there, so gets a thumbs up from me there. Though a bit of a let down with the 2 large SMT electrolytic's on there. Looking at the value , they are 47uF 25V. OK, to be fair, its pushing it a bit for ceramics, and there is a lot of ceramic on there already anyway. I would expect in time the manufacture may swap them for larger SMT ceramics, though might be down to costs, or simply PCB space. So not saying its a bad thing.

Overall, my first impressions are that this PSU has had some good thought gone into it. From what I see overall, I couldn't really grumble about anything.



My first look is at the inductors. Mostly value and resistiance. At this point I have not taken them out of the circuit to test, So results might be a little off, but we shall see if we can get some kind of hints on it. I may take them off the PCB and do more tests on them another time.

The large inducors grey inductors seem to measure about 7-9mR at 6uH, and a smaller green one about 3mR. I wasn't able to get a inductance reading on it (came out at zero). It could be a current shunt resistor, but needs looking into more. I may lift it out to to test another time. Some of the coils I have seen from some manufactures are down to 2mR, but they are much larger size. So sub 10mR values on these tiny inductors I think its pretty good going.

The mosfets I can see on top are Fairchild FDS6690 and FDS6680. Overall these look like good mosfets. Switching times look resonable along with on state resistance and capacitances. The datasheets say 1998 and 2005, So these mosfets have been out of production for some time I assume now. I would have liked to have seen something more recent. Though Fairchild make quality mosfets and I can't really complain about the specs used at a glance.

I can't see the rear of the PCB's without some unsoldering, so I may come back to that another time. Though so far, it looks like some thought has gone into the parts used. So at this point I will assume similar quality of times on the rear also.



I will start using a simple resistive load, its simple and will give a indication of the PSU running at its best. I will power up using a bench PSU, again it will show the PSU at its best. Its not a real world test, but I need a stable and regulated power supply and constant load resistance in order to test the regulation & efficiency correctly. I will mostly be looking at the 5V and 3.3V rails as these are most common used on the Falcon and Atari ST. 12V regulation can be important as its mostly used to power the video circuits. Though these days, there 12V rail isn't put under any sort of high amps load.

12V gets about 1.2 amp load (14.4watts) , and 5V about 2.18 amps load (10.9watts) . 3.3V is not connected (or any other voltages). I am powering it from a regulated 12V supply and the PSU is pulling 2.45amps (29.4 watts.). So total output wattage is 25.3watts. So a loss of 4.1watts which is about 86% efficiency.

So next up I will try higher input voltages to see how it effects efficiency.

12V @ 2.45A = 29.4W
14V @ 2.10A = 29.4W
16V @ 1.83A = 29.28W
18V @ 1.64A = 29.52W
20V @ 1.48A = 29.6W
22V @ 1.35A = 29.7W
24V @ 1.25A =30.0W

So 24V input has the worst efficiency in this test at 84.33%, while the best efficiency was at 16V at 86.41%.

Now lets throw a 1amp load on the 3.3V rail (3.3watt) and see what happens :) So now our total output load is 28.6watts.

12V @ 2.77A = 33.24W
14V @ 2.36A = 33.04W
16V @ 2.07A = 33.12W
18V @ 1.83A = 32.94W
20V @ 1.66A = 33.2W
22V @ 1.52A = 33.44W
24V @ 1.40A = 33.6W

So worst efficiency is again 24V at 85.12% and best is 18V at 86.82%.

Next up, 2amps load on the 3.3V rail (6.6watts). Total output load 31.9watts.

12V @ 3.13A = 37.56W
14V @ 2.67A = 37.38W
16V @ 2.31A = 36.96W
18V @ 2.06A = 37.08W
20V @ 1.86A = 37.20W
22V @ 1.70A = 37.40W
24V @ 1.57A = 37.68W

So worst efficiency is again 24V at 84.66% and best is now at 16V at 86.31%.

Next up, 4amps load on the 3.3V rail. Total output load 38.5watts.

