3x 18mF @ 350V -> What to do? / Electromagnetic Projectile Accelerators / Forums

Forums

4hv.org :: Forums :: Electromagnetic Projectile Accelerators

« Previous topic | Next topic »

3x 18mF @ 350V -> What to do?

first previous 4 5 6 7 next

Move Thread

LAN_403

DerAlbi

Thu Sept 16 2010, 09:18AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Hmmh. The only difference is, that the resistor is at a different place. But i do not see the advantages of your version

The diode is illuminated all the time. Then, when the projectile arrives (this is the moment i want to sense) the transisotr cuts the current through the emitter resistance. Then the resistance is discharging the parasitic loads and the signal goes low in my configuration.

The only thing that i can do is: use a smaller Emitter resistance to avoid saturation and discharge the loads faster. Its just that (falling) edge that need to be generated. Your version has the most advantages for the edge that is generated when the projectile is gone.
Did I miss something?

Edit: I have some updates for now. My work on the skeleton is in progress

Also i have constructed the IGBT half bridge. I decided to use 16 IGBT on every side and balanced it with a tree of 3mOhm ConstantanResistors. The diodepacks are prooven to withstand 9kA discharge. Tested with 18mF @ 300V direct discarge though the diode. All IGBTs together have a gate capoacitance of 540nF.

Lets hope everything works.
IGBTs are IXGK-GX320N60A3

Some news here: I have tested my IGBTs with pulse currents to get an idea about switching losses. With 380Hz i fire 13Âµs pulse on the gates that discarged one 18mF cap @30V. Calculated pulscurrent is about 1kA. All IGBTs seem to share the same current, because they become warm all equally. That means that my Resistor network tree works.. Additionally i have connected one of my thyristors in the short circuit chain. The pulse becomes a little shorter. That means firing the gate through my little transformers is faster than ~7Âµs , So far everything is working.
The comming weeks i will get the PCB for my FiringCircuit. I hope everything will coninue to work so fine

This eventing i was about simulating the currentflow.. i improved the model and added some inductance for the wires.
]asymhalfbridge.zip[/file]
I hope the time for the first shot comes fast..

GhostNull

Fri Oct 08 2010, 11:43AM

Registered Member #2648 Joined: Sun Jan 24 2010, 12:45PM
Location: Australia
Posts: 291

WOW, very nice. Where did you get all those semiconductors?

I hope to see the results soon!

btw, your allowed to make a new post after 48 hour if it has new info. I didn't realise what progress you had made until I stumbled upon it just now.

Great Work, Really Good Job!

DerAlbi

Fri Oct 08 2010, 08:27PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Thanks for your attention

I didnt feel that it was necessary to add a new post.. it would be seen anyway if i would have real news. Currently i am still waiting for my firingPCB. grrr

Getting IGBTs: yeeahh this was hard

I have connections to a local company here. First i looked around who distributes the IXYS-parts.. after a bunch of emails someone answered and linked me to a "local" (german) electronic supplier for train and general highpower E-motors. Parts were only sold in a lot of 25 pieces. 25 were not enought, 50 was to much. So i took 50...
The bridge you see consistos of 32 IGBTs.
Dont talk about money.

EDIT:
Some news (sorry not importenad enough for a double post) I have finished the small PCBs now. It showes that the delay is between 600-700ns. It extremly depends on the light at the diode..
If less then its faster. Exactly like the theory

DerAlbi

Thu Sept 08 2011, 02:17AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

If someone is still interested: i am back

I'll post better information if i have some. Currently i am installing my IGBT module and try to optimize gate-drivers.
I also decided for the barrel: its a tube of 10mm inner diameter. ..bought some 10mm screws.. sliding fine within the tube.

Photos on request

Edit: I am happy to report a 0..15V risetime of 580ns and a 15..0V falltime of 440ns for my IGBT module. ((Dis)Charging the gate capacity 340nF means ~11Amps average gate current

) I used no gate resistor. However i used 50Ohm-Coax-cable to connect to the gates. I have zero over/undershot. Somehow the system is too well damped. i just dont know where. The coax-cable has a resistance of <1Ohm. ~500ns risetime means 2Mhz. I didnt think the 50 ohm impedance would mean anything at this frequency. i just wanted a cable that behaves more capacitive insead of inductive.

Edit2: I am happy to tell that my half-bridge seems to work. I just tried a burst of 20x 10us pulses with 2ms pause between these bursts (so effectively 8.3% duty cycle). I discharge the energy into a small iron wire. It immediately lights up. If Capacitor-voltage is above 100V... at 40V it need a second...

However at approx. 180V my diodes burn out. BURN OUT is the right word. The weakest diode (by design) melts and becomes perfectly a 0Ohm resistor... resulting in an immediately dischartge of the capacitors thorugh the silicon. STINKs. I also witnesst a "something" exploding between my IGBTs. I cannot tell you what it was butr the BOOM was enormeous. I gesst it was a litte hair of wire that causes a spark at 200V... a lot of soldering wire was vaporized

DerAlbi

Fri Sept 09 2011, 01:47AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

I want to discuss my Edit2 of my last post. First some pictures....
So far i have programmed some test routines for my main control interface.

It is now capable of sending controlled signals via a special "Testmode" that is designed into the PCB/CPLD to the half-bridge control:

This control cannot fire SCRs now. But the highside and lowside drivers are working just fine.
So i finshed the IGBT-half-bridge so it can be connected electrically with screws.

1315530057 2906 FT90617 Halfbridgewithscrewterminals

This module fits perfectly it the housing...

After everything works fine a test load must be "designed". Later the load will be extremely inductive with a small resistance. I figured that something that explodes in case of failure would be a good idea. => If the load exlodes the IGBTs doesnt.

