Comments

For a more in depth discussion, start a thread in the forums.

---

Details, details. Aren’t they fun? “Happiness is making progress over not-unknown obstacles towards a known goal.”

Happiness abounds!

In 1973 I worked on a project, for US Navy Department, with nanosecond triggering pulse and synchronicity issues. I learned the following which will also apply to your trigger switching. (Forgive me if I
am telling you things you already know).

Your trigger signals must be sent down ‘transmission lines’. These can ONLY reside on multi-layer PC Boards (with slightly complex geometries) or in coaxial cables: a ground plane is required. Transmission lines have issues, too. But know that is quite possible to get sub-nanosecond synchronicity.
a)The signal lines that connect your ‘FIRE’ signal must be driven by an appropriate power driver transistor to drive the ‘lines’ to get required rise time and power level.
b) The lines, wires, pc board conductors must be impedance matched, equal (LR&C), to the driver transistor (LR&C) else you will get reflections and false triggering, which is one thing that is happening to your rig right now. When reflections happen, the full value of the switching waveform is not achieved on first transit. It is instead divided over the number of the reflections back and forth over the full length of the transmission line.
c) The lines, wires, pc board conductors themselves must be terminated with a load that matches the impedace of the sender (transistor). I doubt the ‘auto spark-plug’ fills the bill here. This
will require some trial and error work with a custom test bench.
d) The time-of-propagation of the FIRE pulses must be the same for all the ignition plugs (same problem for Trinity atomic bomb).
This is easily controlled to the nanosecond by solving the above problems, then assuring each conductor LENGTH, from FIREing transistor(s) is the same.
e) an appropriate test rig/bench (which I have made) can verify the trigger’s electrical performance.

OK, the above is all in my experience. But I have a further thought: is your firing plasma in chamber possibly acting as an old ‘bubble chamber’ where cosmic rays or such can also ionize the gas? If so,
might it make sense to COOL the plasma way down so that only the igniter injectors can cause total ionization?

Best Regards!

LPP should surely contract 4dpf to take a look at their setup and maybe propose some alternative solutions. That could speed up the progress.
If you are reading this 4dpf I would like to suggest to post in the forums, and contact Rezwan or Lerner if you want to help or at least make sure your advice is heard.

We already know how to physically connect the drivers to the igniters: via custom PC board transmission lines (no coax cable at your voltages!). We now need to design igniters of matching impedance and power.

(Keep in mind: if this gets too hard, lasers & fiber optics or microwaves and helicon antenna can also ionize your gas chamber. But I think simple spark is best bet, so far.)

Test Of Igniters

1. Method To Determine the LRC impedance of the driving transistor or other active element.

a. Connect it to drive a test load: a simple variable resistor of value and power approximately known from driver specifications.

b. Set up to monitor both voltage and current independently in real time over the course of a pulse. (Expensive oscilloscopes or PC plug-ins.)

c. PULSE the driver (at MAX) and record results of both voltage and current waveforms at target over pulse-width range.

(WHEW! Everybody OK?)

d. Go back now and digitally measure the phase angle of the current: how many degrees its rise is offset from the voltage waveform’s rise (positive or negative). Remember “ELI the ICE man.” mnemonic?  Voltage leads current in the case of a primarily inductive device. Current leads voltage in the case of a primarily capacitive device. Resulting graph will show whether the driver transistor is neutral (resistive), or primarily inductive (my guess), or capacitive. Measure and record the PHASE ANGLE: the amount in degrees or radians that the current waveform is offset from the voltage waveform. So what is it? Inductive? Resistive? or capacitive?

e. Now we have the specs for our load device, namely the igniter. First thought should be whether it is primarily inductive of capacitive. (I predict at your high frequency (short pulse width) it will be more inductive: will tend to act like a one-turn coil.) And Voltage will lead Current (ELI).

f. So let’s make an igniter that has the proper phase angle. (A really bad way to do it is to take the existing one, a spark plug, and attach a band-aid compensator: probably a capacitor in parallel with it.) But better, if the igniter is to be primarily inductive we will have a longish barrel-like structure (like automobile spark plug, but of a CERTAIN length). If opposite, capacitive, we will have a DISC: an annular ring with a center electrode keeping all distances short.

g. Now let’s test the new ignitor. We should drive it with pulses like we expect, but some slower and some faster. This test can be at very low power with a custom rig. When we get igniter geometry right we will find that maximum power is transferred at our target pulse width.

h. Tweak igniter design and retest.

End.

Very nice procedure description indeed. I am not an EE but from my minimal theoretical engineering experience I can confirm that it should be correct.
I really hope someone from LLP is reading this? Rezwan? Anyone?

Hi Breakable!  I’m reading.  Eric will have to speak for himself.  Also, what happened to your avatar?  4dpf as well.  Ya’ll should get a picture uploaded.

Great to see you!
I never had a picture, but here it is now.

Back when I was working at the Univ. of Washington NPL Nuclear Physics Lab, I remember they electropolished the inside of the HV resonators to prevent breakdown. You may wish to have the parts electopolished if you are having random breakdown events.

The comments by 4dpf & pulser make me wonder if you are getting non-simultaneous behaviour because you are having a consistent timing problem (i.e. switch 1 & 3 always fire earlier than switch 7 & 8), or if you are having random firings of your 8 switches, such that the time of firing is erratic (sometimes close enough together, sometimes not).  During your testing, it may be helpful to identify this, as consistently seeing 1 or 2 early firing units may give you reason to suspect they are caused by “local” inductive effects (vs. late units by capacitive effects), as 4pdf suggests/implies.  If you’re getting erratic firing (timing bookended by switches 1 & 8 sometimes or 2 & 4 other times, etc), the problem may be inconsistent surfaces, which can lead to random or erratic behaviour, and can possibly be mitigated as pulser suggests, by smoothing the surfaces…  The combination of the pre-firing & simultaneous firing issues suggest to me that it’s the surface preparation that is holding back the process.  Sadly, these will likely get worse as the components age, as local high points will receive the majority of the current, and will be most susceptible to local heating/melting processes, which can cause local inhomogeneities (read high/low points), and exasperate the problem, as per your previous picture of copper electrodes.  Long term testing, followed by SEM or even optical analysis of the components should help answer this question.

One other thought, in terms of the energies involved.  You are attempting to create a pseudo-stable situation where a very high electrical potential exists.  The trigger causes a breakdown event that allows this potential energy to be released, as your desired current flow. 
If this is not working consistently, you have a series of possible culprits: 
1. The potential energy barrier is too low to prevent pre-firing (e.g. without trigger, or at random times during trigger firings);
2. The potential energy barrier is high enough that it prevents regular pre-fire (in the absence of the trigger), but low enough that the the system can be triggered by minor variations at the trigger (inductive effects, the earliest potential changes, etc).
3. the trigger timing/firing is erratic (pre-firing, bad reproducibility, mis-fires);
4. the energy barrier is at the threshold (read sometimes too high) for the trigger (delays in firing, and/or no firing at all). 
To test 1, charge the system (or better yet, overcharge the system), and disengage the trigger(s) during testing occasionally to see if pre-(or self) firing happens, especially as your components wear or retain heat from previous firing events.  By pulling wires, you can selectively do this for any or all of your switches, in principle. 
To test 2, you can also lower the energy on the trigger & see if your timing improves, and look for the threshold at which the firing cannot be acheived.
To eliminate 4, boost the energy to the trigger. 
However, I think the likely culprit is 3.  The test for 4 may also help as higher energy may help the trigger overcome its own internal energy barriers more consistently.  Switching from automotive plugs in your design may just do this anyway, with any luck.  In the meantime, you can buy those most expensive platinum tipped plugs that have superior automotive performance, and trying that.  (It sometimes seems as if I have a vested interest in Pt, but I don’t!  I just know it works really well in the lab.)

In the end, if you cannot acheive good results, I may be more inclined to use a laser for timing as 4dpf suggests, as I think this would be the easiest to synchronize. With some modification, existing commercial systems could be applied to yours, as many Fourier Transform Spectroscopies use lasers to time events (FTIR being the most well known & widely availabe example).

Hey Pulser and Mr. Keech,

I was re-reading your posts and also the first post on this topic. The first one describes the synchronization error band with 2 TWO numbers: 150 ns and 300 ns. This is a good hint for us, I think. Remembering that a pulse in a transmission line travels at roughly 1 foot per nanosecond, we are talking big distances here. Also, 300 is 2 x 150, which sounds like reflections issues. But what is 150 feet in length? Looking at you lab photos I don’t see 150 feet between trigger source and igniters. Am I wrong? This is a key dimension to the problem IMHO. Raises another question. If you know the 150 and 300 ns data then how do you know it? You have an oscilloscope waveform of the rise time to show us? If so please describe. If you do my guess is that it is ‘terraced’ and not a linear rise, and that it takes about 150 ns to rise to full voltage. True? (Rent a top-end oscilloscope to capture it, if you haven’t already.)

Another point. 150 ns implies 150 feet in variance. Can surface polishing possibly be responsible for that amount? Here I must bow to those with high power pulse experience as my own is only in very low power.

By the way, what IS the trigger power? What is the typical voltage and current of the pulse you are making? Is it possible that every pulse significantly damages the igniter, by electrical machining? If it is an issue, then we’re back to microwaves or laser triggering.

GOOD PROBLEM!

Page 1 of 2 pages of comments  1 2 > 

Post a Comment