Dual linear power supply with switching ATX pre-regulator

Thread Starter

Silvio

Joined Jul 2, 2015
14
Hi I am new here and I am building a dual power supply with an ATX pre-regulator. I have already modified the ATX which is capable of 350 watts. I dismantled the ferrite transformer,output inductor, bridge rectifiers etc, and re- wound it to my needs to provide a dual 36 volts output. My second stage is to be an LM317T based voltage regulator with an TIP2955 transistor coupled with the 317 regulator. The ATX is to follow the main output voltage with 3 volts more in the input. I have bread boarded it and the switcher follows perfectly. This way it will give me 5 amps or more on every output voltage on each supply simultaneously. This way the dissipation is kept low on the pass transistor. My next stage is to include an LM301 opamp on each regulator to control the current individually on each supply, following the 317 regulator data sheet. My doubt is whether I should make 2 positive regulators which i can vary individually and hook up in series or parallel if i want to, or make one positive and one negative regulator. What are your suggestions about this project. Any advice will be considered. I will post my schematics as soon as i draw them.
 

Thread Starter

Silvio

Joined Jul 2, 2015
14
Here is the schematic diagram for the ATX and for the linear positive regulator section complete with the tracking arrangement.

Just to reply #12 well the power was there and why not use it if ever comes the need. I will take your advice and the other supply will be a negative regulator. What do you think about the schematic, especially on the -1.25v reference voltage, do you think it will work that way? my other question is what value shall be the voltage adjusting pot 2k or 5k ohms, how will these affect regulation and adjustment? I was considering a 10 turn bourns pot.
 

Attachments

dl324

Joined Mar 30, 2015
16,846
What do you think about the schematic, especially on the -1.25v reference voltage, do you think it will work that way?
The addition of a -1.25V reference to allow adjusting output to 0V is a nice touch. Instead of a -5V regulator and a voltage divider I probably would have just used an LM113, as shown in the datasheet for a 0-30V regulator; fewer components and you don't need to worry about regulation issues caused by quiescent current changes from the regulator... You might want to put the decoupling cap for the adjust pot to ground. Why did you decrease the value of that cap from 10uF to 1uF? Other than that, the schematic is right out of the datasheet.
my other question is what value shall be the voltage adjusting pot 2k or 5k ohms, how will these affect regulation and adjustment?
5K; 2K won't give you enough adjust range, shouldn't affect regulation.
I was considering a 10 turn bourns pot.
For voltage adjust, I'd go with 2 single turn pots; one for coarse adjust and one for fine. For current limit, I'd also use single turn.

The voltage out of your filter network is going to be too high for LM317 for output voltages <= 10V.
 
Last edited:

dl324

Joined Mar 30, 2015
16,846
Forgot to mention that your meter should be placed on the other side of the divider for setting voltage. As drawn, the meter will have resistance that will have an adverse affect on regulation.

I didn't analyze the BC327 and associated components. Too tedious the way it's drawn.
 

Thread Starter

Silvio

Joined Jul 2, 2015
14
First of all thanks for your comments. here are a few answers that might help out why i decided to things the way they are.
1) regarding the reference voltage 1.25 my local electronics shop does have a 1.2 volt reference (LM385BZ) which cost around 2.5 euro, rather costly i might say. I was considering an LM337T with its adjust pin tied to ground which will also give me the reference i need at half the price instead of the arrangement on the schematic. I am designing my own PCB so its not a problem to change design.
2) I did not mention that I have salvaged a small switcher from an old DVD player which has + and - 12volts for opamp plus a +5volts at around 500 ma. I connected another switching diode this time with the positive to the output of the transformer plus cap and small inductor which this will give me another -5volt out. That was the reason I opted for the voltage divider as reference.
3) As I read the datasheet for the 317 its said somewhere that either a 1uf tantalum or a normal 10uf cap will do the same job.
4) The arrangement for the tracking of the switcher is in a way that the voltage divider with 22K and 220 ohms creates a 1volt reference hence this is driven by the collector of the BC327 with its base tied to the output as a reference and the emmiter with the input via the 3volt zener. This was tried out and works perfectly with 3volts more than the output of the 317.
5) I made a mistake there in the schematic as the digital Volt amp meter is on the negative lead. I will tie the ground of the regulator circuit to the output terminal to compensate for any voltage drop.
6) Don't forget its going to be a dual supply and there will be to many knobs to fiddle with ha ha that's why with the 10 turn pot for V adjust. it should also give fine adjustment i believe. The current pot will be a single turn pot.
6) Lastly about the filter network on the output you mean that a 1000uf cap is to high? The 1.5K resistor is to load the output just a little to prevent any voltage rise with no load. What do you suggest then?
Finally I will upload the schematics and pcb layout also with pics of the whole project for anybody who wishes to replicate.

