DIY vacuum-pump? (for vacuum impregnating)

Those are just the pressures that were used! a lot of the reason for the elevated pressure isn't to crush the bubbles, it's to force the resin into the voids that exist in the winding. If you have a multi-layer coil, possibly with an outer insulation and some inter-winding insulation, you have to overcome the viscosity of the resin.

Mark Rand RTFM

What I have read (not (yet) my experience; need a modified container) is: The vacuum container has an in feed for the varnish (I learned that word) from the bottom through a valve. Start to vacuum the container and open the valve, so the varnish gets sucked into the container. Now the coil is standing in varnish, but just for a few mm. As vacuum increases, the varnish gets sucked upwards in between the wires. Keep the level of varnish low in the beginning. If the vacuum is complete, the level has to be

*below* the top of the coil. As soon as the level gets above the coil, bubbles are trapped. "Bumping the vacuum" worsens things in this case. I *think* it helps to have an air tight outer insulation layer (as I do have). But this requires a better coil body as mine is for now. More holes (or the usual two slits) on the end disks to ease the varnish coming in. Also heating up the coil to 50 .. 100°C (but below curing temperature) helps by decreasing the viscosity of the varnish.

But this is just my understanding and model how things work. I might be wrong.

Nick

Thanks Sandy - The neat part of what Nick's doing is that he is home-brewing the components! Also if you price the Hemingway parts, I think it works out at around £60/$120US so that's quite a budget to build your own and learn something on the way.

So I didn't save a penny. :-))

wire, varnish, cores, isolation material: 100.- EUR geared motor (for the coil winder): 25.- EUR casting compound (PU; 1kg, will last) 27.- EUR rod of POM (2m long, will last some time): 18.- EUR HV-probe for the scope: 40.- EUR

Outcome: Two coils, more to come, new wire to buy, ...

But I don't think I'm stupid. I'm only nuts!

Nick

If you want to recover your costs, I think that Hemingways no longer have a supply. You could be the supplier of coils to all of Europe

(And sell oil pumps to Myfords :-)

Mark Rand RTFM

I already see me accepting 99% returns. "The primary is burnt". I now have

0.08 Ohm on the primary winding, next one will be 0.06 ohms. Doesn't run on a simple breaker.

Vorsprung durch Technik!

Nick

Some odd comments

AC coupled high voltage probes are easy. If you shunt the 1 Mohm scope input impedance with 10,000 pf and couple it to the HV source with a few pf you have a coupling time constant of

10mS which should be low enough for your application. The "few pf" is conveniently a few turns of wire wrapped round the high voltage leadout wire.

I assume that your high voltage winding is the first winding on the cylindrical powder core so that the primary is on the outside and near ground potential. In this case it would be useful to reduce the effective airgap of the open solenoid by adding some iron on the outside diameter. Ideally this would be a thin iron powder cylinder but in practice a few "I" laminations from an old "E" core would serve.

High voltage is more important than high current for effective spark ignition. The RFI suppression series resistance in spark plug leads is typically 10,000 ohms! This means that the resistance of the high voltage winding is unimportant - for maximum voltage in a given volume it should be wound with the thinnest practicable wire - 0.05 or 0.1mm.

Because you are layer winding the HV secondary the interturn voltage is only a few volts so standard enamel or Lewmex wire is quite OK - double insulated or TEX-E wire is not necessary. However, in order to keep down the secondary self capacitance, be generous with the interlayer insulation.

With single insulated thinner wire you've room for a lot more secondary turns. Your turns ratio is only 15:1 - the turns ratio on a 12v car coil is about 60:1.

For experimental trial winds, oil or wax are convenient impregnants that still permit dissasembly

Your're working on an interesting project so keep us in touch with the results - experiments that don't work are just as interesting as those that do!

Jim

That's exactly what I'm doing right now. Yesterday, while falling asleep, I remembered having a book (1977) titled something like "Electronic Motortesters". and I found a neat circuit including a suggestion for a HV-probe. Easy to do. Some POM, some wire and a tube over it. Epsilon of POM is known. So capacitance can be calculated and will be 4.7pF. Just bought a 2.5% capacitor for the divider. I'll add a suppressor-diode for security.

