1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

Solid State Tesla Coil Burning Mosfets

Discussion in 'General Electronics Chat' started by Nissan20det, May 24, 2017.

  1. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    It's not a matter of can it switch it, it's a matter of how well it can switch it and how long it takes it to switch it. With 3x the propagation delay alone, your driver is going to switch the H-bridge too late (every nanosecond counts here), leading to more abuse of your MOSFETs.

    My point is you should be using 5V logic (the 74HC series, specifically). The 12V logic is too slow.

    Good to hear, that's much more reasonable.

    13V is probably still too low to reliably switch the MOSFETs. There is a direct relationship between Vgs and the switching time. Higher Vgs = faster switching. If you're only applying ~12V to the gate-source junction, you probably will not get the switching times specified in the datasheet. I would recommend replacing your TVS diodes with 30V ones and modifying your GDT to supply >15V out.

    It doesn't matter what it is being powered from. You need a reservoir capacitor directly across the pins of the driver chip. This minimizes inductance of the wires to the PSU that would otherwise prevent current from being pulled quickly enough from the power supply. 100uF is probably a bit overkill (1-10uF would probably be more than sufficient, and would probably have a lower ESR).

    One thing you need to keep in mind is that GDTs are very sensitive to winding lengths. If you're off by half a turn on one of the windings, the corresponding transistor will switch at a slightly different time than its complement. This could potentially lead to shorting of the H-bridge supply through the MOSFETs, quite likely damaging them.

    One of the best methods I've come across (and I use this one myself) is to wind CAT5 Ethernet cable (still in its sheath) around the core. Tie all of the white wires together. That creates your primary coil. The remaining 4 twisted pairs connect to their corresponding transistors (take careful note of the polarity!). You MUST keep the lead lengths as short as possible. Added inductance can have a severe impact on your switching times and waveforms. While the GDT you posted the image of may work, I expect it has a very high leakage inductance and mismatched winding lengths (even if only by a few millimeters) which will impact the performance of your coil.
     
  2. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Yes this I paid very high attention too. Every wire was cut to identical length, then wrapped together, Im sure they are equal. but good point, ur right the slightest difference will add a big delay on its corespondent.

    Okay now I just feel dumb. lol . Every data sheet I have ever read will show a faster response as voltage increases. (ie CD40106).. Im I going crazy? Cuz what I learned is 12v is faster than 5v?? Also Okay I have a 7414, this means I must re-wrap my GDT. 12-12 will give less than 5v on gate right? So I need around 6-14 or 8-18? So I get 13ish at gate? these fets will blow at 30V gate. They're max 20v so 3/4 of max input should insure a fast switch right? 14v at gate with 15v zeners should be okay?

    So before next mains test-
    74HC14
    5v
    Re-Wrap GDT (Ratio?) and ya I have alot of Cat5-e around good thought.. I do love the flush look tho, with coil wire.
    Any thing else or give her a shot? I am still stuck on feedback did you see my cores worked as a GDT in last post?
     
  3. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    Ok, let's get something straight. When I say logic I am referring to the schmitt trigger inverters used in the feedback circuit. The driver is NOT logic. The driver should be powered with 12V or higher (to drive the MOSFETs harder).

    A 1:1.5 GDT would be ideal (this would provide you with 18V gate drive from a 12V supply) but that's more difficult to get with shielded CAT5 cable. You'd have to find some way to shorten the primary coil without shortening the secondaries. You might consider changing the entire drive to 15V and use a different driver IC (there was a microchip version that was a drop-in replacement for the UCC27425 that I used on my DR that is designed for a higher supply voltage). This will ensure your MOSFETs switch fast enough.

    Regarding the feedback, you're right that if it works for the drive, it should work for feedback. I see you have your secondary ground connected to your digital ground. That's just a bad idea. Look at Guangyan's schematic -- he suggests connecting the base of the secondary to mains ground or a true earth ground. I suggest you do that too. Otherwise you could be injecting noise into your digital circuitry which can cause undesirable operation.

    One other thing I'm noticing in the waveforms you showed me, they are not very good square waves. You should be seeing a very distinct square wave, not that semi-rounded cr@p in your images. And again, a half-decent scope would go a LONG way here and would really help troubleshoot this mess more easily. That piece of garbage USB scope of yours is not doing the business. It's hiding a lot of important information from us.
     
