First regulator. Constructive critique please

Discussion in 'General Electronics Chat' started by jimmiegin, May 27, 2014.

  1. jimmiegin

    Thread Starter Member

    Apr 4, 2014
    49
    0
    Hello all. This is my first regulator schematic. I have re figured some of the values after some help from another forum and would like to know what you guys think of this schematic. I have not yet included an error amp or voltage divider as when I ran a rc4558 in multisim pro it told me the output was 34KV!! What the hell?
    Anyhow here is the schem' so far

    Regulator 300V.png

    I was wondering if a 317 could be added as a ref for the error amp (when added) to make the output of the HV variable?

    Thank you for reading :)
     
  2. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,542
    1,251
    I assume there is something wrong with your transformer notation. You show a rectified voltage of 335VDC, which means the transformer is 1:1. Is this really what you want? For one thing, Rz is dissipating over 13 watts.

    What is your intended output voltage and current from the completed regulator?

    ak
     
  3. jimmiegin

    Thread Starter Member

    Apr 4, 2014
    49
    0
    There is no mistake :)
    I have calculated the following assuming a 335VDC.
    The Regulated output is 300V 200mA full load
    Zener D5 is (RD5.6F) 5.6V 1W from datasheet.
    Iz=P/V= 1/5.6 = 178mA, less 10% fudge= 160mA. 25% of which is (optimum I) 40mA.
    so for Rz i went 335-5.6V= 329.4 so 329.4V/0.04= 8k2...ish.
    329*40mA= 13.2 watts disipation
    I assumed Hfe of 6 for Q1 and Q2 darlington pair so 6x6= Hfe of 36
    For Q1 to source 200mA to load it would need 33mA base current.
    so Q2 would need 5.5mA at it's base.
    I add 5mA for base drive transistor, 5.5mA+5mA= 10.5mA at Q1 base.
    So for 300V regulated out, Rb must drop 335-300= 35, then minus 1.2V for Q1 Q2 Vbe= 29V at 10mA.
    29V/10.5mA= 2k7...Ish...I used 2.74 for Rb. 0.3W disipated.
    For the base shunt resistor Rsb, the voltage
    accross it is equal to Vbe of Q1?
    R=0.65/0.002= 325ohms Is this correct?

    For Q4 emitter resistor Re... Q4 will drop a
    serious voltage so the resistor is there to limit current
    So i work it out as 306/5mA= 61k2 1.5W
    Aside from real world soa can you see any problems?
     
  4. jimmiegin

    Thread Starter Member

    Apr 4, 2014
    49
    0
    I mention the 317 as I have seen it in floating regulator supplies to 150V which gave me the idea of using a separate 317 regulator circuit within normal operation voltage and hooking it up to the non inverting pin of an error amp to vary the ref voltage so as ref (317 output) voltage increases to the non inverting pin it would drive the error amp out voltage up which in turn would increase current through Q4 of my schematic and so reduce the base current of the pass element and lower the high voltage output. Kind of like a regulator riding on a lower voltage regulator. I thinking of the 317 as it would be a good stable ref maybe? just figure the voltage difference between invert and non invert of op amp to raise error out enough to turn on Q4 to reduce the output of HV section. Any thoughts?
     
  5. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,542
    1,251
    A few...

    Anytime you have hundreds of volts across a single resistor, it is common practice to use two or more resistors in series. This limits the damage caused by a single-component short, and spreads the voltage out across more distance to reduce the voltage stress on the parts, the pcb material, etc. Also it reduces problems cause by contamination of the pcb surface (such as dust and pollen in an aircraft hangar in Georgia, a problem I'm fighting with a power supply vendor right now). But a better way is to eliminate the resistors from the design.

    If you put the zener in the emitter of Q4, so Q4 now has an effective Vbe of 6.2V, you eliminate Rz and its 13 watts. This is a common alternative to a true differential amplifier in the regulator circuit.

    http://circuit-diagram.hqew.net/Discrete-Voltage-Regulator_4655.html

    Resistor Re (which should be Rc since it is a collector resistor) isn't there just to limit the current into Q4. It also reduces the maximum voltage drop across Q4. with Re=0, the power dissipated in Q4 is 5mA x 300V (ish) or 1.5W. 300 V collector to emitter and 1.5 W is trivial for this part, so you don't need the resistor.

    What is the purpose of Rsb? It diverts current away from Q3's base, so doesn't that change your Rb calculation?

    About the LM317 - is the output of this regulator to be adjustable? If so, what is the intended adjustment range? If not, I think a 317 complicates things more than necessary.

    ak
     
  6. jimmiegin

    Thread Starter Member

    Apr 4, 2014
    49
    0
    Eliminating Rz is a good idea. But I'm not sure I follow you with the board contamination. Is it dust build up across the board that creates a kind of parallel resistance? If so might this be the reason I have seen cut-outs of board area covered by the resistor body and leads?

    The resistor Re which I forgot to rename Rc is there as I planned to replace the transistors with lower power rating where possible. As it has been pointed out to me that using BJTs as pass elements is not the grooviest idea I will continue to use this design but for educational purposes as this is my first transistor design.

