Author Topic: DRSSTC Performs only with 1cm Arcs - Troubleshooting  (Read 772 times)

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
DRSSTC Performs only with 1cm Arcs - Troubleshooting
« on: February 21, 2021, 01:48:01 PM »
Hello to all High Voltage Enthusiasts!
I recently build a DRSS Tesla Coil!
I am a bloody beginner to this topic!
Here are the specs:

UD2.7 Driver without Phase Lead Inductor (removed and shorted)

Secondary Coil 11cmx70cm = 4.3"x27.5"
0,32mm Wire = 28AWG
About 2000 Turns
Top load 4cmx16cm = 1.5"x6.3"
Fres = 100kHz

Primary Coil  ID 18cm = 7"
Turns 5 x 7AWG

Tank Capacitor / Primary Cap MMC 225nF 5kV MKP-1

Half Bridge
Bus Caps 2x2200uF 400V
IGBTs ixdn 75n120
440V 1.5KE TVS
1uF 1200V snubber
5Ohm gate resistors with bypass diodes for discharge
24V Zener diodes
Bus voltage 325V each Cap therefore is charged with 160V

GDT Typ77ferit big core with 10 turns cat 5 1:1:1

Interrupter with shielded cable and optocoupler implemented.

The coil is operating and oscillating.
But only 1-5cm arcs are achieved with a screwdriver/soldering iron hold to the coil.

My thoughts why the coil does not work:

Long unshieldet cable from the feedback loop to the Driver Bord. Therefore big parasitic inductors, also big parasitic inductions on the circuit board.
Parasitic inductive on the atypical Primary cap because of its design.
Geometry of the secondary because of 7:1 ratio and very small top load. Therefore very small band for FR.

My thoughts on solutions:

Secondary with 50cmx10cm. (Maybe bigger toroid if needed?)
Using only one path of the primary cap, therefore get rid of parasitic inductance.
Reducing the length of the cable from primary feedback an shielding.

Should I use phase lead?
Maybe because i removed phase lead the system isn't working right? Do i need phase lead (with the ud 2.7)?
Maybe removing the 150pf cap of the input of the driver?

What are the thoughts, as Tesla coil experts?

I would be very thankful if you share them with me and give me solutions how to fix the problem.

Yours Chris
« Last Edit: February 28, 2021, 04:43:44 PM by ChrisBlis »

Offline Mads Barnkob

  • Administrator
  • High Voltage Expert
  • *****
  • Posts: 1663
  • Karma: +33/-0
  • Denmark
    • View Profile
    • Kaizer Power Electronics
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #1 on: February 21, 2021, 06:40:00 PM »
Hi Chris

Welcome to HVF!

It all looks good, that should have no problem working from the specifications and materials you have used.

Since you have a oscillation inverter that also shows end of burst reversal voltage (despite two phase shifts in last scope shot...) I would check out 3 things for starters.:

1) Phasing of the primary coil or feedback transformer, reverse one of them to see if you are simply driving it completely out of phase.
2) Is grounding of the secondary coil firm and correct to its own RF ground or a artificial ground plane?
3) Feedback signal strength, is it powerful enough to start second burst and get itself running?
https://kaizerpowerelectronics.dk - Tesla coils, high voltage, pulse power, audio and general electronics
https://www.youtube.com/KaizerPowerElectronicsDk60/join - Please consider supporting the forum, websites and youtube channel!

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #2 on: February 21, 2021, 07:50:37 PM »
Great to see that you have an oscilloscope.  That will be important to figuring out issues.  Do you have a second probe to allow two-channel measurements?  Do you have any way to power the half-bridge from an isolated source (isolation transformer or DC power supply)?  That would make probing half-bridge output easier.

It looks like CT (current transformer) feedback to the driver uses twisted pairs.  As long as the CT leads are twisted together, length isn't critical.

Trace inductance on the driver ECB may be a problem.  There appear to be narrow traces on source leads of the GDT output FETs and other power supply interconnections.  Still, your fine-pitch SMD assembly and rework skills are impressive!
Secondary geometry and top-load may not be exactly optimum, but should be plenty workable.
Primary circuit wiring inductance is high, but also not likely a critical issue.  MMC construction is fine.  It's the large loop from bridge through CT to primary coil, back through MMC to the other side of the bridge.  Minimizing area of that overall loop is what matters.

Where is the scope probe connected for the images you attached?  Is this on the driver output (GDT input)?  If so, it looks like the scope probe ground must be connected to one of the driver outputs.  Even if the scope is floating, the capacitive load on the probe ground lead can be problematic.

Can you scope the feedback CT (voltage across the 51-ohm burden resistor at the driver input)?  Ideally this would be combined with one side of the driver output on the other scope channel.

BTW, C36 on your driver board appears to be 0402 size.  Do you have a part number, or value and voltage rating?  High-value caps of that size typically loose capacitance rapidly with voltage.  This isn't likely a critical issue now, but may be long-term.

Good luck with debug!
« Last Edit: February 21, 2021, 07:52:53 PM by davekni »
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #3 on: February 26, 2021, 09:07:40 AM »
Thank you Mads Barnkob and Davekni for your solution approach!


I will test changing the phase of the oscillator.

