nice very nice wixx....how bout a part 2 for this to put +5 volts too

like to mod this mod....LOL

all you need for 5 volts is a 5 volt regulator. there are alot of those kind of regulators and power supply designs. i build a 3.3 volt regulator for a laser diode when i was 12. so i think you all can build a simple power supply. good luck peoples

Thanks, guys. The next one will include +5, +3.3, -5, -12...basically we'll be building an ATX power supply from scratch.

The idea is to then extend that idea and make an ATX power supply that accepts +12V as input for use in a car.

im currently building a car pc (project on the overclockers.com.au forums), current design is linear voltage regulators coupled with power transistors for a max output current of 21A on each of the 3.3, 5 and 12v rails, if you'd like the diagram gimmie an email.

im using a 7805 and 7812 for the 5 and 12v rails, and an lm317 for the 3.3, purely because 3.3v fixed linear voltage regulators are hard to come by cheaply.

the current problems :

#1, linear voltage regulators require the input voltage to be 2-3v above their output, this means if you are regulating the 12v line from a 12v battery using a 7812 you will only get around 10v out of it. this isnt as much of a concern when the car is running (and the alternator is outputting) since then your input will be at 14.4v and you wont have a problem. there are 2 ways ive seen around this. the 1st is to step up the voltage before stepping down, but then you have current issues as most step-up ics cant handle a whole load of current. the 2nd is how ive set mine up, 2 x 13v zener diodes in parallel driving a 60A relay for the input to the power supply, this way the pc will only power up when the car's voltage system is at 13v or above, which will only be when the alternator is running, since most pc's will flatten a car battery inside of 15 minutes its not too bad a thing to have =)

#2, HEAT. my current design if i draw the max current (21A) off the 5v rail, which is based directly off the cars supply of 14.4v (for simplicitys sake ill say 15v). thats 10v the supply has to bleed out, at 21A thats 210W of heat that it has to dissapate.

not to mention the 3.3v's max heat output of 233w

And you though keeping an athlon cool was hard =)

Add to this the fact that the pc will probably be in the boot which is warm at the best of times, its certainly not ideal.

Thankfully no pc will draw this much current, and even if it did, voltage regulators will run at peak efficiency until around 110oC.

im considering getting some 100oC thermostats and wiring them up to the zener relay so that the pc will shutdown if the psu overheats. dunno yet...

#3, Power GOOD and PWON signals, although we can just ignore them, ram 5v down the power good line at all times, its certainly not the best way to do things, you really need to design something like a BASIC Stamp that will monitor the voltages and signals and power on the psu when appropriate. its annoying i know, but its what should be done. that said, im not doing it =), my objective is to be CHEAP


#4, negative voltage rails, while getting negative voltage from a center tapped ac transformer is easy, getting them from 12v DC is not. there are 3 ways i know of to do them

1st being a capacitor pump, which consists of 3 capacitors, 4 transistors a few resistors and a 555 timer chip, the idea is to use 2 transistors on each side of a capacitor, the first one opens up the +12v line to the positive side of the capacitor while the second connects the negative side to ground so it charges. then, when the 555 chip triggers it, the first and second transistors unlatch, and the 3rd and 4th kick in, they connect the positive side of the capacitor to the ground line, and the negative side to the output line.

the other 2 capacitors are used for timing the 555 and smoothing the output.

Its an ingenious idea but has a few limitations, primarily being that they cant handle large currents, and secondly that they produce a noisy output.

2nd is to use a switchmode power supply, these will generate negative voltages as easily as positive.