12V @ 3.68A = 44.16W
14V @ 3.34A = 46.76W
16V @ 2.89A = 46.24W
18V @ 2.56A = 46.08W
20V @ 2.30A = 46.00W
22V @ 2.10A = 46.20W
24V @ 1.95A = 46.80W

So worst again is 24V at 82.26% efficiency and best is now at 12V at 87.18watts.

Next up, 6amps load on the 3.3V rail. Total output load 45.1watts.

12V @ 4.45A = 53.40W (PSU was tripping on/off)
14V @ 3.33A = 46.62W (PSU was tripping on/off)
16V @ 3.25A = 52.00W (PSU was tripping on/off)
18V @ 2.91A = 52.38W (PSU was tripping on/off)
20V @ 2.65A = 53.00W (PSU was tripping on/off)
22V @ 2.21A = 48.62W (PSU was tripping on/off)
24V @ 1.86A = 44.64W (PSU was tripping on/off)

Its clear that the PSU has maxed out at 6amps. After just a few seconds run the PSU was turning on/off every few seconds. Inductors seemed pretty hot during this test also. I would assume this PSU would start to have issues after about 5amps. Though long term running could de-rate that to 4amps.

Worst efficiency was at 12V at 84.46% and best is now oddly at 24V at 98.98% efficiency , erm WTF ?! OK, so 3.3V is at 2.64Volts, so clearly things has gone totally screwy now and the PSU is unable to sustain a 6amp load at 3.3V. I'm not sure what to conclude here. I am going to assume the inductors are saturating somewhere around 3amps. Even so, the PSU isn't happy at lower current ratings using higher input voltages either. Likely the inductor is simply saturating at somewhere around 35watts.

In anycase, the 3.3V fell short of rated spec and as the 3.3V rail is dropping in voltage, that my percent calculations are simply going out of the window. So I would suggest the humble reader to ignore the 6amp tests as clearly there are some issues during those tests.




Regulation testing on the 5V and 12V rail are with a fixed load resistors as stated previously. I'm not going to test every combination of amps Vs regulation as it would be time consuming and I don't think its going to be that useful as ST's and Falcons typically use around 2amps on the 5V rail anyway. The 3.3V rail I will test a little more as the 060 CPU uses this voltage so regulation on this rail is important considering the prices of the CT60 and 060 CPU's!

12V rail - 12V input
12V rail - 16V input
12V rail - 20V input
12V rail - 24V input

12V regulation seems to improve with higher input voltages. Though the spikes don't seem to totally disappear. Regulation seems overall stable at 12.2volts and the max spike I saw was to 13.4volts , Which is a spike of 1.3volts!


5V rail - 12V input
5V rail - 16V input
5V rail - 20V input
5V rail - 24V input

5V regulation seems to overall get better with higher input voltages. Voltage seems stable around 4.92V. Spikes I see up to 5.38V, a jump of 0.46V.


3V rail - 12V input
3V rail - 16V input
3V rail - 20V input
3V rail - 24V input

3V rail I used 4amps as the test load as this was probably near the max load the PSU can take. 3.3V worked out at 3.46V or there abouts. Regulation doesn't really seem to depend on input voltage from what I can see. The spikes seem less than the other rails and I saw a spike at 3.92V, a jump of 0.46V.



I can't really fault the overall design of the PSU, clearly a lot of work and thought has gone into it. While I have not yet looked on the rear of the board, I feel there probably isn't any need other than to see what controller it used, or what other mosfets are used. The coils do get a little warm, though they ran a lot cooler than I had expected.

A lot more tests could be done though its taken all day just to test this one PSU out (albeit very quickly at that!) There is some oscillations going on which are more noticeable on the spikes. I have not looked into that at all. Though really it shouldn't be there. Looking at the frequency of the spikes (well oscillations) they seem to be in the 50MHz+ range. So likely are a self resonance of the inductors themselves. Again, its a side effect of how switchmodes work. Though really there should be something in place to prevent the coils from oscillating in the first place. Though that is just my opinion.

