Shop OBEX P1 Docs P2 Docs Learn Events
"best" power MOSFET — Parallax Forums

"best" power MOSFET

yarisboyyarisboy Posts: 245
edited 2014-05-03 14:08 in General Discussion
I am not an EE. I struggle reading part spec sheets. I have some surface mount MOSFETs claiming a 100 amp continuous rating. As I dissect the sheet further and further it appears the authors consider on-times of 100 ms to be proof of continuous ratings. For my application I now reject surface mount designs simply because I can't bolt on extra heat sinks and because the small size /form factor makes a 100 amp true rating even hard to achieve with affordable copper board (even if you could set aside the cruel laws of thermodynamics). Back to through-holes. My application involves draining a 60 Ah battery from 4.12 volts to 2.5 volts through a .1 ohm resistor controlled by a bottom balance Propeller program. Continuous current would be about 40 amps that declines as the voltage of the cell drops. I'd like to avoid having to buy expensive SSRs or paralleling power MOSFETs (although it can be done) if possible. What would be the favorite power MOSFET in such a case. Due to thermodynamics, bigger size and beefier tabs are better. I'm willing to solder #10 AWG solid copper wire on to the board to avoid spending the big bucks for heavy copper greater than 2 oz/ft^2.

Comments

  • tonyp12tonyp12 Posts: 1,951
    edited 2014-04-26 07:48
    >My application involves draining a 60 Ah battery from 4.12 volts to 2.5 volts through a .1 ohm resistor controlled by a bottom balance Propeller program.

    Pics or it didn't happen (e.g show schematic)
    As I don't know what you mean when you say draining. what is this 4.12v to 2.5v conversion all about?
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-26 08:35
    I think your doubts are well founded.

    In general, for electrical material... the higher the volts, the more insulation is required. And the higher the amps, the larger the copper wire diameter.

    I strongly suspect that nothing over about 3 amps should really be on a printed circuit board unless there are some serious modifications to the heavy current runs. And so... it seems to me that a MOSfet that is rated at 100 amps continous on and is surface mount, tends to be paradoxical. The MOSfet migth run cool, but the leads on the circuit board may just burn up under load.
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2014-04-26 09:42
    There was discussion at the Maker space last week to setup an test to determine how much current could actually flow through the package leads before the Solder would begin to melt. I would imagine some of those ratings would be under an extremely specialized setup to handle the heat... while the MOSFET /might/ run cool and efficient, you would still need to take precautions to cool off the PCB and leads going to the mosfet.
  • tonyp12tonyp12 Posts: 1,951
    edited 2014-04-26 09:52
    2oz copper pcb, dedicate 90% of a 6x4" pcb's copper pours on both sides to the leads going to/from mosfet.
    Use a mosfet with extreme low on Resistance, if rapidly switching the mosfet use 4amp gate driver.
    Use SMT heat sinks to aid the copper pours dissipating of heat.

    But until we know what you are trying to accomplish your DIY voltage regulator or what ever it is, may not be the best way to go about it anyway.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2014-04-26 10:14
    It's not a voltage regulator. As stated in the first post, he's simply draining a battery, presumptively to determine it's power curve under heavy load conditions.

    -Phil
  • tonyp12tonyp12 Posts: 1,951
    edited 2014-04-26 10:27
    So nearly short circuit the 60Ah battery with a 60Amp load, to see if the battery last 1hr before it reach 2.5v ?
    Just a hardware switch is needed for that.
    But I guess with some controlled intermittent power burst testing, a mosfet is needed.

    Or is this a under-voltage protection circuit that is gone be in final design?, maybe it's better to implement that on the individual lipo cells.
  • yarisboyyarisboy Posts: 245
    edited 2014-04-26 12:01
    tonyp12 wrote: »
    So nearly short circuit the 60Ah battery with a 60Amp load, to see if the battery last 1hr before it reach 2.5v ?
    Just a hardware switch is needed for that.
    But I guess with some controlled intermittent power burst testing, a mosfet is needed.

