EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Skyrdr on June 11, 2020, 10:30:35 am
-
Hi everyone, I want to start by saying that i'm at the first year of Electronics engineering and so i'm more or less a beginner in the field.
I wanted to start an ambitious project by building a variable power supply using stuff i have already laying around for the most expensive components. My goal is more or less a 0-40v 4/5A power supply with reasonable efficency and output ripple, and with 0.1V 50/100mA precision.
I was thinking to build a minimalistic syncronous buck converter on the rectified output of a 48v transformer by using two N-ch mosfets, an half bridge driver ic (like the IR2111) and controlled by a microcontroller pwm output. I really don't know a whole lot about switching converters, so please correct me if i'm saying something stupid, but it seems that managing the output ripple is a quite a challenge if you want to have a wide range of output currents, because since it is basically an amplified pwm signal with an LC low pass filter, if you want a small precise output current the output capacitance should be low or you risk to blow up stuff, and on the contrary if you want a big current the output capacitance should be higher to avoid an huge ripple to occur.
I searched for different solutions to this problem:
-"Big" output capacitor and a switchable dummy load on the output that automatically turn on if a small current is required in order to provide always a minimum current draw for the buck converter.
I rejected this idea because i would most likely using the power supply for 5v at low currents a lot of the times and so the dummy load would be on causing a terrible efficency.
-Linear regulator built around an LM317HV or a NPN darlington power transistor + an op-amp, controlled trough DAC by the microcontroller, and connect it in on the output of the transformer with a relay to switch between it and the buck converter. The idea being using the linear regulator for small currents and the buck converter for high ones. I also rejected this idea for pretty much the same reason as before since the linear regulator would pretty much always have to drop 43v to produce 5v.
I recently started to investigate more elegant solutions: like building only a fixed linear regulator and regulate the voltage on the primary side of the transformer with a triac, with the possible downside of producing huge noise electrically and even acoustically, or just building a variable linear power supply with a big heatsink and screw the efficency. Even if this last option was tempting, as there are already plenty of designs online, I really wanted to use a microcontroller in order to use rotary encoders,big led displays and dedicated ADC with hall effect sensor for precise current sensing, instead of potentiometers and chinese volt/ampmeter with crappy displays.So since i'm already using it I might just use one of the pwm output that it has and build for a more efficent switching regulator.
I found yet another design solution that has the potential to be the perfect answer to my needs but has little to none documentation online. I saw a tear-down of the MDP-XP power supply by Miniware, wich uses a buck converter to step the input voltage down to a little bit higher of the requested output voltage, and then a linear regulator to regulate the voltage precisely for the last bit. I think this design is very clever as it combines the efficency and high power rating of a switching power supply with the low ripple of a linear regulator, and so i would like to build something similar even if I admit the complexity of such a design scare me a bit.
I want to share those ideas to someone more skilled than me to get an advice on what to do, what do you think it's a good challenging design to replicate for a semi-beginner as I am?
thank you in advance for the answers, and sorry for the long post filled with bad english :P
(I forgot to add it before but yes,I know i need a negative voltage rail in order to get the linear regulator output down to 0v, when I settle on a design I will probably swap the big 48v transformer with something with a center tap, or maybe add another transformer).
-
For the first try I would start with a lower power version, more like 0.5 A and maybe 20 V. Smaller parts release less magic smoke and the learning experience is nearly the same.
The logical solution is to have the SMPs and linear regulator in series. For a first step possibly both separate.
With a SMPS handling the ripple at the output and fast reaction to load changes is a real challenge, especially with a relatively low switching frequency. For the SMPS control there are ready made analog control chips - no need to use a µC for this. A µC would add another source of failure.
-
Do you mean to use an analog control IC to directly drive mosfets or only to generate the pwm signal? because with an Arduino UNO i was able to generate a stable pwm signal up to almost 400khz, wich seems to be a common frequency in commercial power supplies and since I already have a bunch of atmega328 micro controllers to use I don't see why i should replace them with a specialized IC. If you mean to use an all-in-one kind of IC to drive the mosfet directly in an half-bridge configuration you are right, it would be much easier, but I struggled to find any so if you can give me a hint on what IC to use it would be appreciated
-
This design can go down to a few millivolts without needing a negative control rail.
https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/msg3089254/#msg3089254 (https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/msg3089254/#msg3089254)
But it does need an independent positive rail to properly power control-and-measurement circuitry.