The regulation at the spikes (at least on the 3.3V rail) sees a p-p voltage of about 400mV (about 0.4volts). So we are talking a best case regulation "error" of around 10-15%. Excluding the spikes, there is about 40mV PP regulation which is pretty good what I would expect from a modern PSU design. However, voltage bouncing up and down by 0.2volts seems a litlte on the high side. The original ST PSU's would around that mark and while it was pretty good in the 80s, its not really good to see such regulation in todays switching power supplies.

The voltage spikes are in part of how switchmodes work. Normally the spikes get worse as the PSU ages and capacitors start to fail. Though I do not plan on any long term testing, I feel the spikes could become a problem long term. Considering I used a regulated PSU to power the module, plus a fixed dummy load resistor, my tests would be (or should be) best case figures. Powering from another switchmode which will likely have spikes in itself is only going to compound the problem. As to what these spikes will "grow" to over time worries me a little. Like on the original Falcon PSU's, I have seen over 1 volt spikes on aged PSU's. This PSU is already spiking around 0.4volts.

The mosfets which I mentioned before do not seem to get warm. Though I don't know what rails those are powering. As they are near the 2 large coils, then assume they will be likely for the 5V and 3V rails. The coils do get slightly warm after a few moments of use. Though They could be a lot worse. Clearly some good thought into keep the resistance down with the mosfets and inductors.

So would I recommend this PSU ? Well, I am not going to say either way. Simply because I have not tested a desktop PSU to compare this pico psu with. So I feel it would be a little unfair to say it is good or bad in comparison to a desktop ATX.

I will say that it will probably hard to beat in terms of quality and parts used. Though I do feel it needs a little work in solving those oscillations, and also it does fall short of its rated current, at least on the 3.3V rail. Its a small PSU and only so much can be done in terms of regulation. I do also wonder that considering I use a fixed load resistors with a regulated input supply, that if this is the PSU's "best" results, then under a real world test, they are likely going to be worse to the point it starts to worry me somewhat. Its not great, but at the end of the day, it is what it is.




MINI ITX Power Supply 160W Module

A claim of 92% efficiency, no clue under what load. 3.3V limited to 3 Amps, 5V also limited to 3Amps.

160 Watt total output power.

12V X 3 = 36 Watts
5V X 3 =15 Watts
-12V X 0.1 = 1.2 Watts
3.3V X 3 = 9.9 Watts
5VUSB X 1 = 5 Watts

TOTAL = 67.1 Watts.

So where did the other 93watts go ?! From those specs it would work out at more like 60% efficiency.

I have not yet been able to find any more information about this module :-(



Arrived in a jiffy bag in a antistatic bag.

This one has a socket to directly connect a DC jack to it. Fantastic if you need to do that of course. But annoying for me as I can't easily connect to my bench PSU. The PSU cables do not include a floppy power connector which kinda sux.

To my amazement this board also has flat wound inductors! These are stamped at 100, so assume 100uH there.

There is just one single PCB, parts stuffed both sides. 4 electrolytics on the rear, and stuffed full of small SMT caps and resistors.



4438CGM APOWER MOSFET - Low resistance, very fast switching times. N-channel

6679GM APOWER MOSFET - Low resistance,reasonable switching speeds, high input capacitance, P-channel.

92U03GM APOWER MOSFET - low resistance (4mR) 20amps - fast switching considering fairly high input capacitance.

ISL6440IAZ INTERSIL PWM controller - Surprised about a branded PWM chip here! Lots of protection features built in, out of phase operation (reduces supply rail pumping) . A quick look at the datasheet and it seems to use all N-chan mosfets. So not sure why there are P-chans in the mix. But could be not relating to the PWM chip itself.

There is also a R47 part, assume its the current shut resistor. Likely 0.47ohms. Again appears to be flat wound, so thumbs up there.

The 2 large coils measure 24mR and 13.40uH. So likely it means 10uH and probably around 20% tolerance. 24mR seems a little high for a flat wound coil though. As said before, these are measured in circuit, so may not be totally accurate. I will try to find time to take them out to test properly some other time.