If the feedback signal is to low how should i increase it? Decreasing the ratio of the one current transformer to 20x30= 1:600 instead of 1:1150?
I use one big ferrit ring core in order to feed the current transformer for oscillation feedback as also the ct for over current detection, may this be a issue? May also the big air gap between primary cable and ferrit ringcore of the first stage of CT may be a problem (shown in the pictures) ?

At the input of oscillation feedback is parallel to the 51Ohm Burden resistor a 150pf capacitor. I do not use a phase lead Inductor, so should I removed this capacitor in order to get rid of the current divider of C and R? Therefore shuffle of phase shifting and signal weakening, or isn't that relevant?


I do have access to to probes as also to a isolated power source.

I will remove the parasitic inductance on the ecb by connecting the critical parts direct at the inputs oft them via twisted pair cables, instead of leading them across the whole ecb.

I will also remove the big loop from bridge output to primary and back through the MMC.

The probe is in the first picture connected to one Gate of the IGBT-Half bridge. At the second picture the probe is connected to the bridge output.

I will do that to -> Can you scope the feedback CT (voltage across the 51-ohm burden resistor at the driver input)?  Ideally this would be combined with one side of the driver output on the other scope channel.

I mixed up the 2.2uf and the 1uf cap of the output, i will change this. May this be a problem now?

I also use 24V zener diodes at the IGBT-Gates. May this be a problem? I thought 33V zeners on the gates may be very harsh to them. But i can confidently use 33V zeners can't I ? Because I destructed know the half-Bridge, maybe this was due loses in the Zener diodes and therefore destructing them, and afterwards the IGBTs?

Can I otherwise change(hack) the DRSSTC to a SSTC by changing the 225pF Cap with a 5uF(safety for dc) one and add secondary feedback with a 1:50 CT to the UD2.7. Is that realistic or isns't it?


Thank you, I appreciate your help very much!



« Last Edit: February 26, 2021, 10:26:42 AM by ChrisBlis »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #4 on: February 26, 2021, 08:01:54 PM »
1150:1 should work for CT ratio.  You could reduce the ratio, especially for lower-voltage testing.  The red-wire second stage CT looks like it has two turns of the white wire from the first stage.  Is that correct?  That is an easy way to reduce CT ratio without rewinding much.  Your CT looks great as is.  The gap between primary cable and core is ideal - reduces capacitive coupling.  No down-side to that gap.

A bleed resistor  (5k-50k ohms) from half-bridge output to center of the bus caps can help with startup.  The resistor makes the initial charge on MMC consistent so the first half-cycle always produces the same amplitude of current.  (You can get twice the current on the first half-cycle by connecting the bleed resistor to bus- or bus+ instead of the center.  However, only one of those two options will work, depending on which IGBT turns on first.  The bleed resistor needs to be across the other IGBT, the one that is off during the first half-cycle.)

At 100kHz, 150pF is over 10k reactance.  That has insignificant effect on 51ohms.

The ECB inductances I'd be concerned about are traces on internal supply voltages to the output FETs and ICs.  If those are problematic, it will require a new ECB layout.  For now, use it as is.  Scope measurements should indicate if ECB behavior is problematic.  (Actually, I can't see the traces for CT feedback - must be on the back side of the board.  So I don't know if they route near anything that could couple to them.  Wouldn't hurt to run twisted pair back to the 51ohm resistor just to be sure.)

Thank you for explaining which signals were probed for the scope traces.  That makes more sense now.  Yes, do reduce the primary wiring loop area.  Pair primary wires together to minimize loop area.

At 100kHz, it looks like you are OK with the 2.2uF/1.0uF cap swap.  At lower frequency it would be a problem.

Are the 24V zener diodes in series pairs back-to-back (cathode to cathode or anode to anode)?  If so, they may be OK.  You could reduce supply voltage to the driver to 22V to allow for some margin.  Zener voltage increases some when hot, which helps with voltage margin.  If you are concerned about 33V being too high, there are 27V and 30V zeners.  (I use 21V TVS diodes with 19V gate drive.  TVS diodes are zeners designed for high peak power.)

I think UD2.7 could work for SSTC.  With its 51-ohm burden resistor, the secondary CT will need to be lower ratio, perhaps 5:1 or even 2:1.  Perhaps someone will answer who has tried that.

Good luck with the reworking and debug!
« Last Edit: February 26, 2021, 08:15:53 PM by davekni »
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #5 on: February 27, 2021, 01:44:13 PM »
Thank you David Knierim for your methods of approach!

Actually I always asked me what is defined as one turn. One whole 360 degree winding, or just putting the wire straight through the ferrit ring core?

You suggested a bleed resistor with 5k-50k. Have I to figure out this via testing, or just choosing one of them in this range? How big should the power rating be? 10-20W for the 5k one?

I added screenshots of the front and back side of the ECB-layout. At the front side is a ground plane, but not on the back side. May this bee a problem? I also just hooked up every ground to this plane, may this have been misfortune? For example the 51Ohm resistor is at one side hooked up with a wire, the other side is just connected via the ground plane.

Which scope measurements should I take in order to figure out if the ECB is working as expected? Is it at first sufficient to measure only the voltage drop on the 51Ohm resistor and one IGBT-Gate?

