Where is the protection? IGBT? Diodes? Low voltage?

We're not perfect! What can we change?

Where is the protection? IGBT? Diodes? Low voltage?

Postby lancefraterUC » Wed Jul 31, 2013 7:33 pm

From the electrical point of view I am very disappointed in level of protection used in your design and I am therefore not surprised by the number of IGBT, diode and low voltage side failures.

The control and driving of the IGBTs is the most important part, with catastrophic failure if it goes wrong. Firstly there should be protection of the gates and secondly protection of the DC bus. TVS diodes should have been used on the gate to limit the voltage to +/-20V something like a P6KE20A or similar bidirectional TVS. The toriod L and gate C have the resonance potential to exceed these limits. Likewise a large TVS should have been used on the DC bus to prevent over-voltage; with the large electrolytic capacitors the bus could swing wildly. A 1.5PKE550A TVS or similar should be used, able to sink 1500W to ensure the bus voltage does not exceed the IGBT ratings. Also to reduce the effect of the DC cap inductance a thin film cap should be placed in parallel (a few uF).

To further protect the rectifier Diodes a thermistor PTC or NTC resistor should have been used on the AC side. Most likely a PTC to limit the inrush current and limit the current if a fault occurs. The fuse protection alone will not act fast enough to protect the diodes.

As for the low voltage side zener diodes should have been placed on both voltage lines 5v and 15v to ensure these remain in tolerance of the ICs. If the regulators do fail or were inserted incorrectly the ICs, or opto-receiver take the full voltage and become the mode of failure. Also in the feedback circuit the diodes used to clamp the voltage D1 and D2 are not sufficient, there should be parallel Shockley diodes or TVS diodes to ensure the signal never goes above or below the rails.

Adding these components is challenging to the existing PCB.

Lance
University of Canterbury NZ
lancefraterUC
Tipsy Toggle Switch
 
Posts: 3
Joined: Wed Jul 31, 2013 6:54 pm

Re: Where is the protection? IGBT? Diodes? Low voltage?

Postby Bayley » Fri Aug 02, 2013 1:01 pm

Thanks for the suggestions, but we've got good reasons for having the design the way it is.

Through experimental measurements, we've concluded that the gate voltage doesn't ring above 20V during operation, so the gate TVS'es are unnecessary. Furthermore, work in the coiling community has shown that modern IGBT gate oxide layers are capable of withstanding upwards of 30V, which means we have significant margin for error. In fact, many of the larger DRSSTC's in existence run 24V gate voltage, in order to minimize the risk of IGBT desaturation during operation.

Similarly, our design uses a low-inductance laminated bus. With only a few nanohenries of inductance on the bus (the electrolytics we use have a maximum series inductance of 20nH) and zero-current switching, the bus voltage never rings over 600V. Using a dissipative clamping device to compensate for extra inductance is, in my opinion, a suboptimal design if the inductance can be made low enough in the first place.
The fuse does act fast enough to protect the diodes, and often, the IGBT's. This has been experimentally verified as well.

If the regulators were to fail, zener diodes on the rails would not protect the IC's, as the zeners (especially the one on the 5V rail) would dissipate the full input power of the AC adapter and fail short, probably leading to an explosion.
The 1N4148's are sufficient - 74HCT14's are rated for an input from -0.6 to Vcc+0.6V, which is (not coincidentally), a diode drop above and below the rails.

-Bayley
User avatar
Bayley
Lord Protector
 
Posts: 1369
Joined: Sat Jan 12, 2013 6:34 pm

Re: Where is the protection? IGBT? Diodes? Low voltage?

Postby Heidi » Tue Aug 06, 2013 8:59 pm

Its mostly likely true the driven gate does not ring above 20v but coupling and interference from the main coil or the primary could be disastrous. The IGBTs are right next to the gate transformer, electric fields are huge. A few dollars spent on protection can go along way. Protection systems should be active not reactive as replacing out components wondering were things went wrong is expensive


Interference between components on the board is not an issue. There would be many more failures with interference as the culprit if it were a problem. To verify this, we have scoped the traces while the coil is energized (we do not recommend doing this yourself unless you are very experienced, as you may expose yourself to high voltage if you don't know what you're doing). We have waveforms of a variety of logic signals on the board in the manual, taken directly from a digital scope. Those waveforms look identical when the board is in operation.

Fused protection for a couple of cheap ICs is overkill. It is as expensive to put in additional protection as it is to replace the chips, not to mention it adds additional complexity and points of failure for a kit that we tried to keep easy-to-build.

Keep in mind that any active protection circuitry is itself noise-susceptible; the only failsafe protection devices are ones that are based directly on fundamental principles, such as passive low-pass filters and zener diodes. We've confirmed through measurement that no additional filtering or clamping is needed in the design (the gates, for example, are pulled down to -15V during turn-off, so you'd need roughly +20V of noise to inadvertently turn an IGBT on).

Furthermore, in a kit, any protection circuitry is just as likely to be incorrectly assembled as the circuits that are being protected. This is especially true of active overcurrent/overvoltage detection circuits, which typically contain a small handful of parts - plenty of solder joints to mess up!

While the IGBT's are a bit pricey, they don't fail all that often once the kit is working. A little care during assembly, and patience during initial testing, goes a long way to preventing bridge failures.

