mb_6582 Power Supply

[Janis1279]

[... And since I'm here : I received a 1st version C64 power supply and +5v is on pin 4 instead of being on 5... In fact, the plug only has four pins coming out, unlike my other "newer" C64 PSU that has all 7 pins (even though some aren't used). Can I bridge pins 4 and 5 inside my Midibox Sid without making a mess the day I plug my "newer" C64 PSU to the midibox ?]

You can bridge pins 4 and 5 inside your MB Sid

[tupinamba]

You can bridge pins 4 and 5 inside your MB Sid

Great, thanks !

[technobreath]

Hey,

I am scratching my head here too. There are all kinds of more or less exotic variants of linear psu designs out there. Both here and online.

First of all one has to find out how much current u need to have available to the synth, and in my opinion, the brick is in the lower end of this, and without any proof I will say it either way. It may be exactly why it dies on some people. Stressing old components like that to its limits is in my opinion to ask for trouble, and if it hadn't worked for most users I would scrap the idea. Now this is only my humble theories hehe.

As for designing a modern linear psu - linear is recommended for this unless u have a scope and know what u do when designing a switcher - it is fairly simple to make a linear that works well.

Only thing is that u gotta know how much current u need, read some datasheets for the vregs and you should be quite some way to the psu.

The vregs have specified max input voltages, max current rating. This is all based on the assumption of best case scenario with almost always 25c as the ambient temperature. Of course these ideal assumptions aren't the case in real life.

So u need to use some brains here to figure out realistic operating criterias.

Ones that is done I don't see any problems using the components within specs.

There are tricks to do things a bit easier temp wise. Put the heatsinks out of the enclosure. Make it big enough. U don't need the psu to be as small as possible. Easy trick, right?

What u basically need to construct a linear psu is transformers, input smoothing caps - those are the big electrolits u see in most schems around. These can be used to filter input of many vregs, u don't need to have one / several for each vreg.

U need vregs of course, and u get iirc up to 2a max output current (keep in mind this is not realistic, but ideal conditions)

The vregs also need one cap at the input and one at the output. These are needed for each vreg, and in datasheet u find how close they physically need to be to the vreg. Often an inch or so. Of course this is the electrical path, u can't stick the cap right next to the vreg and then route the path around the board and then expect it to be ok. These caps are there to keep the vregs from oscillation aka make funky electrical hi freq noise on the line. The size of them is clearly stated in the datasheet also, just follow that and don't think too much why and how on those.

Then a 2nd smoothing cap is used at the output of the vreg too, a lot smaller than the input caps. Helps stabilize the line more if there is some noise left after passing through vreg.

The input caps need to be big enough for the current drawn from the psu.

To say it simple. The higher voltage u fees the vreg, the smaller caps u need, but also heat dissipated from vreg increase if it has to burn off more voltage. So this is a game to find the golden way between two scenarios.

There is formulaes to get this right, I can't remember them right now. I usually use a simulator sw to load test the circuit, and when watching the scope I can get some clue to how big caps I need. But if u like math instead, we have Google.

These days I have been picking up on designing my psu for the mb6582. It contains no funky resistors etc, and is completely within specs and design rules given in the datasheet of the vregs.

So to make this short, u don't need any funky stuff when using these vregs. U can manage without resiators to lower stress on the vregs etc, resistors also add noise to the line. I can't with my best available fantasy thoughts understand why one would want to go all funky about this when the specs are so easy to follow.

Just keep the heat away from the vregs with a big heatsink... They are made to burn off heat, no need to cuddle them! If they get too hot there are also thermal shutdown thing in them mostly.

I can't promise anything to when my design is gonna be ready and tested, but when I have made it and test it ill make the design available in the wiki pages. I'll also post pics if scope under stress testing, and also pics of temperature on various places in the psu and ambient temp. Info on proper heatsinking etc.

Please do not interpret this as me coming here thinking I'm much better than the rest, but I believe I have some good points, and as I said in the beginning, there as are many funky out of specs designs everywhere, and it only serves two things - to confuse the builder and give the builder a crappy psu. Might work, might not work, but in the end, the smartest and easiest thing is to just follow the datasheets and trust that the ones made the component really knows what they are doing.

/signing off.

[Hawkeye]

What t´breath said is all true.

Just here to let you know that the slightly modded retro-donald design works great and is even better regarding noise than the original psu:

Greets,

Peter

Edited October 11, 2011 by Hawkeye

[technobreath]

Hey. I had a look at that psu now. It looks good on schem, but as I said before in the post, the input ripple filtering caps need to be the right values for current draw and input voltage, else ripple will not be filtered as good as can be.

