A "laboratory" power supply from a PC ATX power supply ...

[RV-P]

- Hello again.
- These days, I wanted to make me a lab supply, so down to OV, from a computer ATX power supply. We find almost "under rock" so there are some who "bloom" in the outlets! Would you like ?
- So let's go !
- So here is an ATX power supply, as it stands before modification:

- As soon as you have a diet of this kind, the good habit to take is the extraction of the diagram! Using my favorite GIMP and layers, I extracted the drawing from the circuit board photo and placing the components behind!
- So here is the diagram:

- What you need to know is that the integrated circuit that "drives" the power supply is either a TL494 or DBL494, or a KA7500B. Do not worry, it's exactly the same! That's why you see the two nomenclatures on the diagram! The color chips (including the white) indicate the voltages that drive the different circuits, as well as the output voltages.
- Right away, you notice components in gray, others in red and others in green.
* For those in red, these are the voltage servo components. The majority of cases, we try to leave them as they are.
* For those in green, we modify them.
* For those in gray, we delete them! But we must proceed in stages!
- Look here at the previous modified deck. I did not represent all the components, but only those modified and removed:

- But I proceed in stages. To then try the power supply, provide you with a "box with bulbs", to put in series with the food to be tested, to avoid having the electrical installation break and AVOID BALLADING YOUR HANDS TO THE PRIMARY 230V !!! DISCONNECT ALWAYS AFTER EACH TEST and wait a few moments before continuing: the large capacitors 330μ / 200V primary can hurt a lot!
- So, ACTION:
* At the top of the red zone (overvoltage / overvoltage protection), a diode and a resistor. Tests: OK,
* Then dismantle all the components in the red zone (including a resistance on the + 5V circled in red, apart). Tests: OK,
* Removal of + 3,3V, -5V and -12V components (orange, white and blue areas). Tests: OK.
- BEFORE replacing the two resistors of 4k70 in the green zone by a potentiometer of 10k, there is a very important manipulation to make: REPLACE the + 5V and + 12V output capacitors by models that are more robust in voltage. For those of the + 5V, replace them with 16V models and for those of + 12V, replace them with 35V models. Indeed, now, the 5V will vary between 0V and 10V, the 12V between 0V and 24V!
- The double diode + 12V and the two resistors and two capacitors, which form a circuit "snubber", call a comment. By removing the power 3,3V, I had recovered the double Schottky diode to replace that of + 12V to increase the available intensity, just in case. But during a test, by raising the maximum voltage to no load, the double diode "slams" without warning: short circuit total! Fortunately, the bulbs in my light bulb box have protected and test feeding and electrical installation and have signaled the fact! But what I did not pay attention to is the nomenclature of the diode. I had replaced an F10C20 with a S20C40, which I thought was stronger in tension. Fault ! The "F" (note the letter) has a reverse voltage of 200V and the "S" a reverse voltage of 40V! Since then, I had replaced it with a reverse voltage diode of 60V before putting a F12C20 (12A, 200V!) Back in! At the same time I modified the snubber circuit 15 ohms / 10nF and doubled it between the + 12V and each secondary output of the transformer!
- Here is the diagram of the modified feed:

- It is noted here a modification of the current limiting circuit which acts by putting in series with the primary a few turns on the "modulation" transformer which transmits the pulses of the TL494 towards the bases of the primary transistors! With a power supply of 400W, it is better to do it, this limitation of intensity!
- And here is the modified feed feeding on 4 bulbs 28V 40W and on 3 bulbs 12V 15W:

- It is equipped with these volt / ammeters Chinese "flower" these days on the Net! It works "nickel"!
- Cordially !

[Petrus]

Interestingly, is the voltage stable at no load or low load?

[izentrop]

Congratulations and thank you for sharing.

[RV-P]

Interestingly, is the voltage stable at no load or low load?

- Voltage control is such that it almost does not change when empty or charged! Before modifying it, you can always try it on car headlight bulbs in 12V, to convince you!
- By the way, the food that I presented the circuit is a brand food XLESS (USA) Model: 400XA ATX 2.03 (P4) (I will later introduce another different brand food and you will see that we can not take "inca" for generalities!).
- But I continue the description because there are other changes in sight!
* You noticed that, in one of the previous diagrams, the PS-ON circuit cuts the 12 pin power of the TL494 / KA7500B, no!? ... Then we will use it to control the fan which, from was plugged into the + 12V. But now that this voltage varies from 0V to 24V, either it does not turn, or it burns! So you have to feed it somewhere else ...
- For this, we will feed it from the voltage that arrives pin 12 of this circuit (before the PS-ON circuit), not without having connected a stabilized power supply, that can be cut or put in tension to part the 12 pin and without disturbing the operation! This is the purpose of the following schema:

