Symmetrical PSU with different voltages and loads

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

barthman.de

Well-known member
Joined
May 26, 2007
Messages
98
Location
Erfurt / Germany
I'm just build a rack for some old preamps. To power them up I need -20V (round about 600 mA load). For LED's and relays I need +12V (round about 250 mA load). Can I use a symmetrical design like the following or should I build two different PSU's?

symnetzteil.jpg
 
 
why not, the additional voltage drop on the positive supply will be at 250mA current, so the loss is easy to cope with... better than to have to use two transformers.... it will add around 2W on the positive, so add good cooling anyway.

- Michael
 
If you have more unregulated voltage than you need for one side, you can consider only half wave regulating the lighter loaded rail, and consider a smaller reservoir cap. While this will generate more ripple voltage, the average unregulated will be lower for lower dissipation in the one regulator vs, FWR and low ripple cap.

JR
 
> Symmetrical PSU with different voltages and loads

Different voltages and loads.... then it's not symmetrical.

What you really have is two loads. Doing VERY different jobs: sweet audio and clunky switching/lighting. I'd think about two supplies. Two transformers best. A dual-winding PT, each winding doing its own thing, is also probably good.

BTW: you could re-think your relays and LEDs to be "positive ground". They really don't care which way ground is. Then you could derive both supplies from a single raw 30V supply.
 
Thank you for your answers. I've build the PSU like shown above test wise. Without load I get the right voltages (+12V / -20 V).  For the LED/relays - part I used the following simple circuit. The preamps are not connected with the -20V rail at the moment. If I switch on the switches gradually the LED's shining darker an darker  :eek: This should not be in a parallel circuit so I think it must be a problem with the PSU.

parallel.jpg



Edit:

Now I tested the 12V circuit with an external 12V PSU and after switching all switches on all LED's shining with full intensity. The problem is definitely the symmetrical design of the PSU above.

How about to make two separate PSU-circuits on the same transformer. Each circuit with own diodes/capacitors/0V? Would this make any difference from the circuit above?

PS:

The orange LED's are ******* bright…  8)

leds.jpg
 
Hi Barthman,

6 MV810 in 1 Box!? Great!!  :D
This is going to be a heavy beast...

And as I see you've picked up my design to hide them fully behind the front panel...  :D

In my build i used a single switching PSU (48V, Sunpower) + JLM TREX, straight on, no problems...

Good luck....

ROCK-ON!



P.S.:
What you really have is two loads. Doing VERY different jobs: sweet audio and clunky switching/lighting
Great!
 
Dr_J said:
Hi Barthman,

6 MV810 in 1 Box!? Great!!  :D
This is going to be a heavy beast...

And as I see you've picked up my design to hide them fully behind the front panel...  :D

Yes really heavy but not heavier then my 4 x V72 Rack  :p 5 of them are with germanium transistors, one is the "2"- Version with silicon transistors. 

The place inside the rack is more than limited this time. To get place on the front panel for on/off, phantom power and phase I had to put the preamps completely inside. 

It was a pain to clean all the rotary switches inside the modules but after maybe 50 years only one preamp needed more repair after that - and this was the newer "2" - Version  ;D

I use my ramped phantom power design and I think I will now make two separate PSU's for 12V and -20V. I hope I get all in the box :)
 
JohnRoberts said:
If you have more unregulated voltage than you need for one side, you can consider only half wave regulating the lighter loaded rail, and consider a smaller reservoir cap. While this will generate more ripple voltage, the average unregulated will be lower for lower dissipation in the one regulator vs, FWR and low ripple cap.

JR
John, this is surprising coming from you! Half wave rectification is not a good idea for transformers, as it results in unbalance of core currents, and can lead to transformer heating. Its fine at very low current, but otherwise not a good idea.
 
radardoug said:
JohnRoberts said:
If you have more unregulated voltage than you need for one side, you can consider only half wave regulating the lighter loaded rail, and consider a smaller reservoir cap. While this will generate more ripple voltage, the average unregulated will be lower for lower dissipation in the one regulator vs, FWR and low ripple cap.

JR
John, this is surprising coming from you! Half wave rectification is not a good idea for transformers, as it results in unbalance of core currents, and can lead to transformer heating. Its fine at very low current, but otherwise not a good idea.
Sorry, you want simple, that's simple. If worried about being kind to the transformer use FWR and small reservoir caps (what no complaints about ripple voltage/current in the caps?).

For a higher tech solution back in the 80s I did a trick PS design where i used an electronic switch (small transistor) in series with the FWR diodes so i could open the path between the transformer and lower voltage reservoir cap when transformer voltage was higher than needed. Conventional diodes fed the full winding voltage to the higher voltage rail.  In fact this approach is energy efficient without the complexity and RF noise from using a DC to DC switching supply.  I did one product with a high current 5V digital supply, and modest current +/- 15V all from a simple one winding (center-tapped) transformer.  Trying to operate the high current 5V supply with pass regulation, would have required using a larger transformer and dissipating lots of waste heat.

JR
 
The problems are not removed till now  :-[ Need help.

Like I wrote, the LED's getting darker and darker, if I switch on more and more. The problem isn't there, when I put on a ready built (bought) PSU on the circuit! So it must be a problem with my self-built PSU. For my self-built PSU I used LM317 to get +12V. Since my first postings I  tried out two different circuit-designs. First the shown design from JLM, second a one rail design without the -V rail and other input capacitors. Both don't work. I changed the transformer too. First I used 25V / 1A second a 24V / 1,25A. With the bought PSU the LED's need not more then 130 mA if all are switched on. That should not be a problem for the transformer (1 A is more then enough). I wonder  too that the voltage looks fine without load. It seems that LM317 (1- adjust; 2-out; 3-in) don't regulate with load? I don't know what should be wrong. Any ideas?

