New to Coilguns - Coilgun Buildlog (Charger apparently fixed) / Electromagnetic Projectile Accelerators / Forums

Forums

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

« Previous topic | Next topic »

New to Coilguns - Coilgun Buildlog (Charger apparently fixed)

first previous 15 16 17 18 next last

Move Thread

LAN_403

Artikbot

Mon Nov 15 2010, 10:48AM

Registered Member #3247 Joined: Mon Sept 27 2010, 09:42AM
Location: Spain
Posts: 137

I'll need to take a look at those dampening equations.

Hope that Maple will draw nice functions so I can see how's it behaving ;)

I have 5 capacitor banks, all of different capacitances. I have some sort of margin to tweak capacitances, so it shouldn't be much of a trouble trying to optimise the functions.

But that diode should be able to absorb the worst of the cases according to the ratings, alright?

Thanks!

Edit:

Doing some calculations based on Barry's page I found this:

The injector stage with the biggest cap bank combination (1.3mF) is slightly underdampened. Calculation throws -0.24402576e-2 as a result.
First acceleration stage is even slightier underdampened, throwing -0.40375e-3 as a result.
And second acceleration stage is a bit more overdampened than the first, giving as a result -0.4067775e-3

So far seems that the accuracy of the system is pretty acceptable, considering all the numbers I intially took were just "thrown into the machine" without even being calculated.

This are the RLC simulation's graphs (used Barry's, Maple is a pain in the ass to work with, especially when long functions including e appear)

Injector's:
Injectors

1st accelerator's:
Accelerators

2nd accelerator's:
2ndaccelerators

Seems both injection and 1st acceleration stages are pertty well critically dampened, and the second acceleration will need a bit of tuning to achieve critical dampening.

By the way, will the diode absorb that -80A spike or it will be killed? I don't know how do diodes behave with negative currents.

Thanks a lot!,
Marc.

Pinky's Brain

Mon Nov 15 2010, 11:57AM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Artikbot wrote ...
To me it sonds like it's gonna be of the same speed, but certainly smaller, since most of the current has already been wasted to heat the coils/projectile. That is... a pair of ms for the diode to say "Hey, we have to block this current!" considering I'm talking about injector, when it comes to the accelerators, it has barely a half ms to react.

The only way the diode could block the current is if you put it in series with the coil and the current changed direction, that should never happen in your circuit because you will have a negative voltage on your capacitors before that happens. The idea is to put the diode anti-parallel with the coil ... it will start conducting as the capacitor voltage has reached 0 and not allow it to drop any further. It lets the coil freewheel.

Which is why the reverse recovery delay is almost entirely irrelevant (you will lose a tiny bit of energy as the thyristor turns on).

Artikbot

Mon Nov 15 2010, 12:14PM

Registered Member #3247 Joined: Mon Sept 27 2010, 09:42AM
Location: Spain
Posts: 137

But if the diode isn't fast enough to swap from blocking to conducting state (definition of recovery time), it will keep blocking, sending current back to the SCR and therefore to the capacitors. Not in my injection or my 1st acc coil, but in the second one there's a -80A CEMF that's very likely to make fly into pieces anything that goes in its path.

So, the idea here is:

Will the CEMF pulse be as fast rising as the main pulse is? If affirmative, a fast recovery diode is needed for sure. If it rises more slowly, a regular diode will do the trick.

Another issue might be ringing. If somehow a coil rings, will the diode be able to absorb that ringing if it's not fast enough?

Sorry for being sceptical, but I need to doublecheck everything

Thanks!

ScotchTapeLord

Mon Nov 15 2010, 12:18PM

Registered Member #1875 Joined: Sun Dec 21 2008, 06:36PM
Location:
Posts: 635

Artikbot wrote ...

Seems both injection and 1st acceleration stages are pertty well critically dampened, and the second acceleration will need a bit of tuning to achieve critical dampening.

By the way, will the diode absorb that -80A spike or it will be killed? I don't know how do diodes behave with negative currents.

I think what you have now is acceptable. You'll just have to play around with capacitances until you get the best results. As for the negative spikes, that's what the antiparallel diode is for. The point of a diode is to block negative current and pass positive current, so you put one backwards ("antiparallel") against the coil so that it sees the currents in reverse-- it'll block the positive spike so it goes through the coil and conduct the negative spike coming out of the coil.

Artikbot

Mon Nov 15 2010, 12:30PM

Registered Member #3247 Joined: Mon Sept 27 2010, 09:42AM
Location: Spain
Posts: 137

Yeah, I'll tune with some spare capacitors I have until I achieve the best results.

I knew what were the antiparallel diodes for, I knew what was the back EMF, just didn't make it to relate both concepts. I guess I'm too much excited with this

So the reverse biased diode absorbs the negative spikes (the ones that are harmful for the bank). Great then! Something less to worry about.

I'll calculate some resistor values, pack them onto my Farnell order and order them! I expect them to be here for thursday/friday.

Will update soon!

Thanks!

Pinky's Brain

Mon Nov 15 2010, 11:51PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Artikbot wrote ...