Resistance of the small iron-cable is 0.3Ohm and inductance of these cables.. i dont know. Not much.

Of course i tried it. Started with 40V. Iron glows at 8% duty cycle as long as the CapLoader is delivering power. If the CapLoader is off the 40V of these 54mF cause not much damage. However causing damage was the goal

So the voltage was increased. After 130V was melting the iron immediately.. vaporizing it was the goal

200V indeed did vaporize something but it wasnt the iron..

1315530057 2906 FT90617 Firsttimeusuallyhurts

What you see is mainly soldering wire. There seemed to be a spark that jumped beween little hairs of the strong copperwires resulting in a low resistance short circuit. BOOM. Adrenaline^3. Next time i clean it better

Funny: the boom came clearly out of the IGBTs... had my eyes verry close to it.. uuh. But they all still work. However the bridge malfunctioned somehow. I searched the failure.
A diode was molten.

After repairing the diode (verry easy because the diolde module is made of 18 diodes.. one of them is designed to be weak so only one melts and protects the others) I tried again 210V. Nothing did explode.The spark wasnt there this time.

Understandable.. it would now need to jump over a much greater distance.:-D
However a suprisingly great amount of black smoke came out of the same diode again...
Here a diagramm. Forgive me.. Paint is not made for painting..

Can anyone tell me why the hell this damn diode likes to smoke? It well known thats not healthy.

0.3Ohm at 200V are 600Amps. The bridge is made of 16 IGBTs on each side. One IGBT can handle this current alone!! The diode module is prooven to withstand a 9kA discharge. WHAT can cause this?

EDIT
FOUND IT.
Explosion wasnt copper or soldering wire.
IGBT:

Verry interesting. Done the same experiments more than once... no other IGBT failed like this.

DerAlbi

Mon Sept 12 2011, 07:06PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Hi guys. Having some serious problems now!

MY BRIDGE DIED AGAIN

I tried a discharge at 320V (54mF) for 500us into a coil of 9.9uH at 0.156Ohms. Simulating this gives me a peak of 1.89kA reached after 175us and then the resistance discharges the capacitor resulting after 500us at 1.85kA.
1.89kA/16 IGBTs = 120A per IGBT. Normally this IGBTs can handle 700A for 1ms.

So obviously the current is not shared equally within te paralleling circuit. . Here is a schematic:

You can see the Emitters are connected together. Gates are connectes in groups of four. Each group is connected to its neighbor through 10Ohm. All collectors are connectes throug the 3mOhm Resistor tree. However this resitor tree could have some tolerances.

HOW TO SOLVE THIS PROBLEM?

Edit: additional info:
1) The bridge died before. The same IGBTs were dead. Last time i thought it was a voltage surge. So i put several 470V varistors between emitter and collector and every block has a 24V supressor diode at the gate connected to the emitter..
2) Opening the dead IGBT shows a molten chip inside. 2 of 8 bonding positions are quite black. Interestingly it are the same that are molten at the IGBTs that died before. They are the closest bondingwires to the gate-bondingwire.
Tomorrrow maybe i can give some pictures where you actually can see this.

Forty

Mon Sept 12 2011, 11:12PM

Registered Member #3888 Joined: Sun May 15 2011, 09:50PM
Location: Erie, PA
Posts: 649

I think you may need to invest in larger igbt's so that you don't need so many in parallel. the slightest difference between each branch (igbt itself, resistor, wire length, temperature, etc) could cause one of them to fail.

DerAlbi

Mon Sept 12 2011, 11:30PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Isnt that where the resistors are for?
I check out how preceice they are. maybe there are some bad connections.

Is this resistor tree even a good thing? If every IGBT would have 16mOhm in series.. would it be better? Any experience with that?

What really bothers me is: why only the 2 bonding connections near te gate melt?

I am considdering this currently:
Turning the IGBT on cant hurt it. Due to inductive load the current is verry low at the beginning.
Turning the IGBT off does hurt it:
Absolute worst case: Gate is confirmed off after 400ns. Datasheet dictates a turnoff delay of 1.5us.
ok. given this: there are 150Amps at the IGBT. There are 320V at the capacitor. It takes 1.5us.
1.5us*150A*320V = 75mJ of Heat. Its nothing!

Why the IGBT fails?

Is it overvoltage? As i have said, I protected it with varistors between emitter and colletor and supressor diodes betwen emitter and gate.
Overvoltage would explain the failure near the gate. Let me see if i can measure the inductance of a string of the resistor tree. The varistor does not cover this.

Edit:
Ok. From the connection point of all Collectors to the connection point of a single collector the resistor tree has an inductance of 55nH.
55nH/2*150A^2 is 0.85mJ however to break the insulation beween gate and collector..? hmmh... Is that enough? Spice says its enough to reach 1kv or so...

DerAlbi

Tue Sept 13 2011, 11:20PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Any idea what caused the 2 left IGBTs to blow? near the gate there are some black points... overvoltage?

The right one dies not need to be discussed.

Inducktion

Tue Sept 27 2011, 02:46PM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

Looks to me like you might want to try one very very large IGBT instead of several smaller ones in parallel. As many have said before me, even the slightest difference in properties between them will cause one to conduct more (since IGBT's are usually negative temperature coefficient), and thus, one will "runaway" with the current and blow up. Then another, and another, and so on.

You should be able to find a large brick or mosfet, and use the ones in side of them instead for your purpose. A half bridge module would work best; Just connect them in paralllel, like how you have them now. Because they're in the same package, you know they're generally made from the same batch, and have very similar, if not the same properties as the other. (doesn't hurt to add a balancing resistor though)

first previous 4 5 6 7 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob