Bandaxall Oscillator

I could never do better than 0.1mm OD wire - 42 awg. Professional coil winders can do a lot better - down to 0.2mm wire, around 52 awg.

Again, sub-contract the job to a professional. It's not a business that lends itself to automation, so get one-off jobs down isn't too difficult.

There do seem to offer quite a few variants on that theme.

That's not likely to be a correct explanation of why it didn't work.

It would have needed to be a very small capacitor with a very high voltage rating

A high winding capacitance is unlikely to be all that helpful.

If you haven't already read J Williams application notes on the subject (AN45, AN49, AN51, AN55, AN61, AN65 on the Linear Technology web-site) you should do so. Jim Williams never called the circuit a Baxandall class -D oscillator but that's what he's describing.

I've got some thoughts on the subject on my web-site

scillator1.htm

along with a copy of Peter Baxandalls's 1959 paper

which wasn't easy to get hold of. I ended up buying a copy from the UK IEE (and got their permission to put it on my web-site).

-- Bill Sloman, Nijmegen

Impossible to use a CFL transformer for that..they are designed to act like a neon display (tube) transformer: high initial voltage for the strike, and constant current for the light (arc).

I did a little snooping, SMT only (by way of the baby bird: goo-gull): Mini-circuits have RF transformers, balanced (CT) to unbalanced up to

1:16 turns ratio. Or Coilcraft S5499,looks like 1:10 turns ratio. There are a number of thru-hole xfmrs, some with multiple windings allowing a wide truns ratio that maybe worth looking at.

Read Jim Williams' Linear Technology application notes on the subject

- AN45, AN49, AN51, AN55, AN61, and AN65.

Cold cathode fluorescent tubes are a slightly different breed of cat from neon display tubes, and Jim Williams' version of the Baxanadall oscillator seems to have driven the back-lighting in a whole lot of lap-top computers.

-- Bill Sloman, Nijmegen

I thought this was achieved more by the series capacitor than by the transformer design. That is, the transformer runs at a more-or-less constant voltage (a couple of kV perhaps). And the series capacitor creates a "constant current" drive. So when the CCFL is initially high impedance it sees a high voltage, and after the arc strikes it experiences a low voltage and a constant current drive.

That is how it seemed to behave when I have made these circuits. (I have made several using various CCFL transformers and had no trouble getting them to oscillate, so I think the OPs problem is yet to be found).

--

John Devereux

It's both -- the voltage on the primary will be constant (assuming saturated switching and constant supply through a series inductor), but the secondary voltage in general will be different due to leakage inductance. CCFL transformers are usually constructed in a way which gives low capacitance (bank wound secondary) and high leakage inductance (one bank for primary, relatively distant from the secondaries).

The equivalent circuit model looks like a series inductor (the LL) feeding an LC parallel resonant tank (secondary LL / self L working into parasitic capacitance). This allows the secondary voltage to ramp up more than turns ratio suggests.

When the tube fires, the series cap acts to limit current and maintain resonance. Q is lower so the secondary terminal voltage is probably below turns ratio, but this depends on how much -- a short circuit puts the lamp C in parallel with winding C and Q and voltage go back up.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Here's a simulation of the circuit I've been working with.

So far, I haven't had to add a capacitor to the primary to get it to work because there is enough reflected capacitance (15pF from 1000 turns on a pot core transformer).

In the simulation I removed the 15pF on the secondary and added a

4.2nF across the primary and got the same results.

I don't know why the circuit didn't work in the lab using the FL2180 (low winding capacitance) with a capacitance added to the primary as shown on Bill's page :

scillator1.htm

but I will keep trying.....

Bill, I looked through those notes....there is an extra winding on the transformer to drive the bipolar transistors, is this circuit the same principle as the Bandaxall osc? Also, I've been using mosfets because I thought this was necessary since they need to conduct both positive and negative current ?

- Jason Kooner

Hello..... Did i not say why? The transformer is made to limit the current for constant conduction (think requirements of fluorescent, of CFL); the turns ratio obviously there to make the required high voltage for initializing the arc.

t k e .

on

There's nothing in the Baxandall circuit that explicitly limits the output current. The CoilCraft transformer will obviously have a relatively high resistance secondary (because it is made with lots of turns of fine wire) and will have a highish leakage inductance - which the OP could measure by shorting one or other of the coils and measuring the residual inductance of the other coil - but I wouldn't think that either feature was designed in as an intentional current limit.

-- Bill Sloman, Nijmegen

t k

Read the original Baxandall paper from 1959, back before MOSFETs had been invented and some 15 years before they became commercially available.

Obviously not true, since the classic Baxandall uses bipolar transistors. MOSFETs are better than bipolar transistors, in that they don't squeg if the feed inductor (L1 in your circuit) gets too big.

-- Bill Sloman, Nijmegen

May i "pete" again (re-pete)? Magic word: TRANSFORMER.

uction

"Transformer" may be a magic word in your vocabulary, but it doesn't seem helpful to impute magical powers to a transformer design that seems to have been shaped only by the familiar requirement to get a high turns ratio while minimising the inter-winding capacitance of the secondary.

If you could tell us why you think that CoilCraft have added something extra, your contribution might become marginally useful, but at the moment you appear to be pretending an expertise that you don't actually possess.

-- Bill Sloman, Nijmegen

l t

he

e 0

but

ook

Bill...sorry for the late response. In Baxandall's paper, fig 9e the collector current is shown as positive (half sinewave). However, in my simulations (see above) and in the lab (where I use mosfets) the drain current is a full sine wave (positive and negative current). In the simulation you have on your site I don't see the drain current. Is it a half sinewave or full sinewave?

thanks,

-jason

l t

he

e 0

but

ook

Bill...sorry for the late response. In Baxandall's paper, fig 9e the collector current is shown as positive (half sinewave). However, in my simulations (see above) and in the lab (where I use mosfets) the drain current is a full sine wave (positive and negative current). In the simulation you have on your site I don't see the drain current. Is it a half sinewave or full sinewave?

thanks,

-jason

oil

pot

The

ire

.

..

300

e but

look

y e

The current you are seeing is segments a full sine-wave at twice the operating frequency - if you loaded the oscillator so that you were dissipating an appreciable amount of power in the load, the current through the drain would look a lot more like a positive half-sine wave.

The component at twice the operating frequency is the reason that the Baxandall class-D oscillator doesn't produce a particularly pure sine wave - there's pretty much always an irreducible minimum third harmonic content in the output at around 1% of the fundamental (plus progressively smaller proportions of the higher odd harmonics).

I've got a way of reducing that by a couple of orders of magnitude, but it drops the efficiency of the device (as an inverter) from around 95% to closer to 50%. E-mail me for details - snipped-for-privacy@ieee.org is a real address.

-- Bill Sloman, Nijmegen