HV capacitor question (Rolling my own) / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

HV capacitor question (Rolling my own)

1 2 3 4 next last

Move Thread

LAN_403

Ash Small

Sun Apr 27 2014, 01:12AM

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

I've been saying I'm going to roll some more HV capacitors for a while, but I've been waiting to find some 'reasonable quality' polythene sheeting. I found what looks like really good stuff in 'Screwfix' this afternoon, much better than the stuff on offer in 'Homebase' or 'B&Q', which are only smooth on one side, and look to be made from re-cycled polythene (which is never 'pure'). (These are all stores in the UK, but I imagine similar quality stuff is available elsewhere if you look for it)

Anyway, back to the point. This stuff is 155 microns thick, and really smooth on both sides.

The question is as follows: What kind of voltage can I reasonably (optimisticly?) hope this stuff to withstand at TC frequencies?

And the second part of the question is: How much 'overlap' should I leave at the edges to withstand the the same voltage at the same frequency as in the first part of the question?

I'm not planning on using any other insulation, just air. I'm 'guessing' around two inches overlap, but I'd appreciate some feedback from those who are more knowledgable than me regarding these subjects.

(Please don't respond that I'm wasting my time trying to 'roll my own', and that it will suffer 'corona breakdown'.)

I'm trying to match the overlap to the expected breakdown voltage of 155 microm thick polythene sheeting at TC frequencies (several hundred kHz).

Polythene sheeting was ~Â£10 for a 4 metre by 3 metre sheet. Aluminium foil was Â£0.54 per 15 metre by 0.3 metre roll (from Tesco's)

At the moment I'm working on ~15 metre by 0.225 metre capacitor plate surface (each plate) by 155 micron polythene dielectric. This gives ~50mm overlap, but I'd appreciate some advice. Also, if anyone has nothing better to do on a Sunday morning, and wants to try calculating expected capacitance.........

Sulaiman

Sun Apr 27 2014, 06:04AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

I can't tell you how to make a reliable roll-your-own capacitor (mine never lasted very long)
but here's what I found;

1) it IS better to use 2 or 3 thin layers of polythene rather than one thick one
(the idea is that any pinholes/flaws in one area have 1 or 2 'good' layers in that area)

2) rolling 1 layer Aluminiun + 3 layer polythene + 1 layer Aluminium + 3 layer polythene
is quite frustrating .. patience and an assistant required
a large CLEAN rolling area is required, maybe get a new dust sheet or similar
(when I did it it was on carpet .. bad idea!)
and wear gloves to reduce contamination

3) HF HV can surface track a surprisingly long distance !
one inch is definitely not enough for an sgtc with single capacitor
two inches with oil may be enough if under oil
if not under oil, two inches is possibly not enough for a single cap with a 15kV NST

4) corona really is a problem;
it shows up along the edges of the aluminium foil
if you cut the foil make sure that the edge you cut is the terminal end, not the internal end

5) scrunch up the terminal ends to make electrical connections AFTER the oil has filled the voids,
if not it's impossible to get the air out (I did not have a vacuum)

6) a vacuum pump would really help

7) the difference between 5kV and 20kV is massive
if possible put several home-rolled caps in series to reduce stress

8) make a small low value capacitor to experiment and learn with
before investing the time/effort to make your desired capacitors
doing this you will quicly learn all of the above!
Observe operation in the dark to see corona etc.
This is my main recommendation
REALLY!
make test caps first, it's horrible when all that effort is wasted !

Good luck

Ash Small

Sun Apr 27 2014, 12:41PM

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

OK, Wikipedia states that the dielectric strength of polythene is 20MV/m.

This reference:

gives several different figures, ranging from 18.9 MV/m to 700MV/m, and states that, "over a wide frequency range, at room temperature, commercial grade polythene can vary from 500MV/m to 700MV/m."

Confusing. Anyway, the 'best figure' works out at just over 100kV for 155 microns, and the 'worst figure' works out at ~3kV for 155 microns. The figure of 160MV/m appears several times, and this figure gives ~25kV for 155 microns.