    Or is this a under-voltage protection circuit that is gone be in final design?, maybe it's better to implement that on the individual lipo cells.
  • yarisboyyarisboy Posts: 245
    edited 2014-04-26 12:23
    I bought a 24 kWh battery pack from the "Better Place" bankruptcy liquidation. The cells cans are made by AEC in Japan. These are the same cells used in the Nissan Leaf. There are 48 cans. Each can is a 2 series, 2 parallel pack sub module. That means I need to drop the energy in 96 cells to a uniform voltage level of, say 2.7 volts before the first balanced charge. It can be done by hand one cell at a time but the cells bounce back up after discharge and must be discharged several times before they stay down. 96 times # of discharges = a lot of time in the shop. You can see why I want to build a 16 switch dis-charger run by a Propeller. The reason for all this is so that the pack can be charged until the least-capacity cell reaches full charge and then stop. This is a procedure patented by one of the power tool companies just after Lithium cells became affordable. After balancing one can "drive out" the pack to low voltage cut-off without reversing a cell permanently destroying it. In the case of the Fluenza pack I would bottom balance 8 cans at a time so the program would have to be run 6 times to get all cells setting stably at 2.7 volts. People using this procedure have drained the entire pack to .1 volts per cell by accident and then were able to recharge the dead pack with out loosing any cells. It's the only way to avoid "bricking" the pack when you really screwed up. A $10,000 screw up really hurts. After all you'll need someone to take your used Leaf to after the warranty runs out without having to take out a second mortgage someday. Right now the pack is in an air conditioned space setting at 380 volts.
  • yarisboyyarisboy Posts: 245
    edited 2014-04-26 12:34
    Yes, that is why the Cray XMP super computers use to run Flourinert over the boards and chips. If flunked out as synthetic blood but found application as an electrically compatible refrigerant so the old Proffs say. Seymour was pretty crafty.
  • yarisboyyarisboy Posts: 245
    edited 2014-04-26 13:07
    The NXP power MOSFET ( PSMN09-25YLC ) uses three of the four leads as the source. One lead is the gate and the drain is the tab. I've got old IGBTs still in the tubes thermally brawny enough but then I need higher control voltages and gate driver chips. An option with less expense than SSRs or relays.
  • yarisboyyarisboy Posts: 245
    edited 2014-04-26 13:10
    Even with the IGBTs the trace widths would need to be .5 inches wide on both sides of the board via-ed through to live.
  • jmgjmg Posts: 15,173
    edited 2014-04-26 14:12
    yarisboy wrote: »
    ... My application involves draining a 60 Ah battery from 4.12 volts to 2.5 volts through a .1 ohm resistor controlled by a bottom balance Propeller program. Continuous current would be about 40 amps that declines as the voltage of the cell drops. I'd like to avoid having to buy expensive SSRs or paralleling power MOSFETs (although it can be done) if possible.

    Just parallel the MOSFETS - it is what eveyone does.
    - they are cheap enough, and splitting the load resistor also helps spread the heat.
    eg 5 FETS with 0.5 Ohm loads will be ~ 8A per fet, and at that level a 10mOhm fet is under 1 watt. (640mW)

    sub 5mOhm fets should be cheap enough, and lower the cluster heating effects.
  • kwinnkwinn Posts: 8,697
    edited 2014-04-26 21:44
    I wouldn't even think of mounting transistors that are going to dissipate that kind of power on a pcb. Look at something like a TO220 package and mount it (or them) on a heat sink. Use 2 awg14 wires on the input and output (1 each if you need 2 transistors) and to the 0.1 ohm resistor.

    The power dissipated by the transistor will be 1600 x Rds on ( I squared x R) rating of the transistor. The 0.1 ohm resistor will be dissipating the balance so it should be at least a 150W.
  • Duane C. JohnsonDuane C. Johnson Posts: 955
    edited 2014-04-26 23:35
    Take a look at this MOSFET for your special application.
    PSMN1R1-30PL
    RDSon = 1.3mΩ
    So at 41.2A
    41.2A2 * 1.3mΩ = 2.2W of power dissipation in the MOSFET

    OK it has a 10V gate.
    But seriously low RDSon is best done with 10V gate drive.