Zeners are wired up kathode to kathode.  I will implement a 20V voltage regulator and 21V TVS-Diodes.T

When testing the resonance frequency of the secondary with a sstc driver with vco I got arcs at 100khz but also at 220khz, is that normal? And how should i tune the primary Fres? To aprrox 90khz?

Should I otherwise create a alternative driver as the ud 1.3, and test the setup with this one? This time without parasitic inductors? Or do you think the setup may work with the existing driver board?

What do you think is the biggest issue till know in your eyes?

With kind regards

Christian Fersterer
« Last Edit: February 27, 2021, 03:58:38 PM by ChrisBlis »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #6 on: February 27, 2021, 08:01:39 PM »
Turns is the number of times the wire passes through the core.  If the white wire is passing through the second-stage cores twice, then the overall CT ratio will be half as much.  Estimating from your picture, I see about 35:1 ratio for first stage and about 33:2 ratio for the second stage (for the red-wire second stage) for a final CT ratio of 1155:2 or 577.5:1.  The blue-wire second stage looks to have only one turn of white wire, so would be 1155:1 ratio.

A bleed resistor to the bus cap center will have +-160V once operating at full 320V bus supply.  This is during enable pulses.  So power for a 5k resistor is 160^2/5k = 5.12W when running.  If you are running at 10% duty cycle, then average power is 1/10th, or 0.512W.  Higher resistances will dissipate less power.  Anything in the 5k to 50k range should work.  Pick something you have laying around.  No need to adjust.

For now use your ECB as is.  There is no magic test to see if it will cause problems.  When an issue in overall Tesla coil operation shows up, then it is time to figure out the cause.  That is where scope traces are so valuable.  If the issue is within the ECB, then it is time to fix the ECB.  (BTW, for ECB layout in general, there may be good tutorials online.  I haven't looked.  Perhaps someone else can respond here.  A ground plane is great, a necessity for many higher-frequency designs.  Usually the ground plane is on the back, with components on the front.  For two-layer ECBs, wiring should be almost all on the front, with only short traces on the back to cross over other traces.  That way the ground plane is close to an actual plane.  Bypass capacitors should be immediately adjacent power pins, with vias to the ground plane close to the other side of the caps.)  Once you get good spark length, removing the optical isolator inputs from the long ECB traces may be the first needed change.

At 220kHz you may be hitting a second node of your secondary with high voltage in the middle as well as on top.  That is not a good mode for running.  Yes, 90kHz primary should be a good place to start.

For first steps, I recommend:
1) Add bleed resistor to half-bridge output.
2) Minimize primary coil wiring loop area
3) Make both CTs the same to avoid later confusion.  I'd make both with 2 turns of white first-stage wire (two passes through center).
4) Test both polarities (swap leads) of CT output to 51-ohm feedback to see which one works.
I suggest doing this initial testing with an isolated bus supply, 50V or whatever you have available.  Leave one channel of the scope on one point, say the low-side gate.  Then separately scope current (voltage across 51-ohm resistor) and half-bridge output.

BTW, looking at your layout, I think the 2.2uF and 1uF GDT output coupling capacitors are in the correct places.
« Last Edit: February 27, 2021, 08:39:59 PM by davekni »
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #7 on: February 27, 2021, 09:59:59 PM »
As shown in the scope shots my circuitary should work in principal, shouldn't it? I Set the C33 of the UD2 to 1nF, would 2.2nF be netter for 100kHz operation? Operating the UD2 driver without phase lead shouldn't be a problem in order of getting big arcs should it? Why may i have to remove the leads of the optical coupler when I get big arcs? How big do you estimate my chances do get the system working, in order to get significant arcs? BTW do you think the big loop of primary coil connection was the biggest issue?

I appreciate your work so much!
« Last Edit: February 27, 2021, 10:13:16 PM by ChrisBlis »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #8 on: February 27, 2021, 10:32:43 PM »
1nF is good for 100kHz for UD2.7 C33.  2.2nF is better for lower-frequency coils.

Phase lead has two advantages, but isn't critical for your design.  First, it reduces switching spikes in the bridge.  Since you are using 1200V IGBTs on 320Vbus, you have plenty of voltage margin.  Switching spikes are not likely to fry your 1200V IGBTs.  Second is reduced switching losses.  Without phase lead, the IGBTs will dissipate somewhat more power.  Again not critical.

Optical coupling is to isolate electrical noise generated by the Tesla coil, both arcs and internal switching noise, from the interrupter.  All other designs I've seen use long optical isolation through an optical fiber.  Using a packaged isolater leaves the interrupter wires going physically to the coil drive electronics.  That makes a long antenna to pick up noise.  Routing the interrupter traces across the ECB adds even more coupling to noise sources.  If you are using a simple 555-based interrupter without a microprocessor, it may handle the noise without failure.

The scope traces show an oscillator startup issue.  So the biggest problem is either reversed CT feedback polarity or the missing bleed resistor.  Once the startup issue is fixed, the big primary wiring loop area may be the next key factor limiting arc length.  Primary resonant frequency will increase some when that parasitic inductance is removed, so may require another turn or fraction of a turn on the primary winding to compensate to keep primary tuned just slightly below secondary frequency.

