1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

Parallel Boost Converters

Discussion in 'General Electronics Chat' started by Edward2107171, Sep 4, 2017.

  1. Edward2107171

    Edward2107171 New Member

    Joined:
    Sep 4, 2017
    Messages:
    5
    Likes:
    0
    Hi,

    I have a project that requires 45V at 100A from a 12V supply, I want the design compact so I don't want to play around winding bulky transformers and expensive IGBT's / FET's.

    I want to go with a boost converter, I can currently build a cheap 5A output version from surface mount components making it very small, of course I would need at least 20 of these modules to achieve my 100A current. Building it modular means I can increase the power easily at a later date.

    I'm no expert at power supplies so I need a little guidance. So, from what I've read it's a big no no to tie boost converters in parallel. I have however seen many modules doing such a thing and that many datasheets show a multiple unit sync circuit. I imagine that if I were to sync them all together they would all still individually manage their own currents so none would be overloaded, the issue I can see is the voltage regulation if different modules are running at different currents.

    The chip I will be using is the UC3843

    Figure 23 in the below datasheet shows using a 555 timer to synchronise the duty cycles, it doesn't mention anything about the voltage feedback which I assume can all be tied together since they are measuring the same source. Is synchronising the duty cycles going to work?

    http://www.st.com/content/ccc/resou...df/jcr:content/translations/en.CD00000966.pdf

    Any thoughts or circuit diagram examples of parallel boost converters.

    Many thanks
     
  2. ronsimpson

    ronsimpson Well-Known Member Most Helpful Member

    Joined:
    Apr 17, 2007
    Messages:
    7,370
    Likes:
    974
    Location:
    Loveland, CO USA
    ONLINE
    LOL ok.......
    45V 100A = 4500 watts = 12V 375A
    There will be expensive FETs and a expensive inductor or transformer.
    That will synchronize UC3843s BUT they will be in phase. (I do this) It will work better if two are 180 out of phase. Or 4 at 90 out of phase.

    Just off the top of my head with out computing:
    The FETs need to be on 3/4 of the time at about 530 amps.
    The output diode(s) need to be on 1/4 the time at 400A.

    Best guidance is "this is a big project".
     
  3. Edward2107171

    Edward2107171 New Member

    Joined:
    Sep 4, 2017
    Messages:
    5
    Likes:
    0
    I understand there is a lot of current, this is the reason why I want to use a lot of smaller modules at a lower current since I would have to go with huge IGBT bricks which are very expensive. If I end up with 20 Mosfets at 25A each they are a hell of a lot cheaper and tend to have a lot lower Rds. The schottky diodes, 500A at 0.3V drop at 25% is still only 38W, probably around 150W total with the Fets which is going to be a lot lower than I could ever achieve with large IGBT's switching slowly.

    I could build a microcontroller circuit to phase them, I have seen a module that is 4 phase and can be stacked in parallel, the reason I know it should be possible. Would it be such an issue if they were all in phase ? Would just synchronising the duty cycles be enough? I guess I could make it 2 phase to reduce ripple which would cut down on this size of the caps.

    I have built a couple of smaller supplies, never anything to this kind of scale before, just don't want to throw a load of money into this project without getting some advice first.

    Thanks
     
  4. dave

    Dave New Member

    Joined:
    Jan 12, 1997
    Messages:
    -
    Likes:
    0


     
  5. ronsimpson

    ronsimpson Well-Known Member Most Helpful Member

    Joined:
    Apr 17, 2007
    Messages:
    7,370
    Likes:
    974
    Location:
    Loveland, CO USA
    ONLINE

    What are you doing that needs 45V at 100A? Automotive?
    Yes. You might want to read up on LTC3871
     
  6. dr pepper

    dr pepper Well-Known Member Most Helpful Member

    Joined:
    Oct 6, 2008
    Messages:
    4,757
    Likes:
    258
    Location:
    North west UK
    If you ran them in 90 degrees sequence peak current would be a 1/4, and voltage drop too so its worth the effort.
     
  7. ronsimpson

    ronsimpson Well-Known Member Most Helpful Member

    Joined:
    Apr 17, 2007
    Messages:
    7,370
    Likes:
    974
    Location:
    Loveland, CO USA
    ONLINE
    If the output is 1/4 the time and you have 4 (or more) boost circuits running 90 apart. Then the output is 100% of the time. There is almost no need for an output capacitor. Also; the input current does the same thing. It goes from pulses to 100% of the time current flow.
    You can not make it perfect but the "AC" current is much smaller.

    Example with one PWM: 100A DC output; 400A at 25% of the time and 0A 75% of the time. The output capacitor must work very hard.
    Example with 4 PWM: 100A DC output; (100A at 25%)x4= if the timing was perfect 100A at 100%. This "perfect condition" will not happen because the timing will change if the load changes or if the input voltage changes. There will be very short times when two outputs are on or when no output is on. So you might have 100A most of the time and for a short time 200A or 0A but the point is the current is much closer to DC.
     
    • Agree Agree x 1
  8. Edward2107171

    Edward2107171 New Member

    Joined:
    Sep 4, 2017
    Messages:
    5
    Likes:
    0
    I looked into something similar to the LT3871, it would be an ideal option if the package they came in wasn't so small, and they weren't so expensive, unfortunately.

    I have done quite a bit of thinking and even if they were sync'd in a single phase like both of you say I would need large capacitors and I would draw huge current pulses from my supply.

    I think I will go with the 3843's just because they're so cheap. Each one will run independently current wise, they have to if I don't want to fry mosfets. I need around 20 modules to meet my current needs (limited by my choice of cheap, small footprint inductors). I will run 20 separate phases, all of them will have their duty cycles clamped as to not exceed their output voltages. Each feedback will be connected to a main microcontroller which will output an equal pulse for each phase in a cycle in order for every phase to equally share the same duty, a single comparator will be used to keep the reference voltage in check with the microcontroller. I think having feedbacks running independently may lead to erratic conditions, maybe ?