3rd is to make an occilating circuit and turn the 14v DC into AC, then use a center tapped transformer to get your negative voltages. the only problem here is that the ground lead from the transformer is isolated from the ground line of the rest of the psu, which wont work, so you need to run your whole psu off that occilator and transformer. good luck building / buying one that will do 60A =)

or you could just do what i have done, dont supply them =), the only thing current pcs use the negative voltages for these days is for the serial ports, and outside of serial lcds i cant see them being used in a car! everything else works fine (the pc im writing this on has the -5 and -12v lines clipped). parallel lcd here i come =)


thats the most of it, other than fighting with crappy diagrams on the net that have the wrong pinouts, its not going too bad.

i'll be making a website for it shortly and will post it here if anyone is interested.

sorry for the essay, hope you make it through it =)

otherwise, good article Wixx, one of the better DIY articles ive seen, nice and simple yet very well explained. cant wait to see the finished product =)

No, this is good stuff

First off in my design I'm ditching linear regulators altogether. Sure, they can handle a lot of current. They also put out a whole lot of heat you got to deal with. Plus, isn't the 7805 only rated at 1A? And you're trying to power an Athlon? What do you have, like 21 of them in parallel?

Maxim makes two parts, the MAX787 and MAX788 switching regulators that take ~12V as input and are fixed outputs at 3.3V and 5V at 5 amps each. The -5 and -12 lines are mostly ignorable, and the solution with the 555 timer is OK to a certain extent. However, you can use the Linear Technology LT1074 to function as both a buck converter and an inverter, so I'm considering using this part for both the 5V and -5V rails. +12V regulation is always the bitch, I'm looking at several buck-boost regulators although I agree with you that a bridge of zeners is probably going to be the way there.

To solve of the problem of draining the car battery, the design I have is powered off of its own +12V, 5 amp-hour buffer battery from PowerSonic. When the car is on a relay is closed to charge the battery, which the PSU ties directly into. Once the car turns off the relay is open so you can still use the PSU for a little while until the battery drains--at least you could power your MP3 player off of it.

Faking the PWR_OK and PWR_ON is also going to be tough. I'm looking at a couple of logic ICs for that, but that portion is still down the road. No sense in implementing any of that stuff until we have a working power supply.

I don't know what your design goals are, but mine is to completely avoid any DC-AC conversions at all. At that point I might as well go to Wal-Mart, buy a 12V inverter and just skip the whole power supply problem altogether.

A good resource I've found is at formfactors.org where you can download the ATX power supply specification and see just exactly what you are required to provide.

Quote: Originally posted by viper-7 im currently building a car pc (project on the overclockers.com.au forums), current design is linear voltage regulators coupled with power transistors for a max output current of 21A on each of the 3.3, 5 and 12v rails, if you'd like the diagram gimmie an email. im using a 7805 and 7812 for the 5 and 12v rails, and an lm317 for the 3.3, purely because 3.3v fixed linear voltage regulators are hard to come by cheaply. the current problems : #1, linear voltage regulators require the input voltage to be 2-3v above their output, this means if you are regulating the 12v line from a 12v battery using a 7812 you will only get around 10v out of it. this isnt as much of a concern when the car is running (and the alternator is outputting) since then your input will be at 14.4v and you wont have a problem. there are 2 ways ive seen around this. the 1st is to step up the voltage before stepping down, but then you have current issues as most step-up ics cant handle a whole load of current. the 2nd is how ive set mine up, 2 x 13v zener diodes in parallel driving a 60A relay for the input to the power supply, this way the pc will only power up when the car's voltage system is at 13v or above, which will only be when the alternator is running, since most pc's will flatten a car battery inside of 15 minutes its not too bad a thing to have =) #2, HEAT. my current design if i draw the max current (21A) off the 5v rail, which is based directly off the cars supply of 14.4v (for simplicitys sake ill say 15v). thats 10v the supply has to bleed out, at 21A thats 210W of heat that it has to dissapate. not to mention the 3.3v's max heat output of 233w And you though keeping an athlon cool was hard =) Add to this the fact that the pc will probably be in the boot which is warm at the best of times, its certainly not ideal. Thankfully no pc will draw this much current, and even if it did, voltage regulators will run at peak efficiency until around 110oC. im considering getting some 100oC thermostats and wiring them up to the zener relay so that the pc will shutdown if the psu overheats. dunno yet... #3, Power GOOD and PWON signals, although we can just ignore them, ram 5v down the power good line at all times, its certainly not the best way to do things, you really need to design something like a BASIC Stamp that will monitor the voltages and signals and power on the psu when appropriate. its annoying i know, but its what should be done. that said, im not doing it =), my objective is to be CHEAP #4, negative voltage rails, while getting negative voltage from a center tapped ac transformer is easy, getting them from 12v DC is not. there are 3 ways i know of to do them 1st being a capacitor pump, which consists of 3 capacitors, 4 transistors a few resistors and a 555 timer chip, the idea is to use 2 transistors on each side of a capacitor, the first one opens up the +12v line to the positive side of the capacitor while the second connects the negative side to ground so it charges. then, when the 555 chip triggers it, the first and second transistors unlatch, and the 3rd and 4th kick in, they connect the positive side of the capacitor to the ground line, and the negative side to the output line. the other 2 capacitors are used for timing the 555 and smoothing the output. Its an ingenious idea but has a few limitations, primarily being that they cant handle large currents, and secondly that they produce a noisy output. 2nd is to use a switchmode power supply, these will generate negative voltages as easily as positive. 3rd is to make an occilating circuit and turn the 14v DC into AC, then use a center tapped transformer to get your negative voltages. the only problem here is that the ground lead from the transformer is isolated from the ground line of the rest of the psu, which wont work, so you need to run your whole psu off that occilator and transformer. good luck building / buying one that will do 60A =) or you could just do what i have done, dont supply them =), the only thing current pcs use the negative voltages for these days is for the serial ports, and outside of serial lcds i cant see them being used in a car! everything else works fine (the pc im writing this on has the -5 and -12v lines clipped). parallel lcd here i come =) thats the most of it, other than fighting with crappy diagrams on the net that have the wrong pinouts, its not going too bad. i'll be making a website for it shortly and will post it here if anyone is interested.

Quote: Originally posted by Wixx No, this is good stuff First off in my design I'm ditching linear regulators altogether. Sure, they can handle a lot of current. They also put out a whole lot of heat you got to deal with. Plus, isn't the 7805 only rated at 1A? And you're trying to power an Athlon? What do you have, like 21 of them in parallel? Maxim makes two parts, the MAX787 and MAX788 switching regulators that take ~12V as input and are fixed outputs at 3.3V and 5V at 5 amps each. The -5 and -12 lines are mostly ignorable, and the solution with the 555 timer is OK to a certain extent. However, you can use the Linear Technology LT1074 to function as both a buck converter and an inverter, so I'm considering using this part for both the 5V and -5V rails. +12V regulation is always the bitch, I'm looking at several buck-boost regulators although I agree with you that a bridge of zeners is probably going to be the way there. To solve of the problem of draining the car battery, the design I have is powered off of its own +12V, 5 amp-hour buffer battery from PowerSonic. When the car is on a relay is closed to charge the battery, which the PSU ties directly into. Once the car turns off the relay is open so you can still use the PSU for a little while until the battery drains--at least you could power your MP3 player off of it. Faking the PWR_OK and PWR_ON is also going to be tough. I'm looking at a couple of logic ICs for that, but that portion is still down the road. No sense in implementing any of that stuff until we have a working power supply. I don't know what your design goals are, but mine is to completely avoid any DC-AC conversions at all. At that point I might as well go to Wal-Mart, buy a 12V inverter and just skip the whole power supply problem altogether. A good resource I've found is at formfactors.org where you can download the ATX power supply specification and see just exactly what you are required to provide.

hehe, ok, the #1 thing with my power supply is that it had to be cheap. Currently i can get a switchmode 550w DC-DC power supply made up for AU$300. To date, the entire psu (and i have all components required to build it), not including car wiring, and heatsink, adds up to AU$36 (US$18), its not as pretty, but thats DIY for you.

unfortunately i have 0 actual training in electronics, i have no idea what components are out there, or what they can do. ive gotten this far with electronics catalogues, google, and a huge gob of common sense.

