pouët.net

C64 PSU upgrade project - Defective PSUs wanted!

category: general [glöplog]
Hey, you. Yes, you! You have an old, broken C64 power supply rotting away in some cardboard box, don't you?

It's one of the shim-shaped ones, isn't it? Like the one on the left in this picture:
BB Image

Alright, you might also have some that are different, but I don't care about those.

So, uh, if I could have that old thing in exchange for a couple of beers at some party that we both attend, that would be great. :)

I intend to be at Revision, and after that at Evoke, Nordlicht, Outline and GhettoScene, and possibly others.

Oh, and would you by any chance be interested in a fully working and very efficient replacement that looks exactly the same?

Background:
The old shim-shaped C64 PSUs by Commodore are not very well built and often fail. However, the transformer itself is very sturdy and can probably last half an eternity. The problematic components are on the circuit board that is located below the incline, which provides rectification and voltage regulation for the 5V output.

I've been working on a new circuit board to replace the original one. It is much more efficient and uses quality components to ensure a stable output voltage with lower tolerance than the original. The components are also a bit overdesigned so that they should last for a long time. The added efficiency makes it possible to supply 2A instead of 1.5A, which means that power-hungry modules won't cause a problem anymore.

On the downside, the components are a bit pricy, and the PSU also loses its function as a footwarmer ;)

I don't want to cannibalize fully working power supplies though, so I'm looking for broken ones. So if you have one you want to get rid of, please leave a message. Thanks!

As for how you can get an upgraded PSU or an upgrade kit, that part still requires some thought. But for now, it would be interesting to see how many people might want one so that I can plan for an appropriate batch size. So if you're interested, please let me know as well :)
added on the 2015-03-22 00:47:46 by MedO MedO
I might have one laying around, but I have no idea if it works. If I can find the space to bring it with me to revision it's yours either way.

And I'd totally be interested in a replacement too.
I don't have my original PSU anymore (it went kaput waaaaay back) but I would be interested in a good replacement should my current one ever fail me.
I have one for you I can bring to Revision. The one on the Left.
added on the 2015-03-23 11:37:32 by tFt tFt
Do you accept czech or polish beers? How about Liquor or schnaps?
added on the 2015-03-23 13:24:33 by Exin Exin
Just to clarify: The beer is for you, for parting with your broken PSUs. Unfortunately the components and PCB production cost money, so I have to take some hard cash (or at least hard schnaps) in exchange.

I don't have any final numbers yet as it depends a LOT on the batch size (especially for the PCB production), but you can expect an upgrade kit or upgraded PSU to be more in the area of 30€. And I'm pretty sure they won't be done in time for Revision either, since I want to collect some "pre-orders", at least informal ones. In the area of tens of PCBs, prices drop very quickly with batch size - you can get 2x the units for ~1.4x the money, so it's important not to order too few - but I also don't want to be left sitting on circuit boards that nobody wants.

As a small update, here is the current design:
BB Image
added on the 2015-03-23 21:59:43 by MedO MedO
Quote:
I have to take some hard cash (or at least hard schnaps) in exchange.

In exchange for the upgrade of course, you don't have to pay me to take your old PSUs :)
added on the 2015-03-23 22:01:41 by MedO MedO
I still wonder how you get those fixed. Not all of them have their electronic parts easily accessible. Many of them have them behind a filling of some kind of industry garbage...
added on the 2015-03-24 08:40:54 by Exin Exin
Quote:
Just to clarify: The beer is for you, for parting with your broken PSUs.


I'll take a random fancypantshipsterbeer from your local brewery.

Alternatively the "KARLSBERG" beer of your area :)
added on the 2015-03-24 14:46:52 by tFt tFt
Would have offered non-professional SMD soldering skills for vodka - but the LMZ12002 looks really ugly to solder without proper equipment. Well, a proper oven / hot air "device" does the job, but I wouldn't like to take the risk of damaging (8+$) parts - especially with such a large ground plane.
added on the 2015-03-24 15:01:00 by las las
Quote:
I still wonder how you get those fixed. Not all of them have their electronic parts easily accessible. Many of them have them behind a filling of some kind of industry garbage...

As far as I know (and from what I have seen), most of the wedge-shaped ones (left) only have the transformer and the voltage regulator embedded in the resin, the area below the incline which houses the circuit board is free of the stuff. The brick-shaped ones (center) are entirely filled in. That's why I specifically ask for the wedge-shaped ones.

As for soldering, if it wasn't for the LMZ12002 I'd have opted to hand-solder everything myself, but after trying for quite a while to solder the thermal pad to the plane with my puny little iron for the prototype, I agree that some better equipment is needed. Luckily, we might be able to sneak this thing into a professional reflow oven while nobody is looking :)
added on the 2015-03-24 15:23:14 by MedO MedO
Huh, didn't know there were SMPS ICs which rely solely on capacitors. Have you tried building this and hooking it up to a dummy load to measure the ripple?

