need help with constant current PWM conversion from 12v to 24v

Discussion in 'The Projects Forum' started by samuraimarc, Mar 29, 2009.

  1. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    Hello all, this is my first post.
    I am trying to convert a 12 vdc constrant current PWM to 24 vdc. I have a 12v cooling fan on the heat sink which I put a 7812 in front of as well as the gate driver board. The negative is still common throughout the system. The fan works just fine and the gate driver seems to work ok. The problem I am having is the Mosfet is getting real hot real fast. when run in 12v system, the frequency is audiable. One the 24v system it is ear piercing and smokes the wires going to the stiffening capacitor which is 12,000 uf. The mosfet is rated for 60v, is n-channel, gate + 20v, 340A. At only 30 amps it get too hot in 25 seconds. At 12v 70 amps it barely gets warm. I don't know if the sense resistor needs to be changed to a higher value or what. I also tried to make a + - power regulator with 7812 and 7912 and had no luck. Any help would be greatly appreciated.
    Thank you!
     
    Last edited: Mar 29, 2009
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    That isn't a "constant current PWM" circuit; merely a PWM circuit. It would need an inductor and a different configuration to be a constant current circuit.

    It's not a very good PWM circuit either; for one thing the LM324 is a VERY slow opamp, for another the LM324 is being used as a comparator AND a gate driver for a MOSFET that has a very high gate charge requirement.

    Your MOSFET is getting hot because it is spending so much time in the linear region; the LM324 can't source/sink current fast enough to turn the MOSFET on/off quickly.
     
  3. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    Hey thanks, I read another posting that said the op-amp is not the way to go. I posted another thread asking about using an NCP1212 but I'm not sure how to use it. any other suggestions?
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I drew up the attached circuit months ago for someone else who found the "zerofossilfuel" PWM circuit to not be very suitable.

    If you want to experiment with the simulation, you can download Linear Technology's LTSpice SPICE simulator, available for free from their site:
    http://www.linear.com/designtools/software/

    The simulation is the bottom file with the .asc extension. Copy that into the SwCad directory under where you installed LTSpice.

    The difficult-to-obtain item will be L1; you'll have to make one yourself from a large toroid and heavy gauge wire.
    [eta]
    Six evenly-spaced turns of AWG 6 around an Amidon FT-193A-J toroid ($10.50):
    https://www.amidoncorp.com/items/21
    ...would give you around 267uH with a fairly low resistance
     
    Last edited: Mar 29, 2009
  5. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    Right on thanks.
    I'm not to sure if I totally understand the schematic, and does it use 2 Mosfet's?
    Is there a way to clip the linear region of the "zerofossilfuel" design for now until I can build the design you suggested?
     
  6. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    I was also wondering: as the electrolyzer warms up it will try to draw more current, will the 555-pwm4 limit that as well.
    Thanks
     
  7. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Try downloading Linear Technology's LTSpice and running the simulation.

    Since I don't know for absolute certain what an "HHO" cell would look like electrically, D1-D4 create a reasonably fixed voltage drop of around 2v, and R2 is a guess at what the rest of the internal resistance would be like.

    R4 is simply a two-foot long piece of AWG 10 copper wire; that happens to measure 0.002 Ohms.

    R7 and C4 act as a low-pass filter to give the inverting input of U3 a relatively stable voltage.

    R8 & R9 act as a voltage divider to set a current threshold level. C2 keeps the threshold stable.

    U3 is a high-speed opamp. An opamp with the ability to sense near ground might be substituted.
    R10 provides gain control for the opamp.

    The output of the opamp feeds the control voltage input of the 555 timer. It winds up changing the output pulsewidth of the astable 555 timer configuration. Note: C1 can be increased to 2nF; when R1=15k this will give a basic operating frequency of 30kHz, which should be well out of audible range.

    The output of the 555 timer feeds the 2A synchronous buck MOSFET driver IC, which handles the synchronization of the MOSFET gates. MOSFET M1 sources current to L1 until the current threshold is reached or the 555 times out; then MOSFET M1 is turned OFF and MOSFET M2 is turned ON. M2 takes the place of D1 in your circuit; it provides a return path for the current that L1 continues to provide when M1 is turned off.

    Yes.
    There's not really much point to doing that.

    With that circuit, when the MOSFET turns on, the only items limiting the maximum current from your vehicle's electrical system is the wiring resistance, the "HHO" cell itself, and the MOSFET.

    When the MOSFET turns off, your "HHO" cell is not generating any HHO.

    When the MOSFET turns on, your cell is getting far too much current, and this excess current is heating up your electrolyte, wiring, and your MOSFET.

    In order to "fix" that circuit, it needs to be thrown away and re-designed.
     
    Last edited: Mar 29, 2009
  8. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
     
  9. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    LTSpice is a bit quirky, but quite powerful, and the price is right. :)

    I have the Eagle Layout Editor, but not the BBS PSPICE add-on module. I like to be able to simulate circuits before building the real thing.

    No, I haven't worked with that particular IC.

    But what you really need is a synchronous buck current regulator, and the NCP1212 is not that; it's merely a buck regulator. The circuit that I posted already should do what you're looking to do.