I haven't read the instructions on this in detail, but if it doesn't tell u to watch out for the input caps value to be high enough, and the ceramic ones to be close enough, it serves as a brilliant example of one of those confusing design instructions I talked about.

The case with the ceramic caps are all in the vreg datasheet also.

It took me a long time to understand what happens in a linear psu, so I don't bug those who hasent read a lot bout it and gone through that learning curve I did. That is why I say, do not trust designs all together, read some sheets and see why it is the way it is, and if there are special considerations to be taken into account when doing PCB layout. Just a tip hehe.

So as you found out P, what u did was actually to correct some design errors that made this psu operate outside specs, and it worked for u then. Veery nice. Good job figuring it out. It is cool when people combine knowledge to address issues like this.

I may be a atypical man with this, but I almost always read the datasheets to better understand the building process, and when I'm too lazy to do that I ask nils hehe, and he always bites me back and do a let-me-google-that-die-you to me hehe. Well, not completely true, as he have helped me a lot getting started on my psu quest, and during the quest smashtv has also been tremendously helpful!

I made part of the PCB layout today for the psu, but since eagle free is so nice to not let me have those big pcbs in there ill have to make the crowbars on another module hehe.

Let me also add that the psu u link to here peter, is slightly different to my approach. My goal is to take every psu component out of the mb6582, so I output no 9v ac. I have 5, 9 and 12v dc outputs. Mainly because I have all the psu components in one box then and not some in the synth and some in the psu box hehe. Makes service easier and also makes heat in the synth less of an issue, because u move all vregs out of there.

Now off to tetris some crowbar module. Hehe. Have fun.

[technobreath]

Shithe, I can't figure out how to edit post using my phone.. I mentioned earlier that iirc u get vregs up to 2a max. Those are the 78sXX vregs. Of course there is higher rated vregs around, but what I meant is that u get up to 2A in a TO220 package. Higher rated vregs seems to be mostly in those metal can packages, and those are not that practical when it comes to heat sinks...

Sorry for the double postings here, but I really found no edit function in the forum on my phone...

[orange_hand]

yeah... the PSU.... an evergreen :tongue:

I have now received all parts for my PSU experiment, a combination of the 220 to 9 V AC transformer together with a 4 V DC switching PSU... I will report, how that works...

If not, I will go for the optimized design from Hawkeye's PSU :-)

Cheers

orange

[Hawkeye]

Yes, using a 5V switcher on the 5V "logic" (not audio) chain seems plausible and energy-efficient, but aren´t the 5V and 9V added together in the 6582, for Power Option "B" (when using 6581 12V SIDs).

In this case, you´d have "switcher" input for the SIDs on the audio chain...

Of course, you could go with power option "D", as technobreath suggested... and supply everything from the outside...

Or stay with power Option "A" and use the suggested solution (but then, only 9V SIDs and it is not a "universal PSU", again).

Greets,

Peter

Edited October 12, 2011 by Hawkeye

[technobreath]

/me haven't studied the different options for supplying mb6582, as it was very clear for me before I started the build that I was going for opt.D. I haven't enough knowledge either to say yes/no to combining switching and linear - since I have no idea how seperate those systems actually are in the synth itself. I would somehow assume without watching datasheets for the SIDs since a long time now - that the rails have the same ground potential / shared ground?

This is just speculations, nothing more, but maybe someone can clarify... Wouldn't a noisy switcher impose noise to the other rails?

[orange_hand]

Hi guys,

here is my prototype, a combination of a regular transformer and a switched PSU:

My first impression is good, but I have to compare it with the original C64 PSU.

I will keep you posted.

Cheers

orange

[orange_hand]

Hi Guys,

I have now compared my prototype PSU with the original C64 PSU. I can’t make out a difference, which means that the switched PSU does not cause noise in the audio path. This is good news, as this power supply will be able to supply two MB652 or two C64 computers at the same time. I just need to update the PSU with some fuses as well as a crowbar on the 5V stream for maximum protection and then I am done. Please note that the small switched PSU is very strong - capable to provide 5V at 5A !! The good thing is that this PSU is easy to build (less number of parts and easy wiring) and does not suffer from any kind of heat issue. It will also be more efficient, as the 5V switched PSU works more efficient compared to a voltage regulator that heats up your studio :-)

You can build it with a minimal soldering effort if you like. I have already prepared the enclosure (holes and PCB board) and will post a picture later today. I will keep you updated.

Cheers

orange

[pinksoir]

Good work dude. It's looking great. I'd be interested in having it as an internal PSU in the MB SID I'm currently working on.