i forget which rail it is, but i know only 1 of the negative voltage rails is actually used, and i believe it to be -12v. and theres only about 200ma of draw from it. in most respects it shouldnt be hard to supply it, but at least for my prototype i dont care about serial ports =)

as for the battery, ive designed my psu for absolute overkill (it cost me 50c more for 21A than 6A), and because of that i like to work from max draw. i cant imagine any pc no matter what you hook up to it drawing 21A off the 12v rail, but its there if its needed. from a battery, outside of the fact that i dont have the time, money, or experience to implement a buck-boost converter (only found out what one was 4-5 days ago), i cant run my system off a battery anyway. but if i could, it at peak it would flatten it much to quickly to be useful. I dont see it as a huge problem, im going to be making up a lead to run to my car while its in the garage which will have all the voltage lines, and ethernet hooked into a standard atx psu and switch inside, this way i dont need to make the hdd removable and can have full access to the pc while its at home without worrying about battery power. as for side-of-the-road listening, i dont have any problem idling my car, cars with big stereos have to.

the large current draw from the 7805 and 7812 regulators is from a circuit i found on the web, its a diode, 3 resistors, a 20A power transistor, the voltage regulator and a couple of smoothing caps. diagram is attached (just dont follow the voltage reg's pinouts in the diagram, its for a different regulator.)

ill await your reply before i get too carried away =)

Attached Thumbnails

 

All ATX power supplies are switching, which means that the AC is converted into high-voltage DC(160v), then feeds a flyback convertor to generate 12v, then buck regulators reduce this to 5v and 3.3v.
BTW, I reverse engineered an old dell supply from a p133.
That dell supply used a LITEON power transformer(at first, I thought they only made CD-ROM drives).
There is only one switching transistor(well, only one big primary transistor, as there is a smaller transistor in the 5VFP supply, and several big transistors on the secondary side), but it is a high-power IGBT boobie(20A, 900v) in a TO-3 case.
The primary caps are 680uF each, for a total of 1360uF(most PC power supplies have only 2 470uF, or 940uF total)!

eek! This is a bad circuit for use in an automobile!

You see, the power transistor is a bypass for the voltage regulator. The regulator will conduct electricity up to about 600mA load. After that the current load becomes too great, the voltage to the base decreases, and the transistor begins to conduct.

What is it conducting? It's input. So if the car is on, and the alternator is running, that's about 14V on the 12V line. If there's an electrical spike (say, from the distributor cap which sends about 100,000V to the spark plugs), the power transistor will conduct that too.

Ideally, we'd like to completely isolate the power supply from the car's electrical system, much in the way that a transformer isolates a standard ATX power supply from the line voltage. Unfortunately, a transformer cannot be used in the design, so we'll have to stick with a carefully regulated output.

star--unfortunately, with a car power supply, we don't have the benefit of getting high voltages--we're stuck with a low voltage DC input and have to somehow filter and regulate that to produce the necessary outputs. I've already ruled out DC-to-AC conversion--at that point you might as well invest in an inverter available at most auto supply stores and run a standard ATX power supply, at which point efficiency drops like a rock.

Quote: Originally posted by viper-7 hehe, ok, the #1 thing with my power supply is that it had to be cheap. Currently i can get a switchmode 550w DC-DC power supply made up for AU$300. To date, the entire psu (and i have all components required to build it), not including car wiring, and heatsink, adds up to AU$36 (US$18), its not as pretty, but thats DIY for you. unfortunately i have 0 actual training in electronics, i have no idea what components are out there, or what they can do. ive gotten this far with electronics catalogues, google, and a huge gob of common sense. i forget which rail it is, but i know only 1 of the negative voltage rails is actually used, and i believe it to be -12v. and theres only about 200ma of draw from it. in most respects it shouldnt be hard to supply it, but at least for my prototype i dont care about serial ports =) as for the battery, ive designed my psu for absolute overkill (it cost me 50c more for 21A than 6A), and because of that i like to work from max draw. i cant imagine any pc no matter what you hook up to it drawing 21A off the 12v rail, but its there if its needed. from a battery, outside of the fact that i dont have the time, money, or experience to implement a buck-boost converter (only found out what one was 4-5 days ago), i cant run my system off a battery anyway. but if i could, it at peak it would flatten it much to quickly to be useful. I dont see it as a huge problem, im going to be making up a lead to run to my car while its in the garage which will have all the voltage lines, and ethernet hooked into a standard atx psu and switch inside, this way i dont need to make the hdd removable and can have full access to the pc while its at home without worrying about battery power. as for side-of-the-road listening, i dont have any problem idling my car, cars with big stereos have to. the large current draw from the 7805 and 7812 regulators is from a circuit i found on the web, its a diode, 3 resistors, a 20A power transistor, the voltage regulator and a couple of smoothing caps. diagram is attached (just dont follow the voltage reg's pinouts in the diagram, its for a different regulator.) ill await your reply before i get too carried away =)

