Connect with us

topology suggestions for 3.5kW batt charger?

Discussion in 'Electronic Design' started by Tesseract, Jun 30, 2007.

Scroll to continue with content
  1. Tesseract

    Tesseract Guest

    Greetings all - just found this site while searching for "boost pfc" and
    noticed that Winfield Hill has posted here before so I figure the site can't
    be all that bad ;)

    I'm just throwing this out as a sort of curiosity-type question. I have
    designed quite a few switchmode power supplies over the years but nothing
    even approaching this power level (600W is the previous maximum). A friend of
    a friend has one of 492 electric chevy S-10 pickups, and the onboard charger
    has lost its magic smoke. Frankly, the design of this charger/motor drive
    pretty much blows, ihmo, so I'm toying around with the idea of whipping up
    something new. "Toying" because there is little financial motivation for
    doing this. However, it is an interesting project and that is motivation
    enough, sometimes.

    There are two banks of batteries in the pack, each containing 26 12V/42Ah
    batteries. The manufacturer, Genesis, recommends charging them at the 0.4C
    rate up to a maximum of 2.4V/cell (with -20mV/C of compensation). As this
    comes out to a rather onerous 6.3kW, I'll either have to go with a less-
    ambitious charge rate or force the guy to unplug the range every time he
    needs to plug in (or, of course, have a dedicated 40A. circuit installed). As
    of now he does have a 240V/20A circuit for this purpose which, if we adhere
    to the NEC, gives us *just* enough continuous amperage to charge each bank at
    0.25C (10.5A) . This means both banks can be sequentially charged in about 8
    hours, which is tolerable.

    A linear and/or design using 60Hz magnetics is totally out of the question -
    you won't be able to move the thing around without a forklift. The next
    simplest thing I can imagine, then, is to use the output of a boost PFC more
    or less directly. This has some serious safety issues, but it ain't like I'm
    going to be submitting it for UL approval any time soon. So, these are the
    various factors/ideas I am contemplating:

    quasi-resonant (soft) switching; ZVS or ZCS?
    IGBTs or advanced MOSFETs
    Boost PFC + full-bridge or some combo to provide isolation

    Current mode control is a natural, here. Voltage regulation needs to be
    accurate but there's no need for a lot of loop bandwidth (ie - fast transient
    response not required). Ideas welcome!

  2. MooseFET

    MooseFET Guest

    In other words, we need a voltage of about 26*24 = 624, or so, as a

    The manufacturer, Genesis, recommends charging them at the 0.4C
    240 * sqrt(2) = ABOUT 340V

    As a result, we need to about double the voltage. This smells like a
    transformer based design not a simple booster.

    Just for fun I'll suggest this:

    ( ! ! !
    L1 T1 ( !!- ! ---
    ----(((((---------+ G1-!!- ! ---
    ( !!--[sense]-GND ! !
    ( ! GND

    At the wareform peak, the G1 and G2 are running almost as a squareware
    converter with only a small dead time when both are off.

    When both devices are off, the L1 voltage springs up to the battery
    voltage to ramp down the current in L1. When either is on (Vin-Vbat/
    2) ramps the current up. In this way, near the peak, L1 is working to
    buck the input.

    As you come off the peak of the mains AC, the duty cycle of the G1 and
    G2 increases. At some point, their conduction overlaps. When this
    happens, L1 begins to boost because the center tap pulses to zero
    volts while both are on.

    The core of L1 never has to store huge amounts of energy. The core of
    T1 also isn't storing lots of energy because it is working as a

    L1 and T1 can be merged into a single core if you are willing to do
    some creative winding.
  3. Phil Allison

    Phil Allison Guest


    ** Not true at all.

    ** Ok - here is your *simplest* solution.

    Use a 240 volt to 40 volt transformer, rated at say 25 amps to boost the AC
    supply to 280 volts rms.

    A 40 amp metal bridge on that will give 400 volt peaks at 100Hz.

    400/26 =15.4 volts a piece = just about right for end of charge.

    Install a DC amp meter ( say 30 amp FS) in series with the battery bank.

    Use a 1kW Variac to control the input to the boost transformer so you can
    adjust the charge rate to a nice number like 15 amps.

    Keep your fingers off the damn batteries ( ie wear gloves) and keep all
    spectators at bay with a safety fence around the vehicle !!!!

    ........ Phil
  4. colin

    colin Guest

    how about a dozen PC power supplies ?
    just modify them to work in current limit mode.
    they should then share the laod ok.

    Colin =^.^=
  5. Wimpie

    Wimpie Guest

    Hello Jeff,

    When a switching topology is a mandatory, mi first thoughts go to a
    full bridge forward converter with zero voltage switching where the
    current limit is determined by the leakage inductance of the
    transformer (probably a very large set of U cores). This is inherently
    short circuit proof and you have plenty of time to shut it down. The
    voltage waveform at the bridge is trapezoidal. dV/dt at the switches
    is limited by the external capacitors.