I can't remenber how much overlap I left last time, but I think the polythene was not as thick (it was at least 12 years ago), but after what Sulaiman has posted I think I'll leave ~3 inches, which will reduce plate area by 0.375m (15m by 0.2m instead of 15m by 0.225m, this also leaves 25mm for the terminals)

I'm planning on using a flyback initially for charging purposes and voltage will be 'controlled' to some extent by the spark gap. While I do have a couple of vacuum pumps, I don't want to use vacuum inpregnated oil insulation, at least not on the first one, although what Sulaiman says is very good advice.

This will give 3m^2 plate area. Relative permittivity of polythene is ~2,25. Distance between plates is 155 microns. Putting these figures (plate area 3,000,000mm^2, distance 0.155mm, dielectric constant 2.25) into this capacitance calculator

gives 386nF. (Theoretical capacitance for a parallel plate capacitor)

I think I should go out and get some more insulating tape, as I don't want to run out half way through (I do use a lot when rolling capacitors). I'll also pick up some PVC tubing to use as a former.

EDIT: The biggest I've rolled before was 50nF, with thinner dielectric, so I'm expecting that this one will be 'tricky' to roll successfully.

Ash Small

Mon Apr 28 2014, 08:17PM

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

OK, yesterday I picked up some 35mm OD polypropylene tubing from Screwfix to use as a former (sold as 32mm waste pipe) and ten rolls of white insulating tape (33m by 19mm). I chose white as it's less likely to have conductive pigments in it.

I've also decided to think it all through again before I start rolling. Firstly, I realised that rolled capacitors have twice the capacitance of parallel plate capacitors (checked by this online calculator:

) so theoretical capacitance should be ~770nF.

Secondly, I've decided to wear rubber gloves for the whole construction process, so as to avoid getting greasy conductive fingerprints over everything.

Thirdly, I've realised that, including the former, 16 metres times 2 times 155 micron polythene sheets wrapped around a 35mm former, plus the aluminium foil is likely to produce a capacitor ~80-100mm in diameter, and I'm wondering if 25mm of 'terminal' will be sufficient, but I think I'll stick with it and solve that issue when and if it arises.

I've drawn a sketch of the layout of the capacitor, using a 500mm length of the polypropylene pipe, 300mm wide aluminium foil and 350mm wide strips of polythene sheeting.

I've also washed the kitchen floor, as this is the biggest area I have to spread out the 4m by 3m polythene sheet to mark it out into 350mm strips and cut it. I'm hoping to start this in the morning, as I expect it to take some time (I have to mark out and cut 32 metres of polythene strips, which isn't going to be a ten minute job.) I've decided to use a CD marker pen to mark it out. I'm still undecided whether to use scissors or a straight edge and knife to cut the polythene sheet, but I think scissors may be easier, considering the size of the sheet and the room I have to work in.

Sketch of layout of capacitor. Blue is PP pipe, black is PE sheet, grey is aluminium foil. (Dimensions in millimetres)

Dr. Slack

Mon Apr 28 2014, 09:00PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Firstly, I realised that rolled capacitors have twice the capacitance of parallel plate capacitors (checked by this online calculator: )

Really? OK, so the online calculator gives answers different by a factor of 2, but is it correct?

Both a rolled capacitor and a multi-parallel plate capacitor electrically stress the plastic dielectric on each side of each conductive plate, so other things being equal, I would argue it's the same capacitance.

Or to go to a geometry/topology argument, imagine a thin rolled capacitor produced on the outside of a large diameter tube. Now concentrate on just a portion of that. How's that different from a multi-parallel plate capacitor that's a bit bent?

Or from an economic argument. If rolled caps were twice as good at utilising materials as parallel plate caps, then parallel caps would not be sold, because rolled ones would wipe the floor with their costs. But both types exist.

Ash Small

Mon Apr 28 2014, 10:07PM

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

Dr. Slack wrote ...

Firstly, I realised that rolled capacitors have twice the capacitance of parallel plate capacitors (checked by this online calculator: )

Well, the 'parallel plate' capacitors used in most examples consist of two plates separated by one layer of dielectric (at least the ones I found on the internet do).

