Yet another power supply

Discussion in 'General Electronics Chat' started by Wendy, Aug 18, 2015.

  1. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    I'm working on. I've taken some old ideas I've drawn and added to them.

    I'm starting with a modular 48V 2.06A power supply. I'm thinking of using it to feed these circuits.

    Basically there will not be a fixed current limit, it should boost the total current available with lower power supply voltage settings. Max voltage out will be 24V at 3A, going to almost 10A when set to 5V.

    There will be an indicator that shows when the power draw exceeds the 48V Power Supply module.

    I've broken it down into 4 boards, I'll probably merge the last two. I'm going to draw up some PCB layouts next.


    Board 1 - On board power supply

    B1 Loacal PS.png

    Signals

    +48V / Input
    +28V / Output
    +51.5V / Output
    -3.5V / Output
    Ground
    +DVM1 PM PS / Output
    -DVM1 PM PS / Output
    +DVM2 PM PS / Output
    -DVM2 PM PS / Output


    BOM

    U1 - LM317
    U2 - 555
    CR1.2 - 1N4454 Silicon Diodes
    D1,2 - Blue LEDs
    ZD1 - 15V Zener Diode
    SD1-8 - Shottky Diodes, Low Power, Optional
    C1,4,6,8 - 0.1uF Caps
    C2 - 1uF Cap
    C3 - 220uF Cap
    C5 - 1nF Cap
    C7,9 - 10uF Cap
    C10,12,14 - 10uF Cap, Optional
    C11,13 - 0.1uF Cap, Optional
    XFM1 - Torrid Transformer, Hand Wound, Optional
    R1 - 120 o
    R2 - 56Ko
    R3 - 2.7Ko
    R4 - 1Ko
    R5 - 10Ko
    R6,7 - 330 o


    Board 2 - 48V 2A Current Regulator

    B2 48V 2A Reg 1.png


    Signals

    +48V / Input
    +51.5V / Input
    +48VLIM Out / Output
    Ground


    BOM

    U3 - LM393
    U4 - 7555, CMOS 555
    Q1 - P-Channel MOSFET
    Q2 - NPN, 2N2222A
    D3 - Red LED, Indicator
    ZD2 - 36V 5W Zener
    SD9 - Shottky Diode, 50PIV 2A Min
    C15,17,18,20,22 - 0.1uF
    C16,21 - 220uF
    C19 - 10uF
    L1 - 220uH Hand Wound Torrid
    R8 - 100 o Variable Trimmer
    R9 - 22Ko
    R10 - 2.4Ko
    R11 - 10Ko
    R12 - 10 o
    R13 - 0.1 o 1W
    R14,16 - 1Ko
    R15 - 4.7Mo


    Board 3 - Preregulator

    B3 Preregulator 1.png

    Signals

    +48VLIM / Input
    PreReg / Output
    PS Output / Input
    -3.5V / Input
    Ground


    BOM

    U5 - LM393
    U6 - 7555, CMOS 555
    U7 - LM317
    Q3 - P-Channel MOSFET
    CR3,4 - Silicon Diode, 1N4454
    ZD3 - 5.1V Zener
    ZD4 - 12V Zener
    SD10 - Shottky Diode, 50PIV 10A
    C22,24 - 220uF
    C23,25,26 - 0.1uF
    L2 - 220uH, Hand Wound Torriod
    R16,17,21 - 10Ko
    R18 - 100Ko
    R19 - 10 o
    R20 - 2.2Ko
    R22 - 120 o


    Board 4 - Main Regulator

    B4 Main Regulator.png

    Signals

    +28V / Input
    -3.5V / Input
    PreReg / Input
    +PS Output / Output
    Ground
    + DVM, Current Meter
    - DVM, Current Meter
    + DVN, Volt Meter
    - DVM, Volt Meter
    Q4 Base
    Q4 Emitter
    Q5 Base
    Q5 Emitter
    R2 +, Current Adjust
    R2 -, Current Adjust
    R6 +, Voltage Adjust
    R6 -, Voltage Adjust
    +D4. Indicator
    -D4. Indicator


    BOM

    U8 - 1465, Fast Linear Op Amp
    Q4 - PNP Darlington
    Q5 - NPN Darlington
    D4 - Green LED, Front Panel
    C27 - 0.1uF
    R23 - 0.1 o
    R24,28 - 10Ko Pot, Front Panel
    R25 - 500Ko Trimmer
    R26,29 - 10Ko
    R27 - 13Ko
    R30 - 10Ko Trimmer
    R31 - 5.1Ko
     
    Last edited: Aug 18, 2015
  2. Dr.killjoy

    Well-Known Member

    Apr 28, 2013
    1,190
    156
    What do you plan on using this on Wendy ??
     
  3. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    I have a one question, is that +51.5V rail really necessary? I think you should be able to reference and level shift everything to be below the 48V rail and eliminate that. Adding a divider to the output of R13 lets you shift that sense voltage safely below 48V and into the common range of the opamp, without the need to up the positive supply rail.
     