No it ain't. I know the argument better cooling of the primary, but it also would increase the length (bigger circumfence(SP?)). Next version will have

1.12mm wire (ca. 30% less resistance) and I get the layers full, so no space lost. Wire already sitting here.

OK, good hint/argument!

So I'll change that too. Will give me room for more winding. But I'd like to reduce the size too in a next step. But then, I only can base a completely new design on results I get with the current one.

No, it's getting close to the 0.5..1kV where start of one layer meets end of second layer. But the double insulated (TEX-E is triple and quite thick) is good enough (with above 2kV and thus 4kV wire/wire).

OK, thanks for that hint!

Winding ratio is 15:1 but voltage ratio is 16:1. I know of the much higher ratios of other coils, but they don't get the dI/dt on the primary I have (500V inductance voltage currently; will investigate in other IGBTs that can accept higher UCE). Anyhow, I'll measure, check and experiment with that too.

No failures until now. :-)

Thanks a lot for your input! Nick

Works and still alive! :-)) Turned probe out of POM (> 50kV break through voltage) and cast in PUR (similar break through voltage).

Nick

Sounds good. A few extra comments.

With very thin wire you would find it a lot easier to work with the small diameter and smooth surfaces of an inner HV winding.

The voltage stress at the alternate ends of the HV layers will be handled by the relatively thick interlayer insulation. The stress on the wire coating will still be pretty low at this point.

A few iron laminations on the outside typically increase the nH/turn by about 50%.

You're driving the core pretty hard (2000 AT). It's worth checking to see if you're losing permeability. Plot HV output against peak drive current starting with very low currents. The HV increase should be at least 80% of the ideal linear increase.

Jim

OK, so I'll try it the other way round in version 4 and see what I get. Copper losses should be in the range of 10W for now, at full spark frequency (about 500Hz, yet to find out).

Problem: No L-meter here. I know that I will be off, no guess how much. I think I can see that on the current drawn. There should be a "bent" as soon as the coil is fully loaded (or the core is saturated). An L-meter for the inductance I'm working with (0,25mH) costs 250.- EUR :-(( Couldn't find a DIY solution. That's why I ignored the suggestion you already made.

I can calculate the inductance with the charge time and the current and the voltages. I'll see what I made.

Yes, I'm aware of that. Is on the to-do-list! Currently have some nasty spikes on the power-supply and will have to change the R in front of the IGBT's gate. Now, it is much too fast. I can adjust both on and off timing. Some tuning required. Anyhow, an engine is running with it, so it can't be that bad. :-)

Spark duration is 0.4ms. Not so bad. But currently only measured with a spark laying on the table. I should get different result in an engine. ->

Scope back to the shop, off the kitchen table.

Thanks for your helpful input!

Nick

Tried something similar: airbrush-compressor (AKA fridge pump) and my injector on the outlet-side: Only a minor decrease in pressure (-0.9 bar; without injector: -0.88 bar or so). Even if I do have a leak on the fridge-pump's outlet-side, I would expect more, because the injector is very good at making volume at about -0.5 bar.

Strange!

Did I make a stupid error, or does that only work with two fridge-pumps in series? My fridge (one I lent to someone) just let escape the magic smoke, so maybe I have a second one.

Nick

I wonder if you are suffering from the fact that the resin is producing condensable vapours. If it is, you may need to introduce a gas ballast (controlled leak) part way into the cycle. This may be hard to arrange without building a pump from scratch. Another project :-)

Mark Rand RTFM

With this test, nothing except the manometer was connected. But an idea I had yesterday was, that some condensed water inside the pump evaporated? No, can't be. Boiling point / temperature (ca.)

200 mbar / 50°C 20 mbar / 20°C

But you are right, the vanish might boil at the (wanted) pressure.