I will start using a simple resistive load, its simple and will give a indication of the PSU running at its best. I will power up using a bench PSU, again it will show the PSU at its best. Its not a real world test, but I need a stable and regulated power supply and constant load resistance in order to test the regulation & efficiency correctly. I will mostly be looking at the 5V and 3.3V rails as these are most common used on the Falcon and Atari ST. 12V regulation can be important as its mostly used to power the video circuits. Though these days, there 12V rail isn't put under any sort of high amps load.

12V gets about 1.2 amp load (14.4watts) , and 5V about 2.18 amps load (10.9watts) . 3.3V is not connected (or any other voltages). I am powering it from a regulated 12V supply and the PSU is pulling 2.29amps (27.48 watts.). So total output wattage is 25.3watts. So a loss of 2.18watts which is about 92.47% efficiency. I should note here the 12V output rail is actually at 11.40 volts. So it is a little low.

This PSU does not support anything other than 12V input so I am unable to test any other input voltages relating to efficiency.

Next up is loading the 3.3V rail with 1amp, 2amps, 3 amps. Note that 3 amps is listed as the max rating here, but I tested 4 & 5amps anyway :)

3.3V 1amp = 12V 2.63A input ( 31.56W), 90.62% efficiency. (28.6watts total output)
3.3V 2amp = 12V 2.96A input ( 35.52W), 89.81% efficiency. (31.9watts total output )
3.3V 3amp = 12V 3.44A input ( 41.28W), 85.27% efficiency. (35.2watts total output )
3.3V 4amp = 12V 3.94A input ( 47.28W), 81.43% efficiency. (38.5watts total output )
3.3V 5amp = 12V 4.00A input ( 48.00W), 87.08% efficiency .(41.8watts total output ) (3.3V dropped a little)

So 5amps the 3.3V rail started to drop out. Efficiency figures will not be correct there. Its probably more realistic to assume somewhere around 76% efficiency.

We can see the PSU seems to maintain good efficiency (about 90%) for 1 and 2 amps. Then drops down sharply to 85% under 3 amps load. I assume 3amps is the point the inductors saturate. 4amps still seems to hold up, though likely 3 amps is the top current rating as listed in the specs.


Same load as before on the 12V and 5V rails. The 3.3V rail was loaded with the full 3amps. As this PSU only supports 12V input then there isn't much testing to be done. I am actually assuming here that the 12V rail is simply directly connected to the 12V input rail making a simpler circuit design and the reason for improved efficiency over the previous PSU.

3V rail
3V rail close up
5V rail
12V rail

According to my voltage meter the 3.3V is very stable at 3.33V. Though on my scope it shows that there is a 1 volt p-p variation in regulation. Like the previous PSU, there is 50MHz+ ringing but there doesn't seem to be those huge voltage spikes either. The 5V rail is a little better, but still suffers from about 0.8V PP regulation errors. The 12V, even though looks pretty bad, is likely directly connected to the 12V input from my bench PSU. There just seems to be ground bounce on that rail causing the noise. There doesn't appear to be any actual switching oscillations on there.



The overall build quality doesn't look to bad for a cheap PSU. Some really good inductors and mosfets and the use of a branded PWM controller was a nice surprise. Overall the parts used seem really good.

When we get down to regulation, its getting in the area of around 30% error. Probably the PSU needs much more ceramics on the board to keep it more stable. The 3.3V rail reaching 4.28V, well, that isn't very good. The previous PSU under test still varied 0.4V, which is still a lot , but this PSU is easily twice as bad.

The efficiency is pretty good with this PSU. If you do not need a wide input voltage range and can use a 12V fixed input voltage, then no reason why not to use it. The 12V rail in and out seems to be a direct link. So 12V regulation will depend on the regulation on the 12V power brick used. Again, nothing really wrong with that either. I actually think thats a better idea, than having a more complex circuit to generate 12V when you can simply use a 12V input to start with.

So would I recommend it ? Well, that is a tough one. If it had some bulk capacitance on the output to keep it more in check with its regulation, then it could be a pretty decent PSU. Though I do not plan in looking into "fixing" these things. As my tests are "out of the box" and using fixed load resistors and a regulated DC input, this PSU will likely do worse in real world testing. So while I like a lot about this PSU, I would have to say its one to avoid.