I have no way to know probabilities.  As Mads said in his initial reply, you have all the basic elements in place.  Some level of success seems likely.  Don't know what arcs you will define as success.
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #9 on: February 27, 2021, 10:44:59 PM »
I destructed the IGBTs know is it a problem if I go with 600V ones or should i use 1200V again?

The cable for the interrupt signal is shieldiet and the shielding groundet, will i still have problems with noise?

I would sucess define with 50cm arcs.

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #10 on: February 27, 2021, 11:30:47 PM »
Sorry to hear that your IGBTs died.  Are the gates shorted, or just collector-to-emitter?  Are the TVS diodes fried?  One thought on what may have fried IGBTs:  The 440V TVS diodes are across IGBT collector-to-emitter and are unidirectional.  They will end up conducting forward current in parallel with the IGBT's internal diodes.  The TVS diode's reverse recovery time may be too slow.  (Even worse, just looked up the data sheet for ixdn75n120.  It shows no internal diode.  That is a critical issue.  Fast-recovery diodes are necessary across IGBT collector-to-emitter.  It is amazing you got this far with just the external TVS diodes.)  Once you replace IGBTs with ones that include internal diodes, I suggest connecting TVS diodes across Vbus, not individual IGBTs.

600V IGBTs are fine for 320Vbus if the bus inductance is low.  I posted an example for TO247 parts that could be modified for mini-block packages:
https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795
There is at least one issue with your existing half-bridge implementation.  The high-side IGBT has both emitter terminals tied together.  The reason for two emitter terminals is "kelvin connection".  One emitter terminal should be used for half-bridge current (current from the collector).  The other should be used only for gate drive return.  Without knowing the construction of both sides of the half-bridge board, I can't tell how low inductance might be.

BTW, phase lead reduces spikes, so can make a less-than-ideal bridge layout work.

I can't predict if you will or won't have interrupter issues.  I had to change to optical cable for my Marx generator due to noise.  My Marx generator makes much more high-frequency noise than a Tesla coil, however, so my experience there may not be relevant.  I've used optical cables since.  Only my initial SSTC is wired, just to a simple hand switch, no microprocessors.

If this project is too frustrating now, it may be wise to take a break.  Set this coil aside for now.  Come back to it later when you feel motivated to take up the challenge again - in a week or month or year.  Whatever works for you.
« Last Edit: February 28, 2021, 04:01:13 AM by davekni »
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #11 on: February 28, 2021, 10:04:36 AM »
All Pins of the dead IGBTs are shorted together.

Ok just hooking the suppressors up across the Vbus as you said, simply two 1.5ke440 parallel to the snubber cap, should work? Voltage transients accour on the bus supply not on the IGBTs itself, do I understand this  correct? Can I also use bipolar/unidirectional TVS then for this task? So there is no need  for  free wheeling diodes?

Your work on constructing a low inductive bridge is so genius! If I use a half-bridge with TO-247 IGBTs should this work for my setup too?

Could I use these TO-247 IGBTs? https://at.rs-online.com/web/p/igbt/8648795/

My alternative plan would be using an half-bridge module.
Would this half-bridge module be suitable? https://docs.rs-online.com/0b99/0900766b81478091.pdf

Still using your low inductive design with IGBT mini bricks would be a very cool option.

What do you recommend ? What would be the half-bridge design I should make?

I also have a 4"x20" secondary with 28AWG-wire, should I this one instead of the 4.3"x27.5" 28AWG?

I don't want to take a break because i think i am so close to my target. Mostly due to your help.
« Last Edit: February 28, 2021, 06:32:38 PM by ChrisBlis »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #12 on: February 28, 2021, 07:17:20 PM »
Yes, two 1.5ke440 diodes across the snubber is fine.  You probably don't need the 1.5ke440 at all since your bulk caps are close to the IGBTs and connected with copper planes.  Either unipolar or bipolar will work here.

Transient voltages do occur on individual IGBTs too.  The best way to clamp those is with the freewheeling diode of the opposite IGBT and low inductance connections to VBus.  Clamping with TVS diodes often leads to the TVS diode failing first (shorted) which then fries the IGBTs.

It is best to purchase IGBTs with internal freewheeling diodes.  If you use IGBTs without internal diodes, then you must add freewheeling diodes externally.  TVS diodes are not good for this use.  Freewheeling diodes must be fast-recovery or schottky diodes with the ability to handle high peak current and turn off quickly.  Soft recovery also helps, especially when not using phase-lead.  I've seen only one successful DRSSTC build using external freewheeling diodes.  Almost all builds use IGBTs with internal diodes.

Yes, FGA60N65SMD should work well.  If using these TO247 packages, remember that the case is not electrically isolated.  Either use two separate heat sinks or insulating pads between IGBTs and heat sink.  Insulating pads add thermal resistance, so reduce total power dissipation capability (ie. limit max duty cycle without overheating).  I usually use separate heat sinks, but that adds mechanical complexity to mount them while maintaining electrical isolation.

The half-bridge module is fine too.  A bit lower peak current rating (150A) than FGA60N65SMD (180A), but easy to mount and connect.

