Why do it for 3 when we can design for 10?! At college we were in charge of designing a 3 Amp power supply, but I was seeking for a challenge, not a grade!
Overcomplicating this kind of “simple” projects can make your life a bit harder but way more exiting, and the rewards just have no price, you will also learn way more than doing a small 3 Amp power supply, fortunately I had a good and experienced mentor to guide me, Lopez Hernandez Jose Antonio from the Investigation and Technologic Innovation Center (CIITEC) in Mexico City.
I completed the project in time thanks to some administrative and economic planning technics , the right tools and a bit of will power
A linear power supply takes the 120vCA 60 Hz mains power, steps the voltage down through a transformer, rectifies, filters, and regulates it. It's simple and robust; the main problem is that it's inefficient. You need a big, heavy transformer to handle the current, high current diodes, huge electrolytic capacitors, and lots of thermal dissipation for your Darlington arrays. It can also become prohibitively expensive to build high power linear supplies, even if they're easy to design. (This one was around $6,500 MXN pesos, around 433 USD considering 1 USD = 15 MXN pesos)
Here you will learn to do what I did right, and how not to not commit the same mistakes I made, mistakes are good because you do learn from them, but on the other hand you can save a lot of time and money by learning from my mistakes instead of doing your own.
Needless to say, this is my first instructable… and this my first project using copper clads… and my first time using Altium CAD PCB software… and my first linear power supply ever… and my first time documenting something in English.... so try to not be so hard on me on the comment section ;), all the feedback is welcome.
The guide is a bit long, as I said please beware, if you find any grammar , calculation or technical mistakes please let me know , and I will review them and see them fixed as soon as possible.
Along the start of each step I will be adding underlined text where I warn about what I did wrong , this bold underlined text can also be an upgrade note to do what I did in a better way, with some other ways I have learned over the time after completing the project
Every single file used to make the power supply is in my GitHub Repository feel free to use it at your discretion, hopefully you can do it better, smaller, cheaper, or why not even bigger!
Step 1: Electronic Component Suppliers
For most of the electronic components i used Newark Element 14, they ship internationally to anywhere in the world in 3 days, with free shipping (its already being charged in the components and yes they also charge the taxes)
For sourcing some other components like the voltmetters and ammeters that are a bit way expensive in Newark, eBay is your friend, however you need to be really patient to get the components since the shipping times can go up to 2 months in the worst case.
Step 2: Enclosure Design
For starters we need to set the objective and decide what we will do, we have many questions;
- How many outputs will our power supply have?
- How will I turn it on/off?
- Where will I get a custom made transformer?
- How can I measure the voltage from the outputs without using a voltmeter?
- short circuit protection?
- How will the chassis look?
Well you can start addressing all this using Corel draw to start drawing your chassis, Corel allows you to input measurements in cm/in so you can print it out and decide where will you put everything in the main panel, to do this you might want to use a Vernier to get the closest real value, and measure the components you have in hand, or you can also use the technical datasheets, in order to have the most accurate visual representation of your final product
As you can see my first draft had an analog ammeter and 2 digital voltmeters, with just 3 voltage outputs, the design was evolving over time as soon as I found a new component I liked, I kept changing the design in order for it to show a preview of how should it look like in the enclosure and even after buying and enclosure, I had to redesign it to fit the size of the front panel of the enclosure
Needless to say I ended up changing the analog ammeter for a digital one, print your design; this will help you keep in mind what or how much space you might need to do your power supply
Step 3: 10Amp Board Calculations
Upgrade Note: I should have asked the technician to make my transformer 10v+0v+10v in the secondary, since 9v is not enough for the 12v Regulators to actually Regulate the 12 volts my OPAMPs power supply can barely get to 11.8 volts, which still can feed OP Amps however I couldn’t have foreseen this since it was my first project of this kind and when I noticed it was too late
Now, we have to define which outputs will be able to provide the 10Amps, i decided to make 2 boards,
- 10A board, with 5v and the variable 0 to 25v both will be able to output up to 10Amp
- OPAMPS Board, with fixed -12v, 12v, and a variable negative outputs, the 3 of them limited to 1 Amp
These last 3 will be limited to the maximum output current of the regulator 1 Amp, since these will be used to feed mostly OPAMPs, so it makes little sense to amplify its current , each regulator can provide enough current to feed several OPAMPs,
