Starfire

Custom Power Supply for mb6582

235 posts in this topic

In a 1U!? Goodness I was worried about fitting everything in my 3U design :) Do you have pics of that as I would love to see how you layed things out. Are you using an MB-6582 baseboard there as well?

I was planning on just using one xformer though I seem to remember reading that having an unbalanced supply can sometimes cause issues. The only thing using the -12V would be the 2044s (and the 2164's? though I am not sure I'll have those...at least not at first - I really want 2044's though :) ). I suppose if I used a smaller xformer for each that would not be so bad as an alternative though. Are there any ground issues to be worried about though when using two xformers?

Share this post


Link to post
Share on other sites

Mine works, I just have to finish the mb6582 to be sure it'll handle the power requirement, but proper voltages are seen on the 5V, 9V and 12v rails in the MB itself.

Just a simple dual coil 9VAC transformer, one is rectified and regulated to 5VDC, the other is 9VAC straight in, of course some fuses just incase.

Share this post


Link to post
Share on other sites

I was originally going to do that with my MB-6582 too, but preferred to do all the regulation outside of the box to keep the heat down. Otherwise that's basically what a C64 PSU gives you so yeah I think that should be fine as long as you have the 5V regulator heatsinked well (and have sinks on the regulators inside the MB-6582 as well of course). I don't think you will have any problem with current draw with that setup - it's mostly making sure the 5V reg doesn't overheat, or at least that has been my experience.

For my rackmount solution, the PSU will be inside the case (at least that's my idea so far) so I guess it doesn't matter where I put the regulation, though I will likely still do all that on a specialized power board/section. I can more easily cool the regulators that way - all the positive regs will likely be screwed or glued right to the chassis with the negative regulator getting it's own standard U-shaped sink (or so is my plan).

Share this post


Link to post
Share on other sites

Hi.

Strophlex, did you make the wiki on the psu? Nice work, however a few comments after reading through.

In the centertap xformer chapter there is some minor stuff that seems not to be correct. It says that 15v transformer is needed because of voltage drop over the rectifier. If one uses a fullbridge rectifier, 4 diodes, this isnt completely correct. Rectfied voltage will be higher than ac, so a formula is needed to calculate the exact output from the rectifier.

The reason that you need headroom is because of the vdrop over the vreg. I know this is details, but if you document a design so good it might as well be correct, no offence at all.

The transformer's ac out should be calculated as close as possible to what dc the vregs need, taking into account the higher voltage on the dc side of the rectifier.

Also the vregs datasheet tell u what headroom voltage it needs. It aint always true that it needs 3 volts. The input volt range of the vregs tell you what it can take, but beware of the extra heat genetated in the vreg if it has much headroom voltage. Best is to calculate everything as close to ideal as possible taking into account the worst case scenarios such as variations in mains voltage etc.

Also one last thing. Psu design isnt just about getting the right voltage out of it. The design in the wiki offer no protections from over voltage or excessive current draw. It needs to be protected itself, and to protect the load also from a short in the psu. A fuse is not enough by long shot. You need some way to eat power in the psu while giving the fuse time to blow. If only a fuse is used it is most likley to kill a lot when there is a fault somewhere before it blows. A way to think of fuses is that they are only there to cut power when spmething fails, the actual fast and real protection are in the electronics. I would never run a mb6582 without this real protection.

The c64 psu design isnt ideal at all for the synth. It has as mentioned a tendency to fail and take sids and other precious stuff with it. Maby use them with success, but i wouldt do it.

Just my thoughts... not meaning to offend any in any way.

Tbreath

Share this post


Link to post
Share on other sites

Keep in mind that that vregs have some protection built in. They will generally overheat if too much current is applied or fail completely in some cases. It's probably not fast enough for an instantaneous issue on the synth side, but may be able to protect against some situations on the AC side.

Also, the formula is the output of the rectified side of the transformer * 1.41. So if the transformer is outputting 12VAC, you will end up with 16.92VDC. I've verified that with my multimeter and 1.41 gets it pretty close. It comes from the fact that AC is a sign wave and that, after rectification, the average voltage is less than the peak of the wave so you end up with a bit more juice. I can't remember how to derive the number - I just remember it :) There's forum postings about how to get to that number however and SmashTV has a good link on what's going on visually though I can't find it :/

That is for full-wave, by the way. Also the input voltage on the main side of the transformer may not always be 110/220 and that will affect what voltage the transformer outputs on the other side. So you need to keep that in mind. For those really concerned, low-dropout regulators may be the answer, though I don't know if they are able to go above 1A and I have not had any experience with them myself.

