Big Power Supply - circuit proposal

[rutgerv]

---MODIFIED: A final working version of the PSU design can be found below!---

Hi everyone,

I'm working on a rather big 8xSID module with 8 seperate Moog filters controlled by AOUT_LC modules. For this project the modified C64 power supply may not be powerfull enough, so I designed my own and I was hoping that some of you could double-check my design and tell me if everything is allright.

The circuit is attached to this message as JPEG. Some comment with it:

- To the left the 230V power input connector (pin2 is ground)

- TR1 = Transformer(230V to 9V), TR2 = Transformer(230V to 2x15V)

- The TR1 output drives 2 independant 5 volt regulation circuits, this is because

I want to separate the digital and analog 5V lines as much as possible and I want to distribute the current load over two 7805 regulators (gets less hot than 1).

- The TR2 output drives a +12V and -12V regulation circuit and the +12V output is fed into a +9V regulation circuit.

This gives in total:

+5V    (= 5V line for digital stuff)

+5V/1 (= 5V line for analog stuff)

+9V   (=for SID modules)

+12V  (=for SID modules and op-amps)

-12V   (=for op-amps)

AGND  (=ground for analog stuff)

GND    (=ground for digital stuff)

I have two questions:

1.  Is the wiring and circuit okay?

2.  Is it alright two join all grounds at the power supply? I have joined the relative ground of all regulation circuits with each other and with the earth pin on the power inlet.

Greetings,

Rutger

[Flying Panther]

This is interesting.. I was also thinking of designing my own PSU with +5 (cores/other modules), +9 (sid) and +9/-9 to feed an op-amp for an active mixer. I was going for just one 2x12V transformer, but you made me doubting again. ;)

I think you can join grounds this way (I was thinking of joinging them in about the same way), but I'm not sure myself if this is right, since I don't know much about this kind of electronics.

[rutgerv]

Hi Flying Panther,

thanks for the message! My main reason for this design was to keep down the heat and the physical size of the transformers and still deliver quite some current. At the moment I've estimated that my SID PCB will draw 1500mA on the 5V together, but I'd like to have some headroom so I'll put in a 1.8A transformer or something. Dividing the 5V lines over 2 regulators keeps down the heat on each of them and delivers a clean analog 5V line. On the +12 and -12 rail I estimated that the filters and headphone buffer use about 320mA. The SID's will draw either 320mA extra on the +12 rail or on the +9 rail (depending on a jumper setting near the SID socket). The 2x15V transformer gives about 1A at each rail, so that plenty of headroom.

I'm planning to exchange the diodes with a B40C5000 rectifier and since it can handle up to 3.3A the 5V lines may share the same rectifier (to keep down the costs and number of components). I also plan to add fuses after each transformer to protect the transformers against overheating, melting, etc, instead of any trouble in the circuit. The powersupply will be added to 1 single PCB containing all components of my SID synth. One thing that i'm a little bit scared about is that I have to add a 230V connector and PCB traces to this PCB to connect the transformers. I'm not really happy about that....for safety reasons, but I have limited options in a 2HE rack space.

Greetings,

Rutger

[TheAncientOne]

In the 5V side of your design, I would suggest just one big bridge rectifier and (say) 4700uF, feeding two 7805's. A  100nF ceramic across the 10uF's wouldn't hurt either.

Did you know that you can get 7805S, which will supply 2Amps?

The reason for using 1 bridge is that the diodes may vary in the seperate bridges, and there might be a wierd ground current. It's a bit cheaper too.

I much prefer seperate transformers for the 5 and 12 volt supplies.

Mike

[Altitude]

2nded about the 1 bridge rectifier for the 5 volt rails.  Also, your transformers are rated really high for what you are doing voltage wise.  Remember that voltage rating is RMS so you will have a 1.4V drop across 2 diode in the bridge (7.6 V RMS) and with your smoothing caps that will raise that to almost the peak of 10.6V which is a lot of voltage for the 5V regs to burn off as heat.  Consider using a 6V transformer.  Plus the way you have your grounding set up is wrong.  You can not just tie all the grounds to case, I am sure someone more knowledgeable than I will explain why but the general idea is that with 9v transformer, you are referencing ground to the lowest point of the wave and the with the center tapped one, you are referencing it in the middle of the wave (thats how you get your +/-). With a mixed signal chip like the sid, this will be VERY bad to have 2 different grounds.  Plus your signal ground and case ground should not be tied together. 

[rutgerv]

Hi,

wow, these tips are of great value! Thank you very much. I've already modified the design of the 5V rails to use 1 bridge rectifier and I like the suggestion of the big 4700uF cap and the extra 100nF ceramics.

I have considered the use of a single 7805S (2Amps), but I'm not sure if I can get rid of the heat properly...I'm afraid that regular heat sinks will not cool enough to get the 2Amps from it. Anyone experienced with the 7805S?

Finally I'm open to suggestions about the ground-connections. My feeling tells me there's something wrong with it. Somehow I'm afraid that there will be a difference in potential between the ground that comes from the TR2 and the ground that comes from the TR1. Connecting them seems scary to me.... does someone know for sure how the grounds should be connected in the case of two transformers?

Greetings,

Rutger

[Altitude]

Rutger.

  I updated my msg. Usually, you will want a transformer with dual secondaries, using 2 transformers with one referenced ground can be problematic.  Search this forum for some various PSU designs there are several including mine.

I use the 2Amp 7805 and it works well

[rutgerv]

Hi Altitude,

thank you for the clear explanation in your updated post! I already had the feeling that this would become a problem...but what would the solution be? Is it impossible to use multiple transformers with a common ground?

About the transformer for the 5V you advised to use a 6V/AC rated one...how about the 2x15V transformer? Is this overrated too?

Greetings,

Rutger

[Altitude]

Here is some reading that may clarify some things:

http://www.kpsec.freeuk.com/powersup.htm

[TheAncientOne]

Just noticed something else: at TR2, if the windings are normal, then the two centre points should be joined, so as to get antiphase, about the centre tap, on the two ends. This is sometimes only checkable with a dual trace 'scope! Otherwise you won't get a  smoth +/- voltage ripple. This depends very much upon the transformer used. most 0-12 0-12 rtranformers can be wired as 12-0-12. Check the data on the transformer itself (sometimes there is a picture), and make sure your centre connection is the '0' of one winding connected to the '12' of the other.

Dropout Voltage on a 7805 is usually 2 volts, so the minomum input voltage, with max current draw has to be 5 Volts. You really need some margin here, and I would never feed less than 8 myself.

If you need to read up on this, the very best, "No bull****, here's the electronics" book is called 'The Art of Electronics" , by Horowitz and Hill. There is a website here: http://www.artofelectronics.com/index.html

Chapter 6 deals with power supplies, and you can have fun spotting what is wrong with 'Bad Circuits' at the end of the chapter.

Mike

[rutgerv]

Hi guys,

to keep things going I've been working on a new design today. As Mike pointed out was an error in the connection of TR2, this is now corrected. I also replaced the diodes with a bridge rectifier B40C5000 (capable of 3.3A without heat sink).

So, attached to this message a new design is added. I still do NOT have a solution for the grounding problem. I don't want to use multiple grounds at different potentials within the circuit, so i'm very open for suggestions to make this work with two transformers, one of which is center tapped.

Greetings,

Rutger

p.s. Perhaps joining grounds would not be a problem if both transformers were center tapped?

[TheAncientOne]

Looks good to me. Star point grounding is usually best for this type of kit - keep the main ground junction in one place, and use decent cable.

For some audio uses, a 'ground lift' switch may be needed to stop hum loops. Hard wire the mains safety earth to the metal of your casework, transformer exposed metal etc, but have a switch, (or moveable link) between that and the electronics earth. The switch should not be 'immediately accessible' - like on the back panel or under a cover. Some pro kit has a 'ground lifted' light on the front panel, so if you do get a horrendous hum, you may know the symptom. Stryd may be the best nameity on this one I think.

Mike

[lylehaze]

Sorry I'm late, I've been out of town.

Very nice design, I have a few comments, if I may:

I see no problem with the grounding arrangement at all.

The difference is correct, as the difference between a

Single supply and a bipolar one.

Heatsinks: All of the positive regulators (78xx) have

their center tabs grounded, so no problem mounting them directly to the grounded metal case. do NOT do this with the negative regulator. Use a dab of thermal grease for better heat conduction.

There may be a problem between the +12 output and the 9 volt regulator. Having 2200 uf on the LOAD side of the 7812 may cause a problem when powering down. It may briefly have a higher voltage on the output of the 7812 than on the input. This is not good for the regulator.

2 possible solutions:

1> place a diode across the 7812, cathode to input, anode to output. this will conduct only when output voltage is above input voltage.

2> supply the 7809 from the same V+ as the 7812, instead of feeding it regulated 12 Volts. Yes, the 7809 will get a little hotter, and the 7812 will stay a little cooler.

Personally, I prefer the second solution.

In any case, nice design!

Have Fun,

LyleHaze

[rutgerv]

Thanks for the response, guys (and girls?)!

I've worked on the schematic again. I still got the 9V regulator behind the 12V, but I've taken out the 2200uF cap and the 5V now comes from a single 78S05 (2 amps). I will be testing this powersupply within a couple of weeks and I'll let you know of it works! If it does and behaves as expected I'm willing to share the latest (known to work) schematic and possibly redesign the PSU to fit on a small single sided PCB. This will probably be a nice contribution to the people who want to build a big MBSID V2 and run into the limits of the C64 PSU.

Greetings,

Rutger

[rutgerv]

Hi everyone,

today I did the first test-run of the power supply proposed in this topic. Based on this test-run I can confirm that the power supply is working perfectly! I did not yet have the chance to push it to its limits and see how it handles, but I can confirm a stable voltage on all outputs!

This is the latest schematic:

http://members.home.nl/r.j.vlek/psu2.jpg

It can deliver at most (but that is if the transformers can supply it):

+12V (1A)

-12V (1A)

+5V (2A)

+9V (the maximum depends on the current drawn from the +12V. If the +12V is not used the +9V can deliver 1A maximum.).

Note: X1 is a connector with 6 pins. Pin 1 and 2 are connected to a 1x9V transformer. Pin 3 is earth. Pin 4, 5 and 6 are connected to a 2x15V transformer where pin 5 is the center tapped one. Pin 1 and 4 should be in fase.

In my design the PSU schematic is part of a big PCB with an 8xMBSID + 8xMoog filter. The transformers are on a seperate board because I did not want 230V on my main PCB.

Anyone interested in more details: feel free to send me a private message.

Kind regards,

Rutger

[c0nsumer]

Just curious, why not set up a series of switching regulators for the digital stuff, then some lower amperage linear for things like the SID and op amps?

You'd dissipate far less heat, especially with the amount of power taken up by something like a bank of LEDs, the LCD, and all the ICs. It'd be a lot more efficient too.

I've been thinking of doing this for a while, but just haven't gotten around to doing it.

[rutgerv]

Hi,

that's an interesting idea. However, wouln't it be difficult to keep the noise from the switching power supply away from the linear one, especially when there's little wiring between them (i.e. connected to the same high voltage input socket).

Kind regards,

Rutger

[c0nsumer]

that's an interesting idea. However, wouln't it be difficult to keep the noise from the switching power supply away from the linear one, especially when there's little wiring between them (i.e. connected to the same high voltage input socket).

No, I don't believe so, especially not if you have enough smoothing caps.

Anyway, the SIDs are so damn noisy I doubt you'll notice it much anyway.

(I've had this idea running around in my head for a while, because once I finally get a legal license of Eagle to do open-source PCBs which are the size of a PT-10 I'm hoping to do an 8-SID with full control surface, and all design documents made available to anyone who wants them. (I'd hopefully also do art similar to my x0xb0x art, full-on 8 ins and outs [via stereo jacks], etc.)