But if the diode isn't fast enough to swap from blocking to conducting state (definition of recovery time), it will keep blocking

For all extents and purposes forward conduction is instantaneous as the threshold voltage is exceeded. It just takes a while for the diode resistance to drop to it's lowest point. Unlike the name suggests fast recovery diodes can actually have worse forward recovery because their initial resistance is much higher (unless you use Schottkeys, those are simply fast)

ScotchTapeLord

Tue Nov 16 2010, 02:01AM

Registered Member #1875 Joined: Sun Dec 21 2008, 06:36PM
Location:
Posts: 635

You have a point; you really have to check the graphs in the datasheet to get the full picture. But for what it's worth, I've had better experiences with ultra-fast diodes in applications like this. Besides, the current starts low at the beginning of the pulse and works its way high, so the slightly higher initial resistance doesn't strike me as too terrible of a thing. Not to mention, some people put low-value resistors in series with their freewheeling diodes to shorten the pulse. I don't think it improves performance too much, though.

Not sure if I'm right on all that, as they are nearly baseless conjectures, but it all sounds good to me. =)

Artikbot

Tue Nov 16 2010, 09:53AM

Registered Member #3247 Joined: Mon Sept 27 2010, 09:42AM
Location: Spain
Posts: 137

It's more based on a "We do this, and expect to get this because if we make it this way, it might not work".

It's not really based on any formulae or any calculus, it's just the feeling that it'll work better.

@Pinky's Brain:

Those diodes' graphs were pretty nice, the threshold is low and the power tolerance rises pretty much faster than the graph spike does, so I think they'll be alright.

By the way, I got a TO220 cased 3A 600V Schottky diode for the boost converter, 1.5A without any heatsinking/dissipation pad seemed a bit filmsy, so I got this small beast and there we go. I hope it improves performance further ;)

Ah, and I've gotten about 15 new cameras to scavenge and get their capacitors. I expect to use them not to increase energy, but to help achieve critical dampening to my current coils. After all, the whole point of moving onto multistage design with such fast components and all this optimisation stuff we've been doing has the target of increasing overall efficiency while not compromising power.

I'll leave plain raw power to others, I'm good enough having lower power and nicer efficiencies that lead onto a more cheap, portable and powerful gun

To be honest I started to think about multistaging when i saw a buddy on YouTube that had 1.6kJ (even more than Saz43's nuke coilgun) and had a hard time while trying to stick a nail onto a wooden panel. I thought to myself "Lol buddy you'd better do something efficient, look at such a waste of power and money

)

Artikbot

Sun Nov 21 2010, 12:34PM

Registered Member #3247 Joined: Mon Sept 27 2010, 09:42AM
Location: Spain
Posts: 137

Hokay, optical triggers done and working!

Instead of two separate modules, I've done one merged module sharing voltage divider for non-inv input and sharing +5V supply for LEDs (and any other additional equipment nearby) aswell.

I'm currently soldering the second IC of the replicated circuitry, when it's finished I'll post a photo just for the lulz

I was thinking... The LED happens to turn completely off, but will the SCR stay off? I need to measure the voltage on the output terminal, but I believe it sits around the volt or so.

TYN640RG are my SCRs.

Thanks!

Edit: This is how the board looks. I still need to cut out that excess of board tho

It's suuuuuuch a mess LOL.

Artikbot

Tue Jan 25 2011, 07:08PM

Registered Member #3247 Joined: Mon Sept 27 2010, 09:42AM
Location: Spain
Posts: 137

Back from the break!!

It's been quite a long time since I left this project on the table, but I now look at it and think "Aw dude, so much effort left lying on the table".

So here we are! I'm going to try and finish the thing ;)

So far, we're in the same place I left the circuitry. Optical triggers work... But instead of sourcing current, they drain it.

I'll explain myself:

When the circuit is triggered, the LM211 instead of outputting current, it "communicates" itself with ground and drains current instead of sourcing it.

I've added a pullup resistor between output and ground, 10k one so I don't drain much current, and so far it's been enough. The circuit now outputs current at the same voltage it is fed.

BUT there's a problem. The LM211 is only able to output as high as 50mA, with the TYN640RG SCRs requiring 35mA to turn gates on. So far it looks okay... But if you remember, I'm using two of them paralleled. What happens? Current splits, and there isn't enoug power to feed the gates successfully (25mA is nowhere near the current needed for them to switch on)

What I've thought:

Add a transistor (one that switches as fast as possible to avoid delays), make the LM211 feed transistor's gate, and drive about 100mA of current through that transistor. That would obviously work. But the question is, is out there any fast enough transistor (that can be purchased with a little wallet, or scavenged outta old PCBs)?
With fast, I mean FAST. Less than a millisecond if possible (since the circuit itself sends the pulse in 2.4ms, and that's long enough).

Thanks!

Edit: I've tried the transistor trick!

So far when the phototransistors are completely darkened the current flows towards the gates, but I haven't managed to make them conduct (47Ohm resistor, limiting current to about 110mA for both gates, 55mA per gate)

Reading thru the datasheet of the SCRs... Looks like 35mA is the maximum current to gates. Then why on earth it didn't work in the first place (after installing the pullup resistor)?

I'm starting to think there's some wicked circuitry in between.

first previous 15 16 17 18 next last

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