I agree that a 'multi-plate' capacitor pretty much approximates a 'rolled' capacitor (they have dielectric on both sides of most of the plates).

Part of the reason for this 'mini project' is to determine this by experimental result, as I agree there is far too much BS on the net.

Give me a couple of days and I should be able to answer whether the capacitance of this capacitor is closer to 385nF or 770nF (or maybe they are both wrong. Anyway, I hope to determine something from all of this).

Images for 'parallel plate capacitor' on Google:

Sulaiman

Mon Apr 28 2014, 11:43PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

I have not checked your math but the effective area is 2x the area of overlap if flat,
not considering the first and last turns, so less than 770nF
BUT
unless you roll the capacitor really tight
the gap between the plates/foils will be more than the dielectric thickness, so even less nF

Dr. Slack

Tue Apr 29 2014, 05:42AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Do an experiment by all means to see how close you get to your expected capacitance, but work out whether it's 1x or 2x from first principles please.

A metal plate has two sides. If you waste one of them, then of course your capacitance will be half. Single dielectric plate capacitors exist, when the dielectric is a sheet of window glass, or a salt-filled PET drinks bottle in a bucket of salt. But when you have flexible metal foil, and flexible plastic sheet, you can make multiplate caps. I wonder what the author of the calculator was thinking?

Don't forget a metal plate isn't 'active' in a capacitor, it only affects losses. The important bit is the dielectric, that's where the energy is stored. Concentrate on the dielectric sheets and you'll be OK. In the signature of one of the users on this forum, in a capacitor, the metal plate is the hole, the dielectric plate is the donut.

Ash Small

Tue Apr 29 2014, 11:33AM

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

Neil, I think it's the terminology here that's the issue. While a 'multi- plate capacitor does have parallel plates, it's generally called 'multi-plate' or 'stacked'. As far as I'm aware, the term 'parallel plate' capacitor comes from the days of Leyden jars and simple 'schhoolkid' experiments with two metal plates and a sheet of glass, as depicted in my earlier link to Google images of 'parallel plate' capacitors. E.g. a 'multi-plate capacitor' is a 'stack' of 'parallel plate capacitors', etc. As I understand it, the 'parallel plate' capacitor is the standard test capacitor for measuring the permittivity of different dielectrics.

I do recall, from when I was designing these capacitors in 2001 (before the days of 4HV) that most coilers, can crushers etc. recommended using 'stacked capacitors' under oil as they claimed that these don't have the inherent 'inductance' and 'resistance' usually associated with 'rolled capacitors', however, the 'popular' design for 'rolled capacitors' back then was to have the terminals on the outer layer of the capacitor, rather than on the sides, feeding all turns directly, as my design does.

Sulaiman, I agree that rolling it as tight as possible is important for two reasons, firstly, to achieve maximum (theoretical) capacitance, and secondly, to 'expel' as much air as possible to keep corona to a minimum. While it's impossible to exclude all air from the 'terminal sections', I don't believe this to be a great problem, as I'm under the impression that corona won't, for example, form 'inside a sphere', i.e. the electric field present in these areas will 'suppress' corona, however this assumption is just based on 'intuition' and not on any 'reference'.

As it's a nice day here today, I'm going to get some jobs done outside, but I'll update this thread when I make some more progress (the weather is supposed to deteriorate by the end of the week).

EDIT: Also, if it's not rolled tightly, there will be mechanical movement as it charges/discharges as adjacent, opposite cherged, plates are attracted to each other, etc

Ash Small

Mon May 12 2014, 08:23PM

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

I've thought this through again and realised I have just enough polythene sheet for an extra inch of overlap, 100mm instead of 75mm.

I've re-drawn the 'mark 2' below:

I've re-calculated the capacitance, which now works out at ~674nF (theoretical).

I've also decided to use a template to help mark out the polythene sheeting, a sheet of plywood 375mm by ~1 metre in length, which I'm currently marking out prior to cutting.

1 2 3 4 next last

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