    The "Vlogic" can be created from a vehicle's 14v supply by using a 7806 regulator with a 1uF cap on the output.
    "Vcc" is straight from the battery.
     
  10. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Yes. It will limit the current to limits set by R9 and the min/max PWM achievable by the 555 circuit. The simulation indicates roughly 9A to 36A constant current when R9 is a 1k pot. The lower current is when R9 is below 250 Ohms, the higher when R9 is set to 1k Ohms.
     
  11. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    cool, so if I raise the value of R9 high then 1K I can get more amperage?
    I'm probably wrong, correct me if I am, but it looks like M2 would be the power FET (meaning for the higher amperage) and M1 taking the place of my D1 since I need N-channel for this project.
    I'll throw this other one away and start on your design, Thanks
     
  12. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    M1 "charges" inductor L1. M2 provides a path for L1 to discharge current through the load.

    As it is right now, around 40A is the practical limit for the design when operating from a 12v-14v automotive system. The current limit is due to the way the 555 timer portion operates.

    You will likely find that much more "HHO" is produced with this new circuit at the same average current as the old.

    But, you started off in the thread that you want to put it in a 24v system, and I haven't addressed that yet. What kind of vehicle, a fire truck or a surplus military vehicle?
     
  13. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    It's going on a combine that has a 24v system. If you could help with that, that would be great.
    what kind of capacitor's do you recommend?
    The LT1007 is a regular op-amp (741) ?
    Will I still need a 12,000 uf stiffening cap?
    Is the ctrlV just a label?
    Thanks a ton for your help!
     
    Last edited: Mar 29, 2009
  14. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    OK. PLEASE keep in mind that you risk SEVERE engine damage by introducing "HHO" into your fuel/air mix. If your combine has computerized engine controls with O2 sensors, the computer will be fooled into thinking the engine is running too lean, and will make the mixture more rich to make up for it. If your O2 sensors are of the narrowband type, you will need a conversion kit to change them over to wideband. The ignition advance profile will need to be modified to reduce detonation. The exhaust gas temperature will need to be monitored. This isn't just a "plug and play" type of modification you're attempting.

    The large caps can be electrolytic, the small ones mylar, ceramic, poly metal film.
    No, the LT1007 is a precision high-speed opamp. The 741 is slow, obsolete, and won't sense inputs near ground. It's over 40 years old. It won't work in this application. However, you COULD use a single channel of an LM2902, which is an automotive version of an LM324 opamp. Speed isn't really the issue here; it's the single supply and sensing near ground that is.

    If you're taking power right from the battery, that's about all the "stiffening" you'll need.
    Yes, so it shows up on the plots if you select it. "V(ctrlv)" makes more sense as a plot label than "V(N03)"
     
  15. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    Thanks for your warning, I realize the risks involved. I have the zero design on my '70 jeep and it's does ok, but I'm going to replace it with your design. Good thing the combine is an old one, no puter.
    So will the 7806 handel the 24v?
     
  16. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    OK, so why don't you start with the Jeep with the new design?

    Is your Jeep engine a 4, 6, or 8 cylinder?

    You must know that re-building a Jeep engine is a heck of a lot cheaper than rebuilding a combine engine, right? But even then, a Jeep's engine isn't under constant load like a combine's is; most of the time an automotive engine is coasting along with low intake manifold pressure; a combine's intake manifold pressure would likely be much higher.

    That's sort of a mixed blessing. You'll have to manually retard the ignition timing and lean the fuel mixture. The flip side is that if/when the HHO system fails or runs out of electrolyte, you'll immediately lose power and start overheating your valves due to the late ignition timing.
    At low current draw, it shouldn't be much of a problem. You will need to use a heat sink on it. The MCP14628 IC prefers a Vcc of 5v, absolute maximum is 7v. It was necessary to use 6v for the Linear Technology LTC4447 simulation.

    You'll find that your combine actually generates around 28v when it's running.
     
  17. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    If your combine is a turbo, HHO hurts even more than on a gasoline engine. There are zero reported (let alone proven) improvements with cummins diesel engines (turbo and non-turbo), but turbos have had damage from the water.
     
  18. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    The Jeep is an 8 cylinder 350 buick and the HHO made a huge difference in power. It was pretty gutless before with 170 thousand on it. I have a friend who had real good luck with a power stroke and HHO, but I too haven't heard any thing about a cummins. This has a detroit w/turbo in it. My friend with the power stroke said not much modification was needed on his truck. the biggest edition was a CHT and EGT sensor.
    Yeah we're pretty fuel crazy here In montana. Think I'll try the bio-desiel thing first.
    Thanks
     
    Last edited: Mar 30, 2009
  19. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Odd that you're getting a drop across L1.

    What did you use for L1?

    Did you use the MCP14628 driver IC as I mentioned in the note instead of the LTC4447?
     
  20. samuraimarc

    Thread Starter Active Member

    Mar 28, 2009
    48
    0
    For L1 I used the torroid core you suggested with the #6 wraped around it. Not easy.
    I did use mcp14628 driver. I have tried many different variations with the LTspice program but I don't think I'm getting the right load. I've used a 12v light bulb to test the old "PWM". I can't get any light with this one. Not sure where to go from here.
    Thanks
     
    Last edited: May 3, 2009
Loading...