    I found my main issue is the diodes, even schottky's at 35A are 500ns at 1V drop which is a lot worse than I expected, maybe using a mosfet will be a better choice, cheaper too.

    In all the project is a little more complicated than I first thought so I guess I should start small with maybe only five phases, 25A at 45V is a still a lot of power and should get me going on my other project.
     
  9. schmitt trigger

    schmitt trigger Active Member

    Joined:
    May 24, 2009
    Messages:
    914
    Likes:
    85
    The biggest problem you will face, is making sure that all power supplies share the output current equally, across all load, line and most importantly, transient conditions.

    You can adjust with a 10-turn pot to have the individual outputs within a few millivolts of each other. But then there is a transient, and the different loop responses will cause havoc.

    Unitrode Semi (which was bought by Texas Instruments) had an app note describing the challenges of paralleling supplies.
    The interplayplay between the two separate control loops is, well, significant.

    Using those techniques, I once paralleled a pair of boost supplies.
    I may be wrong, but I don't think you can do 20.

    One way around could be to put ballast resistors on the output, and outside of the feedback loop. But then the load regulation will suffer.
     
  10. ronsimpson

    ronsimpson Well-Known Member Most Helpful Member

    Joined:
    Apr 17, 2007
    Messages:
    7,370
    Likes:
    974
    Location:
    Loveland, CO USA
    ONLINE
    With a 3843, if you connect all the error amps together, so each is forced to have the same output voltage, then the current will share.
    I will say that another way. I have used one error amp like normal. Then connect error amp_1_output to error amp_2_output, then the MOSFETs will share current. (effectively not using error amp_2)
     
  11. crutschow

    crutschow Well-Known Member Most Helpful Member

    Joined:
    Mar 14, 2008
    Messages:
    10,623
    Likes:
    479
    Location:
    L.A., USA Zulu -8
    ONLINE
    Where are you getting the required over 400A @ 12V (including circuit inefficiency) to power this converter? :eek:
    It will require a few large automotive batteries in parallel to provide that power for even a short period.
     
    • Like Like x 1
  12. Edward2107171

    Edward2107171 New Member

    Joined:
    Sep 4, 2017
    Messages:
    5
    Likes:
    0
    A single automotive battery, it's not ideal but it should work just fine. For the majority of the time it will only be drawing around 30A, it will draw 200A maybe 5% of the time and 350A for around 1%, if that, longest time at peak current may only be 10 seconds. It's one of those proof of concept projects that this project is powering, plus I can get a replacement battery within a year if I scramble this one. The project it's powering has been tested at 700W but I ran out of current so hence I want to build this boost converter to aim for 4kW, plus it's something interesting to add to my ever increasing project list. I will share my other project in time, but not until I know it's a success or a failure.
     
  13. kubeek

    kubeek Well-Known Member

    Joined:
    Mar 11, 2006
    Messages:
    1,510
    Likes:
    189
    Location:
    Prague, Czechia (not Chechnya)
    If 8 bucks is too much for you, how do you expect to buy 25 of anything? Or don´t you realize you need just one controller and banks of gate drivers, fets and inductors?
    I mean, if each converter runs on its own it will be a nightmare, as each will influence the others and you will get tons of instability and/or current hogging
     
  14. Les Jones

    Les Jones Well-Known Member

    Joined:
    May 15, 2015
    Messages:
    1,474
    Likes:
    191
    Location:
    Lancashire UK
    Why not just use four 12 volt batteries in series ? They would probably drop to 45 volts with the 100 amps load.

    Les.
     
    • Like Like x 1
  15. Edward2107171

    Edward2107171 New Member

    Joined:
    Sep 4, 2017
    Messages:
    5
    Likes:
    0
    So hostile,

    Ok so the cost for one module would be around 6USD, that's control chip, inductor, mosfet, diode, and some resistors. The cost of a gate drive chip is around the same cost as the control chip and I would need a diff comp to ensure I wouldn't exceed my inductor current, so that works out a little more and increases my circuit size. You did not read my previous post, each module will not run on it's own, the current control will to ensure no inductor is overloaded, but a main microcontroller will control all of the modules keeping them 18 degrees apart from each other while maintaining the same duty cycle in all of them, so they should be more or less balanced.

    Hi Les, that was the original plan but if my project doesn't work I have four batteries laying around, plus they're pretty expensive. My other issue is that the maximum voltage is 45V and fresh car batteries will exceed this, three batteries aren't enough either. RC Li-po's are another consideration but they just don't last very long.
     
  16. crutschow

    crutschow Well-Known Member Most Helpful Member

    Joined:
    Mar 14, 2008
    Messages:
    10,623
    Likes:
    479
    Location:
    L.A., USA Zulu -8
    ONLINE
    Use three cheap 12V batteries, plus one 6V battery in series.
    For example, this battery for U$33 provides 100 cold cranking amps with a 18Ah capacity so should deliver your 100A without problem.

    I think you'll find it a lot easier and probably cheaper than trying to do the inverters.
    Building a 45V, 100A step up inverter is far from a trivial or inexpensive task.
    I think you'll end of with a bunch of blown transistors before you are through. :eek:
     
  17. Pommie

    Pommie Well-Known Member Most Helpful Member

    Joined:
    Mar 18, 2005
    Messages:
    10,161
    Likes:
    340
    Location:
    Brisbane Australia
    Is your project that sensitive that it needs 45V but can't handle 48-52V? A battery with 100 CCA will drop to around 10V when you draw 100A at room temperature.

    Mike.
     

Share This Page