WFC Water Fuel Cell by Farlander

Discussion in 'The Projects Forum' started by Farlander, Dec 22, 2008.

Thread Status:
Not open for further replies.
  1. Farlander

    Thread Starter Active Member

    Oct 14, 2008
    158
    0
    Hi guys,

    I opened this thread with the intention of posting progress as well as posing questions related to this project to minimize congestion on the main board.

    I'm building the Dave Lawton D14 dual 555 timer circuit in the pic. This is my fourth PWM design constructed. I had originally avoided this one because it was popular but apparently had only worked for one guy. Did anyone else do a correct replication? If a few did, would they come forth?

    My current problem is this -- I have 4 variable resisters in this circuit. Mine are 5mOhm linear taper pots, 3 contacts. I crossed the middle and an outside contact, then ran wire from the two outside contacts to the board. I realized I was only getting at max 1.6 mOhms. However, in this method, the adjustable frequency range of the circuit is drastically increased.

    When I broke the contact between pins 1 and 2 on the pot, and ran wires from 1 and 2 right to the board, the resistance range increased to 5mOhms again, but output frequency range decreased.

    What am I missing here? Shouldn't higher range of resistance equate to higher frequency range? How do the pot contacts work?

    Thanks for all help and feedback on this project. I was just fired, and could really use a revolutionary technology.

    Regards

    Farlander
     
  2. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    HHO doesn't work in a car engine and nobody has ever proved it.

    5m ohms is 0.005 ohms. 5M is 5 million ohms. Neither is used in the circuit you posted.
    When the frequency pot is a low value then the frequency is higher.
     
  3. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    First things first.
    It's sad news indeed that you have lost your gainful employment.
    I'll kindly suggest that you spend at least a normal work-day's worth of hours in tracking down a new position each day, so that you can continue to afford to keep food on your plate and a roof over your head. Keep in mind that if the circuits you've been finding were viable, they would have been produced en mass long ago.

    Ok, do you mean 5 milliOhm, or 5 MegOhm pots?
    5 milliOhms = 0.005 Ohms
    5 MEG Ohms = 5,000,000 Ohms

    Generally, the center terminal is #2, and is the "wiper" of the pot. The two outside terminals are the ends of the fixed resistor portion.

    Well, yes - but at a price.

    You're dealing with RC time, which is resistance times capacitance.
    If you replace a 50K pot with a 5 MEG pot, you will extend the lower frequency range significantly (100x); however you will also decrease the resolution by 100x at the high frequency end. A very small movement of the pot will result in a large change of frequency, particularly if the pot is a linear taper.

    Already explained above. If you're unsure, measure between the various contacts using an ohmmeter while moving the shaft. If your meter won't go to 5 megohms, try placing a voltage across the two terminals that are furthest apart, and measure the voltage from one of those terminals to the middle terminal while turning the shaft.
     
  4. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    What you have is a hobby circuit for an inexpensive variable frequency and variable pulse width generator. It's been grafted onto the FET driver for the cells. The 820 ohms to ground is unnecessary, and the 200 ohm resistor is about 3 times too large unless you like to torture FET's with poor drive waveforms. You want less resistance in the gate circuit to make the FET switch faster. In fact, changing the 820 ohm resistor to a pullup (tied to +12 instead of ground) might also help.

    It only "worked" for one guy because he lied about it. The magical effect does not exist.

    The circut as shown is approximately okay. To sensitively look for some effective combination of frequency and pulse width, go back to the original values for the pots, but use 20 turn trimmers for better resolution.

    Get a frequency counter and that way you will at least know exactly what frequencies you have not been able to get to work. Then you can tell that you have carefully run from some very low to a much higher frequency with negative result. Eventually, it will either work, or everyone will realize that it will never happen.

    If the magic frequency for water resonance existed, the circuit would only need a small frequency adjustment to get it right for the cell.
     
  5. Farlander

    Thread Starter Active Member

    Oct 14, 2008
    158
    0
    I like your thinking, because I already tried this. I used 25 turn 10ohm in series with the 5Mohm. I'm afraid I might have broken them, because they click after a certain point. Is there a boundary before damage occurs?