160 Watt total output power . Claims 92% efficiency.

5V X 8 = 40 Watts
5VUSB X 1.5 = 7.5 Watts
3.3V X 8 = 26.4 Watts
-12V X 0.05 = 0.6 Watts
12V X 9 = 108 Watts

TOTAL = 182.5 Watts.

Not really possible to guess efficiency from the output ratings there.

What did catch my eye was "Z2-ATX-200 in the +12 V branch circuit uses two parallel MOSFET, the equivalent resistance is only 2.5 milliohms, greatly increases +12 V slip with a load capacity and reduced heat.". Great news for low resistance mosfets, however assuming the 12V itself is a switching regulation, then the largest loss would actually be in the inductors themselves. Also the problem is with low resistance mosfets, is they switch very slowly which actually generates more heat in switching losses. So its one thing to be careful of when things like "ultra low RDSON" mosfets are quoted. Lower is better yes, but its not that simple. Considering they claim to be using 2 mosfets for even lower resistance, ouch! I have seen 100mR mosfets run cooler than 2mR mosfets. I won't go into all that here though as its a huge complex topic.

"High quality components: The United States AVX tantalum capacitors, U.S. AOSMD FET, Taiwan, Japan, TDK and YAGEO and other manufacturers control chip, FET capacitors and resistors". Fair enough. Though I am not a fan of tantalum capacitors personally.

This caught my eye.. "As shown in the four-layer PCB design, using immersion gold process, the outer layer copper thickness 2OZ, the inner layer is 1OZ, compared to the previous product PCB Thickness doubled. Multilayer PCB + Immersion Gold +2 OZ copper, greatly increasing the circuit's output current capability, and extend the life of the circuit." I actually use 2oz copper on my PSU designs to keep the resistance down. So there does seem to be some attention to detail there!

One thing also I noticed is a bank of SMT ceramics on each side of the board. Aside from the tantalum, I do not see any electolytics on the board at all! What I assume is they mashed the value up (maybe 100uF area) in SMT ceramics instead. Now that is something I really do like. Ceramics will have much better performance than electolytics.

Its difficult to tell what inductors are used. They likely are cheap ones, but once thing I can do when the board arrives is test its resistance. Generally good inductors will have low resistance and be a indication that if the resistance is low, then we also hope the inductance value is constant on varying loads. Testing inductors in relation to its saturation currents can be done, though it is time consuming so I will unlikely delve into that aspect. Though we can probably get a indication on efficiency under load anyway. As once the inductor starts to saturate, the efficiency will often sharply drop off and the thing probably melts :)




Arrived in a jiffy bag in normal bubble wrap. So be suprised if it isn't DOA.

This board has a interesting amount of SMT caps across the top. Assume they are used in place of electrolytics, so gets my vote! There is 2 tantalum on there as well though. There doesn't really seem to be much on the board overall. It still has huge amount of stuff, just seems to be a bit more "spacey" than other designs.

The PSU lead comes with a floppy power connector! yay! There is a huge amount of small via's on some large tracks, so we like things like that! Can't really tell what the inductors are, but will look into that more next section. Overall it just looks pretty sexy!



4856A - VISHAY MOSFET - Fast switching, low resistance, high current N-channel.

MDS3653 MagnaChip (who ?!) - Fast switching, low resistance, high current P-channel.

2 packaged marked 4R7 which measure about 7.6uH @18mR. One marked R36 which mesures 0.1uH @ 3mR.

The PWM controller I can't figure out, seems to be a ST part, but doesn't seem to return any results. So if anyone finds the datasheet then let me know!

The input power has the normal DC jack socket on some pretty chuncky looking cables. They even used the black braiding stuff. So some good attention to detail there. It seems a shame to cut off it to connect it to my bench PSU :(




Same load resistors on the 12V and 5V rails as before. I start off with no load on the 3.3V rail and with a 12V input it pulls 2.31amps (27.72watts) . The output load wattage is 25.3watts as before. This pushes a efficiency of 91.27%.