"a transformer isolates a standard ATX power supply from the line voltage."
Not quite.
It is the flyback convertor that isolates.

Quote: Originally posted by star882 "a transformer isolates a standard ATX power supply from the line voltage." Not quite. It is the flyback convertor that isolates.

It depends on the power supply. Maybe in one you built you had no transformer, but in the design I included in the article and in most regulated and unregulated power supplies, is the transformer that isolates the load from AC line voltage. There is no direct electrical connection between the output of the power supply and the input.

"most regulated and unregulated power supplies, is the transformer that isolates the load from AC line voltage."
The coil in a flyback regulator cannot be called a true transformer.
My friend Christina Mahoney designs SMPSes alot, and she said that a true transformer has instants where power is going in and out at once, but the coil in a flyback convertor never has an instant where power is going in and out at once.
When the transistor is on, current flows through the primary winding, and energy is stored as a magnetic field in the core.
When the transistor turns off, energy flows out of the core into the secondary circuit.
In other words, it is like using a cup to transfer water from one container to another.
BTW, she also explained the switching frequency.
Higher frequencies allow higher power density, because power goes into and out of the coil faster(back to the water analogy: it's like filling the cup and emptying it faster, so a smaller cup can be used).

Hi,
I am writing from Argentina and i wanna know if this circuit may work or not. I was planning to make the 12V from the motorola IC, but i could not find it in my city.
So i was thinking in using a DC - DC switcher, the Lm2577, to get 16V. Then it has a 7812. So at this point i would 've the +12V regulated @ 1 Amp.
To boost the current, there is a power Transistor at the input of the Lm.
I dont know if this is gonna work and i know i 've to calculate the resistors and add some caps.
So i wanted to know if it is possible. Theorically speaking.
Here is the schematic.

Guillermo

Attached Thumbnails

 

Quote: Originally Posted by guille2356 Hi, I am writing from Argentina Guillermo

btw......Welcome to PimpRig............

When you have some time............Check them out..

The Sweeet Stuff............

• http://team.pcapex.com/
• http://gallery.pcapex.com/
• http://articles.pcapex.com/

PimpRig CS Source Server..( )

• http://forums.pcapex.com/showthread.php?t=38613

This seems like quite a project. I sure do love playing with caps though (ow). But it was a very informative article, thanks.

You Do relize How ever, A ATX powersupply is a switching Power Supply and not a linear one.

BTW Ill be building A pelt powersupply soon..Something just to experiment with. ITs going to have a 12 V transformer @14Amps, but what an adjustable regulator @5 Amps per regulator. The caps...are huge...Either 5,600uF OR 40,000uf @25V..Did i mention they were computer grade?

And Did you know there IS a such thing as a cap multiplier?

if u Live in the USA I can tell u where to get excellent parts for cheap prices (mainly caps and stuff)

And also a side note, remeber the cap voltage ALWAYS has to be ATLEAST doule the voltage u want..example:12V woul;d require 24V rounded UP is 25V. Also that 12 V is RMS. To get peak there are 2 ways:

RMS/.707
or
RMS*1.4

(sorry if Im jacking the thread) So 12 V is around 16.8V peak. So if the diode rectifers drop a constant 1/4 then that leaves us with 15.4 V. Thats what we will get after Filtering

Also..There are 2 tpyes of Filtering. Capacitive and Inductive. Capacitive is the most Common and gives the highest Voltage output. Inductive is less common because a big inductor is needed,but allows heavy current Flow.

I was working on an article on PSU's ..but never got to finish it