0 Members and 1 Guest are viewing this topic.
For more 35kV Oil-Immersed Power Transformerinformation, please contact us. We will provide professional answers.
MOSFET vs THYRISTOR for high peak current handling?
«
on: October 08, , 07:08:55 pm »
I'm investigating a circuit to "clamp" a DC bus to close to zero volts under certain specific fault conditions. That DC bus is fed from an relatively high speed power supply with fold back current limiting (approx 1ms response time), but includes a "significant" amount of capacitance, and when clamped, may also trigger an inrush of energy from another source. However, all the energy sources, despite being very low impedence, will either turn off, or have a limited about of total energy (current estimate is around 10KJ to 15 KJ maximum).
If the system is triggered, some form of very low resistance semiconductor switch will clamp the DC+ and DC- together, holding those lines at close to zero volts. Initally, there will be a rather large current inrush, broadly limited by the inductance of the bus, rather than any dc resistance. After a few ms, much of the energy will have been disipated, but the peak power during that time will be very large (kW)
With this in mind, are there any significant advantages between MOSFET and THYRISTOR type switches for their ability to withstand very high peak pulse powers? The system will need to withstand a continuous DC voltage of up to V, and should be "triggerable" in less than 500us, ideally, less than 100us.
The system clamping is non-repetative, and i will accept failure/destruction of the switch element in the worst case, providing it has successfully clamped the bus to zero volts (and hence prevented damage to the device it is protecting)
Any suggestions?
TheMG
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #1 on: October 08, , 07:26:17 pm »
What sort of peak current are we talking about here?
SCRs typically have a much high peak current handling than a comparable MOSFET. MOSFETs rated for V are usually not made in very high current handling, IGBTs tend to be used rather than MOSFETs when high voltages and high currents are involved, but just like MOSFETs the peak vs continuous current handling is not as great as that of an SCR, many of which have a peak current rating 10x the continuous/repetitive rating.
So I would start looking at SCRs and forget about MOSFET/IGBT for crowbar purposes.
Another option is a thyratron. These are still used in applications such as high power UHF broadcast transmitters to crowbar the power supply very quickly in the event of a fault, where even the most robust SCR wouldn't be able to handle the very high voltage and very high capacitance.
« Last Edit: October 08, , 07:30:02 pm by TheMG »
mawyatt
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #2 on: October 08, , 07:50:39 pm »
Long long ago we used SCRs as "crowbars" for the Power supplies in equipment designed to survive a nuclear blast. They were used in X, Y & Z axis to detect a nuclear event and latch the PS to near zero volts, or some large surface area transistor was used in X, Y and Z to trigger a single SCR to latch the supply. What happens in a nuclear event is the PN junctions become forward biased, this include transistors, diodes and ICs. The idea was to quickly clamp the supply before the stored charge on the supply output capacitors and lines had time to damage the sensitive electronics. There were certain guidelines for the amount of capacitance on the supplies and what certain ICs and transistors could handle, and the times involved.
From past experience I think an SCR is probably more useful as a supply clamp since they are designed for very high surge currents.
Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #3 on: October 08, , 08:36:51 pm »
Peak currents? Good question. As it stands i have a maximum voltage of vdc, and the power supply is output limited at around 800A, but the capacitance on the DC link will be the source of most of the pulse current. The impedance of the DC link between that capacitance and the clamp is likely to be somewhat significant,but i will probably include an additional inductor (air cored) to provide some pulse shaping and di/dt limitation, and that can also be used to introduce a bit of resistance to help both calm down the pulse and to absorb a bit of energy too.
Looks like thyristors lead the way for peak capability, but they also bring the (small) limitation of not being able to turn off the clamp unless the current has fallen to effectively zero once triggered (this shouldn't be a problem, be cause the electronics triggering the thyristor will also command the dc psu to turn off too.
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #4 on: October 08, , 08:44:18 pm »
One possible option to limit the total I2T for the switch might be to parallel the silicon switch with a mechanical contactor, triggered at the same time, but that will take something like 10 to 25ms to close, but once closed will deal with a longer term current flow, leaving the silicon switch to just deal with the short term peak pulse.
IME with most contactors, you can significantly overdrive the coil to reduce pull in time, using a peak and hold current type driver, so i could probably get actuations down into the 5 to 10 ms region i suspect.
mag_therm
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #5 on: October 08, , 10:11:35 pm »
It sounds like you should model it on a simulator.
