Electronic ignition thread / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Electronic ignition thread

first previous 10 11 12 13 next last

Move Thread

LAN_403

Ash Small

Sun Dec 15 2013, 02:11AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

paris wrote ...

"the man" is pretty good!! .....been wondering about the ign circuit , I was not keen to see 2n3055 .
in mean time I went and got a splitdorf mag = 4.5kg spark plug :0)

I've a few ideas 'buzzing around', but I need to assemble the rest of the bike, so I still have a 'few months' to finalize the details of the 'prototype' ign system. Thanks for the reply, Paris

Peter02

Thu Jan 09 2014, 02:00AM

Registered Member #1488 Joined: Sat May 17 2008, 10:41AM
Location: Germany
Posts: 18

Hi,
i just read this thread with great interest.
Some years ago i was also experimenting with different types of ignition systems. While researching this topic i came across one patent which intrigued me greatly.
https://patentimages.storage.googleapis.com/pdfs/US4589398.pdf

As is already mentioned in this thread, one problem with conventional ignition systems is the impedance mismatch between the ignition coil and the spark plug which leads to energy transfer efficiencies on the order of 1%.
One advantage of these ignition coil type systems is the relatively long burning arc which works quite reliable if the plug is clean.
Whatever, i decided to test this new scheme and build a relatively high energy (about 4 Joules per Pulse) Capacitor discharge system in combination with a normal ignition coil and a peaking capacitor of about 1nf directly connected across the sparkplug.
All of this was fitted to a 1 Cylinder Honda 4-stroke engine at a friends house driving an 3kw Generator.
The difference between the standard magneto and the new system was very noticable.
The exaust gases lost its smell and the engine ran at a higher rpm at the same throttle. This was also testet in comparison to the cdi system without the peaking cap to make sure that the possibly different timing between the magneto and the cdi was not the reason for that.

Another difference was that the engine was able to run with much much leaner air/fule ratios where it did not even start with the magneto or the cdi alone, so that was a very objetive difference.

The next step was to fit the system to a car. The problem here was the limited space for the capacitors and the energy wasted in the distributor which hindered me from getting it to work with a single coil at all. So i have no idea how the advantages would have translated to the car.

here is another interesting paper on the topic:
http://files.uloziste.com/7b1d3568a8a91418/SAE_02ffl_204.pdf

When i ge the time i will try again to fit such a system into my car with seperate ignition coils to remove the losses of the distributor.

Greetings and a happy new year.

Edit: Sorry for the link issue

Ash Small

Thu Jan 09 2014, 03:41AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Sounds very interesting, Peter. Unfortunately your link doesn't appear to have shown up in your post. Could you possibly either try posting it again, or give me some tips on what to google?

Your suggestions sound very similar to what I'm investigating. Proud Mary suggested a 'peaking capacitor' and I agree that adding some capacitance to the output sounds like a very good idea.

The person I bought this engine from recently told me, when I bought the clutch, that he had the person who owns 'Interspan ignition' at the first race meeting that he tried the Interspan ignition unit at there, and he couldn't understand why it mis-fired.

I've a few ideas I'm working on at the moment, but I'm also researching inductors/Xformers in general. I was thinking about a 'multi-spark' system, but I'd need to 'energise' the primary in ~100nS to make it work, which is 'tricky' to say the least.

As this is a motorcycle the system can't be too bulky, and as it's a 180 degree crank (firing at 90 degrees and 270 degrees) I basically need one ign. system for each cylinder (this was the problem with the Interspan unit, I believe. They did 'crack it' in the end)

The idea I'm working on at the moment is basically to charge a capacitor, then 'dump' it into the primary coil. High compression (~11:1 at the moment) and a lean mixture WILL require a BIG spark to ignite it, (or multiple sparks, which will charge the peaking capacitor until it fires, then continue to provide a 'cleaner burn'.

I'm not too concerned about power consumption as I've managed to fit 1000 Watt alternators to this type of motor in the past.

Please re-post the link.

EDIT:

paris wrote ...

I was not keen to see 2n3055 .

