Announcement

Collapse
No announcement yet.

My opinion on MOSFETs

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • My opinion on MOSFETs

    Hi,

    WIth regard to PI detectors,

    There is the general belief that higher voltage break down, means faster discharge, and therefore sensitive to smaller targets.

    I tried a MOSFET with much 'better' specs : lower resistance, faster turn on and turn off time, higher voltage breakdown (600V).

    And it made the detector sensitive to wet sand, this is because an oscillation occurred (to shield, or ground or something) (see attached) I think.

    With the standard mosfet: IRF2405, the detector is sooo quite and stable, even in wet sand, and in the water.

    However, simply changing the mosfet, can turn it into a noisy piece of garbage.

    Observations:
    The high breakdown voltage, does not really improve the response to smaller targets, it simply means the discharge will be faster initially at the very start., but as soon as the threshold voltage is reached, i.e for most of the discharge, there will be a good rapid magnetic field change.

    Even if the maximum flyback voltage is limited to 75 volts (by a diode), the detector can still be very sensitive to small targets.. intially the discharge is slow, and the voltage is greater than 75V, however when this point is reached, the change in magnetic field, is almost the same as with the higher voltage ...the delta-magfield has only been delayed.

    The most important aspect of target sensitivity and fast discharge is COIL CAPACITANCE..

    If anyone has any ideas what going on with the above curve, to make the detector sensitive to wet sand, let me know.

    Note, with the standard mosfet, I can dig a hole in the sand under the water with the search coil, and it wont beep. (unless there is a target).
    Attached Files

  • #2
    Can you please attach images in some graph-friendly format, such as gif or png? These are small and not scalable, and I have only vague idea about what's going on there.

    Comment


    • #3
      Hi,

      as far as I know, reducing the switch-off time can be an driver/isolation transistor between the driver ICs and MOS-FET (such as Barracuda, Delta Pulse, Hammerhead).

      From the ,,555 Data book,, the driver IC provides enough energy to turn on the MOS-FET, BUT the corresponding off-speed performance of the IC is no longer sufficient, beacuse charging the reelative high capacity of the MOS-FET.

      The result is definitely better than using an low capacity MOS-FET, while the voltage affects the search depth.
      In the sandy beach of the targets 90% of 5-9cm from the top is in, so this is probably enough depth to be 75V.
      To lowest voltage may be use an smaller puffer capacitor or/and lower suppy voltage.
      Result: longer battery lifetime.

      The speed increase would be good to explicitly dedicated RF-transmitter MOS-FET also try to availability and price is good.

      It is very important for the equivalent damping resistance is set, otherwise do not get a satisfactory result. The resistance value of the setting, see this pdf.: Carl Moreland: Hammerhead.

      Ok, from the pictures:

      What are the two oscillogram photo?
      Which IC, which have pin?

      If your figure is from 5534 preamp IC, the your coil is strongly UNDERDAMPED.

      The after spike must be an certain time period of negative phase, an it followed by a smaller peak and the finally an exponential section.

      Good hunting

      Rumcajs

      Comment


      • #4
        Reposted graphs

        Thanks for the replies, I've posted the graphs in nice PNG format.

        These are the actual coil spikes, on the coil (50V per division)

        Notice one has a slight ring right at the beginning, this I believe is what makes it very sensitive to wet sand, and makes a huge difference..
        This ring will occur regardless of coil damping,..increase or decrease the damping, the ring will still be there..
        THe different mosfets required different damping resistance..the damping resistance seems more to do with mosfet than coil.. The fast mosfet needed more damping (250 ohm), the slower mosfet needed 330 ohm.

        damping is selected based on what is happening after the amplifier..which shows up any slight ringing. Again, these graphs are of the coil spike at the coil.
        Attached Files

        Comment


        • #5
          Originally posted by Tec View Post
          Thanks for the replies, I've posted the graphs in nice PNG format.