All the outputs will be feed by a single, custom made transformer that will have 3 outputs in the secondary winding, 9v, 0v and 9v, this will allow us to sum up to 18v from the 9v+9v for out LM317 and 0v+9v for all the other regulators, so we will use this as our starting values for our calculations
We need a passive electronic device to suppress conducted interferenCe present on the mains, for this we will use an EMI filter rated at 6Amps since our primary current wont surpass 3Amps plus this provides dual fuse holder for live and neutral connections and the case of the filter its connected to mains earth which will allow us to ground our enclosure with little effort
18V is the RMS value, to obtain our Peak value we have to multiply it by sqrt(2) so:
Vp= 18v * (1.414) = 25.45 Vp
We will be using this value as our Vmax value from now on
Diode Bridge GSIB2560
To select a diode bridge, we need the maximum power that will flow through the entry of the circuit this is estimated by the maximum current plus a surplus of 30%
Pmax = 10+0.3(10) = 13A
The filtering capacitor to lower the rise is given by the maximum current, period of the signal after the diode bridge divided by the peak voltage minus the Regulator Dropout Voltage of 2.5v
C1= (13A-(1/120Hz))/(25.45-(5 - 2.5)) = 4700uF at 63v
We could also choose a lower value 43v no problem, but since the cost the same we will go with the bigger ones this time
Choke Inductor 7448262510
To protect our circuit against transients, we need inductors at the entry and the output, there is little math involved in the selection of this component so we will just pick up a 25Amp one
Darlington Array MJ11015
To calculate the transistors Base current we need to select a Darlington array transistor first, and then do the calculations for it using its hfe so we will go with MJ11015, since its supposed to be able to hold 30 Amperes of current and work up to 200 degrees C (it doesn’t I actually killed one by getting to 90C), according to its datasheet it has a hfe of 1000, then
Ic= Ib (hfe)
Ir= Ib+Ireg = 10mA + 100mA = 110mA
R1= Vbe/Ir = 3.5/40mA = 33 Ohm at 10watts
Protection Diode RURG5060
D2 is an ultra-fast recovery diode, to protect the regulator from inverse polarity and short circuits at the output
Output Filtering ECO-S1JA472CA and ECO-S1JA682EA
Same value as input filtering to assure a clean DC signal
For this board we will be using the classic LM7805 for 5v Output, and LM317 for our variable positive output the first is cap capped at 1 Amp max Current and the LM317 can output up to 1.5Amps, however since we are using a resistor to limit the current they will handle very low currents while the Darlington makes the hard job in order for our outputs to output 10Amps, while the voltage regulators, are actually handling 10mA currents
Step 4: OPAMP Board Power Supply Calculations
Most of the entry circuit up to the Diode bridge can be reused for our second low power board, this means we will plug a connector in parallel to the output of the diode bridge, very few values will change, the main difference will be the size of the components, and current capabilities
However you need to be careful with the pins of each regulators, they keep changing from model to model, so make sure you check the datasheets every time, also another thing worth noticing is that you need to turn around the electrolytic caps for all the negative value regulators otherwise they will just blow up
Filtering ECO-S1JA472CA and ECO-S1JA682EA
9v is our half secondary RMS value, to obtain our Peak value we have to multiply it by 1.414 or sqrt(2) so:
Vp= 9v (sqrt(2)*2) = 25.45 Vp
We will be using this value as our Vmax value from now on
C2= (1A-(1/120Hz))/(12.72-(5 - 2.5)) = 4700uF at 63v
We could also choose a lower value 43v no problem , but since the cost the same we will go with the bigger ones this time Choke Inductor 7448262510
Step 5: Heat Dissipation
The Power Formula P = IV
If a current I flows through through a given element in your circuit, losing voltage V in the process, then the power dissipated becomes heat product of that current and voltage to make sure the regulatoers and darlingtonscan handle the power. we need to undestrand a bit about heat dissipation
TO-220 package is rated can usually dissipate up to 1 W by itself for 1.5 A output current and up to 35 V input voltage. Naively, you might guess that you can hook this right up to 35 V input, and expect to get 1.5 A of output, meaning that the regulator would be radiating 30 V * 1.5 A = 45 W of power.
But the short answer is no, TO220 can’t handle that much power. If you look in the datasheet under the “Absolute maximum ratings” section, to try and find how much power it can handle, all that it says is “Internally limited”, which isnt very clear on its own.
It does turn out that there is an actual power rating, but it’s usually somewhat “hidden” within the datasheet. You can figure it out by looking at the following
- TOP, Operating junction temperature range: -40 to 125 °C
- RthJA, Thermal resistance junction-ambient: 50 °C/W
- RthJC, Thermal resistance junction-case: 5 °C/W
The Operating junction temperature range, TOP, specifies how hot the “junction”, the active part of the regulator integrated circuit, can be allowed to get before it goes into thermal shutdown. (The thermal shutdown is the internal limit that makes the regulator power “Internally limited”.) For us, that’s a maximum of 125 °C.