As a first approximation of the analytical solutions for a loop discharging the capacitor through the crowbar semiconductor:
For the first half cycle:
I_peak = Ecap * sqrt ( C /L) where L is the total loop inductance, including stray and a series current limiting inductor you mention.
pulse_width = pi * sqrt (L*C)
That will enable an approximation of the I^2 * t rating of the semiconductor and feasibility.
To commutate the scr, an antiparallel diode can by used to conduct the negative half cycle and then an orderly off, with the energy back in C.
But use of that depends on your requirements.
Don't worry - That is , until time for the first test!
I am quite familiar with these crowbars, in ratings up to 10's of kAmp
Circlotron
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #6 on: October 08, , 10:47:23 pm »
Any inductance in series with your crowbar SCR, once the buss voltage reaches zero with the current still close to max, will try to keep pushing the voltage down below zero. Make sure there is some intended path for this current.
KT88
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #7 on: October 08, , 11:01:27 pm »
The source impedance is critical in this kind of application. If the impedance is close to zero the current will be close to infinity...
In lightning protection as an example the source impedance has the caracteristic of a constant current source with - in buildings - usually up to 50kA. If you clamp it the current stays basically the same. This also means the power dissipation of the clamping device is fairly low with a duration of something in the ballpark of a few ms the absorbed energy stays quite low.
A capacitor on the opposite side is mostly a voltage source. If you try to absorb the energy in silicon devices you need a lot of volume...
One way to get around this problem is to use carbon based pulse resistors to limit the current and to absorb the energy.
On example could be found here: https://www.ebg-resistors.com/fileadmin/user_upload/EBG/Products/Carbon-Disc/Produktdownloads/EBG_Series_CSR_rev.1.pdf
They would require some paralleling but there are larger ones avalable. The semiconductors could stay small that way. Even a PCB based approach might be possible. Parallel resistors and swithes would allow reliable paralleling as you could balance the load pretty easyly. It would be very scalable as well.
Cheers
Andreas
T3sl4co1l
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #8 on: October 09, , 12:00:10 am »
By setting the supply impedance (both resistance and inductance), you can control the peak currents. You'll need to control dI/dt to ensure safe operation of the thyristor (or maybe not, since single-use is acceptable..). And yeah, thyristor is the way to go here, or perhaps an ignitron (basically, a somewhat controllable spark gap).
You don't need much resistance to have some useful effect. Compared to e.g. rectified 400V 3-phase mains having a 100kA fault capacity, a mere 100uohm limits your peak current to 10% of that, while dissipating only 100W (a tiny efficiency hit out of the total 100s kW we're talking here). Or even more advantage since we're talking bypass caps rather than mains.
So, 15kJ at V, is 30mF, that seems like quite a lot of filtering for some sort of DC-DC converter, or regulator or whatever kind of power supply it is. It is about right for a single phase rectified mains supply. But wouldn't this be three phase?
Well, 30mF at 1kV, discharging at 10kA peak, is 0.1 ohm ESR, or (0.1)^2 (30mF) = 300uH ESL. Quite a lot either way... probably you'd be happy with a compromise between higher peak current (say 40kA, which a fist-sized hockey puck SCR can handle), and less inductance (saving on copper -- or you might well use aluminum at this scale), say 20uH.
You may also consider a switching current limit. This is feasible with SiC MOSFETs or IGBTs (I mean, it's feasible with Si, but even better with SiC), and can provide soft startup, fault current limiting, and on/off switching. I would guess something microwave-oven sized would be adequate here. If you have a surge voltage rating, you'll need much more switching capacity; still, it's quite tractable. (You'd probably be dumping the excess energy into load resistors, or a MOV stack; or into ground if possible, i.e., as a current limiting buck converter.)
Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life? Send me a message!
NiHaoMike
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #9 on: October 09, , 03:03:58 am »
Have you considered a solid state switch in series with the load to quickly switch off the current?
Cryptocurrency has taught me to love math and at the same time be baffled by it.
Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
perieanuo
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #10 on: October 09, , 07:01:09 am »
well, maybe we can re-invent the wheel
just use thyristors, the guys before you weren't stupid. we can discuss this 2 hours, you can get cheaper with 20 cts and less robust by 50% using mosfets vs thyr/triacs
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #11 on: October 09, , 11:01:03 am »
Broadly the energy that needs to be disipated comes from 3 sources
1) DC Link smoothing capacitance: There is "some" capacitance of very low ESR on the DC link. Some in the PSU, some in the loads attached to the DC link, and potentially some additional capacitance added to help dynamically stiffen the dc link under certain conditions. How much capacitance is difficult to answer because the set up of the system varies. Certainly, it's easily possible to have 50mF or more spread across the DC link, but the further away that capacitance, the more the DClinks impedance will help limit the pulse profile
2) the DC power supply that drives the DC link. This will get "turned off" when the crowbar is triggered, but latencies of up to 20 ms are possible before it does so, and worst case is that this supply sits on its maximum output current limit (but at low voltage, therefore low power / total energy) for that period. This is up to 800A
3) Spinning mechanical inertia, converted to electrical energy. The DC link drives a reasonably complex electric motor test rig, and that includes relatively high speed spinning objects, which therefore have kinetic energy stored in them as they spin. If that KE "gets out" to the DClink when it is clamped to zero volts depends on the precise action of the motor inverters. If those inverters drop into active short circuit, then that energy is disipated within the eMachine and inverter itself, and does not appear on the dc link. However, if the inverters do not drop into ASC (and they are prototype units and so they may well not do this and cannot be guaranteed to do this!), then the mech KE will flow out of the inverters and into the DClink, and therefore have to be absorbed by the clamping device
I think that
1) -> 2) -> 3) are in order of duration and speed
1) is a short term, ultra high current effect (several thousand amps)
2) is a medium term, medium high current effect (up to amps)
3) is a long term, lower current effect (probably around 500A)
They will overlap and add up to a complex pulse profile in terms of energy disipation required!
One idea was to laser cut some stainless steel sheets into some deliberately high inductance "power resistors" and use these in series with the main switching element to limit di/dt. There resistance and inductance is going to be a compromise between clamping the bus as quickly as possibly to a low value, but also limiting the peak current and rate of rise
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #12 on: October 09, , 11:03:58 am »
Any inductance in series with your crowbar SCR, once the buss voltage reaches zero with the current still close to max, will try to keep pushing the voltage down below zero. Make sure there is some intended path for this current.
Is this an issue? The trigger circuit for the SCR will have to be isolated (because we are dealing with up to vdc) and the device will remain "On" if the current flowing is above the devices hold on limit, so whilst the voltage might locally go "below zero" there is only one current path in the system?
Zero999
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #13 on: October 09, , 12:26:58 pm »
Definitely go for a thyristor, over a MOSFET at 1kV. It's much more robust and only needs a brief pulse to trigger, which means a small transformer can be used to drive the gate. If the driving circuit can't handle the gate current, the sensitivity of the gate can be increased by adding a BJT, or smaller SCR. If you're using a gate drive transforomer, select one with a low turns ratio, i.e. 3:1, that way the secondary current will be several times the primary. Note: the maximum negative gate voltage rating of most SCRs is only around 5V, so add a diode to protect against reverse voltage.
KT88
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #14 on: October 09, , 10:51:12 pm »
Thyristors are the best approach in my opinion. Soft switching of IGBTs or MOSFETs would cook the die while switching that than has not enough capacity anymore to absorb more energy in the conduction phase.
The main task is to identify existing ESR that could be utilized and to add additional resistance to absorb the rest of the stored energy in the desired time.
Combining inductance and resistance helps to smooth the discharge curve to reduce cost of the semiconductors.
A sequence of high- to low value resistors switched one after the other could give an almost constant an very defined discharge current. This could be controlled with comparators or open loop via fixed timing.
Cheers
Andreas
akimpowerscr
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #15 on: October 10, , 02:12:32 pm »
In all crowbar applications, there are 3 characteristics of the thyristor which are critical and which must be respected otherwise the semiconductor will be destroyed.
1) Itsm: max peak current (max 10 mS) The resistance in series must be dimensioned not to exceed 70% of this value.
2) I²t: to be checked
3) dI / dt: Value of the inductance in series with the Thyristor. Attention, the dI / dt is low, for example, for the SKT, only 125A / µs
I recommend ignoring the ESR of the condensator or the inductance of the conductors.
The set C + R + L which will be short-circuited by the thyristor constitutes a series oscillating circuit.
It is therefore necessary to provide a fast diode mounted in anti-parallel on the thyristor to avoid any reverse voltage and to allow the damped oscillation until total dissipation in R and in the thyristor of the energy of the capacitor.
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #16 on: October 10, , 02:47:23 pm »
Ok, how do i rate the anti-parallel diode? It's going to carry less total energy than the thyristor, but the peak current could be pretty high?
The idea of using a secondary parallel mechanical contactor to crowbar the crowbar after as short a period as possible is also looking like perhaps a good idea?