  4. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    The current limiter on board 2 requires that the comparator be able to sense to the 48V rail accurately, down to the last uV. The 51.5 V for the power supply to it will allow this to happen.

    It is a general variable power supply with decent output specs. My Veleman power supply is acting up, I've repaired it once but without schematics I'm not willing to reverse engineer it to figure out the schematics.
     
    Dr.killjoy likes this.
  5. Dr.killjoy

    Well-Known Member

    Apr 28, 2013
    1,190
    156
    Hey if you have a model # or something for the PS so I can try to find a schematic for it ,if you want me too and it wouldn't be much trouble at all..
     
  6. Lestraveled

    Well-Known Member

    May 19, 2014
    1,957
    1,215
    Wendy
    On board 1 are you using blue leds as 3.5 volt zener diodes??
     
  7. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    The current limiter just measures voltage on R13, so if you move the setpoint a bit lower, and divide the 48V after the sense resistor by adding a resistor from junction R14-C15 to +36V, then you still have the same accuracy without the need for that supply rail.
     
  8. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    R13 must be on the rail, all else follows from that. It is part of the main current path of the supply. Given I will have the negitave voltage generator (-3.5V) the extra parts count to add the voltage drop resistors for board 1 is about the same.
     
    Last edited: Aug 19, 2015
  9. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    Yes.
     
  10. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    Sure, and thank you.

    This one...

    https://www.vellemanusa.com/products/view/?id=522797

    It uses 741 op amps of all things, it is easy to service. I'll give it that.
     
  11. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    Something else I'll mention in passing, this scheme I don't have to compensate for resistor tolerances.
     
  12. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    That doesn´t make much sense since either way you are setting the trip point with a trimmer pot so the absolute value of those reistors doesn´t matter.
     
  13. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    If I have to move the trimmer to 1Ko just to compensate it becomes a lot more finicky. 5% may not sound like much, but it adds up.
     
  14. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,290
    1,255
    Seems like a lot of "stuff" to make a power supply.
    I haven't tried to decipher it. Does it go to zero volts on the output? Is there current limit on the output as well as the 48 volts?
    The extra transformer is for some non- isolated gauges?
     
  15. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    While I'm not totally comfortable with board 2, the rest is pretty standard. It also has a full current limiter / regulator, which if works could be used directly for LEDs without the usual surge issue. The current is adjustable, but it doesn't have a rated maximum. I figure it can handle up to 3A at 24 V, or 9A at 5V, give or take.

    The extra transformer is indeed for panel DVMs, most of which require a isolated DC power supply to power (isolated as in isolated from the voltage being measured). If you use analog panel meters it is totally unnecessary.

    I am expanding on an idea I had a while back. The problem with analog op amps is the input rail to rail issue. By powering the op amps with a greater power supply voltage, both positive and negative, than the inputs will never be near a power supply rail. This is done by board 1.

    The analog power transistors will only see 15W max. This is done with board 3.

    Board 3 addresses a problem usually handled by multi-tap transformers, which is not feasible for a external sourced power supply. It drops the voltage down to a point where the analog transistors will only dissipate 15 watts worst case. If I just feed 48V and 2A through the transistors and short the output one of the transistors would dissipate almost 100W. With multi-tap transformers you switch which windings to use to bring the apparent wattage down, usually using a relay judging from the sound my Velleman makes.

    Board 3 drops the input voltage with a switching power supply to 5V of what is needed for the output. Even when powering a circuit at 20V at 4A it should run very cool temperature wise. Heat is always the enemy.

    Board 4 is very basic and very simple. A really simple current regulator followed by an equally simple voltage regulator. They would normally run very hot, except for the accessories.
     
    Last edited: Aug 19, 2015
  16. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,172
    397
    Do C6 & C8 need a path to ground via diodes?
     
  17. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,290
    1,255
    @Wendy
    Thanks for the edit, but to answer the question - a handful. :D
    I don't mean to yank your chain, but it seems like with 2 integrated regulators. 2 switching regulators and 2 power stages to make a 24 volt supply there must be some room to simplify it.
    I suppose we could start a contest. :rolleyes:
     
  18. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    I'm after a certain type of performance. Low heat, analog performance do not usually go together.

    That and current boosting. Like I said, board 2 is a bit of overkill, but the rest should be pretty decent. I could make simpler, done that with a LM317 decades ago.
     
  19. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,766
    2,536
    Not sure what you mean, I only need the AC component, which is rectified to create a low voltage.
     
  20. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    Well, yes, but the functions of board 2+3 and a bit from board 4 could be merged into one switcher, that would be current limited to 2A at the input and provide a voltage limited current source, and feed that to the final linear reg.
    Also, you mentioned that the linear regs should not be dissipating more than 15W, but I think that at high output current and low load impedance you might be getting more than that in Q4.
     
Loading...