    Duane J
  • kwinnkwinn Posts: 8,697
    edited 2014-04-27 00:21
    Nice find Duane. Much simpler to provide a 10V gate drive than to deal with high power dissipation and temperatures.
    Duane C. Johnson wrote: »
    Take a look at this MOSFET for your special application.
    PSMN1R1-30PL
    RDSon = 1.3mΩ
    So at 41.2A
    41.2A2 * 1.3mΩ = 2.2W of power dissipation in the MOSFET

    OK it has a 10V gate.
    But seriously low RDSon is best done with 10V gate drive.

    Duane J
  • yarisboyyarisboy Posts: 245
    edited 2014-04-27 13:10
    Duane C. Johnson wrote: »
    Take a look at this MOSFET for your special application.
    PSMN1R1-30PL
    RDSon = 1.3mΩ
    So at 41.2A
    41.2A2 * 1.3mΩ = 2.2W of power dissipation in the MOSFET

    OK it has a 10V gate.
    But seriously low RDSon is best done with 10V gate drive.

    Duane J

    Thanks, I will. I've also found out Spec sheet writers are a lot like math teachers. The really bad ones have you thinking you are the thick one. I've found that International Rectifier has better spec sheets. IRF1324PbF is in a TO-220AB package. I've got some new solid copper bar stock here I can bolt all the tabs to. They list a "package limited" current of 195 amps. I fully agree on the 10 volt gate charge-to-voltage. I'm going with two wall worts driving a +12 regulator and a -12 volt regulator so I have sources I can run through optos to the gate pins. Yup, it's a poor man's DC-DC isolation. The package limit or wire bond limits listed more fairly address the issues Beau Shawb was referring to above. Silicon limited on these spec sheets doesn't matter if one burns off the leads.
  • kwinnkwinn Posts: 8,697
    edited 2014-04-27 21:11
    In most cases data sheets have become worse over the years. I have quite a few old data sheets, and when I compare them to a more recent data sheet for the same part there is a large difference in the amount of information provided. I find the older sheets will have two to four times the number of pages, more details, and sample circuits as well.
    yarisboy wrote: »
    Thanks, I will. I've also found out Spec sheet writers are a lot like math teachers. The really bad ones have you thinking you are the thick one. I've found that International Rectifier has better spec sheets. IRF1324PbF is in a TO-220AB package. I've got some new solid copper bar stock here I can bolt all the tabs to. They list a "package limited" current of 195 amps. I fully agree on the 10 volt gate charge-to-voltage. I'm going with two wall worts driving a +12 regulator and a -12 volt regulator so I have sources I can run through optos to the gate pins. Yup, it's a poor man's DC-DC isolation. The package limit or wire bond limits listed more fairly address the issues Beau Shawb was referring to above. Silicon limited on these spec sheets doesn't matter if one burns off the leads.
  • yarisboyyarisboy Posts: 245
    edited 2014-05-02 05:09
    IRF1324S-7PPcF Id (package limited) 240 amps and surface mount to boot. Now I'm researching heat sinks for this. It has 5 source leads. Buy it, built it, test it. The only way to relate a lab spec to the real world.
  • tonyp12tonyp12 Posts: 1,951
    edited 2014-05-02 08:54
    Here are two 2DPAK SMT heatsinks:
    http://www.mouser.com/ProductDetail/Ohmite/DV-T263-201E-TR/?qs=sGAEpiMZZMttgyDkZ5WiurC9qN77dzSraAkYy3aT9dM%3d
    http://www.mouser.com/ProductDetail/Aavid-Thermalloy/7109DG/?qs=sGAEpiMZZMttgyDkZ5WiujTlBOILjl7xi4aoVUO77j4%3d
  • Mark_TMark_T Posts: 1,981
    edited 2014-05-03 14:08
    yarisboy wrote: »
    What would be the favorite power MOSFET in such a case. Due to thermodynamics, bigger size and beefier tabs are better. I'm willing to solder #10 AWG solid copper wire on to the board to avoid spending the big bucks for heavy copper greater than 2 oz/ft^2.

    Forget PCB, use proper packages with screw terminals for that sort of current:

    http://uk.farnell.com/ixys-semiconductor/ixfn200n10p/mosfet-n-sot-227b/dp/1427322
Sign In or Register to comment.