- The LaST Upgrade -

PART 36 - EXXOS PSU 2018-2023

exxos 2018 - Last updated March 3, 2023


The latest PSU from exxos. The "red edition". Featuring highly efficient high power modern switch mode chips from one of the leading manufacturers, Texas Instruments.

PCBs optimised for maximum cooling and uses heavier weight copper than most general PCBs. This not only keeps resistance down which helps increase efficiency and reliability. It also greatly aids in the cooling of the switch mode controllers as the heat can be transferred away much more efficiently with a heavier weight copper. As the PCB is used as a heatsink, there is no need to bolt the regulators onto the metalwork like my previous design.

I have done away with all the electrolytic's apart from the large bulk capacitance. This means there are practically no components to fail or degrade over time. I now use ceramic capacitors which give much better performance and they are practically indestructible.

I have moved over to the much larger flat coil inductors. These is practically no voltage drop as the resistance is near zero and this increases efficiency and reduces heat output. While these inductors are a little overkill in this design, manufactures does not currently make a smaller version. However while these are rated as 30amp parts, the inductance only remains stable up until about 10amps. This of course still gives the design a good amount of headroom as the max output on the 5V rail is 6amps. These are certainly not "toy inductors" which just about every PSU I have seen uses. The inductance will remain ultra stable and not impact reliability or stability of the power supply regardless of load conditions.

Another good feature about this design is the operating frequency is a lot higher so we can have much tighter control over regulation and use smaller value inductors and capacitors. A huge advantage of this is that lower inductances is that the reaction time of current surges is reduced dramatically. Generally we are forced to use higher value inductors which inherently give better regulation figures, but they also become more sluggish to instantaneous power requirements (such as floppy motor powering up etc). So this power supply manages to get the best of both worlds.



The inductors are large to maintain a constant inductance value under higher amp loads. Even though they are "30 amp rated". You can clearly see above that the 10uH inductor will start to saturate after 10 amps. Unfortunatly there are no physically smaller flat wound coils at the time of manufacture. The flat wound coils give the best efficiency and regulation over older coil designs.


The four leftmost holes are 5Volts output.
The six middle holes are 0V.
The two rightmost holes are 12V output.

The 4 PSU mounting holes (each corner of the PCB) have been increased in diameter to help aid in fitting to various metal power supply bases.

As the power supply is more efficient it can run higher amperage easier and more efficiently . Typically 5V rail can deliver up to 6 Amp RMS, the 12V rail can deliver to 3 Amps RMS. However the total output VA is not exceed 40VA.

For example:

3A x 5V = 15va
2A x 12V = 24va
15 + 24 = 39va.

4A x 5V = 20va
1A x 12V = 14va
20 + 14 = 38va.

5V can deliver 6amps max. (30va)
12V can deliver 3amps max. (36va)
Total output must not exceed 40va.

Note: A typical STF/Falcon pulls about 1.8amps on the 5V rail and generally almost no load on the 12V rail.. For example 5V at 1.8amps is only 9va and we have 40va available. So there is plenty of headroom to power just about anything.

(PSU must be operated with minimum 1amp load else fuse may blow due to the peak voltage charging of the main reservoir capacitor)

The power supply has many modes of protection. It has short circuit and overcurrent protection where it will rapidly enter hiccup mode where there is a cycle by cycle current limit until the fault is resolved. The chips will also shut down upon seeing any under or over voltages on any of the inputs or outputs.

Secondly we have the classic crowbar protection circuit which monitors the 5V,12V rails and the main DC supply coming from the transformer typically 15VDC (20V max). Should either the 5V or 12V rail increase more than approximately 0.5V it is assumed there is some problem and the crowbar activates which blows the fuse and immediately cuts power to the whole board. So equipment attached to this power supply have maximum protection possible.

While this power supply does not have a ripple filter which is normally customary like a my previous design, this new power supply topology does not require a ripple filter and can achieve extremely good regulation like my previous design.