    I have an 8Mhz oscope sitting on my bench with automatic frequency counter. I can sustain about 15.5 khz max with the proper balance of pulse width/frequency.

    Since I'm not formally trained in electronics I'm still learning about FETS. Would doing this mean I change the type of fet to a P channel instead of N channel?
    I have a ton of FETS layin around from old UPS systems. The largest quantity is of IRF3710. The datasheet says Threshold Gate Volage Min 2 - Max 4. How do I determine what FETS use 12V at the gate for 100% on?
    Thanks all
     
    Last edited: Dec 22, 2008
  6. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Trimmer capacitors weren't designed for continuous operation. You might be able to run them back and fourth all the way perhaps 200 times before they fail. They are also rather delicate, due to their small size.

    A number of vendors make full-size 10-turn pots with 1/4" shafts, but they're rather expensive at $10/ea and up. The really decent versions (those with dials) run around $40 or so. Sometimes you can find them at surplus houses.

    What you might do instead is to use one relatively large and one relatively small pot in series; IE: a 50k pot and a 2k or 5k pot. Then you can make a small change on the large pot, and "fine tune" with the small-value pot.
     
  7. Farlander

    Thread Starter Active Member

    Oct 14, 2008
    158
    0
    One more thing -- would lowering the 100 ohm resistors in line with the pots help? I was thinking of changing them to 47 ohm. Once I removed them completely and it blew the 555.
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    No, 100 Ohms is far too low as it is.
    Pin 7 in a BJT 555 timer is an open-collector transistor that has a maximum sink current of 15mA. Since your supply voltage will be roughly 14v, the 100 Ohm resistors above "Frequency" should be at least 620 Ohms, and the 100 Ohm resistors above "Mark/Space" should be at least 1.5k Ohms.

    If you want higher frequencies, you'll need smaller value capacitors.

    As it is, with caps that large you're going to be putting a lot of current through your pots. That's probably why your 25-turn trimpot is "crunchy" now; it got fried due to the power being dissipated in it.
     
  9. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I just noticed the exchange about the 820 Ohm resistor to ground proposal as a pullup.

    This is a bad idea. The idea of the pull-down resistor is to keep Vgs=0 to ensure that the gate doesn't float high when the timer isn't powered. If the gate is allowed to float high, maximum current will flow through the MOSFET, likely burning up the attached circuit, or at least draining the battery.

    A bjt 555 timer can sink or source around 200mA when Vcc=15v. The output (on pin 3) can approach within 0.2v or so of ground under light load, but it's high level voltage output is limited to Vcc -1.7v (approximately).

    So, if your Vcc=14v, the highest it's output can get is around 12.3v; plenty to fully turn ON the gate of a standard MOSFET.

    That 220 Ohm resistor needs to be reduced to 62 Ohms, or even somewhat less. 47 Ohms would probably be safe, as the maximum voltage difference will be 12.1v, and the gate won't take long to begin charging/discharging. 12.1v/47 Ohms gives an instantaneous current of 257mA, but that's just an instantaneous value; average current would be far less.

    820 Ohms is perhaps somewhat small for a pull-down resistor. It will increase turn-on time, because it represents a fixed 15mA load to the 555's output.

    There are much better designs for a pull-down circuit.
    [eta]
    See the attached. On the left is a simulated signal generator with an 0.2v to 12.3v pulse at 43.1kHz. The signal on the gate is still reasonably square. The 2N2907A's base gets pulled low by R3 when the signal generator's output goes low, causing rapid dumping of the gate charge. Peak current from the signal generator is less than 180mA.
     
    Last edited: Dec 23, 2008
  10. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Concede the point of the gate floating up with power off to the driver. I would probably use a higher resistance to ground - like 20K or so.

    But the trimmer will go clock at either extreme of adjustment. It is built that way so no damage is done by turning too far. If turning the other way does not click and the expected resistance change takes place, then it is in fine shape.

    Using a meter to look to voltage changes is a good idea.
     
  11. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I was contemplating this just a few minutes ago, as a matter of fact...

    One thing I left off was a safety pull-down. Should Q1 (the 2N2907A transistor) fail OPEN, the gate of the MOSFET would be left to float ON.

