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 10 11 12 13 next last

Move Thread

LAN_403

Artikbot

Fri Oct 22 2010, 09:53AM

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

Alrighty, I'm gonna check it tonight. Maybe the drop on the OpAmp is too high and the voltage is not high enough to reset the 555?
I'll feed the board with 11.1v tonight and see what happens. In the worst of the cases, I have a spare 555 so...

The diode is an 1N5408 rated 3A at 1kV. If it turns out to be slow, can this cause the circuit not to work or just makes it inefficient?

Btw, the led didn't light up... BECAUSE IT WAS NOT CONNECTED TO MASS >.< dammit...

Beware about the project... because once it fires propery as a single stage... I'll evolve to a 3-stage one :P
I already have a working ultrafast optical triggering circuit (as fast as 2uS according to datasheet) that's gonna help this to be self-calibrated and way more efficient.

Will keep posting!

Thanks a lot!

Edit: Searching thru the net for people having recovery time issues, I've found someone experiencing voltage spikes due to a slow recovery time diode. As I deduce, the +24VDC value I read on my voltimeter can be due to a simple spike coming from the inductor that the diode isn't able to absorb, right?

So, in this case, wiring a Schottky diode would do the trick I guess.

I've checked too the RST pin wiring coming from the OpAmp and no issues there, pins are wired properly, no shortcutted stuff, no anything. So, it's the diode, or otherwise my op-amp output is so low that the timer doesn't read the normal working state (current thru reset below 5V?) and it stays reset forever.

Will try tomorrow morning with higher voltages and see if it burns down.. or otherwise it works :P

Edit2: I've thought about getting some sockets and soldering them to the board instead of direct ICs... this way I minimize the probabilities of burning one of the ICs during the soldering process, or otherwise if one gets burnt, I can easily replace them. But I'll need a new PCB, this one is kind of messy (cables are well placed, but they overlay each other several times). If it works, it'll stay as is. If it turns out to be a fail, I'll be redoing things properly

Edit3: Playing at 11.1V charges the bank to 31V, but this time it's not a spike, voltage actually builds up inside the banks. Then when current is removed, they slowly (well, fast, but way slower than last time) discharge until they're empty.

BUT the mosfet keeps heating up like hell and the voltage doesn't build up yet. Everything's properly wired.

I've taken three conclusions out:
1: Slow diode>slow charging speed. Faster diode>faster speed>possible complete charging?
2: MOSFET heats up because circuit actually drains A LOT of power. It's the only explanation. But a lot of this power is lost in switching losses due to a slow diode. Btw, the IRFP450 is rated a 0.4Ohm, kinda high. But it switches at the speed of light, so I don't care much about that.
3: Something on the op-amp is not working properly. I have to check whether it's a wrong wired zener or what is it.

Artikbot

Wed Oct 27 2010, 09:02AM

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

This is afaik what I know about the circuit, and how will I fix it:

Problems:
1:Voltage cuts WAY too early.
2: Non-inverting voltage sample is ripply.
3: Diode switches WAY too slowly.
4: Mosfet burns fingers.

Solutions:
1:Change the Op-Amp (99% chances it's fried) and replace it with a CM393 comparator. (note: This way I can get 4 or 5 and make the triggering circuits aswell). Also recalculate resistors (put ten fold values to reduce current and protect the comparator) and get better quality pots (Maybe just a low value pot and a fixed resistor for increased accuracy).
2: Throw away laziness and put a friggin' voltage/power regulator IC to power the non-inverting input and the 555.
3: Swap diodes and get a fast recovery FR305, regular silicon diode but 250nS recovery time (therefore 4MHz max frequency). 1N5408 has recovery times sitting around the millisecond.
4: Install a larger heatsink, attach the SCR to it as well and put a fan on it. Seems the circuit drains somewhere between 4 and 6 Amps, way more powerful than I expected. This way I can also attach the two additional SCRs from the two additional stages I'll be installing in the future.

Additional comments: Why do i know it is the UA741 that has failed and not the 555? Because voltage does rise, but then it stops at 31V that, oh surprise, is three times the voltage I supply the circuit with.
To explain myself: If I feed the circuit with 11V, the voltage stops at 31V approximately. That is nearly 3 times feeding voltage. If I feed it with 7.4V, voltage stops at around 23V, that is AGAIN around 3 times the supply voltage! If I feed it with 6V, it stops at nearly 19V!!

Conclusions: The Op-Amp is doing whatever it wants with voltage samples. Or otherwise my inductor is saturating itself too early and blocks current to flow through it. But I don't think this is the problem, it is a bit overrated (around 130uH, circuit needs 75uH)

I'll go and order all this stuff tomorrow and see what happens :P

GhostNull

Wed Oct 27 2010, 10:35AM

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

It always seems that assumsions and lazyness is the cause of all problems.

To fix the ripple which might also be effecting the 555 circuit, you need a "Decoupling Capacitor" This is a capacitor, typically a Tantalunium or Aluminium electroytic capacitor that you put as close as possible to the thing you want to decouple. It will smooth the rippling.

I think you should also consider the MOSFET and its ratings if it heats up so much.