Sorry I didn't pick up on this point sooner, Paris, but some 2N3055's are a lot better than others (I've done my research). I'm not sure I'll use them for this project, though. I'm thinking of 'maybe' using SiC devices, although at least one person here (Steve Connor, if I remember correctly) has suggested this may not be the best option.

Peter02

Thu Jan 09 2014, 12:39PM

Registered Member #1488 Joined: Sat May 17 2008, 10:41AM
Location: Germany
Posts: 18

Hi,
Sorry for the missing links, i have used the Link option in a wrong way and forgot to check because i was nearly sleeping already.
If you want to build a multispark system with 100ns risetime (0 to 20kV) on the peaking cap (if that is what you meant) and you have a peaking capacitor of 1nf for example, you would have peak currents of 180 Amperes on the High voltage site. Translating to a primary of about 1000V would require about 2000 amperes peak current with a current rising speed of over 30kA/Âµs.
These conditions are not so easy to achieve.
The patent i posted earlier discusses methods to do this in some detail. In my experiment i tried many different phase control SCRs in combination with saturable inductors to speed up the switching of the SCR using the method discribed in this patent:
http://patentimages.storage.googleapis.com/pdfs/US4266148.pdf
It worked surprisingly well and i achieved speeds of up to 5000A/us with 150A/us SCRs but the SCRs still overheated very rapidly when i increased the pulse frequency.

The speed would be managable with Mosfets, but they dont tolerate these high peak currents, and neither do normal IGBTs. They are commonly only rated for Peak currents of 2 times their average currents.
The only good solution i know would be to use Pulse SCRs which do exist with these ratings, as discribed here:
http://mdk2001.web.cern.ch/mdk2001/Proceedings/Session13/Glidden.pdf
in the last years i gathered about 400MB of Documents concerning Pulsed Power and Ignition systems But the Problem is still the same: finding the right switch.

The reason why i wanted a fast rising pulse was not to be able to do multispark but to overcharge the gap of the sparkplug to increase power density and reliability (Same spark energy independant of cylinder contitions like pressure, temperature, mixture etc.) as explained in the first patent. Because that was not possible for me i went for the normal approach with a slower pulse and a normal coil.

In my tests with lean fuel i also experienced the problem with increased breakdown voltage of the gap which went as far as the spark jumping outside of the plug across the insulator for a gap electrode distance of only about 0.5mm.

If you want to deliver anything over a few tens of milijoules to the plug the 2n3055 will surely not be the best choice even if you have a very good one. I would start testing with a high power CDI+peaking cap and single spark to see if you can reach your performance goal with that already.

Edit.: I just saw, that the 3055 was only used in that article in the power supply which is not th critical part. Otherwise it is just a standard CDI which uses an old and obsolete 400V 5A SCR (2n3525). If you want to build this i would try to use more modern and readily available parts.

For Pickups i used a hall sensor and a magnet which worked fine, but why not use the existing pickup coil?

Greetings and good luck

Ash Small

Thu Jan 09 2014, 07:06PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Thanks Peter, that's quite a bit of reading and should keep me busy this evening

By the way, at least the first link in your first post still doesn't work, but I did manage to 'copy and paste' it into google. This site does have a few 'idiosyncracies, but we've learned how to live with them.

I agree with your points regarding 2N3055's, I was just pointing out that some are a lot better than others, but none would be suitable as the main switches in anything other than a more conventional transistorised ign. circuit.

Regarding the point you make about "Same spark energy independant of cylinder contitions like pressure, temperature, mixture etc.", as far as I'm aware, breakdown will occur at whatever voltage suits the prevailing conditions 'at that time', and these don't remain constant under all conditions, so voltage rises, due to the capacitance of the plug, assuming no peaking capacitor, which is ~15 picoFarads, if I remember correctly, until breakdown occurs, resulting in more current for lower breakdown voltages. A multi-spark system would also charge the capacitance of plug, plus any peaking capacitor until breakdown occurs, where after any subsequent sparks would be of higher current/less voltage as breakdown has already occured, hopefully resulting in more complete combustion. As you've pointed out, this would be difficult to achieve, owing to the need for very low ESR capacitors and very low inductances for both primary and secondary, although the secondary inductance could be considerably higher than the primary (modern monitor flybacks, from CRT monitors work in this fashion, and produce a 'more or less' continuous output).

