PSU -12V rail fried, easy fix?

[LittleAnimal255]

Hi all,

Just about to start making an ATX PSU ready for my new Blizzard stuff and noticed that the -12V pin of my bog standard 25W amiga supply was bouncing around, I've put it on my poor man's oscilloscope as per the pic and it looks to me like maybe just one component is fried, such as a reservoir capacitor or something.

Long shot but could anyone with electronics experience give me any hints as to what components are involved as I'd like to test the sound and serial port on the board before I start my tower project, which is apparently all the -12V rail is for.

[Daedalus]

There are a few different versions of the Amiga PSU, even with the same ratings as other units. So they're less well documented as the machines themselves. To be honest though, rather than repairing Amiga supplies, it's usually simpler to replace the guts with a small PSU module.

I would check though that there's not a short on the -12V rail of the board. A simple resistance test should be able to tell that.

If you're embarking on a tower project, perhaps just test it with the PSU that you're going to use in the tower.

[LittleAnimal255]

I haven't found an ATX PSU which stays on yet, I think all 3 I have are failing due to an uneven load as the Amiga LEDs illuminate but then nothing, even with no extra Amiga hardware attached, and tried with a dummy load across one of the molexes, but still nowt. I'm going to try the newest modular one with a whole motherboard as the dummy load later on, now that my connectors have arrived, but indeed I'd like to do a guts replacement project once I've found a working one.

[Daedalus]

Yep, powering Amigas with modern PSUs can be tricky as they use so little power, and on the wrong rails too. A modern ATX PSU will expect a heavy load on the 12V rails. If it has two 12V rails (see the specs for the PSU), you'll need a dummy load on both of them. Also, check the Molex for a sensor wire - some use the 3.3V rail for voltage sensing, and it will have an extra, thin wire into one of the connectors along with the main conductor. If this is the case, you'll need a dummy load there too.

For such low loads, it might still be worth considering a module like a Meanwell, maybe fitted inside an old AT or ATX PSU case for neatness.

[FOL]

Originally Posted by Daedalus

Yep, powering Amigas with modern PSUs can be tricky as they use so little power, and on the wrong rails too. A modern ATX PSU will expect a heavy load on the 12V rails. If it has two 12V rails (see the specs for the PSU), you'll need a dummy load on both of them. Also, check the Molex for a sensor wire - some use the 3.3V rail for voltage sensing, and it will have an extra, thin wire into one of the connectors along with the main conductor. If this is the case, you'll need a dummy load there too.

For such low loads, it might still be worth considering a module like a Meanwell, maybe fitted inside an old AT or ATX PSU case for neatness.

Yep, 3.3v sensing is a pain. Ive had to wire it up on new PSU's, else they switch off after a second of bein on.

[LittleAnimal255]

Yeah that's exactly what my Hyper-V modular one does. Added a motherboard to it so it stays on, but it won't boot the Amiga from a Molex (Just with GND,+12 and +5) so I'm considering biting the bullet, especially after reading about noisy graphics with Pico PSUs and dying clones and such, and going for this https://www.ebay.co.uk/itm/New-power...IAAOSwVEdbLOMq

[aeberbach]

This looks like a pretty good replacement to put inside an A1200 brick housing - what do you think?

https://au.element14.com/artesyn-emb...t=power+supply

[Daedalus]

Originally Posted by LittleAnimal255

especially after reading about noisy graphics with Pico PSUs and dying clones and such, and going for this https://www.ebay.co.uk/itm/New-power...IAAOSwVEdbLOMq

Yep, a few people have said those ones are decent - I believe they're based on a Meanwell RPT-60B, which is a common module for people replacing the guts of original PSUs too.

The noise in video or audio from a PicoPSU is a common issue, but isn't caused by the Pico itself. The Pico doesn't include a massive amount of smoothing or filtering, leaving that up to the supplying power brick. So noise in the supply from the power brick will pass straight through to the Amiga. This is almost always an issue because people use low quality power bricks with their PicoPSUs. Clones are another issue, but is the same across all tech really - low-quality gear will give low-quality results.

In reality though, PicoPSUs aren't an ideal choice unless you need something that small. For normal Amiga purposes, a PSU module like the ones above or below will be cheaper and simpler, and don't require a separate power brick.

Originally Posted by aeberbach

This looks like a pretty good replacement to put inside an A1200 brick housing - what do you think?

https://au.element14.com/artesyn-emb...t=power+supply

Yep, that looks like a good option for a replacement alright - specs are ideal for a moderately expanded Amiga.

[LittleAnimal255]

After more hunting through MeanWell designs I decided to go for the RPT-60B and do the mod, since 1) No pain, no gain and 2) RS Components do free next day delivery in UK so it only cost £21.47 inc VAT.

The PicoPSU build-in idea would be my first choice but my Amiga bits are too dear to me to be sending noisy or glitchy power through Thanks for the replies all.

https://uk.rs-online.com/web/p/embed...s-smps/6447219

[LittleAnimal255]

So I got my RPT60-B all made up, crimped the JST VHR connectors for mains and the Amiga cable, did the same as the 2 old ATX power supplies...HD LED on the A1200 lights up, along with the red LED on the CF adaptor inside, but nothing else.

