Multi-Tap Transformers Versus Voltage Regulators?

m00dawg · December 27, 2008, 5:46pm

It seems like most people here tend to either use C64 power supplies or build their own using a wallwart and voltage regulators with some filtering capacitor magic for powering the MBSID. I’m ok with that approach (I’ve design a few of my own this way) but I was curious - has anyone used multi-tap transformers?

Someone at work told me you can buy a 120V transformer that can produce multiple AC outputs (9 and 5, for instance). I’m thinking of using that to build my own power-brick. The idea is to have it generate +9 and +5 DC for the MB-6582 board. I was thinking of making this into an external power brick of sorts, but one that sucks much less than that terrible C64 PSU (yes, it works great until the damn thing breaks and you’re SOL).

Trouble is, I can’t seem to find these things anywhere. I’ve looked at Allied and Mouser and I’m a bit confused with getting the right part. I’ve seen multiple ones that have multiple outputs, but they are all the same output? Anyways I’m a bit confused and if anyone knows what to look for or where to find these, that would be awesome!

On an aside, I’m also having trouble finding heatsinks online that match what is in EAGLE. My original power design used regulators with a huge heatsink on top. My prototype just used one I stole from a motherboard, but I can’t seem to find anything close to that on EAGLE. What do other people do? Just measure out the holes on the heatsink and match it to EAGLE or?

All that said, I’d be happy to share my design once it is complete I know a few designs are out in the wild, but most seem to go the vreg route, and also seem to leave out fuses in their designs (something I’m definitely going to be including).

Futureman · December 27, 2008, 9:07pm

I’m sort of facing the same problem / solution.. at the moment I’ve got a C64 P/S, but they are too bulky, and I want to put mine inside my synth.

Yes, a multitap transormer is the solution, (Which is pretty much inside the C64 power supply)

Edit - I don’t know why this is Multi-Taps versus Voltage Regulators? It’s not one or the other.. lol.

I just want to make sure that it’s gonna supply enough juice to keep a fully stuffed 6582 + LCD backlight + Sids running nicely..

To quote Wilba-

"Based on datasheet calculations, the 5V supply usage is ~1000mA.

PICs are 23mA each, 6582A SIDs 70/100mA each, LEDs 100mA max, LCD ~280mA.

If the SIDs are at the max current, then it’s ~1200mA.

9V supply usage is ~320mA."

So, we’ll need the 5v to be >1200mA, and the 9v >320mA

At my local electronics shop, they have quite the assortment of power transformers, including a few meaty (+1500mA) multitaps… I’ve got to investigate a bit more, but I’m gonna try for a toroidal transformer if I can (Lower profile + less EM radiation)… maybe a 9+9 (The specs are always in AC, so it’s 9vAC - 0 - 9vAC)

I just gotta see how easy it will be to get something that is not too big physically, while supplying enough juice.

Not sure if what I’ve said is much help (It’s early & I’m tired…) but I’ll keep you posted on my workings

Regards

Mike

lylehaze · December 27, 2008, 9:45pm

As mentioned above, getting multiple transformer windings does not eliminate the need for regulators.

According to the T.I. datasheet for the 78xx positive regulators, you must provide 2 or 2.5 volts higher input voltage than you expect for the output.

For a 7805, you must give no less than 7 volts (vout +2)

for a 7809, no less than 11.5 volts (vout +2.5)

for a 7812, no less than 14.5 volts (vout + 2.5)

So the common way is to choose a transformer that will supply the highest positive voltage you need, and let the lower ones burn a little extra heat.

We need multiple secondaries, or center-tapped secondaries, if we want a bi-polar supply, with both positive and negative voltages. Doing it with two transformers works fine too.

Most transformers with multiple secondaries at different voltages are custom made for a specific product. That’s why you don’t see many of them at your supplier. It’s cheaper than using multiple transformers, as they can all share the same primary.

Heatsinks: I often use the metal project case as a heatsink for the positive regulators. I just insert the legs from the back of the board, and don’t solder it until the board and regulator are mounted. Positive regulators (78xx) have the tab grounded already, so no problem there. negative regulators are NOT at ground, so either insulators or separate heatsinks are required.

You can make your own parts for Eagle too, if you want to make library parts that match your suppliers heatsinks.

Adding up all your power needs is important.

100ma per LED? That sounds pretty bright.

Mine operate at 5 to 20 ma each. I suppose if they are multiplexed you might be pushing higher currents..??

