SOLENOID QUENCH / Electromagnetic Projectile Accelerators / Forums

Forums

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

« Previous topic | Next topic »

SOLENOID QUENCH

first previous 2 3 4 5 next

Move Thread

LAN_403

DerAlbi

Fri Apr 12 2019, 01:17AM

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

If you keep refusing to tell what goal you want to achieve, i cant really help. Now you want randomly short circuit your capacitor. I mean.. go on. Do you think this will impact your coil current or even shut if off or impact the magnetic field at the instant you remove the voltage?
No offense, but i really cant follow all your misconceptions without you giving context.

Signification

Fri Apr 12 2019, 02:23AM

Registered Member #54278 Joined: Sat Jan 17 2015, 04:42AM
Location: Amite, La.
Posts: 367

@Albi:
My primary goal is maximum velocity. I wanted to weaken the capacitor charge for minimum suck-back as a first step. As I move on, I will want to improve other aspects of the coil-gun, depending on what happens with 1st priority being muzzle velocity. I was thinking, first: max. velocity by MINIMUM suck-back (with a single coil), then a 2nd coil to ENERGIZE at the right moment, for a max. velocity 2-coil gun, etc. Perhaps a max. of four to six stages...don't know at this point.

DerAlbi

Fri Apr 12 2019, 09:54AM

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

A one dimensional goal is always easy to optimize. You ficus on the wrong thing and utilize the wrong methods to accomplish your goals. Maximum velocity is equal to maximum energy transfer. That means you achieve your goal by scaling up. You need at least a multistage (n>=2) design in order to optimize for the zero velocity start and high velocity exit. Also you need to solve the problem at a concept level and not get lost in circuit implementation details.
You overestimate the effect of suckback. Really, just ignore it. Any currentwaveform that maximizes Energy transfer will have a suckback phase, because the high current peak is much more important than the low current tail that decelerates the projectile. It is just a limit that comes with the finite current rise/fall times in an inductor.
Removing energy input (by short circuit or whetever) is far less efficient than actually actively recuperating energy, since this actively removes energy from the coil instead of just wait for energy dissipation in the coil. This governs your circuit topology by compromising at least a second dimension of your optimization goal, namely build size:
a) You use a halfbridge which gives you the best current waveform control and will achieve the best performance in a multistage design, however it is the most complicated to build and the applied currents are limited by IGBT silicon limits.
b) You just scale up your capacitor storage capability and build a multistage SCR design. An optimal SCR design will have more suckback than a Halfbridge, however the current peak, and therefore the acceleration from a given coil is potentially much higher due to less critical silicon constraints. This is also the bigger compromise for the build size - the additional energy storage you will need for this less efficient approach will be very visible.

Again, i want to stress that suckback is really not a too bad thing. Do not focus on avoiding it.
Of course suckback has to be controlled by timing, but if you choose good timing with the natural current waveforms of a given topology, you wont get much benefit if you somehow could remove the suckback portion of the acceleration. And even if you experience gains, you have to weigh those against the additional circuit complexity you have to manage. At some point, it is easier to just add another stage if you are just after velocity.

I also want to mention that there is a slight possibility that you suffer from premature optimization syndrome. How will you ever know that your current waveform does not produce suckback? Measuring force on the projectile is really, really hard. Without measurements you actually have no idea how big the actual problem is nor if you solved the problem. So anything you will do to your gun in order to optimize the issue you are focusing on will be snake oil. At the end of every shot you only have the exit velocity and if you change timing and you get a better result you will never know if this was due to less suckback or a more optimal placement of the current peak which transferred more energy than the additional suckback removed.

To sumarize: live with the topology you are comfortable with. If you dont want to build a haldbridge, accept the behavior of an SCR design and put your work into adding another stage. If you want to add complexity to the SCR design, dont. Invest in a halfbridge instead.
That is the best advice i can give. Sorry.,

Signification

Sun Apr 14 2019, 02:12AM

Registered Member #54278 Joined: Sat Jan 17 2015, 04:42AM
Location: Amite, La.
Posts: 367

@Albi+
Perhaps I still haven't given enough information about optimization:
In school, I loved to apply differential calculus to finding the actual rates of change for applications in real devices. Where I worked in the 90's, I would plot (on very large paper sheets about the size of a large screen TV) "rotary encoders" (digital cameras messed that up--and sooo many people thing the digital screen has so much more resolution than film!). This saved the company a fortune. I would make the drawing and photograph it with old B&W film--the simple developed 24 x 36mm negative would be the encoder! Custom encoders cost $$$.
I can rig a linear encoder that can read encoding rings in the black projectile and determine it's position, velocity, and acceleration (and additional derivatives, I doubt I need). I can get 'lots' of information this way from the single firing of the first stage, and where things are not optimized, make adjustments. "This is with just the 1st stage in place" Then I can add the 2nd stage and try to maximize velocity by adjusting parameters with stages 1 and 2, etc. I haven't done it yet, but from previous experience, it can be VERY effective. I hope this is enough detail to show the kind of multi-testing that can be done. One more important thing..."What's a halfbrdige?"
EDIT:
Just one more thing...Suppose you could have a -perfect- (R=0) LC circuit (PURE UN-DAMPENED SINUSOIDAL RESPONSE), if you now add any resistance in the LC circuit (now a damped series LCR circuit), will the natural frequency of the of the LC circuit decrease (in addition to the amplitude decay)?
Thanks all...

