Thread Tools
Feb 03, 2011, 03:22 PM
Registered User
LeszekJ's Avatar
Thread OP
Discussion

PC ATX power supply unit: 18V 400W


Hello,

I've just finished the conversion of my old ATX power supply for iCharger 208B. As you remember, this charger needs at least 17V on input in order to give 350W charging power. At 12V it provides max 250W so the primary goal was to increase the PSU voltage. It can be done quite easily if your spare PSU contains S494 PWM control chip (or equivalent, e.g. KA7500B).

Testing maximum power:
(0 min 37 sec)


Conversion procedure:
1) test if PSU can hold expected current continuously,
2) check if output capacitors are rated high enough for increased voltage,
3) remove internal load resistors,
4) connect rotating fan between +12V and +5V lines or use one which can run at increased voltage,
5) disable PSU overvoltage protection,
6) disable PS_ON signal sensing or just ground PS_ON green wire,
7) load +5V line with 6.2 Ohm 10W resistor,
8) replace voltage control resistors (voltage divider adjustment),
9) make a full stress test of modified PSU before charging lipos.

Voltage in my case: 12V --> 18.5V (5V --> 7.7V, 3.3V --> 5V).
When a small voltage increase is needed (12V --> 14.4V) you'll just have to accomplish points: 1, 6, 7 and 8.

The details will follow if necessary.

#3
PSU output lines are loaded with resistors. If you rise output voltage more than 10% you can expect them to overheat. Find and remove load resistors connected directly between output lines (-12V, -5V, 3.3V, 5V) and ground. In my case the line +12V was unloaded. These resistors are easy to find cause they are bigger than those in the control circuit. They are rated for power 0.5W, 1W, 2W. And they are situated very close to the output wires.


#5 and #6
Find the S494. Cut the trace to pin #4 of S494. Ground this pin with Rx = 3.9 or 4.3 kOhm. The OVP and PS_ON signal sensing should be gone. PSU starts as you power it. If it doesn't you have to ground PS_ON wire as usual.

#8 (v. 1)
The most tricky part. Find the voltage divider connected usually to pin #1 of S494. It consist of three or more resistors (dashed line). Manufacturer sometimes cuts out a resistor to adjust output voltage. You have to change the ratio of the voltage divider to cause higher duty cycle (wider pulse width) in order to get higher output voltage. I suggest to add a parallel resistor Ry. The approximate formula for Ry may look slightly crazy at a first glance:
V = 18.5 (if you like this voltage level)
Vref = 4.93
G = V/Vref*(1/R1+1/(2.4*R2))
Gy = G-1/R1-1/R2-1/R3 and -1/R4 if exists
Ry = 1/Gy
Output voltage should be changed. If PSU shuts down immediately after powering it means the overvoltage protection is still operational.
If you forgot to accomplish points #2, 3 or 4 expect some heat and smoke

#8 (v. 2)
There is another version of the voltage divider described above. The branch of the R3 may contain a fixed value resistor in serial connection with a potentiometer (potentiometer permits for voltage adjustment in some limits). You can go down with the value of this fixed resistor or try to use additional parallel Ry. Above formulas still hold. In such case R3 = Rfixed + Rpot. You have to measure or guess the actual (set) value of the variable resistor.
Last edited by LeszekJ; Mar 27, 2011 at 05:39 AM. Reason: Some details
Sign up now
to remove ads between posts
Feb 03, 2011, 04:07 PM
Ugly planes do not fly
_Sergey_'s Avatar
That's very easy! Thank you! I can easily do #4. Can you elaborate on rest of items?
Feb 03, 2011, 04:24 PM
Registered User
LeszekJ's Avatar
Thread OP
Quote:
Originally Posted by _Sergey_
That's very easy! Thank you! I can easily do #4. Can you elaborate on rest of items?
I will prepare the description and... a disclaimer.
It's for those who can do #1-4 and #7 without hints
PSU can kill inexperienced hobbyist.
Last edited by LeszekJ; Feb 03, 2011 at 07:20 PM.
Feb 04, 2011, 07:22 AM
Registered User
bmutlugil's Avatar
Good job, supplies with the 494 chip are easily found.