Have Fun,

LyleHaze

m00dawg · December 27, 2008, 10:04pm

Thanks both!

@Futureman:

The current draw is higher than I originally expected, though I did find some PCB mount transformers that might work (they may even be toroidal transformers). The problem there is that they take up quite a bit of room which makes printing a PCB expensive, not including the fact that it’s easy to run into the board limitations of the free version of Eagle At least, that’s when using two transformers. If I can get it down to one, I shouldn’t have a problem.

@lylehaze

Oh yes, I’m aware regulators are still needed, although the input voltage is a but confusing I’ve found. 9VAC can result in +/- 9VDC if I’m not mistaken, depending on load, actual input voltage, and the fact that converting AC to DC gives you slightly more volts? I got that by consolidating all the information I’ve found from other posts so I’m not sure if that’s 100% correct. In other words, I’m not sure if I should get a 9V/5V transformer, or a 11V/7V one. The higher one shouldn’t hurt anything, of course, if the regulators are sufficiently cooled.

Either way, my plan was to have a 7809 and 7805 in my own design. The difference is that they won’t be in series. So the 7809 won’t be supplying the 7805. Instead, they will use different rectified voltages from the multi-tap or multiple transformers.

The fact that multiple voltage output transformers are hard to find is a bit concerning though Mouser has a few options, but they are either too big or too small. I usually use Allied Electronics because their parts get to my doorstep in sometimes the next day, but they have even fewer choices as far as transformers are concerned.

I suppose if I can’t find a multi-tap, I can just go back to supplying both regulators with 11V, but that’s a lot of voltage for the 7805 to handle. That means I’d have to go back to putting them in series and that just seems a bit gacky

Connecting it all together is another issue. My plan is to use a power supply chord and wire it up that way. But since I’m dealing with mains, I have a feeling I’ll need beefier wiring on that side and I haven’t done that before. Heck, I’m not sure I’m even wiring the thing right in my schematic Correct me if I am wrong, but the AC fuse goes on the positive AC line and I would ground the board to the 3rd grounding pin of the AC cable?

m00dawg · December 27, 2008, 10:09pm

Dunno if it will help much, but here’s my current board design. I just threw it together for funzies so I don’t expect it to be a final product or anything, but perhaps it might help with the conversation. Actually, it’s already flawed since it sounds like I’ll need bigger fuses (I only put 250mA).

[powerboard-xformers.zip](< base_url >/applications/core/interface/file/attachment.php?id=4760)

[powerboard-xformers.zip](< base_url >/applications/core/interface/file/attachment.php?id=5813)

m00dawg · December 27, 2008, 10:57pm

I hate replying to myself but I just had a thought…the MB-6582 uses the 9VAC and 5DC connections of the C64 right? So, all I really need to provide is a nice, smooth, 9VAC and 5DC. Thus, I could use a single 9VAC transformer, smooth that a bit perhaps, and supply it as-is to the MB-6852. Then I could regulate that down to 5VDC. That regulator might get a bit hot, but at least this way there’s only one of them and no having to do weird regulator chaining or anything like that.

Thoughts?

Futureman · December 28, 2008, 4:22am

I hate replying to myself but I just had a thought…the MB-6582 uses the 9VAC and 5DC connections of the C64 right? So, all I really need to provide is a nice, smooth, 9VAC and 5DC. Thus, I could use a single 9VAC transformer, smooth that a bit perhaps, and supply it as-is to the MB-6852. Then I could regulate that down to 5VDC. That regulator might get a bit hot, but at least this way there’s only one of them and no having to do weird regulator chaining or anything like that.

I’m not 100% sure what you are saying, but seems you mean… 9VAC -> Rectified -> Smoothed -> Regulated to 9vDC -> Regulated to 5vDC ?

I’ve heard that chaining regulators is a bad thing, (Never had any first hand experience with things going bad)

Give it a go… I do agree with you that that 5v regulator IS going to get damn damn hot.. = BIG heatsink… but you would avoid that if you used a multi tapped powersupply…

Regards

Mike

m00dawg · December 28, 2008, 4:27am

Actually, I’ve chained them before. It does get hot and requires a big heatsink but it works well. You might want to add more smoothing capacitors after regulation though. As for this project, I think I have things figured out:

I’m going to use a center-tap 12VAC transformer. That will give me +12 and +6, which I can smooth and rectify separately to give me +9 and +5DC. Regulating +12 down to 9+ is a bit of a step, but I think it should be fine since it will draw less current than the +5VDC. I opted to go with 12/6 over, say, 10/5 to allow more headroom. Though, to be fair, I should probably figure out the math to back that up I’ve heard conflicting opinions about how much AC voltage headroom one needs to feed to regulators so I still need to research a bit more on that.

