How can Trigger LARGE SCR (100% rookie please help me!)

[dantherookie]

Hi there, my names Dan. I'm currently working on a little project

My project is a 3kj~ Capacitor Bank (yes laugh, I'm rookie, doing rookie things )

I will be using the capacitor bank to discharge into coils and into nearly short circuits (maybe a rail gun huehue)

I have 6 Capacitors Rated at 4700uf 450v arranged in parallel, Im using a SCR (P/N: C451E)

I am sadly that rookie I do not know of a safe way to fire this SCR, the SCR max gate voltage is 5V 200ma (probably irrelevant but you might ask), I already know basically how to fire the SCR but in the past I have fried 2 SCR's , which i think they died to either not being fully switched on OR overvolting due to back emf (Im not sure tell me what I'm doing wrooonnggg )


FYI: I know this capacitor bank is outrageously deadly and that if you brush up against it you dead, kaput fullstop. I have a full understanding that this is NOT a toy and cannot be played around with. And yes anyone with half a brain would stop playing with this stuff but I really cannot help myself, I just must) my previous capacitor bank was rated at 2.5kv @ 400uf, that capacitor bank scared me too much, I could only get the courage to charge the bank up when I had a watcher near by and I only managed to charge it less then a dozen times because it made me that scared) My new capacitor bank is much easier to manage

The Black Capacitors are the 450v Caps I am using now, the grey Bosch Capacitors are the ones I don't go near anymore (too scarey) they are oil filled capacitors so they spend their life shorted out so they don't unexpectedly kill someone, They will one day power a gas laser i want to build (looking for parts now)


Excuse my bad English and thank you for reading 

[ConKbot]

The faster you can turn on the SCR, the less localized heating it will experience.  For a puck like that, it wouldn't be uncommon to use a cap and smaller SCR to create a nice sharp pulse to drive into the gate to turn on the main unit quickly (Through a GDT if needed).  Enough reverse voltage to fry one of those would be impressive, but isnt the most likely cause.  Now, on the other hand, your cap bank wont appreciate being reverse charged, so a catch diode can help spare them, though the current pulse will be lengthened, so suckback could be a problem in a coil gun.  Tweak your bank/coil/charge voltage appropriately.

You do have an appropriate clamping setup, correct? Without enough pressure on the puck SCR, that could be a source of failure too. They like approximately one metric shit ton of clamping force, so get the datasheet, and figure out how to clamp it right.

[dantherookie]

Thank you ConKbot, I was unaware that they needed a metric shit ton of pressure clamping on the SCR ( I shall get out the datasheet!) I did confer with my technical friend about using another small thyristor or even Capacitor to trigger the SCR but he was concerned with applying to much current to the gate, I had a brain fart of using a Zener Diode and Capacitor to trigger SCR but I think my limited technical knowledge in this field may be straying me down the wrong path.

I have some a large quantity of 10A10 diodes, would they be sufficient to snuff back emf. Also I have these diodes on their way, sorry about the ebay link. Sadly i cannot find datasheets on those diodes but i have my fingers crossed that they will be able to handle the voltage/current

http://www.ebay.com.au/itm/141679212421?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

because I'm so basic in this field I was wondering if someone could draw me up a SCR firing diagram, I would immensely appreciate it

[dantherookie]

I'm not sure if you mean GDT as Gas Discharge Tube, but I'm just going to assume (rookie mistake) I have some GDT's I have found, they are 470V. Could I also use these as over volt protection? and do GDT have a short life? or can they be used over and over (I'm concerned about the insides melting together)

[alsetalokin4017]

I used to work for a laboratory that did lots of big capacitor bank discharges into water-arc chambers (typically  between 0.3 and 3 uF at 15 -20 kV.)  We blew quite a few of those SCR pucks, even when rigidly clamped in fixtures as the manufacturer requires. We finally settled on a TAG (triggered air gap) firing system and did away with the SCRs as being too expensive and unreliable. The TAG was made from two aluminum half-spheres with an automotive spark plug embedded in one of them, so that firing the spark plug with a pulse from an ignition coil would cause the main gap to fire and dump the cap charge into the load. The mechanical support and gap adjustment system can be imagined. We took great pains to minimize inductance in the discharge path by using heavy busbars, thick ribbons of copper, a ground plane, and having the cap bank, TAG and load all in a very tight, minimum distance loop.
The system was pretty scary to operate. We had multiple safety systems, HV dump relays controlled by a deadman footswitch, etc. The bank was kept shorted by a NC HV relay that was held open by the footswitch. The operator stood at the charging power supply (a monster that would provide 150 mA at 15 kV), stepped on the switch, slowly brought up the charging voltage until the desired value was reached, then pressed the Trigger Button, and when the shot fired the footswitch was released to short the bank again. Hearing protectors, polycarbonate blast shields, and the presence of safety personnel were all required.

[dantherookie]

400v triggered air gap would be hard to make and even harder to find. the arc distance of 400v is miniscule  I have to use SCR 


sooo according to my datasheet, my SCR requires 5500-6000 lbs of mounting force............ I don't know how I'm going to be able to measure the clamping force  NOOOOOOOOOOOO another problem 

[alsetalokin4017]

400v triggered air gap would be hard to make and even harder to find. the arc distance of 400v is miniscule  I have to use SCR 


sooo according to my datasheet, my SCR requires 5500-6000 lbs of mounting force............ I don't know how I'm going to be able to measure the clamping force  NOOOOOOOOOOOO another problem 

Heh... that's about 3 metric shit-tons then. 

We used big heavy endplates and 3/4 inch threaded rods with nuts and washers to pull the endplates down to squeeze the pucks. Yes, it sounds to me like the mechanical setup is your main problem not electrical. To fire the SCR is almost trivial, all you need is a pulse, you could do it with a 555 one-shot or even a pushbutton to 5volts. What is not trivial is getting the SCR to survive the huge currents and ringing that will happen if you have any appreciable inductance in the discharge path. You might be able to snub it with diodes but think "low inductance" for everything in the discharge circuit. Use flat copper ribbons instead of wires, for example, and keep the circuit loop as small as possible. I don't know if your GDTs will stand up to the strain, there is a _lot_ of energy involved in your system and you'll be dealing with very high instantaneous current values.

You might be surprised at what you can do with a TAG, though, even at 400 V. Once you have ionized the gap with the trigger spark, there is very little resistance there and most of your cap bank energy will find its way through.

[ConKbot]

GDT=Gate Drive Transformer, sorry about that. You should be able to find homemade SCR puck clamps online, this isnt the first time that cat's been skinned, so you should be able to find inspiration somewhere.

[dantherookie]

I can make a clamp pretty easy in the workshop, with some 3/4 alloy plate, its just 2.4ton of clamping pressure seems like alot

2.4ton makes me kinda nervous, its quite some force

I wonder if I can apply like maybe 1.2-1.6ton and be in the safe zone since this SCR is major overkill

[ConKbot]

Belleville washers can be used to help control the force applied if you're nervous about using just bolt clamping formulas.  As as far as turning it on, alsetalokin is correct, a 555 or switch closure will work, and the SCR I mentioned earlier (wasnt suggesting to exceed the gate voltage, just as a way of making a guaranteed sharp pulse. Definitely overkill though ) The important thing is the sharp pulse with a healthy drive behind it.  A 555 has a surprising amount of grunt to the outputs, and obviously a switch closing to a power supply will too.   If you have some older series logic with a wimpy sub mA output, or something else with low current and high resistance output, the weak turn on drive wont put the SCR into saturation as fast as if you gave it a healthy push.  Unfortunately I cant find the reference again, that went into more detail why this was a bad thing and the actual physical things happening inside the SCR.

[T3sl4co1l]

Threads are just coiled up ramps.  An M12 bolt has a pitch of 1.75mm, so one turn advances 1.75mm.  If 100 N is applied to a 1m arm, it delivers 100 Nm of torque, and 2*pi*100 Nm of work for one turn.  The thread moves 1.75mm in that turn, so the force is 2*pi*(100 Nm) / (0.00175 m) = 359kN, or an equivalent weight of 36 tonnes.  100Nm is easily applied to a modest size wrench by hand, so you can see the clamping force is quite easily achieved.  (Hrm, the units seem to work out, but that doesn't seem hardly right; the bolt should snap at that tension, yet the torque seems way too low.  Whatever, look it up!)

That assumes the thread is frictionless, which is a terrible lie; only ballscrews (very specialized precision nuts) can achieve that.  In reality, the clamping force will be less, by a wide range of uncertainty.  Still, with some grease and empirical formulas, getting close enough isn't hard to achieve.