- Just 3 transistors (including a low-power MOS-FET), a Zener diode between 10V and 12V, a resistor handle, a thermistor and a capacitor to reliably drive the fan . As the voltage supplying the TL494 is a little too strong (between 16 and 17V), we reduce it with the stabilized power supply. Then, with the 2N7000 or BS170, we set or cut this power to turn off the fan when we turn off the PS-ON! The third transistor with the CTN thermistor is there to vary the fan depending on the temperature! Just stick the CTN on the power transformer (the biggest)!
- Here is a picture of the change:

- You will notice, from left to right and from top to bottom:
* The fan control circuit,
* the CTN stuck on the power transformer,
* The back of both Chinese volt / ammeters,
* The connection of the two potentiometers of tension and limitation of intensity!
- On the right, the fan of 12cm, consuming only 0,12A under 12V!
- That tells you? I think that as recycling, we do a lot better than throw them in waste so that the "Indians of India" recover the components for us, no!? ...
- Cordially !

[izentrop]

Hello,
Do not bother to make a power supply. A DC-DC buck converter Step-down Chinese costs almost nothing.

There is even an 5 pin command entry http://www.ti.com/lit/ds/symlink/lm2596.pdf

[RV-P]

- Perhaps ! But where I live:

Delivery:
No delivery to Reunion

- Bernique! And then ... what intensity is being charged? The adaptation of an ATX power supply can easily go up to 10A, with a breathtaking performance (over 80%)! It can serve, sometimes!
- I finish another diet that I intend this time to make a power supply SYMMETRIC ± 0 / 24V! As soon as it works, I will publish the plans, especially since there is an adaptation for Chinese volt / ammeters because they only measure the current in the negative part!
- And then ... Have you thought about recycling?
- "If you are wise", you will have all the details!
- Cordially !

[izentrop]

Your fan consumes only 0.1 A.
Otherwise a 7812 or 7805 that is often found on printed circuits, associated with a zen and a knob http://www.electroschematics.com/7015/v ... regulator/

[RV-P]

- Ah! You were only talking about the fan's power supply!? ... Sorry to scorn me!
- But what about thermal regulation?
- EDIT: no need for a Zéner! The potentiometer alone is sufficient to adjust the voltage, the slider connected to the tab "regulation", one end to ground and the other output. In addition, if the cursor reaches the + exit, the Zener heats up enormously! It is better a resistance-heel value 1 / 10ème that of the potentiometer!
- Cordially !

[RV-P]

- Hello again !
- As promised, I "gear" on another power supply 450W max that I am going to modify in symmetrical power supply ± 0 / 24V, which will make 48V in total.
- I have not finished completely because I still have volt / ammeter Chinese to receive ordered on ebay and I wait for them!
- The mark of this diet is a TOPELITE 450 MAX by CWT Model ISO230. You will realize that the diagram is totally different! That's what I told you when you should not take "inca" (said the Aztecs !) for generalities!
- It is, "as if by chance" that I found an English-speaking website dealing with the subject and the author of this article has exactly the same food I am going to show you the changes! It can help some who might believe that I "joke"!
- So, truce jokes, "at work"!
- Here first of all the circuit board of this power supply:

The color zones identify the different functions and tensions.

- Right now, you see several color ranges. It is that I want to make the reading of this photo as clear as possible so that we are not mistaken! Here are the functions:
* In red, the protection circuit, as well as the PS-ON function,
* In light green, the resistance resistors of the voltage control circuit and the intensity limitation,
* In dark green, the resistor resistors of the voltages to be monitored,
* In the orange zone, the + 3,3V and the -5V,
* In red, the + 5V,
* In yellow, the + 12V,
* In blue, the -12V.
- Here is the diagram as it was extracted from the previous photo:

It is "bushy", is not it!? ...

- So, let's start, always test on "box with bulbs", NEVER approach the hands of the primary, unplug after a test and always wait a little later to continue! Here is the modified circuit board:

There are significantly fewer components!

- This specified, "forward"!
* Disassembly of all components of the red zone, EXCEPT a 680 ohms resistor and a strap that are non-colored,
* You will notice, circled in red, a short-circuit to do by orienting the leg of the neighbor strap so that it comes to touch the edge of the hole where there was a resistance of 100 ohms and where is connected the eyelet of the wire PS -WE ! This allows the PS-ON to work as before, without the PS113 circuit! Tests: OK,
* Disassembly of all the components of the orange zone (+ 3,3V and -5V - except the coil which is the same as that of the + 12V), the zone circled with red (+ 5V) and the zone circled with blue (- 12V), including diodes + 5V and + 3,3V on the radiator (S45C20),
* We continue in the dark green zone by removing the resistances which enslave + 3,3V and + 5V, now absent (we leave that of the + 12V!) And we replace the three "low" resistors (connected to the mass) by a only 6k2 resistance. Tests: OK!
*** And now, we will make this food variable! For this, replace the two resistors in the light green area near the dark green area by a value potentiometer between 4k7 and 10k. This is what will serve to vary the tensions. At the same time, the capacitors of the + 12V are replaced by capacitors that are more "robust" in voltage (35V)! Tests: OK,
* Then, we "attack" at -12V! For this, after removing a coil that drops the alternating voltage 5V, we cut the other part of the printed circuit to isolate this part of the 5V ~ (see red line), we mount a double negative diode (the "-" in common!) made from two FR302 diodes which are mounted on the + 12V of other power supplies to which are added two other FR302 diodes in parallel:

The quadruple diode -0 / 24V!

*** This diode is mounted isolated because the radiator is connected to + 0 / 24V and is at the potential of this voltage! ***
* One does not forget to mount capacitors identical to those of the + 12V, the + to the mass, and one replaces the large self-toroid in common mode by a double because the coils of the + 5V will not be satisfactory to well filter the -0 / 24V! Here is the modified self:

Modified common mode coil, wound in Ø1mm wire! Hard-hard for the fingers!

* You will notice in the lower right a burgundy red capacitor: it is the capacitor 0,15μ / 100V which replaces a strap to isolate the secondary mass of the case in order to avoid short circuits of mass for the good functioning of the Chinese volt / ammeters we will connect thereafter! Tests: OK!
- Here is the diagram of the modified feed:

Compare with the previous diagram!

- The sequel to the next post! Corcially!

[RV-P]

- So let's continue the description of the changes, because there is still "taf"!
- In the previous article, you noticed on the photo of the common mode self, two green and yellow LEDs. These are the LEDs that indicate whether the feed is in stand-by (yellow) or on (green)! These LEDs were taken from an old 56k modem, with their plastic support that I kept. Check out the second diagram to see how they are connected!
- And now, Chinese volt / ammeters! They have a defect: they take the intensity only in the negative part! It is therefore necessary to "trick" to connect one on the positive part of the diet!
- So, I tried to connect the positive volt / ammeter only on the positive line, because I saw this diagram:

Ammeters.jpg (32.5 KIO) Viewed 25923 times

where it could be connected to the ground line. But that's a problem!
* If you connect two loads whose mass point off the ammeter positive (blue), the current consumption is the same as the restored intensity and positive 0A ammeter will show even if you pull him 10A !!! The negative ammeter is not concerned because it measures the intensity on the negative! Or installing a current mirror to measure the intensity of the positive pole, or one branch shunt intensity (blue / black wire) to the positive pole and installing a "voltage mirror" to ensure that the negative voltage between the positive line and the ground is converted into a positive voltage of the same value!
- This is the second solution that I retained, knowing that I do not change the volt / ammeter and it is disarmingly simple!
*** In the meantime I had installed on my computer the "LTSpice" software that allows you to do simulations! At first I thought about using an op amp:

- But you will notice that there is a very uncomfortable non-linearity when approaching 0V! Moreover, trying with a real circuit, it does not work! I had to use a TL062 (quadraFET) for it to work. But with this non-linearity towards 0V Grrr!
- Then I tried with a FET kind 2N3819. Again, I got a good operation, but still this non-linearity in low voltages:

- Then, all of a sudden, I thought about using an OPTOCOUPLER! In the LTSpice simulation, I had some common models. I'm doing the simulation and ............

............ draws a very beautiful linearity, which does not move a "centipoil", even very close to 0V! Then I build this montage, with different optocouplers such as the CQY80NG of which I have three copies! Here is the montage:

- Meanwhile, I had rewound "quickly on the gas" a small transformer into a galvanically isolated power supply and designed to supply as the volt / ammeter positive than negative! It has three secondary: two of 7V ~ and one of 23V ~! For it must supply the positive voltmeter with a TOP voltage to the voltage + 0 / 24V, but not by much: 26 / 28V enough! As it does not consume "masses", a small transformer taken from an old model phone charger power supply will be suitable!
- So, I am waiting for the delivery of my volt / ammeters and I stick to it dare-dare! Other photos, once finished!
- Cordially !