Either way, this has certainly simplified my design! I did have to remove the transformer from my Eagle design, however. I couldn’t find a center-tap transformer object in Eagle. Even if I did, the board would have been bigger than it needed to be I think. So I’m just going to be a chassis-mount transformer and screw that into whatever case I end up using.

I’ve got the board already designed but it’s getting late so I’ll send that out perhaps tomorrow.

Thanks for all the help everyone! Will keep you posted!

Futureman · December 28, 2008, 5:00am

You sound like you know what you are doing, but for clarification for others that don’t…

You rectify, then smooth … not the other way around… feeding electrolytic caps with AC will eventually kill them.

Also, watch out with the actual values… 12volts AC, once full wave rectified & smoothed out is going to be much larger than 12volts DC.. more like 18-20volts DC…

Like wise, the 6 volts AC, once rectified & smoothed will be more like 9-10volts DC..

Regards

Mike

m00dawg · December 28, 2008, 5:31am

Turns out I couldn’t sleep. Too excited about all this stuff So I’ll venture a reply…

You sound like you know what you are doing, but for clarification for others that don’t…

Let’s not get carried away now Transformers are somewhat new to me, although I’ve done a few power boards at this point using a single wal-wart transformers.

You rectify, then smooth … not the other way around… feeding electrolytic caps with AC will eventually kill them.

Absolutely. Actually, the full steps should be:

AC -> Rectify -> Smooth - > Regulate -> Smooth -> DC

The values of the capacitors used for the smoothing seem to be magic I use 2200uF, 330nF before the regulator and 110nF, 10nF after. That basically covers most of the types of capacitors and should help filter out low and high frequency noise.

I’ll point out here that I’m no expert, however Those values were grabbed from the forums, experimentation, and just because I had extra capacitors I wasn’t using at the time. The current board I’m using (which is using chained rectifiers) does seem very quiet with this setup. I would say, however, that the average person wanting to make an MB-SID should just try to grab a C64 PSU

Also, watch out with the actual values… 12volts AC, once full wave rectified & smoothed out is going to be much larger than 12volts DC.. more like 18-20volts DC…

Like wise, the 6 volts AC, once rectified & smoothed will be more like 9-10volts DC..

Now that’s the part I’ve been confused about because I have heard it both ways. Rectifying AC actually causes loss of voltage, but the input AC voltage could be more than whatever is labelled depending on load. I’ve seen multiple equations (trying to find a good one to use as a base), but there’s a PDF I found from Hammond’s site (which I grabbed from Mouser while looking at transformers) that offers some numbers:

http://www.hammondmfg.com/pdf/5c007.pdf

I assume there is a relationship here between the rated current of the transformer and the current I’m actually using. So, I’m just guessing here, if I get a 12V transformer that can handle, say, 3 amps, I’m guessing it will actually give me more than 12VDC once I rectify it assuming I’m only using, say, 250mA. *shrug*.

The one thing I’m still curious about is how multi-tap secondaries work in this case. If I have 2 6V secondaries, I’m curious as to if I can combine them into 12V *and* tap off one to give me 6V as well. Basically, I’m wondering if I can use a similar configuration when using multiple-secondaries as when using a center-tap. I haven’t quite found good information on that yet.

stryd_one · December 28, 2008, 6:56am

The overheads needed for the regulators that LH mentioned are correct.

Single tap:

9V *1.414

AC -> Rectify -> 12VDC

|

Regulate to 9V -> Regulate to 5VDC (HEATSINK!)

| |

Smooth Smooth

| |

9VDC 5VDC

Multi tap:

(5V also maybe OK)

6V *1.414

AC -> Rectify -> 8VDC ----------------,

\

9V *1.414 \

AC -> Rectify -> 12VDC |

| |

Regulate to 9V Regulate to 5VDC

| |

Smooth Smooth

| |

9VDC 5VDC

HTH!

lylehaze · December 28, 2008, 2:36pm

Moo:

You are asking all the right questions. I don’t have “proper” training, but I’ll take a shot at a few answers for you, and if I’ve got something wrong, I’ll hope that a smarter person will correct me.

Capacitor values are “magic”: It’s true. To choose the right capacitors, always choose a voltage rating higher than your expected voltage, and a capacitance higher than you need. This is especially true for electrolytics, as they will lose capacitance over time. How much you need depends on load current, allowable ripple, and supply impedance. I’ll get back to current and ripple later. There is math involved in getting it figured out, or just shoot high and test the circuit.

Does rectifying AC cause a voltage loss or a voltage gain? Yes.

You will “lose” the forward voltage of the diode in the rectifier. This is usually 1 or 2 volts. But your meter will tell you the voltage is higher! To understand this, picture an AC wave. This is a sine wave going up and down in nice smooth waves. the centerline will be ground. half the time, the wave is abnove ground, half the time it is below. Now we will add a diode to “half wave rectify” it. Two things happen: The positive voltages will be about 1.5 volts lower (diode forward voltage drop) and the negative voltages will flatline at ground (reverse biased diodes do not conduct). It is no longer AC, as it has no negative part. It is 0 Volts half the time, and somewhere above zero half the time.

With me so far? Now we add a big capacitor to “smooth” this pulsing voltage. What happens now depends on the load we are supplying. If there is no load at all, the cap voltage will rise to the top of the voltage peak, and there will be (almost) no ripple, as there is no load to drag it down. But as we add a load to the supply, the capacitor will discharge during the times between the voltage peaks. This will cause some “ripple” in the voltage readings. The greater the load, the more the voltage will fall between the peaks. Also, a larger capacitor will give less ripple in the voltage.

Obviously, any transformer will only reach it’s “peak” voltage for a small percentage of time. It spends most of the time swinging back and forth between the peaks. so the voltage rating of the transformer is not the “peak” voltage, but the “Root Mean Square” of that peak. This is a more realistic description of the total power available. That’s why converting AC to DC never comes out as easy as you think it would. The results depend on a lot of details that are different for each project.

There was a java applet somewhere that draws this all out.. I don’t know where it was though.

I hope this helped..

LyleHaze

m00dawg · December 28, 2008, 6:25pm

Yes that does indeed make a bit more sense! I knew transformers were measured in RMS, but I never thought to actually look up what that meant in terms of this particular instance So based upon your information, and the magic number (1.414) that stryd_one provided, it looks like I should be good to go with a 10V center-tapped transformer. If I went with a 12V, that would be almost 17VDC I would be feeding the 7809, which seems a bit high I think

Thanks for everyone’s help! I’m still working on the power board design and how to cram all this inside a power-brick style case. And I still need to order the parts. But I’ll try to keep everyone posted as I progress! Seems like something like this would be worlds better than a C64 power supply and it’s an excuse to build something else

stryd_one · December 29, 2008, 7:52am

There was a java applet somewhere that draws this all out.. I don’t know where it was though.

Was that on smash’s site? I have this vague recollection…

Can’t remember if it was nils or bugfight or smash or who told me, but I can’t take credit for that *1.414 magic number… That said, if you google it, you can confirm that it’s frequently-used voodoo (and the math behind it which is nice to know if you’re a geek), so I’d say it can be trusted

m00dawg · December 29, 2008, 1:55pm

I believe you I’ve seen that number used before. As for the Java applet thinger, Smash has an AC->DC graph thing that’s pretty neat, but it’s probably not what lylehaze was talking about. Smash’s just shows the current in various states after rectifying, filtering, etc. Still, it’s a really nice site. Unfortunately, I don’t have the link to that one either

Jack · December 29, 2008, 9:30pm

Hi,

I am planning to build a MB6582 power supply, too.

I used this “NT-25-09” and “SPR-2” from “www.thel-audioworld.de” professional as References design:

http://www.thel-audioworld.de/module/NT1525/NT1525.htm Transformator Modul

http://www.thel-audioworld.de/module/spr/SPR-2.htm Audiophiler Voltage Regulator

See attachement for my design suggestions.

RingCore Trafo Talema 2x115V primary and 2x9V secondary
78S05 and 78S09 Regulators
Cap for smotthing 4x Panasonic FC 1000ÂµF/63V

Best Regards

Jack

m00dawg · December 29, 2008, 10:52pm

So let me preface by saying I tried to translate that page to English and Google sort of failed on that one So I apologize if it’s already answered on the provided links:

Why did you opt for so many 1000uF capacitors, over, say, few larger ones? I know that electrolytics can be slower to react than other caps (which means that you still end up with some wobble, hence the need for additional caps) so is it just smoothing it out further? Not that there’s anything wrong with more caps, mind you!