Test was done using a x10 probe with 4.5amps loading on the 5V rail.

Typically 2mV (0.002V) ripple which is as good as things are likely ever going to get. Noise figure is approximately 18mV (0.018V). To put it in perspective the original Atari power supplies were easily over 200mV (0.2V) noise. Note that this was under heavy load, so results here are considered "worst-case" figures.

So while my previous design was totally awesome, I have managed to make this power supply even more awesome by increasing efficiency and amperage output while also reducing the amount of electrolytic's used and increasing performance.

See this link for a more in-depth look into how this PSU was designed and the vast amount of research which went into it all..




2021 BATCH - Ver 3.20
Almost identical to previous batches, but this one has a adjustment for the 5V. It will be set to approximately 5.00 by default. It can generally be adjusted between around 4.6V - 5.8V . Please note the higher the voltage set the more risk of the overvoltage protection kicking in and blowing the fuse. There has been some other minor changes but nothing noteworthy.

Because of parts shortages I had to use a smaller value output capacitor than previous batches. So if you are putting more than about 0.5A load on the 12V rail (Which I doubt anyone would really do anyway these days) you'll need to add something like a quality Panasonic 680uF minimum capacitor on it. There is of course no harm in doing this by default anyway. Otherwise the voltage may be lower than expected.

I'm no longer selling individual 230 / 110V versions. The transformers are "dual primary" and need to be configured correctly for your mains voltage.




NOTE: Some power supplies were shipped with 2 primary wires temporarily twisted and taped together. This was only done for 240V testing before shipping and the tape should have been removed and wires disconnected before shipping. Please use the correct primary voltage for your country as noted in the links above. The primary wires should be soldered and heat shrinked or joined using bullet connector etc.



You should ignore what is written on the transformer as it says 0-115 0-115 and people wrongly assume its 115V transformer.These are "Dual Primary" transformers. Important thing is it lists 0-115 TWICE! Because it has 2 primary coils! So series configuration is 115V + 115V = 230V.

They can be configured to 110V or 230V as outlined in the links above. 230V will have brown and blue wires free as per UK standard colour coding for 230v mains wiring.


This comes up now and then so let me explain why the fuse can blow. The fuse only generally blows because of some fault condition relating to how it is being used, not because the PSU itself has a fault! These are protection circuits, there are there to keep things from exploding like your £1,000+ Falcon under huge mains surges etc! Of course most issues are generally "self inflicted" by people not reading the instructions :)


If you power up the PSU without load resistors or the ST connected, then its possible the over voltage protection will trigger and blow the fuse. This is because without a load, there is nothing to suck power out of the large capacitor on the board and its voltage rises beyond 20volts and blows the fuse. Why the hell does it do that ? Because it's a protection circuit! The regulator chips (or at least the 5V one) has a maximum input voltage of 20 volts. So if I allowed the voltage to rise to that level or above , it could blow the 5volt regulator.

The overvoltage protection could also trigger if the mains input voltage goes very high even under load. Such as 270VAC for the 230V model. At some point the protection circuit has to kick in right ? Where do you draw the line ? 250V ? 300V ? The circuit will blow the fuse if the voltage exceeds a safe level. This circuit is there to protect your PSU and your ST or Falcon from mains surges! So I recommend never operating without a 1amp minimum load. Lower loads may work but the PSU may enter "eco mode" or "light load efficiency mode" which may aggravate the issue.


Some people manage to run the 110V PSU on 230V which doubles the input voltage to the PSU. As such if the voltages happen to be a safe region the fuse may not blow but generally it will. If the transformer is being run on twice the input voltage, it will get VERY hot very fast and will start to smell or even smoke after a couple of minutes. The transformer doesn't generally come to harm if the runtime is just a couple minutes , just let it cool down for a hour before powering it up again.