Next up to load the 3.3V rail. It is rated at 8amps.

3.3V 1amp = 12V 2.63A input ( 31.56W), 90.62% efficiency. (28.6watts total output)
3.3V 2amp = 12V 2.94A input ( 35.28W), 90.42% efficiency. (31.9watts total output )
3.3V 3amp = 12V 3.26A input ( 39.12W), 89.98% efficiency. (35.2watts total output )
3.3V 4amp = 12V 3.47A input ( 41.64W), 92.46% efficiency. (38.5watts total output )
3.3V 6amp = 12V 4.06A input ( 48.72W), 92.57% efficiency. (45.1watts total output )
3.3V 8amp = 12V 4.64A input ( 55.68W), 92.85% efficiency .(51.7watts total output ) (PSU tripping out)

3.03V 7.4amp = 12V 4.64A input ( 55.68W), 85.65% efficiency. (47.69watts total output)

It was difficult to measure the load resistor accurately as I am going down to 0.4ohms! (if anyone has a proper benchtop load resistor then please donate it ;)

I think those figures are pretty good. Odd the 2amps and below seem to have worse efficiency, though I have seen switchmodes behave like that. Just normally its around 100mA type loads, not 2 amps. Nothing wrong with those figures of course. Then we move to 4amps and above and efficiency seems to get better. Though on closer inspecting the 3.3V rail starts to drop around a 4amps load. So this would explain the sudden jump in efficiency. So the 4,6,8amps tests are not accurate. The 3.3V rail actually dropped to 3.03volts under 8amps load. So clearly it is struggling to keep up there.

I did a adjusted retest as highlighted in bold, and we can see the efficiency has dropped by about 5%. I would guess that the PSU starts to struggle after about 4amps with voltage regulation.



Same loads on the 12V and 5V rail as before. I decided to load the 3.3V rail with 3amps so it will tally with the previous PSU tests.

3V rail
3V rail close up
5V rail
12V rail

The 3V rail seems to have a lot less noise than the other 2 PSU's I have tested. At first glance it seems to have about 1V p-p noise, which is pretty bad. Though when looking at this closer, It actually is lacking in a output ripple filter which is causing the bulk of the noise. The spikes again are 50MHz+, so likely ringing in the inductors or bad grounding somewhere. In anycase, its probably not so fair to sat the spikes are bad, as if it actually had a output ripple filter, then likely those spikes would be greatly reduced along with the ripple.

The 5V rail is pretty good, Not really much to say other than it needs a ripple filter! The 12V rail, likely again, just a direct dump from the 12V input. Though I do see output ripple on the 12V output.



Overall I really like how this module performs. It uses some decent parts and does the job. It does not really push the 8amps as it quotes, from my tests it starts to struggle a little after about 3amps load on the 3.3V rail. Though it does maintain good regulation.

So would I recommend it ? Well , regulation wise, I think it beats the previous PSU's. The fact its got a LOT of ceramic caps on there makes me lean towards long term reliability. What really lets this PSU down is the lack of output ripple filtering. Though to be fair, there simply isn't any room on the PCB for anything else. It would have to move over to another PCB and have the parts on there. If it did have a proper output filter, then I think the figures would be pretty damn good and it would be a clear winner.



It looks like each PSU has its good and bad points. Overall, they all seem to use good parts, which I was surprised about. Though each PSU seems to have some issue with one thing or another.

People are going to ask me which one I recommend, Its a tough one since its difficult to compare as they all have good and bad points. I think overall the Z2-ATX-200 is the winner in my view. It doesn't mess with the 12V rail, a simple pass-though from the supply brick I think is good enough.

Though each PSU seems to suffer with spikes. That is the main worry for me. Those spikes are what can get worse over time and kill your computer. Again, the Z2-ATX-200 has a lot of SMT caps, so I would be inclined to side with that for long term reliability. Though it is only a guess and I do not plan to do such tests with any PSU's.