Shop OBEX P1 Docs P2 Docs Learn Events
Multiple rails power supply. Ideas ? — Parallax Forums

Multiple rails power supply. Ideas ?

AleAle Posts: 2,363
edited 2009-09-16 16:41 in General Discussion
I have this crazy idea for a project where I need many different power rails:

1V ~200 mA continuos, 500 mA peak.
1.2V ~200 mA continuos, 500 mA peak.
2.5V 50 mA peak.
3.3V ~400 mA continuos, peak 800 mA
5V ~120 mA continuos, maybe 200 mA peak.
30 V 10 mA continuos, 20 mA peak.

No battery power needed.

If I use linear regulators I'll need : ~1A continuos and ~2A peak.
There are several possibilities and maybe a combination of linear and switching regulators, any ideas to keep part count down ?:

LTC3417 dual DC-DC step-down is a nice regulator but it costs 5€ a piece.
A LM1117-3.3 and maybe a 2.5.
The 30V I was hoping on using a MC34063 (cheap).

Any ideas ?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU

Comments

  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2009-09-14 11:24
    You're right, it is a crazy idea.

    An MC34063 step-down from 30V sounds cheap but the size of the inductor and capacitor doesn't make it so. Besides these currents are accumulative so you need a switcher that can handle around 2amps peak.

    Use a high-frequency switcher such as the LM2596 or better (higher freq) to switch down to +5V. The 3.3 is easily taken care or by an LM1117 whereas you could go down to 1.25V (I think) with an adjustable LM1117 strapped to ground. 2.5V is somewhat similar but just use a little SOT-23 pack. Now the 1V is a bit of a funny one but you might be able to do it with an emitter follower referenced from the 2.5V line (300mw avg dissipation).

    Ok, what gives, what is this crazy scheme?

    *Peter*
  • kwinnkwinn Posts: 8,697
    edited 2009-09-14 19:56
    How about a line operated 5V 3A switcher, a 3.3V linear regulator, and a quad op amp with drive transistors for the lower voltages and the 30V supply. The transformer for the 30V would be very small and cheap.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2009-09-14 23:34
    Sorry, I thought you were stepping down from 30V. What is the input voltage? You only need around 5VA normally but be able to handle 10VA peaks.

    *Peter*
  • CounterRotatingPropsCounterRotatingProps Posts: 1,132
    edited 2009-09-14 23:53
    kwinn said...
    How about a line operated 5V 3A switcher, a 3.3V linear regulator, and a quad op amp with drive transistors for the lower voltages and the 30V supply. The transformer for the 30V would be very small and cheap.
    Hi kwinn,

    what's the purpose of the op amps ?

    Do you know of a sample/example circuit (of e.g. just one voltage) ?

    thanks
    - Howard

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2009-09-15 00:07
    Howard, seeing kwinn is not online the opamps acts as "regulators" with the npn emitter followers as the power pass element. The opamps have their non-inverting input tied to the desired reference voltage (in this case the resistor divider network) and the ouput goes to the transistor base. Normally if you tie the inverting input back to the output the opamp acts as a unity gain voltage follower but buffered. By tying the inverting input back to the actual output which is on the transistors emitter the opamp compensates for the voltage drop of around 0.6V that you would otherwise have in this configuration.

    Just think of opamps as a "very high-gain" servo amplifier which amplifies the difference between it's inputs. By using negative feedback which is normally some part of the signal from the output back to the inverting input the gain can be stabilized and made usable. A voltage follower therefore doesn't give you any voltage gain but it does give you more current at the same voltage. Regulators need to monitor the output and keep the voltage within limits, not unlike the opamp circuit.

    Regulators combine the reference, "servo amplifier", and power-pass element and also include protection circuitry for overloading and overheating.

    *Peter*
  • AleAle Posts: 2,363
    edited 2009-09-15 06:51
    Let's rephrase it a bit:

    I need many different power rails:

    1V ~200 mA continuos, 500 mA peak.
    1.2V ~200 mA continuos, 500 mA peak.
    2.5V 50 mA peak.
    3.3V ~400 mA continuos, peak 800 mA
    5V ~120 mA continuos, maybe 200 mA peak.
    30 V 10 mA continuos, 20 mA peak.