-
This post shows a simple way of monitoring the voltage drop across the linear post regulator to control the Buck pre regulator.
https://www.eevblog.com/forum/beginners/control-two-regulators-with-one-variable-resistor/msg3052304/#msg3052304 (https://www.eevblog.com/forum/beginners/control-two-regulators-with-one-variable-resistor/msg3052304/#msg3052304)
-
For the first try I would start with a lower power version, more like 0.5 A and maybe 20 V. Smaller parts release less magic smoke and the learning experience is nearly the same.
My first switching power supply circuit, that I built when I was in elementary school, did up to 18V at up to 1A. A few amps is well within the capabilities of common integrated switching regulators.
(I forgot to add it before but yes,I know i need a negative voltage rail in order to get the linear regulator output down to 0v, when I settle on a design I will probably swap the big 48v transformer with something with a center tap, or maybe add another transformer).
No need to use a negative supply if you use rail to rail opamps.
-
I was thinking to build a minimalistic syncronous buck converter on the rectified output of a 48v transformer by using two N-ch mosfets, an half bridge driver ic (like the IR2111) and controlled by a microcontroller pwm output.
That's certainly possible, but for a first design I'd recommend using integrated converter chip with analog controller and MOSFETs. I suggest you have a look at the Envox DCP405 power module (https://www.envox.hr/eez/eez-bench-box-3/eez-dib-dcp405-power-module.html). It's a 0-40V 0-5A power supply with a buck preregulator and linear post regulator.
I recently started to investigate more elegant solutions: like building only a fixed linear regulator and regulate the voltage on the primary side of the transformer with a triac, with the possible downside of producing huge noise electrically and even acoustically...
It's a well known technique. You can also do this on the secondary side: use two diodes and two thyristors to make an adjustable rectifier.
-
That's certainly possible, but for a first design I'd recommend using integrated converter chip with analog controller and MOSFETs. I suggest you have a look at the Envox DCP405 power module (https://www.envox.hr/eez/eez-bench-box-3/eez-dib-dcp405-power-module.html). It's a 0-40V 0-5A power supply with a buck preregulator and linear post regulator.
That project is really cool,I will look better into it but it seems to have all I wanted to accomplish plus something more, thank you for sharing that.
About using regulation on the AC side i was thinking if it's possible to pass half wave only using some kind of zero crossing detection circuit and then regulate to the mean voltage over time,in order to avoid sharp edjes that will occur by using phase cutting. but for now i think the buck+linear solution is better.
No need to use a negative supply if you use rail to rail opamps.
I was saying that because all the basic linear regulator circuits i found on a book i have use all an LM317 wich has an internal vref of +1.25v and a negative rail with a zener diode reference to get the output down to 0v, and since they seemed very simple to implement in my design i was thinking to add another transformer. But from what i'm seeing now it also seems not so complicated to build a linear regulator discretely. I will definetely use rail to rail opamps if i can do so instead of adding a secondary transformer.
This post shows a simple way of monitoring the voltage drop across the linear post regulator to control the Buck pre regulator.
https://www.eevblog.com/forum/beginners/control-two-regulators-with-one-variable-resistor/msg3052304/#msg3052304 (https://www.eevblog.com/forum/beginners/control-two-regulators-with-one-variable-resistor/msg3052304/#msg3052304)
won't a simple voltage divider work as well with those ic that have a direct feedback pin? That's however another reason,other then expandability, why I would really like to use a mc, so I'm able to change the control signals for the two regulation circuits and making them work togheter by means of a couple of line of code instead of having to tune an analog circuit.
By the way,thank you all for the suggestions, for now I think I will start to build everything more or less discretely copying something off the DCP405 project as well. If at one point I screw it up I might go more integrated, maybe even with something like an LM2596 for the SMPS and an LM317HV for the linear side :-+.
-
won't a simple voltage divider work as well with those ic that have a direct feedback pin?
I don't think it can be done with just a voltage divider.
That circuit I posted, subtracts the linear regulator's output voltage from it's input voltage and applies the result to the FB pin of the SMPS IC.
So whatever output voltage the linear regulator is set to, the output of the SMPS will track it plus a few volts.