Yes, my parallel-plane layout could be adapted to minibricks too.  It is the concept of parallel overlapping planes that is key, not so much any specific example.

Any of the bridge options are fine.  Have you considered a full-bridge of FGA60N65SMD?  A full bridge isn't much harder, and avoids the need for a center-tap on VBus.  The power planes can be the same.  You can see from my example topic that I turned the initial half-bridge into a full-bridge using the same copper planes.  I used that full-bridge for this low-frequency QCW demonstration:
https://highvoltageforum.net/index.php?topic=1268.msg10067#msg10067

Either secondary is fine.  A larger top-load would help.  Spun aluminum toriods look fancy, but aren't needed for performance.  Flexible aluminum duct is cheap and works just fine, as you can see in the above link.  Of course, if you change top load or coil, remeasure resonant frequency and adjust primary to match (slightly lower).

A bit more on considering a full-bridge option:  The obvious advantage is increased power to support longer arcs.  The second advantage is a bit more subtle.  Higher drive voltage (+-320V total to primary) allows a higher primary impedance.  You can reduce MMC capacitance in half (and increase MMC voltage) by rearranging capacitors, and double primary inductance with more turns.  Then wiring inductance is a smaller portion of total primary inductance, so less critical to minimize.
« Last Edit: February 28, 2021, 08:16:50 PM by davekni »
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #13 on: February 28, 2021, 09:22:04 PM »
You said I do not need TVS at all, does this also apply for the snubber capacitor?

"Transient voltages do occur on individual IGBTs too.  The best way to clamp those is with the freewheeling diode of the opposite IGBT and low inductance connections to VBus.  Clamping with TVS diodes often leads to the TVS diode failing first (shorted) which then fries the IGBTs." -> So if the bridge design is designed low inductive the internal diode of the other IGBT will take care of that aromatically right? If i understand this in the right way you don't have to reply to this paragraph.

The internal freewheeling diodes of the IGBTs should always be designed to do the task (handling peak current and short recovery time), so I do not have to worry about them?

What do you mean with soft recovery?

Do I understand right, that a half bridge of two single FGA60N65SMD at total cost of 2*3.35=6.7€ would surpass the IGBT Brick shown above for 50€. This seems so counter intuitive to me.

But as you said I should build a full bridge I will do that with FGA60N65SMD ones. Should I also parallel two of them up for each string? For the gates I can simply then wire them up in parallel at the input of each gate resistor? Do i need the inductors on the gate shown in your design in order to expand charge time?

Your low frequenzy QCW tesla coil was designed as a SSTC? Very interesting! First time I saw a tesla coil with a ferrit core!

In what extend would a bigger toroid help? Getting arcs away from the coil, or are there any other impacts?

If I change the primary cap to 110nF and 10000V it would result in 8 primary windings and a coupling of approximately 15%. Should that work out fine?

At https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795 you wrote
"The gap in top-side foil between VBus+ and VBus- isn't visible in these latest images.  Please refer back to the original post.  It's also hard to see, but I've added some 10nF 630V 1206 caps across the VBus+ to VBus- gap adjacent and under the three 3.9uF snubber caps." ->Why did you put them there?

« Last Edit: February 28, 2021, 09:50:13 PM by ChrisBlis »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #14 on: March 01, 2021, 05:00:56 AM »
The snubber capacitor is needed.  I usually use multiple snubber capacitors to minimize inductance.  (That is why I added the small 1206 SMD cqapacitors across the gap between VBus- and VBus+, for yet lower inductance.  I'm using that bridge at 260V with 330V IGBTs, so only 70V margin for spikes.)  TVS diodes across the snubber are not needed unless the bulk caps are farther away or connected with wires.  TVS diodes across the snubber do no harm.  If they fry, it doesn't damage IGBTs as a result.  TVS diodes for the gates (21-33V) are a good idea.  They protect from ESD and from excessive ring from GDT parasitic inductance.

Speed of diodes included within IGBT packages vary by 10:1.  Some IGBT applications are more sensitive to diode speed than others.  A DRSSTC with properly-adjusted phase lead does not need fast diodes.  Without phase lead, fast diodes are more important.  Unfortunately, diode speeds are specified under different conditions, so can be hard to compare.  FGA60N65SMD's diode looks fairly good.  The FF75R12RT4 half-bridge has somewhat slower diodes.  But it is a 1200V part, so could handle the resulting switching spikes.  (Also, fast diodes are harder to make at high voltages.)  I use STGW60H65DRF TO247 parts for my DRSSTC.  They have quite fast diodes and high peak current capability.  I'm not sure how cost compares with FGA60N65SMD, but I think STGW60H65DRF would be a great choice for your build.  (I'm using them conservatively at 260A each, but have bench-tested a couple to 500A without failure.)

Yes, you could build a full bridge with paralleled TO247 IGBTs.  That increases load on the gate driver by another factor of 2.  (The full-bridge is already twice the load of a half-bridge.  But the mini-bricks you were using have higher gate charge, so a simple full-bridge of either of the above TO247 parts will be about the same load as you had before.)  If you parallel parts, drive the gates (and emitter returns) from the center with equal wire lengths to each part.  Don't drive one and then continue the wire on to the other.  Connections need to be identical to get even current sharing between paralleled IGBTs.  I suggest starting with single parts (4 total for a full-bridge).  That way there will be fewer parts to fry if something goes wrong, and construction is simpler.  Once you have success, then consider boosting power further.