Are those vertical red rectangles (on the primary side of the xformer) MOVs?

Finally, what dual primary and secondaries over, say, a center-tapped transformer? Reason I ask is that you can get 10V/5V AC with a center-tap, making life better for your 7805 but beyond that, I don’t know the benefits of a center versus dual primary/secondary transformer.

Jack · December 29, 2008, 11:09pm

Hi,

I am not shure because I slept in school.

My Theory:

Those caps have a tolerance up to 20%.

So more caps should make it more smooth as few caps.

But we should wait the answer from stryd_one.

The Red ractangle are 0Ohm Resistors for 230V respectively 115V primary power supply.

What are MOVs?

The overvoltage is not so problemtic because I use 78S05 and a big heat sink.

http://www.reichelt.de/?;ACTION=3;LA=2;GROUP=C85;GROUPID=3382;ARTICLE=22252;START=0;SORT=artnr;OFFSET=1000;SID=31LpgomqwQAR8AAB7VZUQ6b83af634c0161cfab3b9a7d11c2e743

Best Regards

Jack

m00dawg · December 29, 2008, 11:25pm

I am not shure because I slept in school.

At least you’re honest

My Theory:

Those caps have a tolerance up to 20%.

So more caps should make it more smooth as few caps.

But we should wait the answer from stryd_one.

That’s sort of what I was thinking myself, though I slowly stepped mine down. Since the caps in parallel, the total capacitance will be greater either way. So I assume the object of the exercise is to find the best size and number of caps that produces the lowest noise while still providing enough capacitance for stable power.

What are MOVs?

Metal-Oxide Varistors. A friend from work suggested using these before the primary side of the transformer to help protect against power surges. The keyword is help They aren’t foolproof, but relatively inexpensive and would help protect those priceless SIDs. (Power-strips use MOVs if I’m not mistaken).

The overvoltage is not so problemtic because I use 78S05 and a big heat sink.

http://www.reichelt.de/?;ACTION=3;LA=2;GROUP=C85;GROUPID=3382;ARTICLE=22252;START=0;SORT=artnr;OFFSET=1000;SID=31LpgomqwQAR8AAB7VZUQ6b83af634c0161cfab3b9a7d11c2e743

I was planning on using a heatsink similar to that as well, though might be overkill for mine. My current power board actually uses a single 12VAC wallwart and uses a 7809 for the 9V and 7809 passed to a 7905 (probably unnecessary since both needs to be cooled). All that is under a big heatsink I took from a dead motherboard. This solution actually works and is surprisingly quiet but I was hoping to get better efficiency using a center tap transformer.

Regards & Happy New Year Back At You!

stryd_one · December 30, 2008, 7:09am

Well firstly I wouldn’t wait for my answer on this one… Analog electronics (including PSUs) are not my forte. I just share what I’ve learned so far, but I’m no expert.

As was mentioned in the first post, that design is aimed at audiophile use. TBH, it’s probably overkill for a crappy old 8bit analog synth that’s well known for being a noisey SOB. But it won’t hurt The audiophile factor explains the number of caps there, as yeh, it’ll help smooth things out a bit. (also explains the use of the illustrious panasonic blackcaps) I’ve that n-number-of-caps will smooth better than a single cap of the same total capacitance, because the caps don’t all ‘pulse’ outwards simultaneously, rather, they kinda pass the excess electrons from one cap to the next, each time smoothing it a little bit.

Edit: oh, the boxes on the left of the tranny… I’m not sure you’re meant to mount them all? F1 is the fuse, that much I’m pretty sure of… If you mounted *only* the 230V ‘component’ (and *not* the 115V ones), then the two primaries would share 230V in series, which is correct because they’re 115V primaries. If you mounted *only* the 115V* “components” (and *not* the 230V one) then the two primaries would be in parallel, which is right for 115V input… I’m thinking that those little boxes are actually meant to represent a mains voltage switch?

As LH mentioned, getting the right cap values is a bit voodoo. Sure, there’s math to it… if you know the equations please share, because I have no clue I tend to KISS, and just use a well known PSU design that someone smarter than me figured out