Share this post


Link to post
Share on other sites

That's what I said. Typo from me: I said fullbrigde - I meant of course full-wave :).

Yes, the vregs will shut down when they get hot. But this is as you get into here, not good enough. If ideal conditions in mains would always be ideal, and not fluctuate, it would from the formula you mention (yes it is pretty close to correct at least at 5-15 volts) be more than enough with a 12V transformer - you wouldn't need a 15V transformer. Worst case scenarios should always be taken into account when designing psu, so you should be safe with a sec. 15vac transformer yes. You could always build a capacitor bank also to feed the psu when the mains drop too much... just a thought - don't know if it actually works :).

The worst case scenario would be something like a short in the synth's powerrails. This could lead to failure in the psu, and would lead to failure for sure in the synth itself. The synth would draw too much current, and the vregs would shut down eventually because of heat. Another scenario would be that for some reason the vregs would short, and then what happens? no regulation and unlimited power until something else brakes in the psu, and by then the synth is broken. For this you should have a fuse - yes - but also a (very)high power transistor and a sense resistor that can take the watts needed to cut off the voltage at the input of the vregs or at the input of a pass transistor if used and sink it to ground long enough for the fuse to blow.

This isn't exactly very costly, and it adds very few components to the supply, and it makes it a LOT safer for both the supply and the synth, so why not just do it? :o

I work with CCTV systems, and have set up a lot of them myself. When there is a short in the load or the cable to the load - the powersupply I use will blow the fuse. Correct the error in the load - replace fuse - and you're up and running again. This will not be the case where the psu hasn't got the proper protection. Instead you get to spend time searching and correcting the fault in the supply.

If you want a powersupply as safe as possible and one that lasts over time, you should also build it with that in mind.. it won't run of mind-power :). Also, be sure to serve the synth with sufficient amounts of amps...

All I said, the psu design isn't secure enough for use with such precious equipment. And also the C64 brick isn't secure enough either as I stated. This is based on second hand experience - I haven't seen one fail myself yet, but if it does - God be with the synth...

Also if there isn't enough current served from the psu, you could get in trouble with backlit LCD if you are a bit on the heavy side of brightness... If there isn't enough current, the voltage regulator will not work and it will drop the voltage and all will be working very bad...

I only say that when designing PSU one should do it based on the maths and not based on asumptions that might not be correct at all.. It is not very hard to do the math, and therefore you should do the math :D

I can't remember if I said this, but the reason why I say that it the voltage from the rectifier should be as high as you need (with WCS taken into account), is that higher voltage means higher temperature... So there is nothing to loose and all to gain by doing it properly when first doing it. So, as I said about PSU design - getting the right voltage out of it really just takes you half-way there... in my opinion.. :)

Share this post


Link to post
Share on other sites

For this you should have a fuse - yes - but also a (very)high power transistor and a sense resistor that can take the watts needed to cut off the voltage at the input of the vregs or at the input of a pass transistor if used and sink it to ground long enough for the fuse to blow.

Can you elaborate on this a bit? That sounds interesting!

(Agreed by the way that the C64 PSU sucks huge infected goat balls)

Share this post


Link to post
Share on other sites

I have a quick related question that I think belongs in this thread.

The two 9 volt pins on the C64 power supply are -9volt and +9 volt (I'm only using 8580R5's), correct? I'm stuck a little on how to wire up the transformer... What I think is correct is gnd and +5volt on 2 pins, and +9volt and -9volt on 2 pins.

Cheers,

grizz

Edited by grizz

Share this post


Link to post
Share on other sites

The C64 PSU outputs 9VAC so you'd have to convert that to DC before you supply the SIDs. The standard SID module has to convert to DC on the board (as does MB6582, I think, using PSU Option A). The C64 pin-out contains a GND pin so you'd want to use that. If using a modular design (with separate CORE and SID modules), GND would be wired to the CORE module as it provides GND to the SID board via the SID connector.

These may be of some help, though they are built around powering both 6581 and 8580/6582 SIDs so the incoming voltage would be higher than you need:

http://ucapps.de/mbhp/mbhp_sid_c64_psu.pdf

http://ucapps.de/mbhp/mbhp_4xsid_c64_psu_optimized.pdf

When using the standard CORE and SID modules, however, the above would work for you.