No, gate inductors are not needed.  They can help a bit with optimizing dead-time delay while keeping rise time faster, but that is only me being a perfectionist.  No one else I know of bothers with gate inductors.  It adds complexity - especially in calculating the correct value to match gate charge and series resistance.

The advantage of a larger toroid is increased capacitance on top.  That reduces the relative change as arcs grow.  Secondary frequency doesn't change as much, so can be feed more efficiently from the primary with larger arcs.

Yes, 110nF and 8 turns sounds great for full-bridge drive.
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #15 on: March 01, 2021, 12:25:41 PM »
Okay, then I have few last questions.

Do I need to ground my heatsinks? Propably I then would need isolation plates using the TO-247 IGBTs?
Do I need to ground my farraday cage for control electronics? Or should I connect it with internal logic GND? Or both?
Do I even need a farraday cage for control electronics.

In the last night I thought about the primary cap and then I realized that I have know a primary cap of 220nF and 5000V but switching the 2 cap modules of 110nF and 5000V in series would result in 55nF and 10000V not 110nF and 10000V.
So my question is can I just use one 5000V and 110nF module  instead and still handle the higher reactive voltage caused by the higher supply of the full bridge ? I use FKP-1 capacitors.

And how much bus capacitance do I need? Should I use only one 2200uF 400V capacitor, or should I parallel two of them up? Maybe this increases bus inductance,but it won't if I connect on each side of the copper planes one cap? I want connect them via short wires.

Should I use two rectifiers with 35A/500Apeak and 800V in parallel or is one enough? Will house fuses get blown by charging the 4400uF cap without variac?

I will use the FGA60N65SMD because of fast availability in my country. But I will keep the STGW60H65DRF in my mind for my next DRSSTC.

Do I need thermal conductive paste for mounting the IGBTs? Maybe strange question. If you don't answer it I go for a yes.

Could I also use the IGBT-Brick with 650V bus voltage, or wouldn't that be very subtle? It would be much more wise just going with a full bridge of FGA60N65SMD, wouldn't it? This question is just because of interest I will go with the full bridge anyway.

I will also add little caps parallel to to copper board of the supply as you did! Should these one work out? https://at.rs-online.com/web/p/keramik-vielschichtkondensatoren/7236470/
Would a FKP-1 cap also be suitable for this application as snubber addionally?

I have one big snubber with 1200V and 1uF as shown in the picture. Addionally with 25 630V 10nF this should work out fine as snubber for my application?

21V volt TVS at 24V GDT voltage would not be a wise idea, would it?
24V GDT voltage with 24V standoff TVS voltage would be fine. Max voltage of the TVS is 38.9V is that fine to?

Ok with single parts of FGA60n65SMD at each side of the full bridge, i should handle at least 150A. Is 150A sufficient in order to get some typ of arcs at my setup? How long do you estimate my arcs will be and how critical is tuning of the primary?

But I should also get somehow arcs with the small toroid shouldn't I? Is the tuning then more critical?

Last question, how big should the heat sinks for each individual IGBT be?

I respect your knowledge and willingness to help!
« Last Edit: March 01, 2021, 04:35:14 PM by ChrisBlis »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #16 on: March 01, 2021, 07:16:26 PM »
Since you've chosen TO247 IGBTs, electrical isolation is needed somewhere.  Either use thermal compound and four separate heatsinks or use electrically-insulating thermally-conductive pads under the IGBTs and common heatsink(s).  How much heatsinking depends on duty cycle.  Try something and check IGBT temperature at low duty cycle before increasing.

Your existing MMC is 4S6P.  If you rewire that as 6S4P it will be 100nF at 7.5kV.

2200uF bus capacitance should be fine for this coil.  I don't know your breaker details.

Yes, for 650Vbus 1200V IGBT bricks are appropriate.  Most designs at that voltage are higher power and current, so use larger higher-current bricks.

Those X7R caps will do some good.  NP0/C0G caps are much better.  Those X7R caps drop to 3.2nF at 320V.  NP0/C0G caps will remain at 10nF.  They are more expensive, however.  I used C3216C0G2J103J.  C3216CH2J103K160AA is another option.  For future designs, I bought 33nF 630V NP0 1210-size parts C1210X333JBGACAUTO.

FKP-1 caps are great for the snubber, but not so good for AC of the MMC.

The specified tolerance range of TVS diodes is a nuisance.  I recommend buying some parts with 24V nominal, then measuring to make sure they don't conduct much below 24V.  38.9V clamping may be OK, likely below the oxide punch-through voltage of the IGBTs.  I prefer to clamp at a lower voltage.  If you measure the diodes at 22V, just reduce gate drive voltage to 22V.

FGA60n65SMD should work at 250A peak.  Starting at 150 or 200A is wise, especially without phase lead.  I can't guess arc length, but it should be reasonable.  Arcs will be easy to get with the small toroid.  You may be able to hit your 25cm goal with that toroid.



David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #17 on: March 01, 2021, 09:27:38 PM »
Have I to ground my case for the driver board? Or should I connect it with GND of the driver supply itself?