The best idea I've come up with so far is to try something like an inverter microwave oven flyback transformer, as these have a relatively large cross section, plus some other 'trickery' that I don't yet understand, like having a second air gap which is only inside the primary, not the secondary (another air gap is inside both).

The ~100nS rise time is only approximate, it could be several times this in practice, but you'd only have a few mS at most for multiple sparks (this is a fairly low revving engine).

The advantage, as I see it of a multiple spark system over a single spark system is that you'd only need to 'dump' a fraction of the energy into the coil on each pulse, but the output wouldn't be much different to a single, longer lasting spark, in reality, and ferrite does seem to prefer higher frequency, lower energy pulses. You'd still require a high voltage from each pulse, but lower current per pulse to charge the peaking capacitor until breakdown/discharge occurs.

I think any kind of SCR is out of the question for a multi-spark system, as you wouldn't be dumping all of the energy in the capacitors into the primary at once. Silicon carbide FET's appear to be the fastest I've seen so far, but the highest voltage devices I've seen are ~1700V, which would limit the primary voltage to ~500V, allowing 3.142 x voltage as a safety margin, if I remember correctly. Volts per turn of ferrite increases with frequency, I think, but I've not done any maths in that department yet. I also need to read up on turns ratios of flybacks as there is some 'trickery' that can be used here too, but I forget the formulas.

I have a Panasonic microwave oven inverter here (~1kW), and some Litz wire, but don't as yet have any suitable capacitors or switches. The plan is to experiment, and see what happens. Insulating the secondary will present more problems, though.

As I said earlier, the main advantage of a multi-spark system with peaking capacitor is that you don't need to dump all the energy into the primary at once.

Now, back to reading those links you provided.

paris

Tue Jan 14 2014, 08:15PM

Registered Member #3042 Joined: Wed Jul 28 2010, 12:36AM
Location:
Posts: 121

fist ful of questions ....is there room in there for dual plugs?
the pic looks pretty tight.
i dont know the guff behind them but can they have smaller gaps , theyre series? smaller gaps in hi comp is what Im thinking.
some coils the sec coil is isolated from pri so the 2 plug gaps complete the circuit...either that or burn out!

paris

Tue Jan 14 2014, 08:18PM

Registered Member #3042 Joined: Wed Jul 28 2010, 12:36AM
Location:
Posts: 121

aww gawd....I see Ive already asked that question at top of page Dooohhhh!

Ash Small

Tue Jan 14 2014, 11:26PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

paris wrote ...

aww gawd....I see Ive already asked that question at top of page Dooohhhh!

The answer is still the same, I think, Paris, not a lot of room

I'm reading up on flyback theory again, but this time from a reference I can easily understand, I think I'm getting somewhere now

,d.bGE&cad=rja

(Thanks to Finn Hammer)

Ash Small

Wed Jan 15 2014, 03:40AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I was tired and confused when I wrote this, and should have put something like 'high impedance connected accross secondary and low impedance connected accross primary', instead of talking about big and small capacitors. The bit about the diodes is not correct either, but I've a clearer picture of the whole idea now. I'll develop this further down the thread.

Sorry about the double post, but 'new information'.

I've been reading the link above, and I think I'm starting to 'get it'.

Firstly, to Steve Ward, I now realise I obviously will need diodes on the secondary, that's how flybacks work.

Now, to summarise, and I'll try to get a schematic, or the beginnings of one, started tomorrow (it's 3:30am here).

suppose you disconnect the transistor from the primary circuit once it turns off (I know, it's just a thought experiment for now, but it makes it easier to understand), and you have a 1:1 turns ratio on primary and secondary, so that the current on both sides is the same (the transistor does the same thing as a 'blocking diode' does on the secondary when the transistor is off anyway), but you have a 'freewheeling diode' on the primary, that blocks the primary current when the transistor is on, so the 'freewheeling diode' and the diode(s) on the secondary both conduct when the flyback is 'charged' (ie no charging current flowing in the primary), BUT you have a small capacitor connected to the secondary, and a large capacitor connected to the primary. The voltage generated across the small capacitor on the secondary will be higher than the voltage generated accross the large capacitor on the primary, simple.