Share this post


Link to post
Share on other sites

-9v? I didn't think you needed -9V. Haven't looked too much into the specs of the c64 psu, so I can't do more than asumptions... Are you building a c64 psu clone or are you using the brick itself and want to know what pins goes where? In that case I can't help you very much. In case you build a clone - don't :P. If I'm not totally off here, the correct values for 8580s should be +5V, +9V and ground. Where to connect this on the baseboard is well documented in the wiki.

Can you elaborate on this a bit? That sounds interesting!

(Agreed by the way that the C64 PSU sucks huge infected goat balls)

Haha, reminds me of a former collegue of me. He always said "it sucks several meters of infected cock!" :) hehe.

I might have been a bit quick when dragging the vregs into this. I assume when I think about it (have to check the datasheet) that the currentlimiting in the vreg is pretty good. But when you need to draw more current than the vreg can deliver you need to use a pass transistor. It is controlled by the vreg and depending on how "big" the transistor is, you can draw muuuch more current from it than from a 78xx 79xx regulator itself.

To protect all this you use another high power rated (don't remember all the details just now) transistor and a resistor in series with the load to control the base of the protection transistor. Collector of this is connected to the base of the pass transistor, and the emitter to ground. When current draw through the sense resistor raise above a set point (depends on the amps you wanna be able to draw out of it and the value of the resistor) and thus the base voltage (on protection transistor) reaching 0.7 volt (if i remember) it activates and short the voltage to ground. So the protection transistor need to be able to withstand a lot of power - much more than the pass transistor - to withstand this shorting. And when the transistor is shorted, the fuse will blow and everything turns off, and all you have to do is to change the fuse if you're not too unlucky.

Look at it this way. The transistor is a superquick switch that when it activates it "turns off" the voltage at the base of the pass transistor by shorting it directly to ground (the current will always flow the easiest way to ground). Making this short in the psu makes the fuse blow.

So in short - it shorts out the powersupply in order to protect it :) only difference is that you short it on purpose through a component that won't brake.

There are another option (crowbar circuit) based on a SCR (or thyristor or so if I remember correctly) and zenerdiode that function about the same way - only that the short is activated by a voltage threshold instead of current sensing. This voltage is set by a zener diode (and a resistor) and the zener diode should be set a tad higher voltage than what's actually on the rail - thus preventing false shutdowns. I don't remember the exact details without looking in my books, but it is the idea at least...

So if you protect it with both a crowbar circuit and a fine tuned current limiting circuit and a fuse you will be very secure. I'm not really sure you need both circuits, would probably be enough with one of them and a fuse.

At least this is the way I'm told it work :)

Wikipedia has some good articles on transistor based current limiting.

Share this post


Link to post
Share on other sites

*** C64 PSU OUTPUTS 9VAC ***

As far as the circuit protector, I read up on the crowbar stuff too and I think that's the way I'm leaning. I'll have to look into using a transistor though, that sounds pretty clever, thanks!

Share this post


Link to post
Share on other sites

I am also still looking to build a psu. I made a preliminary design:

mb6582-psu.png

The pic is kinda dated. As i wont be using a 78S09 but a 7809. Have to look into a crowbar design as well. It seems the zener option doesn't work so well as it might get trigger by a little peak at startup. The SCR option is better. It seems there are a couple of IC options out there. Someone pointed me towards the Motorola MC3423. However this seems to be no longer made. However i just noticed reichelt has it on stock still.

Edited by Shuriken

Share this post


Link to post
Share on other sites

I've been reading up on the crowbar circuit and it seems to be more for voltage limiting - but isn't that the point of the regulators? If there's a short, I would have thought the thing to solve for is a large amount of current, not the voltage?

As far as your design, I haven't seen a PSU design that uses diodes as you are - what exactly are they meant to do? That the ones around the regulators look to be a feedback loop of sorts?

Edited by m00dawg

Share this post


Link to post
Share on other sites

The diodes around the regulators are there for protection of the regulator incase of input short. The diodes on the output are (old skool) there for the protection of the phenomenom called latchup. As the sid chips are old style CMOS chips they are vunerable to this.

This design is much like Roland's design for the Juno 60. Which holds a lot of 4000 series cmos chips.

When the C64 PSU fails the 5V regulator will put out something like 10V. Since you have 30 euro chips behind it, a 1 euro circuit can make the difference. So i figure you spend a few bucks to safe you from spending a lot of bucks makes sense.

The crowbar circuit is there for when the regulator fails. If i look at the datasheet of the mc3423 it can also help with current limiting.