    Instead of one unit, I would use (depending on components available) 2
    or three units. In that case, main charging is done by (for example) 3
    units in parallel. As the voltage reaching final level, one or two
    units are switched of. When you can make one, you can also make three

    Main advantage for a full bridge circuit is the ripple current in both
    the primary rectifier and secondary rectifier. There is also efficient
    utilization of the semiconductors. High dI/dt is avoided because of
    the leakage inductance of the transformer. The transformer design is
    somewhat tricky. You might use a separate coil for dI/dt limiting.

    Because of the required large sized magnetics, you are limited in
    maximum operating frequency because of risk on bad flux distribution
    due to field propagation issues inside the magnetics.

    You can drive the IGBT's or MOSFETs with one transformer (with 4
    secondary windings).

    Best regards,

  6. Thanks for the excellent replies, everyone.
    Sorry - I wasn't clear about that: it is, in fact, two independent 312V nom.
    battery banks that, as far as my "eyeball" reverse engineering reveals, are
    selected via contactors. I can only *guess* at the reasoning behind this, but
    I imagine thermal management is a factor...

    This is a very interesting idea, Moosefet. It sort of looks like half of a
    phase-shifted bridge. L1 makes it inherently a current-fed topology which is,
    of course, immune to short circuits. Any references, or at least a "proper
    name" would be most appreciated.

  7. An even simpler but similar idea is my "backup" solution. Specifically, I
    would connect a 120V input Variac to one leg and neutral of the 240V line
    then take the output from the Variac tap and the other leg of the 240V line
    to get anywhere from 240 to 360Vrms. Add a bridge, a good chunk of
    capacitance, season to taste and voilla, a cheap, quick and dirty battery
    charger. The filter capacitance is a nod towards battery life - they
    apparently really hate pulsating DC... Big downside here is the terrible
    power factor which limits the actual charging current that can be delivered
    to the batteries before tripping the breaker.

    Colin - you're idea is nuts, but not totally out of the question. If the PC
    power supplies have PFC and use current-mode control then this is plausible.
    No PFC and the same problem as above creeps in. Voltage mode control instead
    of current mode and paralleling is a dicey situation no matter what. I doubt
    any halfway reliable PC power supply made today uses voltage mode control,
    but you never know what an OEM will stoop to when the very depths of the
    price/performance ratio are trolled.

  8. I agree completely - when you need more than about 300-400W of *isolated*
    power then the full bridge topology, in any of its many variants, is the way
    to go. Setting aside line isolation considerations for the moment, though, I
    don't see any advantage over a much more simpler boost PFC, especially if
    designed to operate in quasi- or fully resonant modes.

    Transformer design is *always* tricky, especially with regards to managing
    leakage inductance. Current-fed might be the way to go here, but I suspect
    that (quasi-)resonant operation will still give you the best combination of
    low-emi and switching stress and excellent current-limiting response.

    Once again, if I was going to design the charger for approval then full-
    bridge would be the way to go, and in order to make maximum use of the
    available RMS current PFC circuit is pertty much mandatory. In fact, if you
    eliminate the bulk filter capacitor from the output of a boost PFC then feed
    a full bridge you automatically get short-circuit-proof current-fed mode,

    Excellent comments, though, Wim.

  9. Phil Allison

    Phil Allison Guest

    "Tesseract.... "

    ** Err - did you want ideas supplied from others OR others to verbally
    masturbate over your insane ones ?

    Just be upfront, pal - it saves everyone lotsa time .......

    ** If you have 60 Hz, 3 phase, 240 volts available - you failed to tell

    BTW the phase to phase voltage is 415 - not 360.

    Add a bridge, a good chunk of

    ** Dirtier than Dirty Harry

    How lucky do you feel - punk ??

    This man is ON the edge ....

    ......... Phil
  10. Oh come on - it's a *given* that you already *have* lotsa time or you
    wouldn't be reading (and replying) to the posts here.

    Actually, I really want ideas from others. No pud-pulling necessary.

    Thanks for pointing this out - helping identify stupid, fireball-forming
    mistakes *before* I make them is the whole point of posting. This *is*
    residential 240V, so your implying that 3ph. would be needed caused me to go
    back and actually sketch out a schematic. Once I did that the fatal flaws in
    my backup plan became obvious. I should've kept TANSTAAFL in mind...

    Now I agree that your suggestion - wiring a transformer in standard boost
    configuration - makes the most sense as the simplest solution. Unfortunately,
    I don't have any 240V Variacs so I may have to go with a voltage doubler.
    Fortunately, I do have two 6,000uF/400V caps that will serve nicely here.
    Well, once I make sure the dielectric is fully formed on them as they haven't
    seen voltage in quite a while.

  11. colin

    colin Guest

    Yeah I hadnt given it a thought that it would be high voltage,
    was thinking of car battery charger voltages lol.

    I did say to modify them to current mode,
    becuase most PC supplies go into burp mode if they sense overcurrent,
    rather than refering to the PWM topology.