I had modified some supplies with the Weltrend WT7514L chip, those are not so frequently found. I made one adjustable, with 4-10V and 10-20V ranges. I used the 5V output for the lower range and the 12V for the higher. I am preparing two of those for a 24V 20A output supply.

Bulent
Feb 04, 2011, 08:44 AM
Registered User
LeszekJ's Avatar
Thread OP
Quote:
Originally Posted by bmutlugil
I am preparing two of those for a 24V 20A output supply.
Now I'm thinking rather about parallel setup and current sharing.
Two S494 can be configured as master-slave.
Feb 04, 2011, 09:46 AM
Ugly planes do not fly
_Sergey_'s Avatar
Great update LeszekJ! Where to look for #3?

When you say about two S494 can be configured as master-slave how it will work? Do you want to parallel two modified 17V PSUs for current sharing?
Feb 04, 2011, 09:59 AM
Registered User
bmutlugil's Avatar
I thought of paralelling them as well.

I think parallel operation is not so trivial, an opamp based circuit may be needed to share the load equally - making one the master is a good idea.

If load is shared well with two PC supplies, 12V-17V at 60A (supplies paralleled) or 24V-30V at 30A (supplies in series) could be obtained.

Bulent
Feb 04, 2011, 12:38 PM
Registered User
LeszekJ's Avatar
Thread OP
Quote:
Originally Posted by _Sergey_
Do you want to parallel two modified 17V PSUs for current sharing?
Exactly.
Small obstacle: I have to redesign control circuit of the slave unit
Step #3 explained above.

About parallel setup of two or more PSUs. S494s don't need to be precisely synchronized. It's necessary to modify their voltage control (and maybe overpower protection). I see two options at the moment:
a) professional approach: to use load share control chips (there are many available types),
b) naive solution: to disable voltage error amplifier in any slave S494 and connect feedback pins (3) directly to the pin #3 of the master S494.
Last edited by LeszekJ; Feb 05, 2011 at 07:00 AM. Reason: step #3 and parallel setup
Mar 26, 2011, 01:33 PM
Registered User
robin123's Avatar
LeszekJ, I have a 420W rated ATX PSU. I opened it and found a SG6105D circuit (although it is very crowded, so I have to dismantle it first pretty much in order to do some reverse engineering and obtain the schematic around SG6105D).

Previously I founded this tutorial: http://www.users.on.net/~endsodds/smps.htm. The overvoltage and overcurrent protection gives both the shutdown signal on pin 4 of TL494 (attached schematic). The author modified also the overvoltage "IN" divider to leave the overvoltage protection in place. In steps #5 and 6 of this tutorial the pin 4 of the S494 is disconnected from the original circuitry and grounded through a resistor, this is why I imagined that also the overcurrent protection is disabled. Of course I would be glad if I was wrong.
Mar 26, 2011, 04:03 PM
Registered User
LeszekJ's Avatar
Thread OP
SG6105D is a different story. Several modifications are needed.
You can find a diagram in the datasheet: http://www.sg.com.tw/semigp/data/610...-datasheet.pdf

---
Inside S494 you can find only two error amplifiers.
First is for output voltage stabilization (PWM), second one is for overpower (overcurrent) protection (it also controls PWM).
Undervoltage protection, overvoltage protection and PS_ON sensing are realized by an additional circuit (transistors or IC) which controls voltage at pin #4 of S494. If you disable it the overpower protection stays intact in most cases.

I don't know why but one of the error amplifiers of S494 is unused (your diagram). Someone decided to use one of the S339 comparators instead.
Option A - It is possible to reactivate the second error amplifier inside of S494 for OPP (input pins #15, #16) and disable all comparators of S339.
Option B - To disable comparator of S339 (pins #7, #6, #1) and to modify the voltage divider R39-R41.
Last edited by LeszekJ; Mar 26, 2011 at 04:43 PM.
Mar 26, 2011, 05:22 PM
Registered User
robin123's Avatar
Quote:
Originally Posted by LeszekJ
SG6105D is a different story. Several modifications are needed.
You can find a diagram in the datasheet: http://www.sg.com.tw/semigp/data/610...-datasheet.pdf

---
Inside S494 you can find only two error amplifiers.
First is for output voltage stabilization (PWM), second one is for overpower (overcurrent) protection (it also controls PWM).
Undervoltage protection, overvoltage protection and PS_ON sensing are realized by an additional circuit (transistors or IC) which controls voltage at pin #4 of S494. If you disable it the overpower protection stays intact in most cases.

