An (almost) linear power supply for my music PC

[quad]

Here's a variable voltage/current circuit using easily available
components - LM317 and 2N3055 . You can increase the current handling
capacity by adding more series pass elements (2n3055 transistors).

0-28V 6-8A Power Supply (LM317, 2N3055)

Here's a similar one with an Eagle PCB layout too.

http://www.cuteminds.com/index.php/en/powersupply

 

[thatguy]

Here's a variable voltage/current circuit using easily available
components - LM317 and 2N3055 . You can increase the current handling
capacity by adding more series pass elements (2n3055 transistors).

0-28V 6-8A Power Supply (LM317, 2N3055)

Here's a similar one with an Eagle PCB layout too.

Power Supply 1-30V 10A with LM317

Thanks quad. The circuit is good for a bench power supply. I would probably not use it if I am looking for a high quality fixed voltage output. It is a voltage reference followed by an emitter follower stage, not very different from what I did. The calibration is open loop, i.e. you only know the output voltage by measuring it (one could in theory have a fixed value resistor instead of P1).

Also, I would not drive a pair of 2N3055 directly from a voltage reference if I need high current output. The DC gain of 2N3055 is very unreliable (specially at high collector current) and may end up overloading the reference.

I didn't (try to) understand the use of P2. But it seems rather crude. Why not throw in a proper current limiter.

Audiophiles may even want to complain about diode noise from D7, D8 and D9

But I know I am nit picking. I just want to try out different power supply topologies and collects lots of data before picking one up. I am rediscovering electronics after a long time and trying to do it the textbook way rather than copying circuits.

cheers

PS: The guy forgot to account for the voltage drop across the base-emitter of the output transistors. His highest output won't be '30v - (3x 0.6v) = 28.2volt' but '30v - (4x0.6v) = 27.6'.

 

[thatguy]

With some circuits, it is a love at first sight. I encountered this one in The Art of Electronics: Paul Horowitz,Winfield Hill: 9780521370950: Amazon.com: Books while reading the chapter on power supplies. Just three components and a very smart use of feedback. I knew I had to make it. The book didn't mention any specific components, so I just picked up whatever I could find on S P road. Here is the circuit

Got a veroboard and soldered the components. First attempt was a disaster. I forgot to connect the transistor's base to the input of 7812. When I increased the load to about 1 amp, the 6 ohm (0.25 watt) resistor started to smoke.

I replaced the resistor with a new one and things seem to work. However, it was terribly unstable. Even putting the multimeter probes across the output could bring the voltage down to sub 1 volt. Sometimes it would recover and sometimes it would just stay stuck. Tried another multimeter with same result.

I first thought it was some soldering mistake, but then remembered that almost all power supplies put a small capacitor at the output stage. Never knew why. Still don't, but I tried it and it worked. So this is the circuit I ended up with:

The capacitance of C is 0.1 uF.

With the capacitor in place, it was stable. The output was 11.73 instead of the 12 volt I was expecting (need to investigate) but the regulation was very good.

I liked the circuit and will try out a few more versions. One with a darlington pair and another with multiple transistors in the output stage to boost the current capability. Will also test the transient response to varying load and input voltage.

Here are some action shots:

Input = 15 volts, load = 9 ohm, output = 11.72 volts

Input ~ 21 volts, load = 9 ohm, output = 11.73 volts

Input ~ 30 volts, load = 9 ohm, output = 11.73 volts

Input ~ 21 volts, no load (blurry picture, the output voltage was 11.76)

Here is the circuit on the veroboard.

 

[thatguy]

The voltage budget
-------------------

One of the key specifications for a linear regulator is the dropout voltage, the minimum difference between the input and the output needed for the device to regulate. The most popular of them all, LM78xx, has a dropout voltage of 2. LM7812, the 12 volt version of LM78xx, will need a minimum of 14 volts to function properly. If you want to attach a power transistor for high current, you must also account for the drop across a couple of diodes.

For the circuit mentioned in post #41, one needs a minimum of 17.5 volts (LM317 has a dropout of ~2.5 + drop across 3 diodes + BE drop for the output transistor) to get an output of 12.