    I doubt if they have PFC though,
    wich i gues is needed for the full power.

    however for the high voltage parallel isnt desirable.
    and series conection is fraught with problems.

    maybe if you put one psu acros each 1 x 12v bank if you had access to mid

    At least PC PSUs are abundant in landfill sites.
    its quite easy to rewind some of them so they have just one very high power
    and strip the unused components.

    as was said elsewhere multiple phases would make things easier,
    and add some backup.
    boost PFC followed by resorvoir cap, followed by fwd resonant phase shifting
    full bridge,
    folowed by full wave rectifier (posibly synchronous if you want to be fancy)
    + LC filter.

    but you can go through a LOT of components getting a high power SMPS design

    also the output voltage is still so high as to be lethal,
    so it needs to be heavily protected from careless fingers etc,
    so why bother with isolation wich is 80% of the work ?

    just have a step down converter after the resorvoir cap wich does the
    current/voltage limiting.

    Colin =^.^=
  12. Phil Allison

    Phil Allison Guest

    "Tesseract via"

    ** Wrong answer, pal - that one labels you as a troll for sure.

    The saved time alluded to would be that time span between one merely
    suspecting YOU were yet another totally mad troll and being completely
    certain of it.

    ** Not a familiar acronym to me....

    ** You have now acquired near genius status.

    Asking for exact details of my proposed anti-spectator HV electrical safety
    fence would have resulted in mucho bonus points.

    ** Sloppy impedance cap dominated regulation = baaaaaaaaddddddd !!!

    ** Those spectators need to be wearing safety goggles and steel lined flack

    Bits of flying electro & lead acid battery are significant shrapnel.

    ........ Phil
  13. MooseFET

    MooseFET Guest

    How about calling it "KenSmith's Idea"?

    This circuit is certainly not immune to a short. There is a DC path
    from the Vin to the Vout so short circuit currents have a path through
    even with the transistors off.

    It is a "current fed" design in that the L1 limits the current rise.
    Basically it is a current fed forward converter where the transformer
    in an autotransformers. The sliding between the overlapping and non-
    overlaping conduction requires a bit of custom controller design.
  14. I was involved in the design of the 10kWt subwoofer amplifier (powered
    from the primary power source of 14.4V). Actually, I designed the
    DSP/PWM controller for it; the power part was designed by the other guy.
    There is absolutely nothing fancy about the topology: a classic MOSFET
    push-pull voltage mode buck running at 24kHz. The magnetics for it was
    quite a challenge. The size of the whole thing is 50x30x10cm. It works
    just fine. For the battery charger, you will need to output the
    regulated current rather then voltage; it should not be a problem.

    Vladimir Vassilevsky

    DSP and Mixed Signal Design Consultant
  15. Wimpie

    Wimpie Guest

    Hi Jeff,

    I forgot Power Factor Correction.

    You may design an active power factor correction circuitry. Your
    application probably doesn't require power factor of 0.99. When a
    power factor of about 0.9 is sufficient, you may use the CLC passive
    topology where the first Capacitor is at the AC side of the bridge,
    the inductor is between the bridge and the DC output storage
    This document describes various passive and active power factor

    I think when I was in your position; I would use the passive PFC and
    fully concentrate on the zero voltage full bridge design with leakage
    inductance. I like this thing because you don't have to measure the
    momentary current, the transformer may have large leakage inductance
    (so safety insulation is easy to obtain) and there is low dV/dt and DI/
    dt (radiated emission).

    Off course he disadvantage for passive PFC is the bulky L and Mains
    capacitor. Maybe you can get some inspiration from the document. I do
    not have experience with zero voltage or zero current switching PFC.

    Best regards,

  16. The solution to this seems obvious to me: don't post to this thread.
  17. I'm liking this idea of your's more and more, Colin. I found a 600W PC power
    supply with active PFC on eBay for $25, incl. shipping. That's definitely
    cheap enough to play around with. Heck, the Litz wire I'll need for the
    secondary will probably cost more than the power supply itself!

    If an SMPS uses current mode control, though, paralleling two or more
    together is trivial, even if Vo is high. I totally agree that no matter what
    the control scheme you are asking for trouble when you try to put supplies in
    series (dueling feedback loops).

    That's the truth, and the components that blow up tend to be expensive ones...

    This was precisely my reasoning for just going with a straight boost PFC

    Thanks for the input - I'm definitely going to try this on one supply even if
    the total output from one PC supply isn't enough to make a practical charger
    (I need to keep the total charge time for the two banks under 12 hours for it
    to be at all practical, which means I need at least 2.6kW). Still, as a quick
    and dirty charger it has lots of merit (and it will be isolated, anyway).

  18. Phil Allison

    Phil Allison Guest

    "Tesseract via"

    ** YOU do not get to contol a thread or any person on a NG - ASSHOLE.

    ....... Phil
  19. Phil Allison

    Phil Allison Guest

    "Vladimir Vassilevsky"

    ** Oh dear - that is very poor advice.

    ** Explains a lot.

    ........ Phil
  20. Fair enough. Still, with no prior art to study one would have a lot of pain
    and misery to look forward to before getting a new topology working, much
    less optimized.

    What I meant by this is that a short circuit will not cause immediate
    destruction of the switch(es). L1 will saturate, of course, and a circuit
    breaker will trip, but once the short is removed and power restored the
    circuit should resume operating.

    I still think you could probably use half of the outputs of a phase-shifted
    full bridge controller IC to manage this topology (eg - UCC3895).

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day