The thermal resistance junction-ambient RthJA (Often written as ΘJA), tells us how hot the junction gets when
- The regulator is dissipating a given amount of power
- The regulator is sitting in open air, at a given ambient temperature.
Suppose that we need to design our regulator to only work under modest commercial conditions, that will not exceed 60 °C. If we need to keep the junction temperature under 125 °C, then the maximum temperature rise that we can allow is 65°C. If we have RthJA of 50 °C/W, then the maximum power dissipation that we can allow is 65/50 = 1.3 W, if we are to prevent the regulator from going into thermal shutdown. That’s well below the 4 W that we would expect with a 1 A load current. In fact, we can only tolerate 1.3 W / 4 V = 325 mA of average output current without sending the regulator into thermal shutdown. This, however, is for the case of the TO-220 radiating to ambient air– almost a worst-case situation. If we can add a heat sink or otherwise cool the regulator, we can do much better.
The thermal resistance junction-case, RthJC. This specifies how much temperature difference you can expect between the junction and the outside of the TO-220 package: only 5 °C/W. This is the relevant number if you can quickly remove heat from the package, for example if you have a very good heat sink hooked up to the outside of the TO-220 package. With a big heat sink and perfect coupling to that heat sink, at 4 W, the junction temperature would rise only 20 °C above the temperature of your heat sink.
You can then verify the couplings and relative temperatures of each component with a spot-reading non-contact infrared thermometer ot ar a thermocouple.
Step 6: TO-3 Darlington Heat Disippation
We need heat dissipation for our Darlington’s, since the currents above 500mA will be flowing through them, a dargliton array has a base, emitter and collector, in the TO-3 package the case itself it’s the collector, meaning the case its part of the circuit
We will need to buy some specific, items in order to insulate them from the heat sink itself, the chassis from the pins and, to be very careful of not short circuiting anything around them, this is probably the most complicated part, in the entire design, since not only we have to solder AWG 14 wiring to each pin of our Darlington’s, we also need to be able to insulate the pins that go through the enclosure box itself, and we also need to add an insulator to the TO 3 case, here’s how to do it.
The metal package can be attached to a heat sink, making it suitable for devices dissipating several watts of heat. Thermal compound is used to improve heat transfer between the device case and the heat sink.
Since the device case is one of the electrical connections, an insulator is required to electrically isolate the component from the heat sink.
TO 3 insulating mica
Allows us to insulate the Darlington from the case itself, it may be made of mica or other materials with good thermal conductivity in order to dissipate the heat
These allow us to insulate the screws that hold the TO3 in place to theheat sink and to the chassis by using metal screws they become part of the collector, by touching the case, an alternative is to use nylon screws, that wont be conductive, however they are not as strong as metal screws
We can’t actually see it, but there is air and lots of holes between the mica and the chassis, we need to fill that air gap between the mica, and the case with thermal compound this will allow the system to have a better thermal conductivity and the heat sink will take most of the heat
We will have to solder the emitter and base pins together along a AWG 14 cable, there is a lot of stuff going on inside the case, so to be sure we don’t get any kind of shot circuit problems with the pins we might as well insulate them using heatsrhink
Step 7: Prototyping and Testing
Now that you have gathered all your components you might want to spend some time in the breadboard sadly the size of the components can actually damage and deform the internal pins and the plastic holes, this might be a problem but nothing we can’t deal with a bit of creativity, remember that the maximum current that the breadboard can handle goes around 3A, after that it just starts melting
To test the Darlington’s you might want to buy the following resistors
- 2Ω 25W
- 1Ω 25W
- .5Ω 50W
The higher the wattage the higher it can hold together at higher currents
Let’s have an example using the LM7805 that can output 5v
Using Ohms law we know that
V = IR
We can calculate the theorical output current using every resistor
I = 5v/2Ω = 2.5A
I = 5v/1Ω = 5A
I = 5v/.5Ω = 10A
To do this test you might want to use a good quality silicon Banana-Alligator cables thick enough to handle the current that will flow through the resistor and an ammeter in series with fuses big enough to handle the maximum output current, if you don’t have this kind of ammeter, then you can just avoid testing the .5Ω resistor.