  4. dave

    Dave New Member

    Joined:
    Jan 12, 1997
    Messages:
    -
    Likes:
    0


     
  5. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO

    okay so ur saying leave circuit how it is voltage wise, but add a 5v rail and swap the CD40106 for a 74HC14. Your saying Me running the gate driver and multivib at 12v is okay right? other than going higher to 15V.

    I was mistaken I only have CD40106 as well as HD74LS04 Will that work? I think it is a logic one as well its faster than the CD40.?

    I only connect them in the circuit to inform people they are electrically the same point. My PSU is powering the driver, the ground rail of that is connected to earth ground. So connecting my secondary to Earth (as I have always done)Makes the secondary earth the same point in my digital ground.
     
    Last edited: May 31, 2017
  6. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    Correct.

    Even if the ground rail of the power supply is connected to earth ground, it will be a roundabout connection with high inductance. Use that rail to ground your digital circuit, and connect the base of your secondary to a metal stake that is pounded into the ground.
     
  7. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Okay so you saying u can run these inside lol??? I cant connect to mains ground??? thats a stick in the ground lol

    I was mistaken I only have CD40106 as well as HD74LS04 Will that work? 15ns 9ns..I think it is a logic one, as well its faster than the 74HC.?


    And thank you so much for the help If the HD74 will work Ill try it out now!!
     
  8. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    Proper earth ground is definitely preferable, as mains ground is technically a very long wire connecting to ground after a long distance (adds lots of inductance). Again, when working with Tesla coils it is imperative that wire lengths be kept as short as possible.

    The 7404 will not work as it is not a schmitt trigger inverter. You must have a schmitt trigger type to make a clean square wave out of the feedback signal. Otherwise, LS should probably work, it'll just take more power to operate. Speed is comparable though.
     
  9. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Im sorry but that confused me lol, the 7404 is bad but LS should work? Im talking about one chip HD74LC04
     
  10. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    A 74LS14 would work. A 74LS04 will not. A 74LS04 is just an inverter, the 74LS14 is a Schmitt-trigger inverter. You must have a Schmitt-trigger inverter.
     
  11. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    ahhhhhh... okay I get it lol. sorry

    This is the wave out of the CD40106 with antenna..So since the 74HC14 be something I must order, Could I at least give it another go with antenna? Ill drop a turn on GDT primary, Try to get above 12v ill shoot for 14 with my zeners?

    Also If making another order
    Is there a faster schmitt then 74HC14 in your opinion that would be better?
    Ill order some 18V zeners as well
    Known toroid value, ferrite. What specs for feedback?
    Any other things to order?
     

    Attached Files:

  12. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    The coil would still work with an antenna after you switch to the 74HC14 (which is plenty fast, by the way), if not better. Drop 4 turns on the primary of the GDT to give you ~18V out, and get some 30V bidirectional TVS diodes for your FET protection. If you're ordering new toroids, get the ones I linked you to earlier. Don't know what you mean when you ask "what specs for feedback".

    Maybe also try ordering some more IGBTs. Even if the MOSFETs work, the IGBTs will work better and will be more efficient.
     
  13. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Okay for sure I have 4 IGBT's, at 5$ a piece for 600W isnt bad at all. so yes I will order more for sure.

    I meant what spec on toroids, to insure a working feedback. Your links are probly my answer there.

    How does this not fry them being max 20v? So if 18v is to low abut is under 20v. How is 30v, over 20v okay for protection?As well arnt Bi-directional diodes cathode to cathode? how does anode to anode work Like I have my zeners now?

    I understand that'll work with the 74HC's but Im asking if with PCB and GDT re-wound Okay to try? I dont care to fry more fets but am wondering the point of giving it a shot?

    One other question, what to far of a distance for an antenna? 1FT 8 FT? obviously I can understand power output would depict this. So say 80V 2.5amp of mains?
     
  14. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    The most important spec is the permeability, which should be around 10,000 for a good GDT/CT toroid.

    Max Vgs rating for the MOSFETs you linked me to is +/- 30V, so you just need to make sure the clamping voltage of your TVS is less than that (note this is the clamping voltage, not the breakdown voltage). As for bidirectional diodes, since the transients you're protecting against are AC, whether the configuration is cathode to cathode or anode to anode makes no difference.

    I need you to work on refining your grammar skills. I'm finding it more and more difficult to understand you.
    Anyway, feel free to play around with an antenna, but using slow logic will prevent the circuit from switching when it should. You're more likely to damage components that way.

    I usually see the antennas placed less than 1 meter away from the Tesla coil. It may still work further away, but this is very coil-dependent. Just find what works best for your particular coil and stick with it.
     
  15. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Sweet, Okay I must have seen some other datasheet cuz I swear that I saw +-20v. But Okay 30V max, then yes above 18-20v would be ideal.

    I apologize for the grammar. As well I am excited about this working. I am trying to reply quickly to ensure a response.
     
  16. shortbus=

    shortbus= Well-Known Member

    Joined:
    Mar 23, 2008
    Messages:
    1,230
    Likes:
    77
    Location:
    youngstown, oh
    That statement is true in the case of the CDxxxx series of logic. But only that series.
     
  17. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Okay I get it! Ya I have tons of CD series, so that explains it. Thks


    Hey, I had a thought about this. If it where a issue of lag, how is, one, my antenna working on the CD series, and two, wouldn't I see an output from the feedback transformer? I understand this series will cause lag but I dont think thats my issue for the transformer not working. If it where a lag issue, I should be seeing a signal, although by the time its implemented to the FETs it'll be so many degrees out of phase but I should see something there right?

    If I scope the output of my antenna or the transformer with one leg floating, I get a clean sine wave. The Second I ground that floating transformer pin my scope reads dead flat. Could this be due to my digital ground being earthed? Therefor sharing with the secondary ground? I attached a scope view of my 50 turn toroid with the secondary ground ran through it. The scope is probed on one leg and grounded on the other.(NC to circuit)

    So it works! But then does not once it touches my digital ground.
     

    Attached Files:

  18. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    I never said using the slow logic was the reason your feedback transformer wasn't working. I was simply stating that using slower devices will decrease the efficiency of your setup, thus reducing the effectiveness.

    What is the capacitance of your scope probe? Also, have you checked D10 and D11 to ensure they're connected the right way around and that neither of them has been damaged? You should also have a DC blocking capacitor in series with R7.

    Is it still working with the antenna?
     
  19. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Sorry, I just figured that's why you brought it up.

    D10-D11 are working- diode drop of 0.654 and 0.652, they are connected as schematic(wrong way would short diodes)

    I did not add the feedback DC blocking capacitor because the guy who designed it (Loneoceans) said It introduces lag and can be omitted. I have tried with and without.

    This does work every time with the antenna.

    Probe Specs:
    Attenuation - 1:1 +-2%
    Input Capacity - 50pf +-10pf
    Internal Resistance - 32ohms +-5%
    Bandwidth 10MHz +-3dB
     
  20. DerStrom8

    DerStrom8 Super Moderator Most Helpful Member

    Joined:
    Aug 9, 2010
    Messages:
    5,514
    Likes:
    507
    Location:
    Vermont (GMT-5:00)
    Can you please probe the node between D10 and D11 when using an antenna? And show it side-by-side with the waveform from the CT?
     
  21. Nissan20det

    Nissan20det Member

    Joined:
    May 24, 2017
    Messages:
    95
    Likes:
    1
    Location:
    Parker, CO
    Okay here it is. Yellow is the Node between D10-D11 @X1 probe, Blue is off CT X10 probe, voltage went past 40vpp. Scope ground is on digital ground, other CT pin is also to digital ground. Is this to make sure I'm seeing an output? Not picking up RF like antenna? Yellow was 12v Blue was 55v ish

    I also changed up My GDT as you suggested, I used Cat5e cable, So I noticed a huge difference on the scope. I was talking in the beginning of this thread, when I probed a high side and low side gate(oppositely switching) I couldn't see dead time due to one being much higher voltage then the other. After the GDT re-wrap I can probe any of the 4 and they are identical voltages with the exact same swing negative. If I probe same switching fets, high and low, They lay right on each other beautifully.. This is a good sign right?

    Edit: After actually running it I could no longer read the highs and lows like in the beginning, this does makes since. Why it randomly allowed me to after soldering on the new GDT, no clue. So back to how it was other then wave forms looking much closer to each other now.
     

    Attached Files:

    Last edited: Jun 1, 2017

Share This Page