Thanks for your help.
Best regards,
Silvio.
 

dl324

Joined Mar 30, 2015
16,846
3) As I read the datasheet for the 317 its said somewhere that either a 1uf tantalum or a normal 10uf cap will do the same job.
I'd increase it to dampen any effects of adjust current deltas affecting the output voltage.
5) I made a mistake there in the schematic as the digital Volt amp meter is on the negative lead. I will tie the ground of the regulator circuit to the output terminal to compensate for any voltage drop.
You need to have the ammeter in the feedback loop or it will affect output voltage (regulation).
6) Don't forget its going to be a dual supply and there will be to many knobs to fiddle with ha ha that's why with the 10 turn pot for V adjust. it should also give fine adjustment i believe.
I have a supply with a 10 turn pot for voltage adjust and it's a pain in the a** to adjust voltage if it takes more than a couple turns.
6) Lastly about the filter network on the output you mean that a 1000uf cap is to high? The 1.5K resistor is to load the output just a little to prevent any voltage rise with no load. What do you suggest then?
I was referring to the rectified voltage. 36VAC RMS will give you around 50VDC. LM317 allow a maximum in/out differential of 40V. If you set your supply to less than 10V, that maximum will be exceeded.
 

Thread Starter

Silvio

Joined Jul 2, 2015
14
Thanks Dennis for your reply, I see what you mean about the 10 turn pot. Well in that case it should be a 4.7k and a 500 ohm pots if i am correct.
I take your advice and change to 10 uf cap on the voltage pot with an addition of a decoupling cap to ground (0.1) correct?

The full rectified voltage is actually 36 volts and the switcher actually goes to around 31 volts at the designed load (5-6) amp each channel, tried it out with car headlamp bulbs 2 X 60 watts in series.
My input to output differential is always going to be 3 volts the switcher will track the output.
I hope I understood what you mean for the amp meter to be in the feedback loop. I know there will be a slight voltage drop due to the shunt. That's why i said i will tie the negative lead going to the circuit board on the output terminal of the PSU as this will be after the ammeter,

Thanks and regards
Silvio.
 

dl324

Joined Mar 30, 2015
16,846
I see what you mean about the 10 turn pot. Well in that case it should be a 4.7k and a 500 ohm pots if i am correct.
Unless something has changed, pots use a 1, 2, 5 sequence; so 5K for coarse adjust and 100-500 for fine.
I take your advice and change to 10 uf cap on the voltage pot with an addition of a decoupling cap to ground (0.1) correct?
Yes. You want to dampen the effect of any current variations in the 27 ohm resistor.
My input to output differential is always going to be 3 volts the switcher will track the output.
How is that being done?
 

Thread Starter

Silvio

Joined Jul 2, 2015
14
I have uploaded 2 schematics one showing the ATX switcher arrangement and another with the linear arrangement. I don't know if you saw them both or only one. The BC 327 will drive the voltage control of the switching IC the TL494.
 

Thread Starter

Silvio

Joined Jul 2, 2015
14
schematic 1 psu.PNG schematic 1 psu.PNG schematic 2 psu.PNG I have uploaded 2 schematics one with the ATX arrangement and the other with the linear arrangement.
 

Thread Starter

Silvio

Joined Jul 2, 2015
14
Unless something has changed, pots use a 1, 2, 5 sequence; so 5K for coarse adjust and 100-500 for fine.
Yes. You want to dampen the effect of any current variations in the 27 ohm resistor.
How is that being done?
Hi Dennis in my bottom schematic showing the LM301AN op amp arrangement i am to use a separate supply for supplying the opamp with a +/- 12 volts with the common line tied to ground. Most of the circuits shown tie the positive supply of the op amp to the V in. Thus these all have 15 or more volts in the V in line. Mine will have a tracking pre-regulator and not sure if it works at low voltages. At 3-4 volts on the input of the regulator there may not be enough voltage to supply the opamp. will my choice for feeding the opamp work? that is a separate supply with minus and plus 12volts.

regards,

Silvio
 
Last edited:

Thread Starter

Silvio

Joined Jul 2, 2015
14
Hi folks I am back again with finalizing my power supply. I got it all set up and its working fine. I have two problems actually which I like to solve if anyone can help me here. Please keep in mind that I have a switching ATX power supply followed by 2 linear regulators. These are composed of a re-wound ferrite transformer thus now giving two outputs of 35 volts. The linear regulator is composed of a 317t with a TIP2955 to enhance current. The other half of the power supply is exactly the same but having a negative regulator with 337t and TIP3055. Due to having a switcher which controls the input both outputs from the switcher are the same and vary by the same amount. The atx switcher is modified and can now be controlled from a minimum of 1v to a maximum of 35volts. This switcher is made to follow the linear regulator by 5volts over the output voltage of the master linear regulator thus this way dissipation on the linear regulator is kept to a minimum at all times. However the linear regulators are controlled separately with individual pots and current limiting thus having actually 2 outputs in one power supply which can have different voltages with a limitation that the negative regulator must be equal or less than the master positive regulator, as the switcher follows the positive master linear regulator. These can be made to work in series or independent.
Problems are as follows:-
1) The set voltage drifts to around 0.5 volts and then stabilizes after say 3-5 minutes. Why it does this?
2) The volt pot on the master is not linear at all though having a linear 5K pot and its rather finicky to adjust to lower voltages, also it works to around 50% and the other half is useless. On the other hand the other channel behaves well in this regard. What could be the problem here?
I used a 220R resistor with the output to the adjust pin and the VC pot. The output capacitor is a 1000Uf for each channel. Regulation from 0-5 amps is according to data sheet specs.

Your help is much appreciated.
 
Top