          These are the actual coil spikes, on the coil (50V per division)

          Notice one has a slight ring right at the beginning, this I believe is what makes it very sensitive to wet sand, and makes a huge difference..
          This ring will occur regardless of coil damping,..increase or decrease the damping, the ring will still be there..
          THe different mosfets required different damping resistance..the damping resistance seems more to do with mosfet than coil.. The fast mosfet needed more damping (250 ohm), the slower mosfet needed 330 ohm.

          damping is selected based on what is happening after the amplifier..which shows up any slight ringing. Again, these graphs are of the coil spike at the coil.
          Hi Tec,

          Don't forget you have to look at the output capacitance (Coss) specification of the mosfet as well, lower is better. From the damping resistance you required it sounds like your 'faster' mosfet has a greater output capacitance which is what is slowing the whole system down. The sensistivty to wet sand is probably because with the extended decay curve your MD is no longer taking the samples in the correct place.

          Midas

          Comment


          • #6
            Midas, I think you are onto something. Coss is usually higher with MOSFETS with lower Rdson, and in fact Rdson is ridiculously small compared to the coil resistance already, everything below 1/10 of coil resistance is insignificant and it is not going to improve anything by getting any lower. It is the off state that generates pulse, not the puny Rdson.

            Tec, you have ringing, and it surely messes up your reading. Trouble with ringing is that resonance changes due to ground proximity effects, and you get false reading. Hairy parts are moving left and right and samples are wandering, just as Midas says.

            So far I thought Rdson is ridiculously small for MD purposes, so what?!? Now I know better. I'll seek some older MOSFETs for my builds. Thank you.

            Comment


            • #7
              Another question about MOSFETs, and an expensive one at that...

              Hey all,

              I had to dig into this old post to get some thoughts about the Coss thing. I am under the belief that the capacitance of A: the coil + B: the cable + C: the MOSFET Coss = the capacitance which leads to the damping resistor value. I am also under the belief that reducing the capacitance in ANY of the three will result in potentially faster sample times.

              Am I correct in this?

              If one ignores the cost of the MOSFET I am considering (WOW$$), it looks promising for my Litz coil and Litz coax attempt at very short delays and small targets.

              The MOSFET I am going to try is the Cree CMF10120D: 1200v, 24a (49a-pulsed), 134w, Coss 63pF. ~$21 usd.

              CMF10120D.pdf

              The Coss 'savings' over an IRF740 is -157pF. I am unfamiliar with how the other specifications of the part compare to the IRF740 and as to how they relate to the needs of faster sample times.

              So, what do you folks have to say (other than the cost!) about this "Silicon Carbide N-Channel" MOSFET? And for that matter, the role that MOSFETs play in the bigger picture of fast sample times?

              I'll try one out anyway, just to see what I can get for shorter timings.

              Thanks, GTB

              Comment


              • #8
                Originally posted by GT Blocker View Post
                Hey all,

                I had to dig into this old post to get some thoughts about the Coss thing. I am under the belief that the capacitance of A: the coil + B: the cable + C: the MOSFET Coss = the capacitance which leads to the damping resistor value. I am also under the belief that reducing the capacitance in ANY of the three will result in potentially faster sample times.

                Am I correct in this?

                If one ignores the cost of the MOSFET I am considering (WOW$$), it looks promising for my Litz coil and Litz coax attempt at very short delays and small targets.

                The MOSFET I am going to try is the Cree CMF10120D: 1200v, 24a (49a-pulsed), 134w, Coss 63pF. ~$21 usd.

                [ATTACH]24535[/ATTACH]

                The Coss 'savings' over an IRF740 is -157pF. I am unfamiliar with how the other specifications of the part compare to the IRF740 and as to how they relate to the needs of faster sample times.

                So, what do you folks have to say (other than the cost!) about this "Silicon Carbide N-Channel" MOSFET? And for that matter, the role that MOSFETs play in the bigger picture of fast sample times?

                I'll try one out anyway, just to see what I can get for shorter timings.