    Rsb is included to bring down Q3 base when collector base leakage current increases with heat. As Q3 requires 33mA at its base to source 200mA load current at Hfe (worst case) 6, and Q1 requires 5.5mA at its base Hfe (worst case) 6 I doubt that 2mA (Rsb) is worth calculating for Rb as it is only .33mA.

    The question of the 317 was really just theoretical. If I connected 317 out to the non-inverting pin of the error amp and varied the output voltage (317) would the output of Hv voltage increase and decrease in sympathy?

    Could you please tell me what the considerations are when designing around a MOSfet as a pass transistor ?

    Thank you for your time and knowledge.
     
  7. Lestraveled

    Well-Known Member

    May 19, 2014
    1,957
    1,215
    I would like to suggest a process for you to go through when you design circuits, especially high power/voltage circuits. When you have an initial design, stop, and do a quick "what would happen if" analysis based on parts failing. As an example:


    If Q1 shorted while the power supply is under load or shorted:
    - Q3 would go to full conduction sending the output to max voltage or current.
    - Q2 would see very high B-E current but since Q1 is shorted Q3 can not be turned off and the output will stay ay max output, (no over current protection). Q2 will probably burn out from excessive B-E current.

    The next step is to adjust your circuit to minimize the bad effect of component failures.

    An example
    When Q1 shorted it sent the output to max and disabled the current protection.
    - The collector of Q2 should drive the base of Q3 directly. In other words, do not design protection circuits that have to depend on other circuits in order to protect.

    Also related:
    Q2, your over current protection transistor, will see excessive B-E current in an over current event. (Design Philosophy: Your protection circuit should not be stressed when it is sensing the parameter it was design to protect against.) A resistor in series with the base of Q2 will prevent excessive B-E current.

    The above assumes Q3 is the only BUX33b and all the other components are correctly sized for their power dissipation.

    I hope you don't mind my critique. It is to assist you in being a better designer.
     
  8. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,542
    1,251
    I disagree with your other advisers. A BJT is the better pass element for most linear regulator applications. Among other things, they are more reliable during handling and assembly. Also, while BJT base current is higher than MOSFET gate current, the voltage overhead is significantly lower. To turn on a power BJT you usually need less than one volt Vbe, as opposed to 5 to 10 volts Vgs for a MOSFET.

    The ultra-low gate current is nice if you have the overhead, but that overhead appears in the pass transistor as heat unless you have a separate low current gate drive supply. Also, most power BJTs have hfe much larger than 6. Even the old 2N3055 has a gain of 20 or more, and many of the TIP series have gains of over 100 at a collector current of 1 A. In a Darlington arrangement, the base current is measured in microamps, not milliamps.

    About the LM317, it sounds like you want to use it as an adjustable reference. It will work, but a simple zener and pot will do the same thing and let you regulate down closer to zero volts.

    ak
     
    Last edited: May 28, 2014
  9. jimmiegin

    Thread Starter Member

    Apr 4, 2014
    49
    0
    Thank you for the advice. For this part of the design process I would like to get to grips with basic considerations calculations and so on, I have been thinking about some sort of current controlled cut off switch to prevent catastrophic failure, also a temperature controlled switch. I consider over current protection and not blowing up my projects, or myself, to be a primary consideration so your input is gratefully received and I will adjust my schematic accordingly and make notes from your reply.
     
  10. jimmiegin

    Thread Starter Member

    Apr 4, 2014
    49
    0
    In post seven, it was suggested that placing a series resistor at the base of Q2 would reduce base emitter over current. This now leads me to wonder.. What are he advantages of foldback current limiting? I guess limiting of excessive base emitter current through the safety circuit transistor itself is one? Can someone explain please?
     
  11. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,542
    1,251
    With constant-current (CC) limiting, there still is power flowing into the load, and depending on the current limit value it can be larte enough to sustain flames and increase damage beyond the initial failure that sent the system into limiting in the first place. With Foldback (FB) limiting, power to the load is reduced significantly or completely shut off.

    The tradeoff is that FB is basically a latch, and the usual way to reset the latch is to cycle power to the system which might force a reboot of other things that still are running ok. There are some power supply designs that go into "hiccup" mode, where the output tries to restore itself periodically by making a short output pulse and seeing ifthe output current still is excessive. A much more complicated regulator design.

    FB - no continuing circuit damage, usually manual intervention to recover
    CC - some continuous output power possible, automatic recovery

    ak
     
  12. Lestraveled

    Well-Known Member

    May 19, 2014
    1,957
    1,215
    AnalogKid described foldback verses constant current very well. It is basically up to you how your protection circuits function. You get to choose the mix of complexity verses cost verses performance.

    We are just here to help you tighten the bolts to the right torque.
     
  13. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
  14. jimmiegin

    Thread Starter Member

    Apr 4, 2014
    49
    0
    Yes the supertex lr8 is a good way to go for the hobbyist like myself and I do have an lr8 I ordered a few weeks ago which I shall be using for a supply some time soon. My main reason for the discrete schematic was just to learn how as I have not designed with transistors before. Only glassware and op amps for small audio signals. Thanks for the links and help guys I shall post and as for guidance further in the development of my discrete psu. :) Thanks all
     
Loading...