Wondering about PSU circuit / 78xx behaviour

[Flying Panther]

I was about to test my 'optimized PSU' board (for a MB-SID) with the C64 adapter, but the C64 adapter was kind enough to fail on me. Therefore, I changed the DIN-connector with two connectors that will fit standard wallwarts. Schematics from my PSU-board are attached (Addition to the schematics: The capacitors right after the bridge rectifier are 2200uF/25V (electrolytic) and 330nF (ceramic). Also the capacitors between the 15V and 7812 / 7809 may not be essential, but that was something I realized after soldering the board.).

I also managed to find a 5VDC and a 10VAC wallwart adapter. The 5V one can deliver about 1A and delivers 5.20V exactly. The 10V one however can deliver about 700mA (which is not enough for all 8 SID boards, but enough for testing purposes), but gives me 12VAC when I measure it. This causes the (heatsinked) 78xx's in the power circuit to get very hot (which isn't strange, because the 7812 needs to dissipate 17-12 = 5V and the 7809 even more, 17-9 = 8V!) When the power is switched on, the voltage of the 10V adapter drops to about 11,5V, which seems normal to me, since it was made to deliver about 10V.

What I found to be strange is that the voltages after the 7812 and 7809 read 13.3 and 11.1 respectively, where I expected it to be around 12V and 9V. The 5V output reads 5.16V so that's correct.

Is this because the 78xx's get an input voltage that is too high? Or is it because I'm not drawing any current from the circuit, and will the voltages lower when I connect some of the SID boards? Or did I just mess up the schematics and did something else wrong?

[lylehaze]

A few problems:

Running 5 volts straight out of the wall wart without any local regulation may be a problem.

Few of those supplies are regulated as well as would be needed. Since that 5 volts is

your logic supply AND the "bottom" regulator for all other voltages, it should be as stable as possible.

The problem is, before you can add a 7805 there, you'll need at least 7 or 7.5 volts to feed it. depending on the manufacturer, there is up to 2.5 volt dropout across the 7805.

Another problem: The LED you have across the 5 volt supply needs a resistor in series, or it won't last long.

Finally, for the voltage tests: If you are providing reasonable input power, most 78xx series regulators are fairly precise, being within 5% of their rated voltages. Try measuring each ones output voltage compared to it's own center pin, and you'll probably get just what you were looking for. The rest is up to your circuit.

If the readings are still high, check your meter batteries.

Have Fun,

LyleHaze

[Flying Panther]

Thanks for the in-depth reply!

About the regulator in the 5V path, that is something I completely forgot about when changing from the C64 adapter supplied 5V to a 5VDC wallwart. I have to reconsider that part, thanks for pointing that out! The plan was to use this simple board (with C64 adapter) to test the Core and SID modules when they are completed. When I get to the enclosure and control surface (as I know by then what currents are pulled from the power circuit), I want to make a transformer-based power supply that convert 230VAC to a lower value (15 or 12V) and then regulate those voltages to the needed 12, 9 and 5V.

Do you suggest adding a 5V regulator also for these 'temporary' purposes? Or could I leave this for now and do it right when making the transformer-based power-supply?

I used a resistor with the LED, but somehow I seem to have forgotten about it in the schematic (I drew this schematic after I made the board, since I figured no-one would have understood my hand-drawings :P).

About the voltages, when I measured between the 78xx output pin and center (which is connected to the common ground), I read the same voltages (13.3V and 11.1 V). But I also discovered something else (when closer inspecting the 78xx in and outputs) that is a bit more trivial. I switched the in- and outputs of all 78xx when soldering them onto my board. That probably also explaines why they generate so much heat, while no current is drawn from the circuit. I'll try to fix that and measure everything again. Hopefully that fixes my problems ;)

[lylehaze]

Do you suggest adding a 5V regulator also for these 'temporary' purposes? Or could I leave this for now and do it right when making the transformer-based power-supply?

Myself, I would not power ANY chip that I cared about without a decent regulator.

I used a resistor with the LED, but somehow I seem to have forgotten about it in the schematic

Happens all the time.. :-)

I switched the in- and outputs of all 78xx when soldering them onto my board

That explains a lot. Many brands of 78xx specifically state that applying a higher voltage at the output exceeds "maximum specifications", meaning that if it's toast, it's your fault.

Congratulations, you have just voided your warranty!! :-)

In the long run, it might be cheaper/safer/easier to build the "real" power supply first.

Have Fun,

LyleHaze

[Flying Panther]

Myself, I would not power ANY chip that I cared about without a decent regulator.

That's a fair point.

That explains a lot. Many brands of 78xx specifically state that applying a higher voltage at the output exceeds "maximum specifications", meaning that if it's toast, it's your fault.

Congratulations, you have just voided your warranty!! :-)

I guess I couldn't return them anyway.. :P I switched in and output pins (which made a complete mess of the whole board..) and now got 5.20V, 8.98V and 11.89V. That's like it should be, so the board now works as it was intended.

In the long run, it might be cheaper/safer/easier to build the "real" power supply first.

You're probably right.. I already had some ideas, so I better put some more effort in those ideas and have the power supply done (instead of fixing this board with a properly regulated 5V). I think I'll have some questions about the "to be designed" power supply, but that's something for 'Design Concepts' I guess.. But first, there's more reading to do :)

[m00dawg]

+1 for building a "real" power supply. Not only is it a fairly simple project which offers some nice learning opportunities, but you can build a linear (as opposed to switched) supply, which means it'll likely work better (and sound better).

At some point I'm going to get a nice printed board for taking a 9VAC transformer and producing 9V and 5VDC. When I do, I'll be sure to post it. But you can make one fairly easily on a protoboard.

[nILS]

I feel the need of writing that everything I hear "transformer": Careful with mains voltage! 9V kinda tickles, 110/220V hurts. Bad. Do that only if you really know what you're doing and make sure to include fuses in the right spots.

[m00dawg]

Well yes that's important, but all you need to do is be mindful. Make sure to always check and have the power disconnected (from the wall) whenever working on the power supply. If you have to work on the PSU live (hint: don't) always use you right hand only and put your left hand in your pocket. That way, if you get zapped there's less chance it will, you know, do things like stop your heart.

I put fuses on both the positive and neutral wires before the transformer (even have an extra fuse on the power plug if the others somehow fail). For me, I chose 1 amp fuses which seems to work well for preventing damage when I do something stupid that short something while I'm testing things with my multimeter.

To prevent doing something stupid, I used the flat crimp connectors that fit the prongs on my transformer and my fuse holders to help prevent accidentally coming into contact with one of those guys (of course, if you follow the suggestions above, that would be a non-issue).

After that, it's your typical power supply. Rectifier -> Capacitors -> Regulators -> Capacitors. The forum is full of information about how to build one of these and there is no shortage of design suggestions, so I'd do some searching if you're not sure how to proceed.

As a hint, I tried doing crazy things with a center-tapped transformer. I'd recommend not trying to go that route and, instead, get a 9 or 12VAC transformer and regulate down to 9V, 5V, and 12V (if needed). In my case, I'm feeding each regulator directly from the transformer. You could stagger them (9V regulator's output is the input for the 5V regulator) but that seems like that would put a hefty load in the 9V regulator and it's not doing anything for heat - you still need to heatsink the things.

I did that by grabbing a spare heatsink I harvested from a north bridge of a PC motherboard. Seems to work well, though for my final design I haven't decided what I should do. Making printed boards gets expensive and so I don't really want to waste space on a board to fit a huge heatsink. Either way, do be sure to heatsink the things.

Have fun and good luck!

[Flying Panther]

I have combined all the thoughts I had and information I harvested (from the MB forum, wiki and some other sources) about power supplies and components and the attached schematic is the outcome (please see attachment). The following parts will be used:

SW1: 250V Power Switch

F1: 250V 6.3A Fast-Acting Fuse (standard fuse already in the power-connector, replace with a 230V 0.3A?)

F2: 250V 0.1A Fast-Acting Fuse

F3: 250V 0.1A Fast-Acting Fuse

F4: 250V 0.1A Fast-Acting Fuse

T1: 230V-9V 16VA (1778mA) Print Transformer

T2: 230-12V 16VA (1333mA) Print Transformer

T3: 230-2x15V 16VA (2x533mA) Print Transformer

BR1: 2A 400V Bridge Rectifier (2W04G)

BR2: 1.5A 400V Bridge Rectifier (W04G)

BR3: 1.5A 400V Bridge Rectifier (W04G)

C1: 4700uF/16V Electrolytic Capacitor (maybe even 6800uF/16V or 4700uF/16 and 2200uF/16 in parallel?)

C2: 330nF/50V Ceramic Capacitor

C3: 10uF/50V Electrolytic Capacitor

C4: 100nF/50V Ceramic Capacitor

C5: 4700uF/16V Electrolytic Capacitor

C6: 330nF/50V Ceramic Capacitor

C7: 10uF/50V Electrolytic Capacitor

C8: 100nF/50V Ceramic Capacitor

C9: 2200uF/25V Electrolytic Capacitor

C10: 330nF/50V Ceramic Capacitor

C11: 10uF/50V Electrolytic Capacitor

C12: 100nF/50V Ceramic Capacitor

C13: 2200uF/25V Electrolytic Capacitor

C14: 330nF/50V Ceramic Capacitor

C15: 10uF/50V Electrolytic Capacitor

C16: 100nF/50V Ceramic Capacitor

IC1: 78S05 Voltage Regulator

IC2: 78S09 Voltage Regulator

IC3: 7812 Voltage Regulator

IC4: 7912 Voltage Regulator

I think some explanation is required, maybe not so much for the schematics, but more about the thoughts behind them.

First of all, the power supply needs to supply 5V (1A-1.5A for cores, SIDs, LEDs and LCD), 9V (800mA for SIDs) and +12/-12V (200-300mA for op-amps in a mixer circuit and maybe some AOUT modules).

I'd like to have the 5V on a separate transformer, instead of using the voltage supplied by the 7809 for the 9V rail. This because currents drawn from the 5V rail are relatively high, and all that current has to run through the 7809 as well. The +12/-12V circuit works best with equal loads on the + and - rails. Therefore it's best to give them a seperate transformer as well, and not regulate the +12V down to +9V for the SIDs. This means the 9V rail gets a personal transformer, which I don't really mind. It's a bit bigger using 3 transformers, but I'm not planning on a small case so that will be fine.

Then, I also want the transformers not to run at full capacity. Therefore I chose the ratings a somewhat higher, so they will run at about 2/3 of their capacity and have a safe, happy and hopefully hum-free life :)

For the rest of the components used, I tried them to match the transformers, not the actual current drawn from the circuit. This way all components have some 'safety headroom' and might make me sleep better at night. Also I don't have to worry about adding some extra buttons / LEDs, since I have enough capacity anyway.

Also I have some other considerations:

1st: what about (more expensive) toroid-transformers? They are said to be better than regular transformers when working with audio, because they induce less ripple/hum. Is it worth looking at those (and redesigning the scheme, because 3 toroid transformers is too much), keeping in mind that ripple and hum is already reduced (a lot) by the regulators and capacitors, and knowing that the SID is noisy anyway?

2nd: Why not save me all the work and buy a nice switching power supply (for at least the 9V and 5V rails) I came across a while ago on Mouser. Judging from the datasheet and the amount of capacitors on the board they are regulated quite well, but I also read a lot about more high frequency noise. They are a bit more expensive compared to transformers, but also save parts and work. However when additional ripple reduction should be needed, the same amount of work and more money is involved, hence it's not worth it. Still not sure though...

Pfew, that was a whole letter, if you're still reading: thanks!! All comments and opinions on the above are welcome :)

[m00dawg]

In regards to a switching supply, I used to use one without issue and others on the forums have done so as well. However, you're dealing with an analog audio generator (the SID) so high frequency noise can be an issue. If you opt for a switching supply, I would still regulate it and add in plenty of filtering capacitors to smooth out and hopefully eliminate the HF noise.

As for torids, I haven't tried these before because they are so goddamned expensive :) I bet they'll give you superior clean power but will you notice that from a noisy vintage analog audio chip? Probably not :) I'm sure there is plenty of room for debate about it. If cost weren't an object, I'd go for it, but I don't have any concrete evidence to prove one way or another if they will help.

[lylehaze]

Everything I have heard so far is good.. I'll add a few more opinions, you are welcome to take or leave them. As with most opinions, some people may not agree. I'm not always right, but I try to stay open minded.

Three transformers seems like a bit much, but your reasons why make good sense. I like making the 5 volts off a "private" transformer for a few reasons: it lets you reduce heat, by not stepping down so far from a higher voltage. Also, since the 5 volts is often "digital" and the others are "analog", keeping those separate reduces audio noise. Finally, you often need to support more current for 5 volts than all others combined, so if one was to be separated out, that would be the one.

toroidal transformers? I believe (and I could be wrong) that they put off less of a stray magnetic field,  but I don't think it's an issue that is worth going after. I do NOT believe they offer less electrical noise, and your power supply will be the factor that determines that.

Switching supplies? great technology, but they can add high frequency switching noise to the power rails.

If you're building an all digital and MIDI device, like maybe an MB64 that has no audio circuits, they are a great choice. but if there's audio being made, well.. I guess I'm old fashioned that way.

Analog regulators make heat, but they don't add noise to the power supply, and another bad thing about switchers is that they are not usually repairable. (at least not easily)

Deliberately over-rating all the parts in your power supply? I thought you were new at this. THAT sounds like someone with lots of experience. I think it's a very good idea.

Power supply safety? Absolutely. Be VERY cautious anywhere near mains voltage. I like to build the power supply first, and once it's working and tested, I enclose it, or insulate it, or do whatever it takes to make it completely consumer-safe (good enough to let grandma use it). Once that is out of the way, then I can relax a bit as I build out the rest of the project.

One last, tiny detail. I understand that your "mains" are done different than ours, but if you can get a "DPDT" or "DPST" switch for the power supply, then you can cut-off BOTH sides of the AC, just in case some fool electrician didn't connect it right (or the plug is in backwards).. that may not apply with your mains, but here in the US, I trust nobody.

Oh, one other detail (you'll love this) 78xx regulators (positive) and 79xx regulators(negative) have different pinouts.. On the 78's it's  IN, GND OUT, but on the negative regulators it is GND, IN OUT.

Since the middle pin is connected to the tab, that also means that the heatsink of the 7912 will be at some negative voltage.. Check your datasheets, do NOT trust my memory.

Sometimes the devil is in the details, but you knew that already. :-)

(Now I'm waiting for a post telling me what an arsehole I am.... It's true!)

Have Fun,

LyleHaze

[Flying Panther]

I feel the need of writing that everything I hear "transformer": Careful with mains voltage! 9V kinda tickles' date=' 110/220V hurts. Bad. Do that only if you really know what you're doing and make sure to include fuses in the right spots.[/quote']

230V is to be watched from a safe distance ;) I will try to put all parts that involve 230V together and isolate that part from the rest to avoid accidental 230V touching (while keeping in mind that sufficient cooling is provided to the transformers).

[m00dawg]

Ooh that board looks really pretty for a protoboard nice job! Those heatsinks do look a tad small. My plan for my next power board is to use the heatsinks with the regulator standing up. They make ones that sort of wrap around the regulator that seem to work well.

Having said that, I really like using my harvested heatsink from a motherboard. It's but can is able to cool both my regulators. You might look at going that way, but it does consume a fair bit of space. I was also thinking of trying some heatpipe stuff. If your case is metal you can try to use it to expel heat if the heatsinks themselves won't do. I dunno how well that would work but something to consider.

I hadn't though of splitting 5V and 9V to improve noise. The transformers I have found seem rather large though so I don't know if that will work for me, but seems like a really good idea! Curious to see how things end up working out if you go that route!

[Flying Panther]

After reading both your comments on the considerations, I strongly lean towards the transformer-based power supply, and not using switching power supplies and/or toroid transformers. The advantages over regular transformers are obviously not that high, and still additional regulation / smoothing is needed. About the three transformers, I agree that it is a lot of transformer :P I'll let that comment sink in a bit, maybe I can find a solution to regulate 9V out of the same transformer as the +12/-12 without disturbing the equal loads and without needing a transformer that is supplying above 1.7A.

One last, tiny detail. I understand that your "mains" are done different than ours, but if you can get a "DPDT" or "DPST" switch for the power supply, then you can cut-off BOTH sides of the AC, just in case some fool electrician didn't connect it right (or the plug is in backwards).. that may not apply with your mains, but here in the US, I trust nobody.

Great idea! We can plug in however we like (no earth-pin to secure that you'll always plug in the same way), so I'll take that suggestion. Switching one side off should do the trick (even when plugged in backwards), but for safety-reasons it's still better to switch them both. Makes me wonder though, should I fuse both sides as well?

Oh, one other detail (you'll love this) 78xx regulators (positive) and 79xx regulators(negative) have different pinouts.. On the 78's it's  IN, GND OUT, but on the negative regulators it is GND, IN OUT.

Since the middle pin is connected to the tab, that also means that the heatsink of the 7912 will be at some negative voltage.. Check your datasheets, do NOT trust my memory.

Sometimes the devil is in the details, but you knew that already. :-)

*makes note to self to tripple-check pin-outs on the datasheet and connections on the board*

(Now I'm waiting for a post telling me what an arsehole I am.... It's true!)

I'll even that one out with the comment about over-rating the parts ;)

Ooh that board looks really pretty for a protoboard nice job! Those heatsinks do look a tad small. My plan for my next power board is to use the heatsinks with the regulator standing up. They make ones that sort of wrap around the regulator that seem to work well.

Thanks :) A lot of planning was involved to fit it all on the board.

Having said that, I really like using my harvested heatsink from a motherboard. It's but can is able to cool both my regulators. You might look at going that way, but it does consume a fair bit of space. I was also thinking of trying some heatpipe stuff. If your case is metal you can try to use it to expel heat if the heatsinks themselves won't do. I dunno how well that would work but something to consider.

Interesting idea! Heatpiping is a nice way to move heat through a place where more cooling is possible. Although I don't think that will work if I connect the 7912's heatsink to the 78xx's (if I understood lylehaze correctly)

[m00dawg]

Interesting idea! Heatpiping is a nice way to move heat through a place where more cooling is possible. Although I don't think that will work if I connect the 7912's heatsink to the 78xx's (if I understood lylehaze correctly)

Correct. You can use a single heatsink for the 78xx's as long as they share a common ground. You can't do that with the 79xx since you're dealing with negative voltage here. So the 79xx might be a special case.

As an aside, the C64 Optimized PSU design supplies 5V to the ground pin of the 7809 to bump the voltage up to 14V so, in that case, you also would not be able to share the same heatsink between the 7805 or 7809.

[lylehaze]

if you want to share a big heatsink with the negative regulators, there are ways..

There is a LOT of good power supply information in this thread..

and it seems to be a common discussion around here.

I need lunch.

LyleHaze

[Flying Panther]

Those mica insulators look really nice :) Might be worth using them as well when only combining heatsinks for 78xx's to avoid groundloops. (Maybe keep one 78xx un-insulated to have the heatsink grounded).

Regarding to the power supply, I think I'll stay with the three transformers. Still two things that I'm not sure about. What about the fuses? Will 0.1A be enough? When you convert secondary current to primary current, you'll get about 70mA, I've read somewhere that fuses should 'work' at about 75% of the total allowed current, so for a 0.1A fuse that would be 75mA, which is close enough to the 70mA I have. But I don't know if there will be peak currents when for instance powering on the circuit that'll go over the 0.1A rating and blow the fuse. Or should the 0.1A fuses be okay (in theory)?

And what about the ratings for the capacitor on the 5V line? is 4700uF enough for a maximum current of 1.5A? or should I better go for 6800uF or the 4700uf / 2200uf in parallel?

Editing reason: nF / uF is not the same ;)

[lylehaze]

For the fuses, you have the right idea. It's not something to lose sleep about.

Each fuse needs to be rated above the maximum operating current, and somewhere

below 300% to 500% of the operating current, as long as it is still fused below any

safety issues.(not a problem with ma DC levels)

Fuses are meant to protect from extreme overloads, as might be encountered during

a dead short or complete failure of some system.

If your system normally draws 70ma, then a 100 ma would be great. If all I could find

was 0.25 amp or even 0.5 amp, I'd use those without worry as long as the power

supply was capable of blowing them if something extreme happened.

Capacitors? Enough is good, a bit more is better. Electrolytics will derate with age,

Tantalums will explode if voltage spikes.  going crazy over-filtering the supply will

create large inrush-currents when you power up. Past that you're on your own.

:-)

Have Fun,

LyleHaze

[Flying Panther]

Thanks!! I think I'll draw a final schematic, and then buy me some parts and see how things work out :)