Edited by Shuriken

Share this post


Link to post
Share on other sites

Gotcha. Just haven't seen that design before. Functionally, I'm not exactly sure what happens :) If the regulator fails and outputs 10V, how does the diode on the regulator help exactly? The crowbar circuits I have looked at online (and that have been discussed a bit on this thread) are based upon a voltage max that triggers the crowbar - how does this design differ (since I don't see a trigger)?

Share this post


Link to post
Share on other sites

Gotcha. Just haven't seen that design before. Functionally, I'm not exactly sure what happens :) If the regulator fails and outputs 10V, how does the diode on the regulator help exactly? The crowbar circuits I have looked at online (and that have been discussed a bit on this thread) are based upon a voltage max that triggers the crowbar - how does this design differ (since I don't see a trigger)?

As i said, i have to look into a crowbar design. So that's not included into the current schematic. So there is no OVP or OVC protection atm. Just protection for the regulator and for latchup.

Share this post


Link to post
Share on other sites

The latchup I think I get (looks similar to a voltage clamp, though how those work is still a bit alien to me). But how does the diode protect the regulator exactly? I've just never seen that and it seems like it might not be a bad idea to implement which is why I'm curious :)

Share this post


Link to post
Share on other sites

Someone correct me if i am wrong. But incase of a latchup. The parasitic structure (eq to thyristor) can change the polarity. So the GND is changed to a -DC and the +DC is changed to GND.

For the other diode have a look at this datasheet and scroll to page 11.

Edited by Shuriken

Share this post


Link to post
Share on other sites

Aha! That's super helpful, thanks! It sounds like it's only required if there's large caps on the other side, though that's certainly possible (you can use the 2200uF power cap from the MB-6582 in a custom PSU design, for instance). So, yeah, that seems like a good idea :)

Share this post


Link to post
Share on other sites

I assume I need a diode for a negative regulator as well? I did some poking around on National's website and did not see a diode used on the 79xx datasheet, though I suspect it'd be basically the same thing?

Share this post


Link to post
Share on other sites

Hi. Did u see the design i posted in the chat?

I assume u dont talk about the rectifier diods...

There is no other diods in the designs appart from the zener diode. The fourth one that looks like a double diod is a SCR. it is the shorting component in the crowbar. The zenerdiode is a 6.5v and it works this way;

When the voltage reaches 6.5 volt the scr will turn and thus opens connection to ground and shorts the psu long enough for fuse to blow.

But the crowbar depends on current limiting, such as a fuse or a fuse in combination with current limit transistor.

Even if a crowbar is defined as a overvoltage protection, voltage and current has its relationship...

Share this post


Link to post
Share on other sites

Hi. Did u see the design i posted in the chat?

No, although it would be great if you could post it here. That way those that might be searching for an answer to these questions in the future will have a place to look :)

Share this post


Link to post
Share on other sites

I think he means this one:

gallery_3579_6_16029.png

If you adapt this to 2 separate windings then you need to add the diodes again.

edit by nILS: Added a link to the picture's source ;) This design is from the wiki, where you can find some explanation as well. Sadly I ran outta time to brush up the other designs as well. It's back on my ToDo list now.

Share this post


Link to post
Share on other sites

No, this isnt the design I posted - I thought I had posted it in the forum too, but I might have forgot :) But I'll make a new version of the schems tonight and post it later on since new information has surfaced since I made that design.

I will post as soon as I have made the new version, and you guys can tell me what you think.

My design has no -12v rail yet. I haven't incorporated it yet because I have to keep my head straight when doing the design... I figured that if I do the positive side first, the -12v rail would be pretty easy to do afterwards.

Many thanks goes out to nILS and SmashTV for helping me out being my "tutors" in designing the psu :) - you guys really pointed me in the right direction :D.

The one thing that sets me back now are xformers. I have none and I'm struggling a bit to deciding what to use. I think I will decide on a design tonight and order in the part so I can finally start making a real prototype, and stress-test it a bit to see if it actually works.

About crowbars:

There is a thing with all simple current limiting circuits - as you mentioned above in some post - they are vulnerable to false triggering on startup. I think the main reason to this is large capacitors on the power rails. They charge up really quick, and when they demand some power at startup to get fully charged that can actually trigger the crowbar and the fuse will blow. I guess if the fuse doesn't blow you will get a hickup from the psu turning on / off / on / off and so on. The solution might be to use not too big caps in the psu and also set the crowbar's zener to a reasonable voltage so that it doesn't trigger the scr or thyristor too soon.

The diodes for latchup is interesting part I haven't looked into yet... I will read some more :).

I'll be back later on with my new version :)

See you.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now