DerAlbi

Sun Apr 14 2019, 11:31AM

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

Using linear encoders to analyze a coilgun is an interesting thought but i doubt it will work. You would need a very rigid connection to the encoder. Suppose you have a lot of suckback and massive deceleration, the inertia of the connection (lets say you use a string dragging over an encoder wheel) will not follow the slow down. But give it a go, if it works its interesting.

What is a halfbridge? Google: "asymmetric half bridge" -> go to images. In the german version of google, the first find is this picture:

which is very instructive.

RLC-Circuit: Bringing an R into the equation is a weird thing. the self resonant frequency will actually be lowered, yes. fr = 1/(2pi) * sqrt[ 1/(LC) - (R/L)^2 ] (with R being a series resistance)
However if you excite the RLC circuit with an external sinusoidal (from a signal generator maybe) you will still find the minimum/maximum impedance (=R) at the uncorrected LC-frequency.

Signification

Sun Apr 14 2019, 12:10PM

Registered Member #54278 Joined: Sat Jan 17 2015, 04:42AM
Location: Amite, La.
Posts: 367

Thanks Albi:
But it looks like I have have been a bit unclear on the encoder--this is linear with shiny shallow reflective circles spaced a bit closer than 1/4" along the black metal projectile: It is all ONE rigid piece. The detector is fast and accurate.

I was asking about the RLC circuit because a youtube playlist included a video with a series RLC circuit of R=100 ohm, L=0.5H, and C=8uF. The zero resistance (R=0) natural LC frequency is 500 Rad/s. and the oscillation frequency came out NEGATIVE. In particular -9,500 Rad/s (10,000 Rad/s - 500Rad/s). I worked it out one way (according to Barry's site-which I am yet to study in any detail) and got: 489.9 Rad/s a 10.1 reduction in Rad frequency. In the video, he changed the 100 ohm resistor to 20 ohms (only x5) and got 100 Rad/s. So I guess I was asking: if this is done properly with the 100 ohm resistor, could the answer possibly be this NEGATIVE frequency?

...Extra thanks for the reference: The schema's look very helpful. I am going to study them now.

DerAlbi

Sun Apr 14 2019, 01:26PM

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

Frequencies in general can be negative. For a sinus, this is the same as the non-negative frequency with a phase shift of 180Â°. For a cosine, there is no difference at all since it is time symmetric. However negative frequencies do usually not occur outside signal processing math.
In context of a RLC circuit, negative frequencies do not make sense. You can basically answer this your self with the formula i have given you: fr = 1/(2pi) * sqrt[ 1/(LC) - (R/L)^2 ]. There is simply no way this formula will ever produce a negative value, because the square root is always positive (or 0).
[In actuality, this formula issnt even correct but is only applicable as a simplification for very small R
(A damped oscillation must have harmonics; it cannot be describes with one single frequency) ]
BULLSHIT. I corrected my self below.

Ideally you describe such a system in its differential equation form, solve it and discard the solutions that make no physical sense. I guess this is what happened.

Signification

Sun Apr 14 2019, 08:25PM

Registered Member #54278 Joined: Sat Jan 17 2015, 04:42AM
Location: Amite, La.
Posts: 367

Here is the video I was speaking of:

Also, when it is convenient for you, look at the comment next to the orange icon by Dale Nassar (me). Is this guy right??? You can see where he surprises himself at about the 6:00 timestamp.

DerAlbi

Sun Apr 14 2019, 11:22PM

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

I just comment along:
- I dont like his circuit definition. There are no current arrows or voltage arrows. His Minuses are pure choice. Not obvious.
- Holy shit, he goes the q-route

- he just pulls a solution out of his ass. wtf.

- Ok, i was wrong above. The oscillation frequency formula is valid for all resistance values until the square-root becomes 0.

- Yes your comment is correct, even 100R is low enough to make this under-damped. He made a mistake. The circuit makes very visible 5 oscillations.

Signification

Tue Apr 16 2019, 02:17AM

Registered Member #54278 Joined: Sat Jan 17 2015, 04:42AM
Location: Amite, La.
Posts: 367

It's OK for the radical to be ZERO (0), in which case the circuit undergoes "CRITICAL-DAMPING"?
Radical = POSITIVE gives "OVER-DAMPING"
Radical = NEGATIVE gives "UNDER-DAMPING"

I would love to see what the curve looks like JUST as the radical:

1) crosses from "0" to "+" and:
2) crosses from "0" to "-"

I am still rebuilding my shop and have no easy way to graph this. Anybody know where such a plot can be seen? I'm sure it won't be long before I can plot them.
The way I see it is that all these variations MUST produce REAL results, and produce REAL plottable curves since they are actual circuits that can be built.

What is most amazing to me is that (for this application) you have to go through -complex- or -imaginary- (i) numbers to rigorously derive them at a most basic level. Nothing mathematically invalid was done in this derivation process which yields totally REAL (no complex) solutions!

first previous 2 3 4 5 next

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