    I am not planning on using batteries to power this circuit but I'd like to use just one input voltage, say 5 V unregulated or similar.

    I was planning on using:
    1V & 1.2V 2x TPS62040 or 1x LTC3417 (requires more components)
    2.5V & 3.3 V 2x LM1117 or similar
    5 V TPS61070
    30V TPS61161

    Any other suggestion (a transformer with 5 output is just not an option, too big, too many diodes/capacitors).

    thanks for the input again.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Visit some of my articles at Propeller Wiki:
    MATH on the propeller propeller.wikispaces.com/MATH
    pPropQL: propeller.wikispaces.com/pPropQL
    pPropQL020: propeller.wikispaces.com/pPropQL020
    OMU for the pPropQL/020 propeller.wikispaces.com/OMU
  • CounterRotatingPropsCounterRotatingProps Posts: 1,132
    edited 2009-09-15 20:54
    > By tying the inverting input back to the actual output which is on the transistors emitter the opamp compensates
    > for the voltage drop of around 0.6V that you would otherwise have in this configuration

    Ahh! Thanks, Peter. That makes sense.

    Would this configuration work well for high-amperage, power transistors? (Apologies Ale, I know this is a bit a thread drift as you're looking at lower power - danke !)

    - Howard

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • AleAle Posts: 2,363
    edited 2009-09-16 05:25
    CRP: Linear regulators work but they dissipate the energy they do not give. But are simple and straightforward to understand, like in this case. Even for high amperage, do not forget a second or even third transistor in D'arlington configuration, transistors amplify current. You can always use (lt)spice to simulate a couple of circuits and see why it works.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Visit some of my articles at Propeller Wiki:
    MATH on the propeller propeller.wikispaces.com/MATH
    pPropQL: propeller.wikispaces.com/pPropQL
    pPropQL020: propeller.wikispaces.com/pPropQL020
    OMU for the pPropQL/020 propeller.wikispaces.com/OMU
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2009-09-16 06:07
    @Howard - yes you can do it this way but extra circuitry is required to protect against overload and also take into account transient response to line and load variations. If you have a simple load and the input is fairly regulated then you could use this approach easily.

    @Ale - There's no problem using the 34063 or similar for the step-up to 30V. The TPS62040 looks good for switching down to the low voltages as well, it's a pity you couldn't just drop 200mv directly from the 1.2V line to get your 1V. If you run from a standard +5V mains-switching supply then you certainly don't need any regulation for that unless the boost converter is simply there to maintain the +5V which I think is unnecessary. I'm scratching my head in the meantime to find a way to drop 200mv effectively.

    *Peter*
  • AleAle Posts: 2,363
    edited 2009-09-16 11:11
    Peter:

    The TPS61161 (for 30V) is well suited for white led backlight, so in this case is a perfect match and needs fewer components than a MC34063 and has a better enable and not the hack I normally use (A HCT gate output to power the collector of the first transistor in the D'arlington pair).
    I need to switch the 5V on and off, so either a MOSFET if I use 5V as supply or some other switching reg if I use something else. It does not add too much complexity, so it is not bad.

    Maybe a shottky diode can be used... but well. If I could use a Spartan6, there is a 1V version, I could have just one rail. Well one cannot have everything.
    The TPS62040 is really tiny and the coil needed is around two times the area!. Do not worry I'll show this when it is ready and built. smile.gif. it may take a month smile.gif

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Visit some of my articles at Propeller Wiki:
    MATH on the propeller propeller.wikispaces.com/MATH
    pPropQL: propeller.wikispaces.com/pPropQL
    pPropQL020: propeller.wikispaces.com/pPropQL020
    OMU for the pPropQL/020 propeller.wikispaces.com/OMU
  • kwinnkwinn Posts: 8,697
    edited 2009-09-16 16:41
    Sorry for not responding sooner. Working in the boonies with limited internet access. Peter is correct. Use the 5V switcher to produce +5V, use a 3.3V linear regulator with +5V as input for 3.3V. Use the quad op amp to control a series pass transistor for the other voltages. Use the +5V as input to produce 1, 1.2, and 2.5V. Use a small transformer/rectifier/capacitor and series pass transistor for 30V. The 3.3V or 5V supply would be used as a reference to the op amp for regulating the 4 voltages.
Sign In or Register to comment.