Now, obviously you don't disconnect the transistor, but the freewheeling diode and large capacitor will protect it, and the capacitor effectively re-cycles the power it absorbs, and it is available to magnetise the core the next time the transistor switches on. Also, you don't have a 1:1 ratio, primary to secondary, that's just to explain a point.

A small capacitor on the secondary (spark plug capacitance plus 'peaking capacitor' will reach a much higher voltage than a large 're-cycling capacitor' on the primary, and the large capacitor on the primary 'protects' the transistor.

At least, that's where I am at the moment. I'll draw a schematic, but I think I'm nearly ready to start blowing up some silicon.

Of course, if anyone can spot any problems with this, I'd appreciate your comments, and thanks again to Finn for the above link.

Peter02

Fri Jan 17 2014, 07:03PM

Registered Member #1488 Joined: Sat May 17 2008, 10:41AM
Location: Germany
Posts: 18

Hi,
Maybe i dont understand what you mean, but i think you mixed up the flyback and the forward converter.
The flyback converter works the same way as the boost converter but with a transformer instead of a simple inductor.
Lets look at a the basic flyback circuit in the picture with primary switch / transistor Q1, the transformer made from inductors L1 and L2, secondary diode D1 and Output capacitor C1. C2/R2 are only included to dampen the ringing a little bit.
We have a primary Voltage Supply V1 of 100V DC. When the Transistor Q1 switches on, its Drain to Source Voltage (V_DS in the picture) sinks from 100V to 0V, so we have 100V across inductor L1 with the negative potential at the dottet side of it. This voltage drop creates a Current in L1 which rises linear while the switch is on (as predicted by Faradays law of induction : U=-L di/dt or solved for di/dt: di/dt= -U/L). This means that the currents rises faster with higher Voltage on the inductor and also when the inductance is lowered. Since L1 and L2 are coupled and of the same Value, they form a 1:1 Transformer.
Therefore while the switch is on we get 100V across the seccondary as well, again with the negative potential at the dottet site of the coil. This induced Voltage will not find any way to create a current since D1 is in its blocking state for this polarity.
So all that happens ist that current is rising and simultanously a magnetic field is created inside and around the 2 inductors storing energy E=1/2 L * I^2.
Now the Transistor Q1 switches off, but we still have that energy trapped in the magnetic field which needs the current to exist. So the current will continue to flow, driven by the voltage induced by the collapsing magnetic field. This Voltage is now of the opposite polarity as the voltage which created the magnetic field in the first place, so it will be positive at the dottet site of the coils.
Looking at the positive site we see that now the diode will be forward biased since it's anode is connected to the dotted side of the inductor L2 where a positive Voltage appears. Now the current has found a new way to close a circuit and continue to flow. It will charge up C1 starting with the same current which was in L1 at the point Q1 switched off. This process takes energy out of the magnetic field, so the current ramps down again until it reaches 0A and the Whole Energy which was stored in the magnetic field is now transfered to C1. Its Voltage can be calculated with the formula E=1/2 C * U^2.
So we see that while the Magnetic field collapses we have not a fixed voltage on L2 but a voltage as high as necessary to support the current which is needed to hold up the momentarily magnetic field (which is shrinking while losing energy, so the current shrinks too).
This is one complete cycle. It is important to look at the voltage across the Transistor Q1 while the Coil/Transformer discharges. There will be the same voltage across L1 as there is over L2, so at the Drain we will start with a voltage of 100V rising up to the final voltage Of C1 + 100V. This of cource only holds up for a inductance ratio of 1:1.
In the plot you can see the voltages and the primary current of a simple simulatio of the circuit.

i hope that this explanation does not confuse more than it helps.

So one important property of the flyback converter is that the energy is transferred while the switch is open and not while it is closed. To get the forward converter one only has to reverse either L2 Or D1, then the energy would be transferred while the switch is closed and the voltage on the secondary would be fixed according to the turns ratio.
To get a basic understanding of this i would start with looking into the boost converter.

Greetings

PS.: V(n002) in the picture is the voltage across the secondary L2.

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