As for the SCR itself, electrically, it can only withstand so much dI/dt (even with a solid drive circuit), and peak current.  These values can be calculated from the RLC equivalent of your cap bank and load.  The dI/dt will set a minimum total inductance requirement (because V = L * dI/dt), and the peak current will set another inductance minimum, or resistance.  You may also need a reverse protection diode (available in puck form), to protect either the capacitors (electrolytics don't appreciate reverse voltage) or SCR, and maybe some transient clamping (MOVs in parallel?) to protect the SCR against overvoltage triggering / retriggering.

Tim

[Ian.M]

There is also the approach of precalculating the strain (elongation) in each fastener that will occur at the required force, and after snugging up then easing back to finger tight to make sure all joints are fully seated, tightening them through a precalculated angle (which may be more than one turn) to produce the required elongation and therefore the correct force.   This is more practical and accurate for long thin fastners than short fat ones and you also need good mechanical data for the fastener so no cheap chinese cr*p.

[matseng]

What is the reason for the required clamping force for the puck SCRs?  Is it purely to get a low resistance or do they physically expand/contract during operation?

[ConKbot]

What is the reason for the required clamping force for the puck SCRs?  Is it purely to get a low resistance or do they physically expand/contract during operation?

From http://4hv.org/e107_plugins/forum/forum_viewtopic.php?43185

If you have a 'dead' one it's interresting to open them up.
(By cutting through the thin copper surround)
(Quite hazardous unless you wear thick gloves)
(We get plenty dead ones where I work)
The two ends are thick copper slugs, plated for anti-corrosion
The slugs are held in place by a thin copper corrugated sheet
Often the gate lead internally is just held by 'springyness' both ends are not soldered/welded !
The high current is meant to pass through the copper slugs, not the side connection(s) (if present)
For something like a launcher/coilgun/railgun the heatsinking will be unimportant at low repetition rates.
The clamping force really should be applied properly however.
Experimentally you could just have a single pressure point in the middle where the dimple is.
There are two main clam types
- The bolts need to be tightened with a torque wrench/accurately set
- An internal spring washer 'snaps' once correct torque is applied.

Pressure contacts both internally and externally. 

http://www.vishay.com/docs/95544/mounting.pdf gives specs on flatness and surface finish (0.001/inch and 2 um/80uinch respectively) for the surface youre mounting it on. Note that the suggested 'thermal' compound (listed as an antioxidant compound)  is loaded with metalic zinc particles, as is 'Noalox' which is used for aluminum on aluminum electrical joints)

Also, for bolt clamping force, www.spaenaur.com/pdf/sectionD/D48.pdf   Note that four grade-2 1/4-20 bolts (grade 2 = commonly available stuff, though hardware stores should generally have 5 or 8 available also) would do ~5000 lbs at 66 in-lb on dry (and clean!) threads  Belleville washers of approprate size would add springyness to tightening it, and prevent the clamping force from changing as much as the assembly heats and cools. As noted in the 4hv quote, cooling isnt an issue on a coilgun/railgun or other pulsed application, so you could do without. 

Definitely not the easiest to use package, module style is easier for general mucking around, but doesnt get the voltage or current capacity that the pucks can achieve.

[T3sl4co1l]

Something about using a metal stack to retain the silicon die.  Molybdenum is involved for thermal matching reasons, and then copper for mass and thermal, or something like that.

I've not actually seen inside one before.  This and a couple of other articles,
http://www.researchgate.net/profile/Ulrich_Schlapbach/publication/4099585_4.5kV_press_pack_IGBT_designed_for_ruggedness_and_reliability/links/0046352de74f03c506000000.pdf
seem to talk about Mo platelets retained by column springs, I guess inside a ceramic braze package.  If the springs are stiff (stacks of cone washers, perhaps?), external clamping force would be a requirement to minimize fatigue of the chip contact.

Also suggests that heatsinking should only be dominant on one side, but I don't know if that's still true of possibly simpler diode/SCR construction.

Tim

[Dave]

I'd like to suggest a safety feature for your capacitor bank. Put a depletion resistor (properly rated) across the terminals in series with a relay, wired as normally closed. Unless you apply power to the relay, it will keep the capacitor bank drained. Saves you the danger of forgetting that you charged the bank, as well as getting bitten by dielectric absorption.