I don't know why but one of the error amplifiers of S494 is unused (your diagram). Someone decided to use one of the S339 comparators instead.
Option A - It is possible to reactivate the second error amplifier inside of S494 for OPP (input pins #15, #16) and disable all comparators of S339.
Option B - To disable comparator of S339 (pins #7, #6, #1) and to modify the voltage divider R39-R41.
OK, I see what you mean. About SG6105 I think the tuning can be done easily (as you did for S494) up to 14.5V which is the Overvoltage protection - input pin 7. I will test this first, then I will try to find a way to "cheat" it.

Thank you,
Robert
Mar 26, 2011, 06:16 PM
Registered User
LeszekJ's Avatar
Thread OP
Quote:
Originally Posted by robin123
About SG6105 I think the tuning can be done easily (as you did for S494) up to 14.5V which is the Overvoltage protection - input pin 7.
In your case all lines are protected: 3.3V, 5V, 12V and negative voltages.
Input pins 2, 3, 6 and 7.
Don't forget to test unmodified PSU. Manufacturers cheat quite often about total output power. A few days ago I disassembled PSU labeled as 400W which PFC subcircuit was rated 350W. This PSU was able to output 300W continuously
Jun 22, 2011, 10:15 PM
Registered User
spoke2570's Avatar
LeszekJ,
I am also interested in modifying the +3V output to +3.8 or so for testing of single cell powered micro gear, while leaving the +5V line intact for standard gear. Is this a possibility in these PCUs?
Jun 23, 2011, 12:21 AM
just look at it smokin'
z-matrix's Avatar
hii

The computer power supply is not designed for 17V operation, that high operating output voltage, it will most likely need rewinding of the transformer. (adding a few turns to the 12V side)
In my spare time i have made mods to computer power supplies too, but only went up 14.2V, and i use some at 13.8V.
Note if you increase voltage and current, there is a possibility to saturate the transfomer core, and/or melt the wires, then the supply will blow up.

The 3.3V to 3.8V will probably work OK, if not, just use the 5V line...
Btw.: some power supplies have a little 12V to 3.3V voltage regulator module with trim pots to adjust voltage they work from lead acid batteries too.

But if you really like hacking things i give an idea, you can connect 2 power supplies in series to give 24V or 27.6V, you should include a reverse schottky diode in parallel with them from a bad power supply (output rectifier), and disconnect the high side power supply ground and it would be wise to place them in same metal case.
.
Z
Last edited by z-matrix; Jun 23, 2011 at 12:27 AM.
Jun 23, 2011, 06:29 AM
Registered User
LeszekJ's Avatar
Thread OP
Quote:
Originally Posted by spoke2570
I am also interested in modifying the +3V output to +3.8 or so for testing of single cell powered micro gear, while leaving the +5V line intact for standard gear. Is this a possibility in these PCUs?
How much power do you need for your micro gear?

Quote:
Originally Posted by z-matrix
The computer power supply is not designed for 17V operation, that high operating output voltage, it will most likely need rewinding of the transformer.
It has been working great for a few last months. No rewinding - just duty cycle adjustment.


Quick Reply
Message:

Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Question PC Power Supply parallel connection withing single unit _Sergey_ DIY Electronics 1 Jan 31, 2011 11:26 AM
For Sale PC Power Supplies Codegen 400W 500W rcgreaves Non R/C Items (FS/W) 0 Nov 21, 2010 07:54 PM
Yippee! Dynex 400W ATX Power Supply Williwaw Batteries and Chargers 1 Jun 03, 2010 12:49 PM
Discussion Triton Power Supply from a PC ATX Power Supply Transition Batteries and Chargers 19 May 09, 2005 07:44 AM
PC ATX power supply, Gray wire? RPatrick Batteries and Chargers 2 Jul 01, 2002 07:27 PM