Aside from high mains surges, generally this has been realised to be relating to the old ST PSU connector. While I mention to recycle the old ST power connector, I generally do not recommend this anymore. I myself had a issue of the fuse blowing (its actually the same as the OPERATING THE PSU WITH NO LOAD topic). As the bad connection causes a "no load" condition. The fuse blowing is a safety feature to protect the PSU and your machine.

A bad PSU connector will have a bad contact and likely cause your ST to intermittently crash along with intermittently blowing the PSU fuse. I had this issue with one ST connector, I tried cleaning it with IPA etc, it did not help. I replaced it with one of my new ST PSU connectors and the problem did not return.


The PSU has automatic overcurrent protection circuits so its unlikely you will blow the fuse by shorting it out. When a short condition is detected the PSU will enter a low current state until the short is removed.

If you short out the 12V output to the 5V output, then its likely the 5V over voltage protection circuit will kick in and blow the fuse. Of course we don't want to feed 12V into the 5V rail on a £1,000 falcon do we ? So again the fuse will blow to protect from such a fault.


The fuse is just a generic 20mm 4amp quick blow fuse.
DO NOT USE ANY OTHER TYPES. Placing a higher current fuse or a "anti-surge" fuse will not help with fuse blowing issue at all! If anything you may well just end up causing damage to the PSU. The fuse is there to protect your PSU and machine from various fault conditions. So don't screw with its values!


The PSU has a current limited "soft start" up to prevent ramping quickly to full power as the motherboard powers up. This is to help protect your machine if you power up with a shortcircuit etc. When the reset circuit caps go bad (which they all are these days) they must be renewed as outlined in the mandatory fixes. Once fixed the machine will power up normally.


Some people manage to run the 230V PSU on 110V. Generally the 5V will output fine, but the 12volts will likely be low around 8volts.


Generally these can take up to several seconds to trip. A fuse can blow in a quarter of a second in general. So unless there are some very high speed auto reset fuses out there, I will not be adding them to the PSU.


A fused mains plug with no greater than a 3amp fuse. If a fused mains lead isn't an option, then a 1-3amp antisurge inline fuse can be used in series with the hot / live mains wire going to the switch. Also note that the transformer already has 2 thermal fuses on the primary side.


Later PSU batches had a 5 volt adjustment trimmer fitted which can adjust the voltage higher or lower by about 0.5V. Generally this is because some machines are unstable on 5volts. So lowering to 4.8V like most PSUs operate at can help resolve such issues. Use with caution as to low or to high voltage could trigger under voltage or over voltage protection circuits. Generally 4.8v to 5.1v are the normal operating voltages.


There are multiple variations of the power supply metal framework where unfortunately I could never produce a "one size fits all" PCB to fit. Atari themselves had this problem and ended up drilling a new hole in some PCBs in one corner. I made the holes as large as possible to accommodate as much as possible because of these variations. Though another problem is the tolerances on the frameworks plus they are easily damaged when removed and fitting.It all adds up to things not aligning properly. The simple solution is to just simply bend the mounting lug on the metal framework inwards slightly so the PCB holes aligns correctly.


This was done during PSU testing. The primary has to be configured to your mains voltage. The tape is normally removed and wires cut after testing, but not always. Linking wires should be done in a suitable manner. For example, solder and heatshrink tubing or a bullet connector.


All power supplies up until the later ones with the 5v trimmer , were all built and tested by exxos himself. The later batches had to be part assembled in China due to lack of time and health issues. The familiar UK sticker is no longer placed on the later edition batches to reflect the change. This was accounced in THIS thread. The previous ones were built using THIS oven.


I regret to say after the current stocks (at the time of typing in 2023) will be the last power supplies I will produce. Mostly because they are just incredibly time-consuming and costly to create. I was previously looking at doing a revised design using a different chipset a couple years ago but parts shortages and also lack of time prevented that from getting off the ground.it is also questionable how long people will be wanting to purchase them for as well. The Atari market is only unfortunately declining.