KT88
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #17 on: October 10, , 03:08:04 pm »
Could you provide a simulation with at least some crude models? That would help to get a better understanding of the challenge. I did something similar a while back at lower voltage where I shorted a one Farad/16V capacitor. The ESR did most of the current limiting job as well as absorbing the energy. For the low voltage MOSFETs were the best solution.
Knowing the impedances involved it becomes much easier to find the appropriate settings.
LTSpice would be preferred as it is the most commonly used tool.
Cheers
Andreas
akimpowerscr
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #18 on: October 10, , 03:48:40 pm »
Ok, how do i rate the anti-parallel diode? It's going to carry less total energy than the thyristor, but the peak current could be pretty high?
The idea of using a secondary parallel mechanical contactor to crowbar the crowbar after as short a period as possible is also looking like perhaps a good idea?
For an oscillating circuit without losses, the peak current in the anti-parallel diode reaches the same value as that of the thyristor.
But here we are in an oscillating circuit damped by the fact that there are several resistances in the circuit: the resistance R, the ESR of the capacitor, the internal resistance of the thyristor and the resistance of the inductor.
The peak current in the diode will therefore be much less than that of the thyristor.
Of how much ? it will depend on the damping of the circuit.
The ideal would be to operate at critical damping, but there are many parameters to respect, including the maximum time after which the voltage stabilizes at 0V
For tests, one could choose a peak current of the diode 60% of that of the thyristor.
For the peak current of the thyristor, the resistance R is dimensioned such that the maximum current is approximately 70% of the Itsm of the thyristor.
As the power contactors are slow, I don't see how they could be useful in this application.
In a project where very high voltages and currents are involved, one should not rely on elements that can vary like the ESR of a capacitor ... They can be replaced by others that have a lower ESR.
Attention, a crowbar circuit is a "killer" of capacitors ..... Check if the capacitors will withstand this very high discharge current.
sandalcandal
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #19 on: October 10, , 04:00:06 pm »
The idea of using a secondary parallel mechanical contactor to crowbar the crowbar after as short a period as possible is also looking like perhaps a good idea?
I think you'd again need some modelling to check load sharing when the mechanical contactor closes. Overdriving is likely to cause more bounce than normal and could be counter productive. Bounce is going to affect total actuation time as well as the likelihood of contact welding which you probably don't want. A search for "crowbar contactor" does reveal some crowbar contactors intended for use in MW level wind turbines so something for your usage requirements is probably out there if you can fork over the money (but so is anything I guess).
Disclosure: Involved in electric vehicle and energy storage system technologies
mag_therm
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #20 on: October 10, , 06:23:42 pm »
Hi Max,
I agree, don't speculate for days about it, run simple calculations that I mentioned in # 6 above and then simulations of the DE in a circuit simulator.
To get you started:
Capacitor energy = Joule
Capacitor bus voltage = V DC
From those, I calculate 30,000 uF
You specify triggerable in less than 500 usec
Let us assume you mean the bus is taken to 0 Volt in 500 usec, to give some basis for calc.
Neglecting damping for the moment,
pulse_width = pi * sqrt ( C/L)
From which L = 0.844 uHenry
This inductance is necessary , because without it the di/dt at gating would be infinite
And then
I_peak = Ecap * sqrt(C/L)
giving I_peak = Amp
Add some resistive damping. Just to get a ballpark,
R = dynamic_impedance : Is that critical damping?.. I forget.
R = sqrt (L/C)
R = microOhm
Now a quick plug into Qucs transient solver: ( Hope the .png come up)
Note that the damping resistor has reduced the peak current as we expect. Peak damped current = Amp
That will need maybe a parallel set of big distributed gate scr.
Also may need specially made capacitor set to safely handle these pulses over their lifetime.
The di/dt inductor looks feasible, needs to be stiff enough mechanically.
The current crosses zero at a longer time than the undamped case, as we expect. Zero cross time = 580 microsec
The voltage trace is the capacitor voltage which rings to minus 180 V (The voltage on the other side of the inductor is clamped to zero volt)
The diode will commutate off at about microsec, at which time the capacitor is nearly discharged.
Please check the above calcs, which are just done for fun as interest on the forum.
Are you intending to build this from scratch?. Looks like a major R & D project?
« Last Edit: October 10, , 06:31:58 pm by mag_therm »
The following users thanked this post: akimpowerscr
akimpowerscr
Re: MOSFET vs THYRISTOR for high peak current handling?
«
Reply #21 on: October 11, , 11:26:22 am »
One more small construction detail: the inductor can be made of aluminum so that the resistance R is included in its own resistance.
All Comments ( 0 )