Just make sure you do the math before doing this , and be careful with the time you leave them plugged since the breadboard can start melting and the resistor can start getting to way hot to touch up to 150C
Eventually something will start burning, the current its way to much for the little AWG 20 wires i used and for the breadboard, so you will barely have time to see if the circuit is actually working as intended, you can start by testing the Vreg only to make sure you have all your connections fine, and then you add the Darlington to test it using your resistors individuality, but remember you need to be fast and always ready to deactivate the input power to your breadboard
Step 8: PCB Design
Now for the design I used Altium Designer, you can use your preferred PCB Design software, however no matter what software you use, there are few “rules” to might want stick to for this 10A design
- Fiber Glass 1oz copper clad (Can handle more heat)
- 40Mils per Amp
- 10mils between tracks
- Ground Plane
- One side PCB
From the 1st PCB you can see that the tracks are huge, it mostly depends of the thickness of your pcb but if you don’t know what thickness does your copper clad have then its most likely a 1oz copper clad (the thinnest) , you can’t trust the auto-router to do this job, you will literally have to route every single line manually, in order for it to stay 1 side PCB, this will take a considerable amount of time in order to get it right, you might want to start by drawing your ground line around the area of your PCB this will allow you to quickly get to the ground no matter where your components are, but still try to remember that PCB design is 90% placement and 10% routing
I did my print my circuit board using a single face PCB, since as I said at the start of this guide this was my 1st PCB I had ever done, and I had the theory and information to do a double side board but I decided not to in an attempt to keep it as simple as possible and lower the chances of error
As you can see from the pictures I had to bridge using an AWG 12 wire, it took me around 2 days to come to the conclusion that it can’t be done, completely in one face and so I decided to “cheat” and bride it,
Another thing you need to take into consideration is that you might want do measurements one by one your components using either a Vernier or the manufacturers datasheet , this will allow you to draw the footprints of your components, after you think you have it as it should be, you can print it out use this foam rubber and star plugging your components there, this way you can decide if you have enough space to solder, or just verify if you have enough flexibility, this is quite useful, to have a “physical preview” of your PCB
Step 9: Testing Phase
This is the time of truth, all the theory finally has become a physical project, in order for you to test the circuits you will end un using the high power resistors and your ammeter
To test your circuits might want to use a good quality silicon Banana-Alligator cables thick enough to handle the current that will flow through the resistor and a multimeter in series with fuses big enough to handle the maximum output current , if you don’t have this kind of ammeter, then you can just avoid testing the .5Ω resistor.
Just make sure you do the math before doing this , and be careful with the time you leave them plugged since the breadboard can start melting and the resistor can start getting to way hot to touch up to 100C
Eventually something will start burning, the current its way to much for the little AWG 20 wires i used and for the breadboard, so you will barely have time to see if the circuit is actually working as intended, you can start by testing the Vreg only to make sure you have all your connections fine, and then you add the darlignton to test it individuality
Step 10: Preparing the Enclosure
We need to drill the enclosure, in my case this was probably the hardest step since I was using a really big, and thick metal enclosure
To start the drilling phase I used tape to stick the printed predesigned panel made in Corel to the chassis, I had to drill first using 1/32 Drill bits, to “draw” the contour of the components, slowly and every 2 millimeters, and then I used a ¼ drill bit in order to take the metal piece completely off, if you follow my steps you really need to have a lot of patience if you use a metal enclosure like mine, in my case it took me 2 entire days to complete drilling the entire panel, and my arms were ended up in pain since im not used to do this kind of work,
It might be easier if you use an acrylic enclosure or any other kind of material, is up to you to this point, after ending the drilling and sanding the sharp things, in order to take off all the residues of the drilled edges, this will is probably the most dangerous part, you might want to use protection glasses, a mask and gloves to do this, I ended up using a sandpaper and sometimes a dremel however the dremel has problems with its own disposable metal sanding bits, they run out really fast, so I ended up using a sandpaper for most of the edges, if you have any magnets around you can use them to pick up the metal residue from your work area, just make sure you don’t have any kind of electronics or fans while you work with metal since it can cause havoc around your work area
Step 11: Panel Devices
Panel Led Indicator 127vAC
How to make sure that the EMI filter fuses are actually ok? Well you do it by adding a Panel LED indicator light to the main panel, this green led will turn on as soon as the EMI filter gets AC power this allows you to tell that the rear fuses are ok and that you can turn on the power supply, right below it, it’s the turn on sptd switch, this one also has a back light in order to tell if it’s on or off,