As the original 25W Amiga PSU still works and even runs the Blizzard1230 and SCSI with no trouble *and* is only 3A on the 5V rail while the new RPT60B is 4A, I figured something else must be up, so I cleaned all the pins on the square DIN connector with a screwdriver, and after that got Power LED and HDD lights, but only black screen. Cleaned it some more with screwdriver, and now the Amiga boots up without the Blizzard1230 kit attached, black screen with it attached. This leads me to believe that the ATX PSUs could have also worked and that my made cable is at fault, either cable is too long, my soldering can't get enough current across (looks ok though), or connections on square DIN socket are grubby. Thought I'd report back as the black screen bit is interesting, many posts think that means a dead CIA or leaking caps.

[Daedalus]

A black screen is a symptom of a million different things. It seems very odd that it doesn't start up, but if the 5V rail is too low, you'll get a power LED but nothing else as the reset line is held low until the 5V rail is in spec. Check what the 5V rail reads at the floppy power connector.

[LittleAnimal255]

Originally Posted by Daedalus

A black screen is a symptom of a million different things. It seems very odd that it doesn't start up, but if the 5V rail is too low, you'll get a power LED but nothing else as the reset line is held low until the 5V rail is in spec. Check what the 5V rail reads at the floppy power connector.

It is a bit confusing that the one with the broken -12V is the only one that works so far!

I measured the voltage across the floppy connector and the new (RPT-60B) PSU reads only +4.53V, while the original one reads +4.99V, so something I don't understand is going on It reads +5.05V on the end of the cable, without a load. I'll try reducing the length of the cable tonight as it's 3 metres long, not sure what the plan is after that! Borrow the cable from the original half-working one, I imagine.


Thanks for your advice

[LittleAnimal255]

It works!

After a quick check for what turned out to be FIVE metres of 0.75 mm^2 copper on a DC voltage drop calculator I regained half a volt by chopping off 3.5 metres. The extent of the drop over a matter of metres has very much astounded me. It's now running fine even with my Blizzard 030 plugged in, even though the voltage across the floppy is still slightly low at 4.83V.

Voltage drop calculator here for fun, amazing what a metre can do to 5V rail
https://photovoltaic-software.com/so...age-calculator

And thanks again, I wouldn't have found that without your help. I'm going to get the mod underway after my 3D printer's working again, hopefully mount an LCD current/voltage meter in there if they don't use much power.
https://photovoltaic-software.com/so...age-calculator

[Daedalus]

Yikes... 5 metres of cable?! Yep, there's your problem I would never have considered even suggesting that since they never came with cables that long as standard - even 3m is quite long. If you need a lot of length from the mains socket, it's better to fit the long cable on the high voltage side, as the higher the voltage, the lower the loss of energy. Which is why national power distribution is done at many thousands of volts

Glad you're up and running anyway. The Amiga won't start up until the 5V rail is within range, and 4.5V is certainly too low. Around 4.7V is where you start getting problems so you're a little close to that. But if it works reliably then no problem.

[LittleAnimal255]

I actually thought the P=I^2*R thing meant voltage wasn't an issue for power loss, which is why the voltage is bumped right up so the current is less than 1 (as anything less than one squared gets even smaller), and thus voltage would remain the same for hundreds of metres, not drop to the point of being unusable over 5! It shows even a 1mm or 1.5mm core cable isn't much better - There go my dreams of running a solar-powered battery-backed 12V rail around the house to do away with all my 12V chargers and PSUs as well! Grid tie inverter then I suppose...

Just wondering how the original PSU caters for such a thing, does it pump out say 5.5V so it's bang on 5 when it gets to the other end of the cable? Lol

Every day's a school day, hey!

[Daedalus]

Yeah, that's more or less the same thing, but there's a bit more to it. That equation ignores the voltage drop due to the resistance, but that doesn't mean voltage isn't involved. After all, P=I.I.R is the same things as P=I.V, since V=I.R. Essentially (ignoring superconductors), there will always be resistance in a conductor, so there'll always be a voltage drop. Reducing the current reduces the voltage drop (as you saw with the difference between loaded and unloaded measurements), but also restricts the amount of energy that can be transmitted. So increasing the voltage to compensate lets you have a lower voltage drop and still end up with the required amount of energy at the other end.

The original PSUs tend to be slightly over 5V, maybe 5.1 or 5.2V. This is enough to overcome the losses in the original cable (which IIRC is only about 2m or less), and the contact resistance of the connectors.

[LittleAnimal255]

Interesting, I guess I've never noticed the difference due to only really working in milliAmps, and on PCs where 11-point-something is always close enough. It seems the higher the current draw, the more the voltage suffers too, which means there's still a chance it won't run my Blizzard PPC kit and Micronik board, unless I shorten the cable to the point where it needs to be inside the case I was unable to squeeze the cable grip into the connector with the thickness of the cable so I doubt I'd even get a 1mm^2 core in there even if I needed to. I bet the ATX power supplies I tried would have worked after all!