"Your existing MMC is 4S6P.  If you rewire that as 6S4P it will be 100nF at 7.5kV."
Yes but if I don't want to rewire this one may it also work with a cap of 5000VDC and 110nF (the half of the mmc) or might this be a very bad idea? Would otherwise work 55nf and 10000V out to, or is this way to low in capacity?

I should wisely rewire it right as you said? Yes!

I read, that AC voltage rating at FKP1 caps is approx the half as DC. 7500VDC therefore leads to 3000VAC. Is that enough? 5000V therefore 2500V is probably way to low?

Why are the FKP1 capacitors not well suited for this task? What would be a alternative?

The fuse is a miniature circuit breaker (MCB) with the characteristic B for 16A. Line voltage is 230V. I think it might blow because it still is blown at 1100uF. How could i limit the inrush current without variac later on?

Why do you prefer clamping the gate voltage at a lower value? 19V GDT voltage and 21V TVS?

Now I use a 7 AWG wire. Can I replace the primary coil with a 9 AWG one in order to get more turns? Or even a 13AWG wire, or is this a bad idea?

You said my ECB has high inductance at power supply rails and GDT ouput. Is this proplematic because of stable supply or noise. Supply I guess?
« Last Edit: March 01, 2021, 09:39:34 PM by ChrisBlis »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 993
  • Karma: +44/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #18 on: March 01, 2021, 11:07:01 PM »
I prefer grounding cases and boards - any metal that is either at ground potential or otherwise not connected.  You will find some different opinions.

55nF and yet more primary winding turns makes for a high-impedance primary.  Actually not a bad place to start.  That keeps currents low, so lower change of bridge failure.  If current is too low and performance not good enough, then try higher capacitance and lower inductance later.  13AWG is fine for now.  If you eventually try high duty cycles for long times and the wire starts smelling/smoking, then replace it.

55nF at 100kHz is 29 ohms reactance.  At 250A peak, that is 7.25kV peak.  You may be fine with the lower voltage FKP-1 capacitors there.  I initially used 6kV FKP-1 capacitors for my MMC.  They failed quickly at 4kV peak even at low 1% duty cycle.  Split cases open.  The AC rating for 6kV FKP-1 capacitors is only 700Vrms, so my 2.8kVrms was 4x rating.  (Although at 1% duty cycle, the long-term RMS average is only 280V.)  You might as well run your MMC until it fails before bothering to figure out an alternative.  It may never fail.

People typically use +-24V to drive gates of large IGBT bricks.  Smaller TO247 parts may be OK there too.  I'm hesitant to go much beyond their spec'ed +-20V maximum Vge.  There is little additional current capability gained at 24Vge than at 20Vge.  On the other hand, I don't hear of failures due to excess Vge.  So pick what you like.

Search for NTC inrush current limiters.  Common in line input circuits of larger DC power supplies.

IC (and FET) supply inductance causes voltage spikes at switching times when load current spikes.  Occasionally that can actually help by reducing peak current.  More often it causes problems such as disrupting internal IC logic or over-voltaging the IC.  Many IC specifications include a recommended layout showing placement of bypass capacitor(s).
« Last Edit: March 01, 2021, 11:10:15 PM by davekni »
David Knierim

Offline ChrisBlis

  • High Voltage Enthusiast
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #19 on: March 02, 2021, 08:26:30 AM »
May the higher impedance need longer on times in order to get the higher currents?
How many pulses per second might I achive?

"If current is too low and performance not good enough, then try higher capacitance and lower inductance later."
Should I go beyond 10cm arcs in any case, or how show I know then, that only performance is low or something is wrong with mit circuit?

I already tested the coil with the primary cap at 5kV FKP-1 rating. I got discharches of one millimeter on the cap by discharging it with a screwdriver. Is that a indicator that at least anything worked?

I use 0.1uF Caps on every IC for stability. Should that work, or should I increase them?

Otherwise I will just implement a safety cap in the primary circuit and use the coil as SSTC without primary cap. Is just any 10uF PP 500V cap for this application good? For example this one: https://at.rs-online.com/web/p/polypropylen-folienkondensator/8251398/ Or should i go with a lower cap like 5uF?

You said I need a current transformer of 1:2 if I want to use UD2 as SSTC Driver. What If I remove the 51Ohm Burden Resistor and use stage CT of 1:50? For what is the 150pf cap?

May I also use my driver board without farraday cage?
« Last Edit: March 02, 2021, 02:44:10 PM by ChrisBlis »

High Voltage Forum

Re: DRSSTC Performs only with 1cm Arcs - Troubleshooting
« Reply #19 on: March 02, 2021, 08:26:30 AM »

 


* Recent Topics and Posts

post Re: Voltage readings on 13.8KV
[Voltage Multipliers]
Mads Barnkob
March 07, 2021, 08:57:23 PM
post Re: MOT tesla coil: Why put the midpoint to ground ?
[Spark Gap Tesla Coils (SGTC)]
Mads Barnkob
March 07, 2021, 08:53:07 PM
post Re: Identification of capacitors
[General Chat]
Mr_Project
March 07, 2021, 08:26:27 PM
post MOT tesla coil: Why put the midpoint to ground ?
[Spark Gap Tesla Coils (SGTC)]
paulj
March 07, 2021, 05:49:49 PM
post Voltage readings on 13.8KV
[Voltage Multipliers]
lex1969
March 07, 2021, 10:47:55 AM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
klugesmith
March 07, 2021, 02:48:08 AM
post Re: Question about the meaning of "±20V" specification for Vge on IGBT datasheet
[Solid State Tesla Coils (SSTC)]
davekni
March 07, 2021, 02:19:47 AM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
TMaxElectronics
March 07, 2021, 01:49:55 AM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
klugesmith
March 07, 2021, 01:26:52 AM
post Re: Question about the meaning of "±20V" specification for Vge on IGBT datasheet
[Solid State Tesla Coils (SSTC)]
jhorneljohn02
March 07, 2021, 12:30:36 AM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
fh89
March 06, 2021, 11:30:27 PM
post Re: Simplified mutli-stage coilgun idea
[Induction Launchers, Coil Guns and Rails guns]
johnnyzoo
March 06, 2021, 10:13:19 PM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
March 06, 2021, 09:54:31 PM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Mads Barnkob
March 06, 2021, 09:48:37 PM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
TMaxElectronics
March 06, 2021, 09:44:20 PM
post Re: Identification of capacitors
[General Chat]
Mads Barnkob
March 06, 2021, 09:33:17 PM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Hydron
March 06, 2021, 09:21:44 PM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
March 06, 2021, 09:15:09 PM
post Re: Osciloscope
[Laboratories, Equipment and Tools]
rikkitikkitavi
March 06, 2021, 08:55:52 PM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Hydron
March 06, 2021, 08:51:06 PM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
rikkitikkitavi
March 06, 2021, 08:48:52 PM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
Twospoons
March 06, 2021, 08:44:23 PM
post Re: Question about the meaning of "±20V" specification for Vge on IGBT datasheet
[Solid State Tesla Coils (SSTC)]
davekni
March 06, 2021, 06:26:56 PM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
March 06, 2021, 06:24:15 PM
post Question about the meaning of "±20V" specification for Vge on IGBT datasheet
[Solid State Tesla Coils (SSTC)]
jhorneljohn02
March 06, 2021, 06:00:21 PM
post Re: Welcome new members, come say hello and tell a little about yourself :)
[General Chat]
jhorneljohn02
March 06, 2021, 05:53:40 PM
post Re: Aliexpress Yunbao pulse skipping Product Guide manual
[Dual Resonant Solid State Tesla coils (DRSSTC)]
GrantV
March 06, 2021, 05:49:48 PM
post Re: Search for opinions concerning the use or test of a yunbao / Ebay drsstc skp
[Dual Resonant Solid State Tesla coils (DRSSTC)]
JCF
March 06, 2021, 05:31:42 PM
post Re: Baxandall converter strange output
[Transformer (Ferrite Core)]
dreamth
March 06, 2021, 05:16:24 PM
post Re: Aliexpress Yunbao pulse skipping Product Guide manual
[Dual Resonant Solid State Tesla coils (DRSSTC)]
JCF
March 06, 2021, 04:35:37 PM
post Identification of capacitors
[General Chat]
Mr_Project
March 06, 2021, 03:43:41 PM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
TMaxElectronics
March 06, 2021, 12:33:22 PM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
fh89
March 06, 2021, 08:27:54 AM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
alexhanyuan
March 06, 2021, 06:49:27 AM
post Re: Coil destroyed at 4ms on-time.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
March 06, 2021, 06:33:58 AM
post Re: Rocket engine questions
[General Chat]
plasma
March 06, 2021, 06:20:59 AM
post Re: Rocket engine questions
[General Chat]
klugesmith
March 06, 2021, 06:03:40 AM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
davekni
March 06, 2021, 05:59:49 AM
post Re: Large coil experimenting with long on-times.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
fh89
March 06, 2021, 05:08:48 AM
post Re: Rocket engine questions
[General Chat]
Twospoons
March 06, 2021, 04:50:01 AM
post Re: Rocket engine questions
[General Chat]
klugesmith
March 06, 2021, 04:24:44 AM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
hightemp1
March 06, 2021, 03:41:03 AM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
Twospoons
March 06, 2021, 01:20:13 AM
post Re: Etching PCBs with a SLA 3d printer
[Printed Circuit Board]
TMaxElectronics
March 05, 2021, 11:26:36 PM
post Re: Rocket engine questions
[General Chat]
plasma
March 05, 2021, 10:17:39 PM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
rikkitikkitavi
March 05, 2021, 08:48:19 PM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
davekni
March 05, 2021, 07:41:34 PM
post Re: Baxandall converter strange output
[Transformer (Ferrite Core)]
dreamth
March 05, 2021, 07:22:17 PM
post Re: Baxandall converter strange output
[Transformer (Ferrite Core)]
dreamth
March 05, 2021, 07:18:02 PM
post Re: Baxandall converter strange output
[Transformer (Ferrite Core)]
davekni
March 05, 2021, 07:12:26 PM

Sitemap 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 
SimplePortal 2.3.6 © 2008-2014, SimplePortal