You didn't see my welds. :-))

Nick

Difficult to be sure but it sounds to me like leaks. 'Fridge pumps can pull a pretty good vacuum but their displacement is only a tiny fraction of the scraping vane type normally used as backing pumps. Even the smallest leak can be a major problem.

Physically locating the leak is often near impossible and it's simpler to coat the whole of the suspect areas with a low vapour pressure grease or wax. Heat the whole of the suspect area before coating it. If the pressure goes up you've got trapped high vapour pressure crud which needs fixing first.

If the "varnish" you is a true solventless varnish engineered for impregnation it will have a low vapour pressure and is an alternative leak sealant.

Procedures in Experimental Physics. John Strong. ISBN 0-917914-56-2 is a mine of information on vacuum techniques and has lots of other interesting stuff in it.

Jim

Thin wire tends to be torn appart. Don't ask how I know! :-) Tried it with 0.112mm (the thinnest double isolated I got). Works great! For some strange reason, the secondary voltage increased beyond being able to measure it. 8-/ Might be a problem with my HV-probe.

Done. Looks good. And now, I'm also getting the 20A. The Al-value of the core is in fact higher, so I had to reduce the primary windings.

Couldn't do it that way, but I reduced voltage and measured the current. I'm at about 15% off of the linear increase.

Now to less secondary windings. I'm getting above 16kV with a spark plug connected (can't believe) and I'll verify it with less secondary windings.

OD of the coil is currently 22mm. Much slimmer than the first version (30mm). Not much left to reduce, but maybe I get 21mm with my improving winding-skills and better layout of the start/end of the wires.

I'll have a beer for the very kind supporters!!

Nick

Sounds as if you're making real progress!

Successful winding with thin wire is something that improves with practice and an insistence on dead smooth small diameter surfaces without a single wire out of place. The mechanical disturbance of thicker leadout wires can be a problem.

One method that is sometimes used is to abandon the separate leadout wire and simply form a leadout tail by double backing the winding wire back on itself 6 or 10 times. There is only a single winding wire that usefully carries current - the rest are just mechanical support.

Ideally self fluxing wire coating is used so that the end connection can be made with very hot soldering iron. However it's still possible with ordinary enamel coating. Light a VERY SMALL pot of methylated spirit (denatured alcohol) and burn off the coating in the almost invisible flame. Just as the end of the leadout reaches red heat dunk it in the alcohol and then rapidily withdraw it.

Provided you have adequate interlayer insulation (which is also useful in reducing secondary self capacitance) I still think double insulated wire is not essential. If you're pushing for smallest overall size 47SWG (0.08mm) should be entirely practicable.

Now that you're getting decent output voltages a 2 ball spark gap might be a useful cross check on your HV probe

Jim

Sorry, I don't understand that all all. Does that mean to wind some smaller diameter wire in the groves to fill them? I have some stiffer/thicker isolation that I intended to try with between primary and secondary.

They can be soldered at 400°C.

I have read to press the wire onto a "Spalt-Tablette" (umm, that is an headache pill, with ... forgot the chemical name, quite easy to do) with an hot soldering iron. But didn't try it.

You might be right. My intention is to have them withstand an open secondary side. Until now, it works. But I'll try a single isolated wire with the suggested 0.08mm. Seems my collection of wires is getting bigger and bigger. :-)

Did you know that the voltage on the HV side has to be negative? All the scope-pictures (of motor-testers like Sun, Bosch etc.) you see have an inverted display. The center-electrode is hotter and thus can be easier ionzed to start the spark.

The rule of thumb I know is 1kV/mm. But that depends on the shape of the electrodes. Do you have an relyable setup with two balls? 2kV/mm?

Nick

I'm getting a decent vacuum pump in the next days, so I'll see. Thanks a lot to the one! (Don't know whether he wants to be named here).

Will make a new vacuum chamber now. With an inlet valve for the varnish at the buttom, so I can suck the varnish in with already having vacuum in the chamber. Also, I will be able to de-gas the varnish in advance.

The one I have is a PU in water that is cured at 120..140°C. I'll see what it does at a very low pressure. It is called "Aquatherm".

Nick