Which brings us back to the discussion we had at the beginning about running the PC from a battery. A 12 volt lead acid battery's output will vary from 10 to 14 depending on the charge. I can't see how one can regulate that to 12 volt using a linear regulator.

There are regulators in the market with dropout voltages as low as 0.5 (e.g LM2940-12). However, they are very finicky about surrounding components and I decided to stick to 78xx (at least for a while) after reading the horror stories about the 2940.

So why not use a high voltage, say 24, and avoid the complicated mathematics. The input voltage is directly related to the efficiency of the power supply. If you are regulating 16 volts to 12 volts, the efficiency will be 75%. For an input of 20 volts, it would be 60%. For a people who love class A amplifiers running at 20%, this may not be a big deal. However, if the power supply is to get into your PC cabinet you must make adequate arrangements for heat dissipation.

Version 0.3 of the power supply is ready and under test. The circuit and a few photos of the PCB:

I have ordered a 120VA/12 volt toroidal from Miracle. Should be getting it next week.

 

[thatguy]

Finally tried testing the -12V requirement for my motherboard but the experiment didn't get off the ground. The sacrificial iBall PSU I had bought from S P Road had an out-of-spec 5 volt standby line. I am not sure if it would do any real damage to the motherboard but it scared me enough to postpone the test.

I will get around to it after I do some more research. For now, only the ATX PSU tester showing the voltage levels (it was also beeping like crazy ):

 

[sashijoseph]

http://imageshack.us/a/img543/749/hfvv.jpg[/IMG]


I have ordered a 120VA/12 volt toroidal from Miracle. Should be getting it next week.

you may improve the load regulation a bit by letting the 7812 source ~250mA.

I'm sure you understand the risk this circuit poses in the event of an output overload or short...try incorporating some means of o/p protection in your final version.

good luck...

 

[thatguy]

Soldered the 5 volt module. I wasn't expecting any problems and I didn't get any. Just put 7805 instead of 7812 and all seems well.

I am still unsure about how much the output capacitance (C2) should be. I tried 0.1uF that the spec sheet says and the output voltage was flying all over. I over reacted and replaced it with a 1000 uF and the voltage has been stable. Though I am not sure if that is the correct way to solve the problem.

The circuit and some action shots:

 

[thatguy]

you may improve the load regulation a bit by letting the 7812 source ~250mA.

I'm sure you understand the risk this circuit poses in the event of an output overload or short...try incorporating some means of o/p protection in your final version.

good luck...

Thanks Sashi. I will try that. The 5 V that I soldered today was almost perfect. All the 12V regulators I tried were around 11.7 to 11.8, which isn't a problem because that is well within the ATX spec.

Protection circuit is next. I want to put voltage clamping at output (to take care of part failure within the supply) and fuses at the input (overload/short circuit). For clamping, I have been thinking of putting a SCR which gets triggered by the current through a zener. If that doesn't work satisfactorily, it will be a power transistor and zener. I have no first hand experience, so the final version will be decided after some experimentation. Do you have any suggestions? Would love to hear if you have any experience with protection circuits.

cheers

 

[quad]

Has this in my bookmarks - nice overview of the classic crowbar.

Axotron

 

[thatguy]

The transformer arrived yesterday. I wasn't sure how to calculate the derating factor for transformers and ordered one with almost four times the needed capacity.

One of the nice things about an ATX PSU is that they can be switched on and off by software. When AC power is supplied to the PSU, only one of the 5V lines is given power. The rest of the buses get power when the motherboard signals. Since I wanted to power the Xonar ST directly (not through the picoPSU), this posed a problem. I asked a few friends if it was ok to power on the sound card before the rest of the motherboard but no one gave me a definite answer.

Still unsure, I have decided to send the power to the Xonar through a relay which will be controlled by one the picoPSU's 12v output lines. I had never used a relay and asked the folks at Amar to help me select one. They gave me one that looked very different from one I had seen earlier (in transparent plastic cases). I tested the relay and it seems good. The coil resistance is 400 ohms and it needs about 8 volts to trigger. I am planning to run it directly from the 12 volt line, without using a series resistance.