It is true that your kit may fail occasionally, and this is indeed frustrating. However, with proper precautions and good grounding, the kit should run reliably for many months (we've verified this with our own kits - as far as we can tell, the IGBT's and capacitors do not degrade over that period). For traveling demos, we recommend you use measures such as a counterpoise in order to prevent inconvenient failures in new environments, which may not have sufficiently low-impedance mains grounds.
User avatar
Heidi
Supreme Leader
 
Posts: 130
Joined: Fri Jan 18, 2013 6:30 pm

Re: Where is the protection? IGBT? Diodes? Low voltage?

Postby Lightning Phil » Tue Oct 29, 2013 5:33 pm

I've not received my kit yet, but have read the manual and pored over the circuit. Rather surprised about the lack of protection also! However, if it works, great!

On the other hand, being a well seasoned power electronic engineer (usually designing and building waveform generators that are measured by the tonnes they weigh and the MJs of in the capacitor banks), perhaps I'm cautions. Initial runs will be at reduced input voltage and with scopes all over it.

I'll likely add a mains inlet filter and a low value series resistor to limit fault current.

Can't wait to meddle with the whole circuit - but only where it reveals a need!
Lightning Phil
Tipsy Toggle Switch
 
Posts: 5
Joined: Tue Oct 29, 2013 5:10 pm

Re: Where is the protection? IGBT? Diodes? Low voltage?

Postby stegu » Fri Nov 01, 2013 6:21 am

IMO, the main problem with the circuit is that it reacts explosively to mistuning, which is a fairly common issue because of variations in the manufacture of the secondary coils. A current limiting resistor might help protect the power diodes, but the expensive IGBTs that are cumbersome to replace would need a more active limiting of the peak current. Other DRSSTC circuits I have seen use two current transformers on the primary circuit, one to generate feedback and one to limit the current. My instinct would be to use the same CT for both purposes, but perhaps the analog circuit required to shape the signal for two very different purposes (detect zero crossing vs. measure the peak amplitude) would become too complicated and noise sensitive.
stegu
Magnificent MOSFET
 
Posts: 378
Joined: Tue Jan 15, 2013 4:56 pm
Location: Norrkoping, Sweden

Re: Where is the protection? IGBT? Diodes? Low voltage?

Postby loneoceans » Fri Nov 15, 2013 7:13 pm

stegu wrote:IMO, the main problem with the circuit is that it reacts explosively to mistuning, which is a fairly common issue because of variations in the manufacture of the secondary coils. A current limiting resistor might help protect the power diodes, but the expensive IGBTs that are cumbersome to replace would need a more active limiting of the peak current. Other DRSSTC circuits I have seen use two current transformers on the primary circuit, one to generate feedback and one to limit the current. My instinct would be to use the same CT for both purposes, but perhaps the analog circuit required to shape the signal for two very different purposes (detect zero crossing vs. measure the peak amplitude) would become too complicated and noise sensitive.


That's right stegu. Many other coilers who have built more complex coils have indeed added some sort of over current detection etc. However, I'm sure you do know that we coilers are running the power transistors *very* out of specifications, at much higher currents and frequencies that they are supposed to! They fact that they don't blow up in normal use is already something to be thankful for.

OneTesla obviously is able to add a lot more protection circuitry, but it is a balance between a simple kit and a much more complex build. As you have seen in the forums, many people don't even know how to use an oscilloscope, or know how to choose suitable components even like diodes or resistors. Making the circuit overly complex in my opinion spoils the main charm of oneTesla.

Finally, over current isn't always the main reason for IGBT failure. There could be voltage spikes or logic issues (GDT phased wrongly etc), which will not save the transistors. However, the fact that you are now aware of these issues suggests to me that the goal and purpose of oneTesla has been achieved - to bring a bare-bones DRSSTC kit to beginners, and introducing them to the world of hobby coiling! Time to build a big coil, stegu!
User avatar
loneoceans
Magnificent MOSFET
 
Posts: 103
Joined: Mon Feb 11, 2013 12:39 pm

Re: Where is the protection? IGBT? Diodes? Low voltage?

Postby DJKOR » Sun Nov 17, 2013 10:32 pm

loneoceans wrote:However, the fact that you are now aware of these issues suggests to me that the goal and purpose of oneTesla has been achieved - to bring a bare-bones DRSSTC kit to beginners, and introducing them to the world of hobby coiling! Time to build a big coil, stegu!


Wise words there loneoceans. Even though this is still something that goes well beyond my depth of knowledge (I have never been good with maths), the oneTesla has made me feel more confident to approach building another Tesla coil. I have a better idea of the things to look out for and having a someone simple kit like this gives a good starting point to start troubleshooting from rather than starting to build a more complex TC for the first time and not knowing where to start when troubleshooting.

Now I just have to gather all the right pieces to start building my own. It's a shame a lot of the required parts need to be sourced from overseas. If only we had better local supplies.
Image
My build thread: viewtopic.php?f=15&t=998 | 246kHz Primary / 267kHz Secondary
User avatar
DJKOR
Magnificent MOSFET
 
Posts: 193
Joined: Sat Aug 17, 2013 4:12 am
Location: Brisbane, QLD


Return to Suggestions

Who is online

Users browsing this forum: No registered users and 2 guests


cron