Thank's a lot.       
 
> I wonder  too that the voltage looks fine without load. ...? I don't know what should be wrong.
> Any ideas?


Take voltage readings?

 
barthman.de said:
The problems are not removed till now  :-[ Need help.

Like I wrote, the LED's getting darker and darker, if I switch on more and more. The problem isn't there, when I put on a ready built (bought) PSU on the circuit! So it must be a problem with my self-built PSU. For my self-built PSU I used LM317 to get +12V. Since my first postings I  tried out two different circuit-designs. First the shown design from JLM, second a one rail design without the -V rail and other input capacitors. Both don't work. I changed the transformer too. First I used 25V / 1A second a 24V / 1,25A. With the bought PSU the LED's need not more then 130 mA if all are switched on. That should not be a problem for the transformer (1 A is more then enough). I wonder  too that the voltage looks fine without load. It seems that LM317 (1- adjust; 2-out; 3-in) don't regulate with load? I don't know what should be wrong. Any ideas?

Well, an LED's brightness goes down with current through it, and in most cases there's a dropping resistor in series between the LED and the supply rail, so my bet is that the rail feeding the LED is collapsing.

So an LM317 needs at least 15 V at its input to give you a regulated 12 V at the output. So, look at the regulator's input and output. If the regulator input is too low, the output will probably follow the input, or it won't. If it does, then the voltage into the LED/resistor path is probably too low and the current through the LED is limited and it's a dull LED.

Now if the input voltage is "too high" but not exceeding breakdown, you could be into a thermal shutdown/start-up cycle if the regulator isn't heat sinked. If that cycle is fast enough, the LEDs can turn on and off fast enough to look as if they are dimmed. You need to look at this with an oscilloscope.

-a
 
Is your LM317 hot or even to hot to touch? if you are using a 24V transformer you are dropping around 21V over the regulator, corresponding in something around 2.8W heat to get rid of. without adequate cooling the LM317 will go into thermal protection in that configuration.

rectification: 24V * 1.4 = 33.6V
voltage over LM317 33.6 - 12 ) 21.6V
Power over LM317 21.6V * 0.13A = 2.8

an other thing I noticed is the rectifier arrangement you use - with a dual winding transformer I would tend to use two separate rectifiers, one on each winding. it might not even be necessary to connect both grounds at all, depending on your circuit. best way to eliminate noise.

- Michael
 
Ok I looked at the voltage (24V DC at the transformer):

Without load:

input lm317: 13,2V
output lm317: 12,5V

With load (round about 60mA- 3 LED's switched on):
input lm 317: 4,5V
output lm317: 3,8V

After one minute with 60mA load the lm317 keeps cool. So the input of the lm317 seems to low, right?

I use the following circuit design at the moment but 1k at R7 to get out 12V:

24V.jpg


How can I get this work with 24V or 25V DC input - maybe change R5?
 
barthman.de said:
Ok I looked at the voltage (24V DC at the transformer):

Without load:

input lm317: 13,2V
output lm317: 12,5V

With load (round about 60mA- 3 LED's switched on):
input lm 317: 4,5V
output lm317: 3,8V
Are you sure your measurements are correct ? You measure regulator input voltage between C5 legs? and where did you measure 24VDC, between C2 legs?
 
> 24V DC at the transformer

Transformers don't make DC.

If you have 24 Volts *AC* at the transformer winding, and good rectifier, and good first-cap, then you should have 35V *DC* at the first capacitor.

All measured to the bottom line in your drawing (which might NOT be chassis; we can't tell)
 
barthman.de said:
How can I get this work with 24V or 25V DC input - maybe change R5?

Are you sure R5 has the correct value in your set up?
Did you measure the voltage before and after this resistor?

Don't see the point in R5 anyway. And your first post's
schematic did not include this resistor...

Cheers,
Carsten
 
I'm sorry of course there are 24V "AC" on the input of the circuit (= output of the transformer). Maybe it was to late yesterday evening  ::) 

@ PRR; yes there are 35V at the first capacitor. 0V is not connected to the chassis / housing / earth, I'm testing the PSU-circuit on a board for experiments. The components are new and tested. And yes: the black cable of my FLUKE is connected to 0V  ;D

@ culteousness1: R5 is 2.2 which means 2k2 in my opinion there is a point in the schematic. The design of the second circuit is a bit different from the first one but the result is the same: both designs don't work. 

@keefaz: yes, there are 13,2V without load on the input of the lm317 or after R5. With a load of 60 mA there are only 4,5V.
 
2.2k is a totally wrong choice here. Omit that all together. This will drop to much voltage - remember ohm's law or look it up. All the led current will flow through there. 2.2 would be interpreted as 2.2 ohms by the way.

Do you have proper current limiting resistors to the individual LEDs?

I think it makes sense to build two independent supplies from that transformer.
Your requirements are not that special, Look up a standard LM317 circuit without all the extra components (like bypass cap on the set resistor) first to get a stable result. Remember to add a isolated heat sink. Use only half  way rectification to lower the voltage on the first cap for the relais / led supply.

- M
 

Latest posts

Back
Top