    A 10K resistor wired directly from the MOSFET's gate to ground would provide an "emergency backup" in case of such a float condition. During normal operation, it would have at most 1.2mA current flowing through it, which would have minimal impact on the gate's rise time.

    R3's 4.7k value is a trade-off between modest loading on the 555's output when high (2.6mA) and providing a prompt discharge of the gate. It's not perfect, but certainly an improvement over the original.
     
  12. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Well, I strongly suspect that the original circuit was for demonstration purposes only. Any reasonable change should constitute an improvement. Especially with respect to driving charge on and off the FET's gate.
     
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Are you using 10:1 probes? 1:1 probes will have a loading effect.
    No, you would not want to change to P-ch MOSFETs if you can avoid it.
    Generally, P-ch MOSFETs have nearly twice the Rds(on) of their N-ch counterparts. This means more power dissipation in the form of heat. It would also mean that you would have to change the output drive circuit quite a bit, because the 555 won't go higher than Vcc-1.7v, which would be required to fully turn off a P-ch MOSFET.

    Those are very appropriate for your application, if you keep Id (drain current) at 20A or less. Yes, it's specified for 40A @ 100°C, but de-rating components is good practice to ensure long life. As it is, 20A @ 14v across 28mA is nearly 8W power dissipation; you will need a heat sink on the MOSFET.

    [eta] Note that MOSFETs are very sensitive to static electricity. One little "zap" and you'll have a collection of small paperweights. :eek: You should have them stored in either static-safe bags, conductive foam, or the good old hobbyists' standby, aluminum foil.
    The threshold gate voltage is where the MOSFET begins to conduct at a certain Id.
    What you really want to look at is what Vgs (gate voltage as referenced to the source) is when they specify the Rds(0N), and also consult the maximum specification for Vgs.

    Your IRF3710 is specified for Rds(0N)=28 milliOhms when Vgs=10v.
    The maximum Vgs is +20v/-20v
    Therefore, you are using the MOSFET within the gate voltage specifications.
     
  14. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Now for a comparison between the original gate driver and the changes I proposed.

    See the attachments. Farlander, if you'd like, you can download/install LTSpice/SwitcherCad from Linear Technology's site for free, and load up the simulation (the .asc suffix file)

    The simulated 555out signal is a pulse with a PRF of about 43kHz.

    The red trace is a simulation of the original driver's gate signal using a MOSFET that has a similar total gate charge as the IRF3710 Farlander's using (the IRF1405 gate charge is 120nC, IRF3710 is 130nC). Notice the relatively long period of time spent in the 4v-8v linear region.

    The green trace is the gate signal of the proposed modification.
     
  15. Farlander

    Thread Starter Active Member

    Oct 14, 2008
    158
    0
    Wow... Thank you so much. I'm digesting this information so it sinks in. Reading your conversation is teaching me a lot.

    I finished the second half of the oscillator circuit. For some reason, the second oscillator does nothing... maybe the chip is fried, or something to do with the reset pins? Will try and diagnose/repair.

    Sgt
    I will make the changes -- I'm using a 12V 1000mA power supply. It's very clean dc and the brick is heavy. When I plug this straight to my cell, it produces 2-3x more gas for about 5 seconds, then trickles down to a constant level.

    Sgt
    No I do not have 10x probes, only 1:1. Can I build some? I have a decent selection of small ceramic disk capacitors. Do you mean 1:1 probes would draw more current from the circuit? I don't understand current flow through capacitors charged/uncharged though I've read it so many times. An uncharged cap has current, when charged, no current. Wouldn't more current flow when it's charged?

    My foreseeable dilemma is the auto transformer I was going to use to increase volts is actually just one coil-- not an isolation transformer. Luckily, I have 4 (20+lbs) isolated step up transformers, designed to switch between 110-220V. I was going to run them in series-- my goal at this point is 600V at the electrodes. With only 12V input, I can't get there. I think I want to use 100V input. I'm not sure where I'll be able to get the power supply, I planned on using the one at my old job. I could build a full wave recitifier and power right from the mains.

    Is there a Trench FET that will switch with12V and carry100-120V? If I use all 4 step up coils --- 100-200-400-800V at the electrodes? Any problems with doing this? I thought I would use forward bias diodes on all the positive leads between xformers.

    Merry holidays!
    &
    Sincerest Thanks
     
  16. Farlander

    Thread Starter Active Member

    Oct 14, 2008
    158
    0
    Oh yea, that program rocks. I've been looking for something just like that. SWEET
     
  17. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    If it ain't workin', it ain't finished. ;)

    I predict that your power supply's voltage output is dropping significantly in those first 5 seconds. 1A is not nearly enough for what you're trying to do.

    If you can find a discarded computer with an ATX form factor power supply, you can make a decent bench supply out of it for a few bucks. I have a little 250W ATX form factor supply that I converted to a bench supply, just the +12V output can source 8 Amperes. I also get +5V at up to 25A, 3.3V @ up to 14A, and -5V & -12V at less than an amp total, but sufficient for small projects.

    Google "ATX Bench Supply" for tons of ideas. Whatever they recommend, don't cut off the black ground wires; you need them for high current applications.

    x1 scope probes will load your signals quite a bit, so what you're seeing isn't what's happening when your probe is not loading the circuit.

    Yes, you can make probes, but your mileage may vary considerably.
    Here's a thread from elsewhere that has a number of links:
    http://www.edaboard.com/ftopic111372.html
    I purchased some switchable 10:1/1:1 100MHz probes via Ebay that have been serving me reasonably well; they were around $10/ea.

    OK, wait a sec - are you using pure distilled water, tap water, or are you using an electrolyte solution?

    They were designed to run at 47Hz-63Hz with a sinewave input. They will be extremely inefficient if you try to drive them with a high frequency square wave.

    This is not a good idea for you to try at this point. My concern is your safety.

    Yes, the IRF6xx series have a Vdss rating of 200v.
    The IRF7xx series have a Vdss rating of 400v.
    However, you need to stick with low voltage for a good while. When you start venturing above 50VAC, you get into potentially lethal voltage ranges, particularly when powered off the mains.

    Yes, as I've already mentioned, they are designed for mains frequency sine wave input, not high frequency square waves.

    If you want to generate high voltage pulses from 12v square waves, that can be done using wound ferrite toroids, ferrite rods, or even winding wire around a tubular cardboard form (basically in order of efficiency; the latter will require a LOT more wire), but not from laminated steel-cored transformers designed for 47-63 Hz.

    Hope your holiday is happy and safe. :)
     
    Last edited: Dec 26, 2008
  18. eblc1388

    Senior Member

    Nov 28, 2008
    1,542
    102
    With slight change of resistor values, the original driver can be improved a lot on its performance.

    [​IMG]

    Beware though SPICE might be lying.
     
  19. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Don't forget that the signal generator square wave is "ideal"; there is no limit to the current it can output - unlike the 555 timer, which can source/sink a maximum of 200mA when Vcc=15v. My simulation stays within that limitation; I promise you that your simulation does not. If you check the peak current flow through R1, you'll see spikes of around +/-12A when the gate is charging/discharging, which would be impossible for a 555. (I haven't re-run the simulation with your new values for R1/R2; this is just Ohm's Law in action.)

    You really don't want a zero resistance path to a MOSFET gate anyway, because it will "ring" at very high frequency due to the inductance of the lead/trace and the capacitance of the gate. A small resistance will snub the ringing, while not appreciably increasing the switching times.
     
  20. eblc1388

    Senior Member

    Nov 28, 2008
    1,542
    102
    Your remark prompted me to do another simulation, this time with a "real" 555. I increased the series resistor to 10Ω so the 555 is not outputting into an uncharged capacitor and the 12A capacitive charging spike issue is no longer relevant. Here is the result. It still shown faster rise and fall time of the current. So unless SPICE is lying...which it often does.

    [​IMG]

    Agree. So this time I used 10Ω instead.

    The real answer is only to be known by actually building the circuit and get the actual performance data.

    I do need to mention that this example is an exception because 555 can both source and sink current with its output. Many single-ended output stage can only source or sink but rely on a resistor to do the work in the other direction. Your circuit, with its capability to discharge the gate quickly, will definitely out perform the simple resistor used here.
     
Loading...
Thread Status:
Not open for further replies.