Artikbot

Wed Oct 27 2010, 11:21AM

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

So, that decoupling capacitor should sit right before Vcc pin of the 555 right? Anyway, it's supplied straight from the battery, which supplies a very stable voltage value. Decreases with time, but has almost no ripple.
Maybe I need to decouple another signal? Or does the inductor introduce ripple into the 555?

The MOSFET is rated 14A 450V, so it's still overrated for this purpose. It has 0.4Ohm drain-source resistance, can this be the cause for such an amount of heat?

This is the datasheet link if you need to read more on that mosfet ^^

Thanks again Ken!

GhostNull

Thu Oct 28 2010, 06:02AM

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

Yeah, I think the inductive load would cause ripple, I'm not sure though =S but I know that you will still get ripple. Put it between that switching load and the circuit you want to decouple, putting as close as possible to the circuit you wish to decouple.

With the MOSFET, the resistance when the MOSFET is on is about half an ohm right? so:
I = V/R
I = 7.4/0.5
I = 14.8 Amps
so the maximum current (excluding Impedance) through the MOSFET would be just above the maximum rating of the MOSFET and the output voltage set by the Op amp cut off could easily go beyond the rating of the MOSFET. So maybe not quite as overrated as you thought =S

Don't be afraid to do the math, it will save you a lot of trouble and time.

Artikbot

Thu Oct 28 2010, 09:55AM

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

So, capacitor right before Vcc and ripple gone. Maybe ripple is affecting my output aswell and therefore avoiding the circuit to work, who knows :S

Thanks for the tip!

About the MOSFET:
I did the math, but didn't thought about it that way! I was cosindering that value the actual maximum amperage the MOSFET could drive. So, I'm essentially driving that amount of current through the MOSFET when my Op-amp stops switching? Then, that's the cause of such an amount of heat building up on the mosfet!
And even if the circuit kept switching, I'd be driving a lot of current through it because the duty cycle stays around the 97%ish.

...Oh wait...
You said something about a leakage resistor of 100kOhm connecting drain and source directly... Did you say that to avoid the MOSFET shortcutting directly positive and negative and therefore conducting an excessive amount of current through it?

By the way, if I put a resistor in there, a big enough one, wouldn't the current (by the paralleled resistors law) just go through the MOSFET again?

I definitely need to avoid the MOSFET from driving all that current. Resistor in series with the inductor maybe? I doubt there is any 30W resistor out there in the market LOL.

I'll have to think about that aspect.

Wow, seems I need to correct a lot of things. Maybe yenka isn't a good simulator at all xD

Thanks a lot again!

GhostNull

Thu Oct 28 2010, 10:55AM

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

Look back at the previous posts >< thats a fail safe resistor and it goes gate to source!

I also don't understand how you are thinking about the current through the MOSFET ><
The amount of current I said there would be the maximum possible current that could from through the MOSFET based on the DC resistance. The actual current might be lower due to impedance and since it would be AC it would put less stress on the MOSFET. My point was that the MOSFET you choose was not over rated as you thought and that you should treat it such.

Artikbot

Thu Oct 28 2010, 01:45PM

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

Oh god I need holidays.

I'm messing up with stuff because my brain's full of university stuff, and my teacher has asked me to dev a program in c++ and he hasn't even taught us how to start! it takes so much time to understand every single stuff, but i'm managing to get it working.

Okay, let's try to settle down the brain.

My fault. Yes, I've re-read it again, don't know what was I thinking.
Current through the mosfet, I had totally forgotten that I was actually communicating positive and negative rails directly via the MOSFET. My fault again for not looking two times at the schematic I designed.
Oh yes, impedance... As well as inductor's intrinsec resistance, that helps to increase overall resistance and reduce circuit current. So current is going to be lower than the pure DC we calculated. Without taking into account the impedance, inductor+mosfet have a total resistance of about 0.8Ohm. That's a 7A peak, and therefore around 50W gets dissipated on the MOSFET. Well this simplifies things a bit.

Anyways, I need a larger heatsink too. This one can't even dream of dissipating 50W. I'll use an AMD stock sink with a fan attached to it.

To do list:

1: Order a new comparator and new higher quality pots+resistors.
2: Put a capacitor right before Vcc pin.
3: Attach a larger heatsink.

Too good I have holidays tomorrow and the next Morning. I'll have time to relax and think about things properly

Thanks for your patience Ken, seriously.

GhostNull

Fri Oct 29 2010, 06:35AM

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

I think it would be cheaper and more worth while to get a higher power MOSFET than a larger heatsink

Artikbot

Fri Oct 29 2010, 11:21AM

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

Hm...

I've been looking at Farnell for some bigger 500V MOSFETs, but all they have in a reasonable price range is 20Amp, with the same DS resistance, and even in that range they are twice as expensive as my MOSFET was.

The next step is into the 40ish amps, being around 16â‚¬ each. And that is way over my budget for a mosfet.

Would paralleling another identical MOSFET to this one be a solution?

Btw, heatsinks are not a problem in my workshop. I have like 4 AMD Opteron stock ones, one from a socket A Sempron, lots of chipset heatsinks and even a waterblock, so I don't have to spend a penny for a cooling solution.

I was thinking about the possibility of limiting the current to let's say, 3 Amps.

That could be done with resistors, but are there any in the market that can handle dozens of watts?

Thanks!

first previous 10 11 12 13 next last

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