                Thanks, GTB

                Hi,

                I'm interested to know if your mosfet speeds things up. Yes I agree it will, because the capacitance is the most important bit.
                Also it has a high break down voltage so, there will be higher peak voltage in the spike.

                Here's what I think.
                When switch-off occurs, the power in the magnetic field is lost rapidly at the following rate : voltage x current. The quicker the change in magnetic field the better/smaller the target can respond.

                Capacitance momentarily lowers the voltage (slows the discharge rate) and stores some energy, only to push it out later in the form of oscillations.


                Something I am doing at the moment is using a single loop. The lowest possible inductance and capacitance.
                With high current.
                With no damping resistor it rings at 13MHz on switch off... so this is all due to the mosfet capacitance.

                Tec

                Comment


                • #9
                  Tec is that the right number IRF2405 ?
                  they seem to be only good up to 55 volts

                  Comment


                  • #10
                    Hey Tec,

                    Nice to hear that I am not completely crazy.

                    I let the 'magic smoke' out of my first HH a few weeks ago. Rather than diagnose and repair a well worn horse, I decided to pull out a new board and start over. Since I am useing 1% resistors, tantalum caps, and a host of other tweaks (wrong thread), I felt I should go for the lowest capacitance possible from the coil to the front end. To that end, I am building my own Litz wire coax, Litz wire coil, and looking for the least capacitance MOSFET I can find.

                    I hadn't thought about the break down voltage. Very interesting observation on your part.

                    I'll let ya all know how this works out as it moves along (waiting on parts, testing, etc.)

                    If anyone else has thoughts, step in..

                    GTB.

                    Comment


                    • #11
                      Originally posted by GT Blocker View Post
                      ...I am building my own Litz wire coax...
                      For purpose of connecting your rig to a coil you actually do not need a coax. Much easier build is a simple weave.

                      Comment


                      • #12
                        OK

                        Ok, Davor,

                        "For purpose of connecting your rig to a coil you actually do not need a coax. Much easier build is a simple weave."

                        I thought we had covered this in my Litz wire thread (http://www.geotech1.com/forums/showt...itz-wire-thing {of note, post No.12}). Re-read both your and my posts concerning this matter.

                        Just for fun, Davor, how, exactly, would you go about sampling at or below 6uSec? Your target is 4 grains @ 80% pure gold. It lies behind 6" to 8" of solid rock (perhaps mineralized rock). Gonna go for the "simple weave"? No need for coax? Opps, I forgot to mention, you're underwater while you're looking for that gold and so is your detector. Good luck with the "simple weave, no coax" thing. What is your plan for sub-6uSec sample rates?

                        Just me....

                        GTB


                        __________________________

                        Solutions shall forever triumph over opinions. GTB

                        Comment


                        • #13
                          Water makes some difference if you let it permeate your braid, otherwise a simple weave offers less capacitance than coax, and for all practical purposes shielding as offered by coaxial is of less significance at PI bandwidth.

                          You'll do what you must.

                          Comment


                          • #14
                            You'll do what you must.

                            True enough, sir. As will you.

                            (You never did answer the above question(s)...although you post an average of 100 posts per month)

                            May I see a photo of a detector that YOU have built? I wonder about the validity of your opinions. Perhaps wrongly so. I don't know. It just seems to me that you have many opinions and few credentials to back the play. When do you work in your lab if you are posting at least three times PER day? 1600+ posts in 16 months? Really?

                            I don't mean to come off surly, but come on. Sit back and learn for once.

                            My opinion,
                            GTB

                            Comment


                            • #15
                              CMF10120D

                              Hey Tec,

                              I'll just go ahead and buy one of these suckers (CMF10120D) from Mouser. Should be interesting. What are the options for increasing the coil voltage if this puppy works out? I never thought I'ld see a MOSFET at 1200v/134w and 63pF. Hope it works as well as it sounds. Thoughts?

                              GTB

                              Comment

                              Working...
                              X