And while I'm at it: wouldn't it be cheaper/simpler to use a rectifier bridge instead of discrete diodes?
added on the 2015-03-25 14:43:46 by Tjoppen Tjoppen
It doesn't rely only on capacitors, the inductor is integrated. I measured the ripple with C64 connected, and IIRC it was <20mV p-p.

Regarding a rectifier bridge - feel free to suggest one, but since the diodes are only 20 cents each there's not *that* much potential for saving, compared to other components. Also keep in mind that the diodes are Schottkys, so they are more efficient than silicon junction rectifiers, and generate less heat.
added on the 2015-03-25 15:59:18 by MedO MedO
You're missing a fuse.
added on the 2015-03-25 16:19:30 by trc_wm trc_wm
No, *you* are missing a fuse. But perhaps I will miss it too once you explain in a bit more detail.

The old primary-side fuse is still included (though it could probably be replaced with a smaller one now), and the LMZ12002 is current-limited to a bit over 2A.
added on the 2015-03-25 16:27:22 by MedO MedO
Ok, if there is one on the primary side, no problem.

Those footprints for the 10uF caps seem small; which type are you going to use?
added on the 2015-03-25 17:38:17 by trc_wm trc_wm
Are you going to solder it yourself or have a company do it for you?
added on the 2015-03-25 17:43:08 by trc_wm trc_wm
Quote:
Those footprints for the 10uF caps seem small; which type are you going to use?


In the prototype I used these because I had them around, but on the input side the 16V are a bit close. These here could be a good option.

I don't plan to hand it off to a company for soldering - too expensive at the low batch size I'd expect to make. But as I said, we might be able to use professional equipment anyway :)
added on the 2015-03-25 18:14:35 by MedO MedO
If you look in the LMZ12002 example circuit, they use 50V capacitors. This is not just to be safe with respect to operating voltage but X5R/X7R materials experience large capacitance losses when they are biased, in some cases reducing the capacitance to 10% or less of the required value.

Your original capacitors lose around 50% at 5V bias and it gets much worse at higher bias voltages quickly. The suggested 50V capacitor has a much better bias characteristics: see 'characteristics' on this page; no loss at 5V.

There is a whole podcast about this kind of stuff which I found helpful.

In general, the bigger the capacitor, the better the bias properties, so I'd go with something much larger than 0805.

With respect to the solderpaste & vias under the LMZ12002; I had manufacturing issues with little solder paste 'explosions' in situations like this. I was advised to remove the vias in places where there is solder paste, or to tent the vias (expensive) so air/moisture can't get in. This might not be a problem if you hand-solder the board.

Not trying to be a dick, just helping to make a better design.
added on the 2015-03-25 21:30:20 by trc_wm trc_wm
I'll look into your links regarding capacitor choice. And I do appreciate suggestions.

With regard to the vias, I did tent them (= draw solder resist over them) to prevent solder paste being wicked down, and I don't see why this should incur extra cost. Maybe you meant plugging the vias, which definitely does make things massively more expensive.
added on the 2015-03-25 21:40:37 by MedO MedO
Yes, sorry, plugging.
added on the 2015-03-25 21:41:46 by trc_wm trc_wm
Regardless of the diodes being schottkys, don't they still get hot? I mean, the 3A ones will be used fully on a C64. Even more so with expansion cards and whatnot...

I had an Apple 2 PSU that had recitifier bridge done in 3A schottkys as well, and they had an Aluminium plate stuck to it with some kind of wire. It really got hot, though.
added on the 2015-03-26 10:46:26 by Exin Exin
Since everyone is making recommendations I'll make one too.

There is a possible ground loop in the (bottom) background plane, you need to cut the (blue) background plane from the place shown in the picture. Don't know whether it has any real improvement or not though...

(Personally I would as well route XFORM1 to diodes completely on top and avoid the via)

BB Image
added on the 2015-03-26 11:49:24 by ts ts
The diodes don't get hot (I'll show evidence of that this weekend). They only have to conduct 0.5A average each if we get 2A on the output, so dissipation is probably around 0.3W per diode.

Regarding the ground loop, I'll remove that, but I doubt it would have caused any measurable issue even if it picked up some noise from your wifi network :)

About the XFORM1 routing, I don't want to have anything on the top (red) layer in the dotted area, because there is a metal plate in that area that the PCB might connect to.
added on the 2015-03-26 12:15:39 by MedO MedO
Since everyone is making recommendations I'll make one too.

Use a switched-mode power supply :]]]]]]]]]
added on the 2015-03-26 12:15:51 by stfsux stfsux

login