[orange_hand]

Hi Guys,

here is my dual line PSU:

Two jacks on the front side and a LED to indicate the function:

On the back side there is the power connector (contains 2 fuses and a switch) as well as 3 additional fuse holders.

Cheers

orange

[Hawkeye]

Well done - looks very good! :-)

Its near-identical (but more hot-blooded) twin says "Hello ´Bro :ahappy:"

Greets,

Peter

Edited October 20, 2011 by Hawkeye

[orange_hand]

Thanks Peter !

[synaptech]

Wow looks great orange_hand! After some recent c64 psu troubles, add me to the list of interested parties! :D

[orange_hand]

Hi Guys,

here are some additional details:

On one line I added a fues and a crowbar, on the other just a fuse. On this picture you can seen the first line completed:

On the next pic you will see the assembled PSU:

The LED's are indicating whether the 9V and 5V line is available or not. The 9V and the 5V lines are fused, just in case :-)

Cheers

orange

[X11]

nice psu!

Would it be a good idea to make it 10VAC,due to the elimination of humming since the 7809/7812 need burnoff voltage?

Like MissionBrown uses 15vdc at the 9vac input ()

just wondering about this during the weekend staring at my amiga psu thinking would it work.

I used the white brick 5v/1,7A that caused hum.

i had a blackbrick laying around without an connector so instead of wiring the amiga psu i wired the black brick(same specs as moonk)

Seemed that the blackbrick reduced the hum more to the background.

for now the blackbrick is my friend:)

inside my mb6582:

there are 2 8580r5 chips (with white brick bit hum ,with blackbrick almost silent)

There are 6 6581 chips (with white brick a helicopter,with blackbrick stil a bit hummy/hf artifact-sounds)

So i was thinking since the mb6582 adds 5vdc and 9 vac to make 12vdc(if im correct),

would the 6581 chips be more silenced (humming wise) if the 7812 regulator got some more voltage to burn?.

I'm new at psuWorld but probably 10vac would be a good choice for 6581 stuffed boxes(unverified/theoretically)?,

hence the 2200pf 16vcould handle this( since 10vac is rectified around 14,1vdc right? )

or am i just an hummingbird;)?

i'll leave my amiga supply in his glory and go for pro like oranges box!

just need the right juice for a hum-free environment!

greetz Hans

[Shuriken]

nice psu!

Would it be a good idea to make it 10VAC,due to the elimination of humming since the 7809/7812 need burnoff voltage?

Like MissionBrown uses 15vdc at the 9vac input ()

just wondering about this during the weekend staring at my amiga psu thinking would it work.

I used the white brick 5v/1,7A that caused hum.

i had a blackbrick laying around without an connector so instead of wiring the amiga psu i wired the black brick(same specs as moonk)

Seemed that the blackbrick reduced the hum more to the background.

for now the blackbrick is my friend:)

inside my mb6582:

there are 2 8580r5 chips (with white brick bit hum ,with blackbrick almost silent)

There are 6 6581 chips (with white brick a helicopter,with blackbrick stil a bit hummy/hf artifact-sounds)

So i was thinking since the mb6582 adds 5vdc and 9 vac to make 12vdc(if im correct),

would the 6581 chips be more silenced (humming wise) if the 7812 regulator got some more voltage to burn?.

I'm new at psuWorld but probably 10vac would be a good choice for 6581 stuffed boxes(unverified/theoretically)?,

hence the 2200pf 16vcould handle this( since 10vac is rectified around 14,1vdc right? )

or am i just an hummingbird;)?

i'll leave my amiga supply in his glory and go for pro like oranges box!

just need the right juice for a hum-free environment!

greetz Hans

The 78xx series should not hum. Usually humming is caused by switching regulators. The 78xx are linear ones.

These regulators need 3V extra to properly regulate. So a 7805 should be fed with at least 8VDC. Any excess voltage will be dispensated as heat. If the regulator gets (too) warm it will be able to produce less current.

The 6581 needs 12VDC. So 10VAC is not enough. As the 7812 needs 15VDC input to regulate. So i would say you would need atleast 11VAC. Since most transformers don't come in that spec, you usually end up with a 12VAC one. Then the 2200pf 16V is pushed to the limit. Thankfully Tim used a high quality one.

Edited December 19, 2011 by Shuriken

[X11]

The 6581 needs 12VDC. So 10VAC is not enough. As the 7812 needs 15VDC input to regulate. So i would say you would need atleast 11VAC. Since most transformers don't come in that spec, you usually end up with a 12VAC one. Then the 2200pf 16V is pushed to the limit. Thankfully Tim used a high quality one.

Thanks this enlightens a lot

good thing i let my a500 psu live.

:thumbsup: