MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

diy power supply questions
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Author diy power supply questions
fengland
Hi all,
Apologies if these questions have been covered before, I read the topics I could find by searching.

I have a couple of questions about building a power supply with plenty of power for my small-midsize modular needs.

First question - I can use 2 transformers (one for +12, 1 for -12) in place of a center tapped one right? Radio shack has 24 volt 2 amp transformers - I figure with 2 of these I can get 2 amps per rail if I wire their secondaries in series and treat them as 1 large center tapped transformer. The connection between the two transformers would be connected to ground. I could use 1 huge center tapped tranny I guess, but these are cheaper and more readily available. Any problems here?

I plan on putting the transformers and filter caps in an external enclosure and making smaller boards with a 7812/7912 pair per 3U power distribution board (I might build my own on perfboard distribution boards with star grounding). This way the voltage regulators won't be working too hard, each set will just be powering 1 row of modules and I can utilize the full 2A per rail of the transformers with multiple sets of regulators. These regulators need to be fed with between 14 and 35 volts so giving them 24 volts should be fine right? are they going to get wicked hot being fed with that voltage? what kind of heat sinks would I need?

Any other concerns? I've built both of the power supplies (half wave rectified no center tap, full wave center tap) on the MFOS site without killing myself and the second one has been going strong for a couple years.

Thanks a lot!
daverj
The higher the input voltage, the hotter the regulators get. Since the unregulated voltage you feed into the regulator has some ripple on it and also changes voltage as the load changes, the ideal is to have the lowest part of the ripple be about 3 volts above the regulated output at full load. The higher the unregulated voltage, the more power has to be dissipated by the regulator (which is what makes it hot)

Heat sinks should be as large as possible. (or as large as practical). Even with only a couple hundred ma coming out, regulators can get quite hot.
fengland
ok, I'll try to get some 15 or 18 volt transformers and some big heat sinks. thanks!
But there isn't a problem with using a separate transformer for + and - right? It looks like this is what the doepfer diy kit #2 is doing.
Big G
fengland wrote:
First question - I can use 2 transformers (one for +12, 1 for -12) in place of a center tapped one right? Radio shack has 24 volt 2 amp transformers - I figure with 2 of these I can get 2 amps per rail if I wire their secondaries in series and treat them as 1 large center tapped transformer. The connection between the two transformers would be connected to ground. I could use 1 huge center tapped tranny I guess, but these are cheaper and more readily available. Any problems here?


Yes. You are making the common mistake of not reading transformer specs correctly. The secondary voltages are given as Vac with no load and the currents as Iac maximum. To get the dc voltages you have to multiply by 1.414 to get the peak voltage and then subtract the rectifier voltage drops. The equivalent dc current will be the Iac DIVIDED by 1.414 (otherwise you would be getting some power from nowhere) so you cannot get a 2Adc power supply from a 2Aac secondary.

Then there's transformer regulation (not to be confused with voltage regulation) which may be 10% to 25%. This is the drop in the secondary voltage at full load. So you have to multiply the voltage by 100 - regulation% and that voltage has to be at least 3V higher than the voltage regulator output value required.

And then you have to allow for a drop in the mains supply too.

So those transformers will give too high a voltage and not enough current and the excess will be wasted as heat in the regulator, in this case more than the load power. Dual secondary transformers are not hard to find in most catalogues, look for a dual 12V 80VA type.
fengland
ok, there's more math involved than I realized.

12 * 1.414 - 0.7 = 16.268 volts
2 / 1.414 = 1.414 amps

so it sounds like I can get 12 volts dc, 1.414 amps per rail from a transformer with dual secondaries rated at 12 volts each and a 4 amp rating, is that right?
thanks a lot, I thought I needed one rated at least 2 volts higher than the dc voltage I want to make.

I was looking at these but I guess I want one rated at 12 volts: http://www.goldmine-elec-products.com/prodinfo.asp?number=G17913
fengland
I'm looking around at various diy power supply designs (cgs, oakley) and it doesn't seem like any provide more than about 600 milliamps per rail realistically. Maybe that's all I need for a few rows of eurorack modules. I was going to use this paia midi to cv converter which says it needs 130ma from the positive rail so I wanted to have plenty of juice but I think maybe I'm over designing this.
PrimateSynthesis
If you really want to build a power supply that's one thing. I built one when I was a kid. I got everything by riding my bike to Radio Shack. They had this book "So You Want To Build A Power Supply?" or something. Then I etched a PCB that used the plastic box for the PCB etching kit as a little chemical tank.

If I remember it was an excellent book, that explained all the math and gave step-by-step examples of each possible configuration -- half-wave, full-wave, bridge, bipolar, etc. I wish I still had it waah

While I haven't actually looked, I'm sure there is similar information on the internet...

However if all you want is a supply for your modular, you can get a better one for less money if you buy it already made from PowerOne, Condor, etc. That would be my advice unless you already know what you are doing because synth modules don't have fuses Dead Banana
pugix
PrimateSynthesis wrote:

However if all you want is a supply for your modular, you can get a better one for less money if you buy it already made from PowerOne, Condor, etc. That would be my advice unless you already know what you are doing because synth modules don't have fuses Dead Banana


I agree. thumbs up

Modules are precious and very expensive! They need good power. I use all Power One supplies. A typical Power One supply costs about the same as one small module, less than $100 USD. They are well engineered with over current protection, that you won't get with 3-pin regulator.

There's still plenty of DIY fun mounting your supply, providing a switch and fuse, and hooking up a distribution board. Here's how I did it for a portable cabinet.

http://pugix.com/synth/portable-rack-cabinets/
Big G
fengland wrote:
ok, there's more math involved than I realized.


There always is in real engineering, that's what separates the pros from the amateurs smile. It's not as if it needs three dimensional calculus, it's only simple sums, but they have to be the right ones that calculate what is really happening.

Quote:
12 * 1.414 - 0.7 = 16.268 volts

Correct at no load, at full load you have to allow for the transformer regulation. If it is 10% (it can be higher) then the pre-regulator dc voltage will drop to 90% = (12 * 1.414 * 0.9) - 1.4 = 13.87 volts, i.e. the regulator will drop out.
You didn't allow for two diode drops in a bridge rectifier and we still haven't allowed for the mains being less than it's nominal value. Be careful to note whether the transformer is specced at 115/230Vac or 120/240Vac

Quote:

2 / 1.414 = 1.414 amps


You've probably spotted that 1.414 is the square root of 2. This figure is used because it is the ratio of the rms value of a sinewave to the peak value.

Quote:
so it sounds like I can get 12 volts dc, 1.414 amps per rail from a transformer with dual secondaries rated at 12 volts each and a 4 amp rating, is that right?
thanks a lot, I thought I needed one rated at least 2 volts higher than the dc voltage I want to make.


Most voltage regulators need a minimum of 3Vdc difference between their input and output. There are some special low drop out (LDO) types. Any dips below this will appear as a ripple on the output.
A 15Vac secondary will give you (15 * 1.414 * 0.9) - 1.4 = 17.7Vdc, or 15.6Vdc @ 20% transformer regulation.

The product of the voltage across the regulator and the current through it gives the heat in Watts that needs to be dissapated. (17.7 - 12) * 1.414 = 8W, 16W for a dual supply, and that requires a serious sized heatsink.

Quote:
I was looking at these but I guess I want one rated at 12 volts: http://www.goldmine-elec-products.com/prodinfo.asp?number=G17913


Hmmmm. I wouldn't put such a heavy transformer on a pcb without careful support--they tend to tear off if you drop the psu or just transport it. Also rectangular transformers have stray magnetic fields and at those currents you will get steel panels vibrating. Toroids are better.

Quote:

I'm looking around at various diy power supply designs (cgs, oakley) and it doesn't seem like any provide more than about 600 milliamps per rail realistically.


Yes and they all look as if the "designer" has not done the above calculations or worked out the reservoir capacitor and heatsink sizes correctly.
Synthbuilder
Big G wrote:
Yes and they all look as if the "designer" has not done the above calculations or worked out the reservoir capacitor and heatsink sizes correctly.


Are you counting me in that? I hopefully did all the maths needed on my two designs. As well as rigourous testing of real devices.

The problem is that not everyone uses the same AC source as intended nor sticks to the current limitations. If used as directed in the User and Builder's Guides the Oakley PSUs work very well without excessive heat build up.

Tony
Big G
Synthbuilder wrote:

Are you counting me in that? I hopefully did all the maths needed on my two designs. As well as rigourous testing of real devices.


Yes. You claim that your compact PSU will deliver an amp, yet it only has half wave rectification and 1000µF reservoirs and tiny heatsinks. No way. If it does deliver an amp before the regulators go into thermal shutdown the ripple (which you don't specify) would make them unusable.

Similarly you only claim that your larger PSU is tested at 1.2A, but omit what the results are. Anybody can slap the standard voltage regulators application note circuit onto a pcb and get a few hundred milliamps out of them. Delivering over an amp is a different game altogether.

Why don't you publish proper specs?

If you've done the maths show me the thermal resistance equation where you calculated the size of heatsinks needed.

I'm not picking on you particularly as others are just as bad, but you bit the bait.
Synthbuilder
The Compact PSU module's current output using the recommended power pack (and thus using the module's half wave rectification mode) is only 200mA per rail. The C-PSU is only designed to power seven Oakley modules - which is pretty much one row in a 19" rack.

I'll admit that the first paragraph on the C-PSU's webpage could be construed as misleading but I would hope that people read the full documentation before embarking on the project. In light of your comments I will, however, edit the web page accordingly.

The User Manual and Builder's Guides I hope are quite clear what you can and can't take out of the C-PSU with any particular power pack. I would urge you to take a look at these and I do welcome your comments and I am very happy to make amendments if necessary.

In the User Manual you'll find other power packs that have been used successfully with the unit. One is a split secondary output and uses the C-PSU module's full bridge rectifier and can be used up to 750mA.

The heatsinks on the C-PSU are 7.2K/W devices. Allowing for a maximum 43C rise in their temperature this would give scope for a 8V drop across the regulator assuming good air flow and a current draw of 750mA.

The larger PSU I do is specifically designed to be used with the PA-20 supply. As such current is limited to around 800mA per rail. The heatsink in this PSU is the 3U x 5U wide aluminium panel. No calculations have been done on this panel but an empirical approach was taken. Experiments suggested that the unit was quite capable of delivering a continuous 500mA with this panel rising to no more than 60C with ambient at approximately 20C. Since I do not recommend exceeding 75% of the maximum deliverable load I regarded the heatsink as sufficient.

Higher currents are obtainable but not with the 3U panel heatsink. I do make various suggestions in the User Manual about this but I try to encourage builders to stick to the recommended AC sources to avoid any problems.

Tony
Big G
Synthbuilder wrote:
The Compact PSU module's current output using the recommended power pack (and thus using the module's half wave rectification mode) is only 200mA per rail. The C-PSU is only designed to power seven Oakley modules - which is pretty much one row in a 19" rack.

I'll admit that the first paragraph on the C-PSU's webpage could be construed as misleading but I would hope that people read the full documentation before embarking on the project. In light of your comments I will, however, edit the web page accordingly.

The User Manual and Builder's Guides I hope are quite clear what you can and can't take out of the C-PSU with any particular power pack. I would urge you to take a look at these and I do welcome your comments and I am very happy to make amendments if necessary.

In the User Manual you'll find other power packs that have been used successfully with the unit. One is a split secondary output and uses the C-PSU module's full bridge rectifier and can be used up to 750mA.


That's a bit different from the 1.0A and 1.2A mentioned. The description that Maplin give for their power adapters is misleading too. All those things are is a transformer in a plastic case. The N57AT says 15V and 500mA, but that doesn't mean at the same time and they are AC ratings. The clue is that the maximum power is given as 6W which means that at 0.5Aac the output will be 12Vac (ie 20% regulation which is about right for a transformer that size). So at full load the dc voltage on the reservoir capacitor will be (12 * 1.414) - 0.7 = 16.268Vdc. This is not enough to regulate down to 15V.

I don't know what maths you did, but it wasn't the right maths. I'm not singling you out because all modular manufacturers are as bad or worse. I don't see any PSUs that are properly rated with maximum continuous current and ripple. That is the vital information to anybody building an expandable modular system.

All this business about "run it at 70%" just means that it doesn't work properly and the "designer" doesn't know the difference between ac and dc ratings.

To build a +/-15Vdc PSU that will deliver 1A constantly a decent 80VA transformer is required as well as other components an order of magnitude larger than what is commonly used. That is how I can glance at a picture of any PSU and say instantly that it is not going to do the job.


Quote:

The heatsinks on the C-PSU are 7.2K/W devices. Allowing for a maximum 43C rise in their temperature this would give scope for a 8V drop across the regulator assuming good air flow and a current draw of 750mA.


That's the wrong maths too. I expect better from someone who puts B.Eng after their name even if the qualification is in another branch of engineering.

Quote:

I do make various suggestions in the User Manual about this but I try to encourage builders to stick to the recommended AC sources to avoid any problems.


Let's be honest here. The problems you are avoiding are CE LV testing and certification for your products and you have passed the buck onto the user who normally won't have the necessary knowledge.

So my advice to you is derate your specifications to what is absolutely verifiable and don't extrapolate as you have done. If you don't promise what isn't deliverable you will earn respect. There is nothing to be gained from trying to outspec competitors when they are lying, just let them hoist themselves on their on petards.

Most of the problems people complain about on this forum and elsewhere are caused by poor PSUs that are hovering below regulation due to inadequate design of both the PSU and the power distribution. I am sick of trying to help people that have spent a lot of money building up a system and have got short changed. When I explain that they have to spend more to get what they thought they had in the first place they are understandably reluctant, especially on something that doesn't make a sound. That is the heart of the problem, PSU design is seen as "boring" or "not exciting enough" by people that design synthesizer modules. In fact it is probably the most vital component of a system as all else depends on it and certainly not something that should be economised on.
fengland
wow - so it sounds like maybe their aren't any properly designed higher output diy circuits out there?
I guess I'll stick with a pro one for higher output. In the mean time the CGS one is working fine for smaller loads - I ended up building one of them for my 6u case.
I'd love to see someone design a power supply that delivers an amp that can be DIY'd. Or is that much power just unneccessary for a few rows of modules? sounds like a controversial topic.

Big G - Thanks a lot for walking me through this. Really educational.
Big G
fengland wrote:
wow - so it sounds like maybe their aren't any properly designed higher output diy circuits out there?


Not many non-DIY either...

Quote:

I guess I'll stick with a pro one for higher output.


Pro-Ones are not perfect either. Read their data sheets very carefully with special attention to the word "derate". Derating is the game everybody is playing, they give one spec to get your attention and then in very small print qualify that figure so that it is not immediately obvious that you are not going to get the performance you think you will. Open frame linear supplies are intended to be bolted to large metal surfaces and fan cooled.

To my eye the components in a Pro-One still look too small and they use rectangular transformers that will have quite large stray magnetic fields at high currents. The best serious power supplies use huge toroidal transformers that alone cost more than a Pro-One. You get what you pay for, the question is why economise on the single most important part of a synthesizer that everything else is dependent on working well?

OEM PSUs are a cut throat market, it is based on low cost not performance. Pro-One seems to have cornered that market by having simple products at the right price. They cannot change the laws of physics though.

Quote:

I'd love to see someone design a power supply that delivers an amp that can be DIY'd. Or is that much power just unneccessary for a few rows of modules? sounds like a controversial topic.


It isn't necessary. A few rows of modules may be up to 700mA which is the most you will get out of most of the PSUs pretending to be 1A or more, for reasons already explained.
It isn't really a DIY project, just be grateful if anybody makes one. If you look how serious PSUs and large power amplifiers are built there is a lot of skill and craftsmanship involved.

I suggest you put an oscilloscope on the reservoir capacitors and see what happens when you starting pulling high currents. You get a sawtooth of several volts amplitude as it charges up and discharges and if the lowest point drops too far the regulator drops out and you get a serious ripple on the power rail. At currents over 1A every 0.001ohms of resistance in your wiring will cause 1mV drop, now measure some of your power wiring and pcbs traces and you will find they are higher than that.
Synthbuilder
Well this has been an eye opener for sure. I've just spent the last day re-learning what was clearly forgotten. oops

Big G wrote:
Not many non-DIY either...


Clearly a gap in the market. Fancy filling it for us Big G? I'm not being funny there. You clearly know your stuff and we, that is the whole synth community, need a decent and safe solution. In the 5U section there is certainly a need for a nice ready made 19" rack solution with some sort of built in distribution.

Big G wrote:
Pro-Ones are not perfect either.


Do you mean Power Ones? I've also seen problems with Power Ones latching up - where one rail fails to come up. The foldback limiting circuit appeared to not cope with the capacitance on the rails. I got onto Power One and they confirmed it was a problem and suggested a considerably bigger supply.

Tony
Big G
Synthbuilder wrote:
Well this has been an eye opener for sure. I've just spent the last day re-learning what was clearly forgotten. oops


I see you have corrected your website. Now if only other synthesizer manufacturers were as honest...
Your avatar looks like a lot of their PSUs.


Quote:

Clearly a gap in the market. Fancy filling it for us Big G?


No, but I know someone who is in the process of doing this. I'll have to check with them.

Quote:

Do you mean Power Ones?


Sorry, Freudian slip. Yes, read Power One.

Quote:

I've also seen problems with Power Ones latching up - where one rail fails to come up. The foldback limiting circuit appeared to not cope with the capacitance on the rails. I got onto Power One and they confirmed it was a problem and suggested a considerably bigger supply.


They are just a 723 built as low cost as possible and not intended for precision audio use. The people who use them have never seen a quality linear PSU so that they have nothing to compare against. The sort of gear that fills a 6U 19" chassis and takes two people to lift and is full of computer grade capacitors the size of jamjars with copper bussbars joining the terminals. There is not much of a market for these now as most of the industrial applications use switchers and the high end audio people build their own.
analog604
Hi Fengland,

The approach of using several current regulated smaller PSUs in the modular synth seems to be recommended and it is what I'll be attempting to do even though I've built stuff like this.

Personally, I'm in favor of staying away from perf boarding any PSUs that are meant for production or that may be left energized unattended.

In the end it will be far less trouble, work, and most importantly safer to simply get a PSU that has already been designed and built.

take it easy!
-J

fengland wrote:
Hi all,
Apologies if these questions have been covered before, I read the topics I could find by searching.

I have a couple of questions about building a power supply with plenty of power for my small-midsize modular needs.

First question - I can use 2 transformers (one for +12, 1 for -12) in place of a center tapped one right? Radio shack has 24 volt 2 amp transformers - I figure with 2 of these I can get 2 amps per rail if I wire their secondaries in series and treat them as 1 large center tapped transformer. The connection between the two transformers would be connected to ground. I could use 1 huge center tapped tranny I guess, but these are cheaper and more readily available. Any problems here?

I plan on putting the transformers and filter caps in an external enclosure and making smaller boards with a 7812/7912 pair per 3U power distribution board (I might build my own on perfboard distribution boards with star grounding). This way the voltage regulators won't be working too hard, each set will just be powering 1 row of modules and I can utilize the full 2A per rail of the transformers with multiple sets of regulators. These regulators need to be fed with between 14 and 35 volts so giving them 24 volts should be fine right? are they going to get wicked hot being fed with that voltage? what kind of heat sinks would I need?

Any other concerns? I've built both of the power supplies (half wave rectified no center tap, full wave center tap) on the MFOS site without killing myself and the second one has been going strong for a couple years.

Thanks a lot!
Bassphix
Hi All,

I too have a DIY power problem...

I've got a 230v single input, 15v dual output transformer set up for centre tapped. This is going into a full wave rectifier.

I'm getting +24.6DC and -26.2DC from the rectifier. not sure why or how they would be different?


That output goes in to a 1000uf cap on each side.

From there it goes into a ST L7812cv and a 7912cv reg.

There is a 10uf cap on the input and a 0.1uf cap on the final output.

I get +12.08v and -17.56v!

I've checked the ground line and it seems fine.

Any help much appreciated!


Ant.
PrimateSynthesis
zombie

I mean, congratulations on searching before posting thumbs up

You have a 100uF cap on the input?

Are you running into any sort of load?

Anyway, I'd go back to secondary before the rectifier, are the voltages even? I'm going to take a wild guess that your ground isn't at ground.
CeeJay
Did you note the different pinout of the 7912 compared to the 7812. The 7912 has the input at the middle pin and this is also connected to the metal tab. So do not connect the same heatsink to both regulators without insulation.

PS: thank you for revive this 8 year old thread! Really good information!
Bassphix
PrimateSynthesis wrote:
zombie

I mean, congratulations on searching before posting thumbs up

You have a 100uF cap on the input?

Are you running into any sort of load?

Anyway, I'd go back to secondary before the rectifier, are the voltages even? I'm going to take a wild guess that your ground isn't at ground.


Hmm, think I checked the secondary and they are OK, maybe 0.1v different. Will check again though. However, the out DC from the rectifier is different between + and -.

However, as far as I can see, the reg should take up to 35v (it's no where near that) or so and still rectify it?

1000uf cap on the output of the rectifier, then
10uf on the input and 0.1uf output of the reg. I have tried different combinations though.

I thought that so I've checked the grounds and they seem fine.

I have a 0.2mA load LED on the + side because that seems fine, but didn't want to out anything on the - until it reads right.

Thanks,
Ant.
Bassphix
CeeJay wrote:
Did you note the different pinout of the 7912 compared to the 7812. The 7912 has the input at the middle pin and this is also connected to the metal tab. So do not connect the same heatsink to both regulators without insulation.

PS: thank you for revive this 8 year old thread! Really good information!


Yes, I have checked and tripple quadruple checked that.

I haven't added heatsinks yet as there is no load. But I have also checked that nothing is shorting the tab.
Bassphix
Hopefully this works!

Black is ground, white + and yellow -

Thanks,
Ant.



Bassphix
The small black blobs by the rectifier are just fuses.
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Page 1 of 2
Powered by phpBB © phpBB Group