SPDT Switch 120v 15A
Right below the Led indicator, in series we can find an rocker switch, this one will help us to turn on or off as fast as possible the supply in case of an emergency (mostly electronics getting burned), this one also has its own back light, you need to make sure the rocker switch can handle the 120vAC in order for it to work, (I plugged a 12v one at my 1st attempt)
Digital voltmeter Four Digit 0.36" Red and Blue LED digital voltmeters
We have 5 voltage outputs, but we only need to have displays for 2, the variable positive and negative ones, these 2 we will use red voltmeter for the positive voltages and a blue for the negative ones, as you well now voltages are measured in parallel, however these voltmeters also need to be powered up, how do you power up meters? Well... for once you can use the inputs of your +-12v power supplies,
Digital ammeter Three Digit 0.56" Green digital voltmeter
Current is measured in series, we only have 1 ammeter and we will use it for the biggest output power supply (the positive variable one LM317)
Temperature meter Yellow 3 Digit
In order to make sure the darlingon transistors that already have a heat sink do not die (I blew one by reaching the maximum temperature) we need some kind of meter hanging around back there with our heasink, this will work as a warning when the temperature goes to high, and we can either shut down the supply or let it go on, the probe needs to be between the heat sink and the Darlington, to have the best possible measurement of temperature, but keep it safe and use a 10% tolerance in the measurement you see, it will update in real time but the Darlington array I killed died when the meter had a 79 in it, (so it probably got to 100C)
Most of the meters are being feed by the 5v power supply, since it can also handle 10amps, some other need more voltage so they are being feed by the 12v power supply
Multiturn potentiometers Bourns 15 Turns
This one is a must for any kind of variable power supply, the multiturn potentiometers will allow us to set any voltage to both variable voltage regulators in 15 turns, and this means we can have exactly the value we want with minimum effort saving time and energy,
Multiturn potentiometers knobs H-22-6a Bourns 0-15
If we have spent all that money in the circuit, we better get fancy with the potentiometers, these caps also allow the user to set the presition of the turn by making them easier or harder to turn, you can also set them to a locked position, and they have a tiny little indicator for the turn number
Step 12: Fuse Blown Indicator Circuit
We need protections for our outputs, as well as indicators to tell us when such protections have been activated typically a short circuit will destroy our fuses, in order for it to protect our load and our power supply, and prevent it to get damaged (or burned)
You might be thinking about using a polyfuse which is a resettable fuse (I did) however the response in time of the resettable fuses is not as good as a simple fast acting glass fuse, and they also decay over time
In short, stick to the fuse holder and our trusty glass fuse, it might be annoying to change them constantly if required but it’s probably the best protection to stop something as catastrophic as a short circuit,
You have 2 options here, Simple indicator and MOSFET indicator, whatever you chose will have its fuse holder in the chasis anyway
- Simple indicator.- the easiest of the fuse blown indicator circuits, in this circuit, when the fuse blows the current needed to turn on the circuit flows through the load, meaning that if you unplug the load from your power supply it will turn off, it will also be as bright depending on the impedance of the load
- Mosfet Indicator In the mosfet circuit you don’t need an actual LOAD for the led to turn on, as soon as your fuse blows, you will see the led on with or without load at maximum brightness assuming you calculate the resistor right using ohms law, about the mosfet basically gets activated by a voltage above 2.5v between Source and gate, this allows the current to flow from Source to Drain in the circuit, using the PMOS for positive voltage and NMOS for negative voltage, the zener diodes are there to protect the gate from any kind of transient that can kill the gate, in the NMOS the led goes fliped since im using negative voltages from the source
Step 13: Tunning It Up
Our circuit works, its protected, however its missing a bit of art, I was getting all kind of jokes about burning the building down with the power supply so i decided to make it symbolical and add flames to the enclosure, using adobe illustrator i designed a couple vectors, printed the outlines and just used scissors to carefully cut them as intended and finally paste them to the enclosure along with my college logotype at the top of it
Step 14: Conclusion
If you got to this step congratulations, you have a lot of patience and a lot of stamina to keep reading up to this point, hopefully you might learn something from this guide and do an even bigger amperage power supply, maybe a 25Amp, it will be a lot of work and sacrifice but the reward is just as big if not even bigger, a power supply is one of the projects that you have to do, because it will teach you the basis of electronic design and if you keep going in the electronics way it will be useful, i will keep up reading the comments and updating this guide going into deep detail of everything that i havent explained yet slowly every weekend new content should arrive, until every single thing in this guide can be replicated or upgraded
Step 15: Costs, Mistakes and Something More
Lucky few mistakes were made, many had something to do with fans, be sure to add a protection to the fans so no one can put a finger in there and break a wing!