Baxendall Class-D oscillator with a big source inductor

he

I don't see any good reason for the topology you posted to want to squeg. An increase in the center tap voltage biases the two MOSFETs on without any delay. Are you sure you didn't leave out some detail of the circuit.

in

ut

I(L4)=3D-0.03

de

ors

the

Obviously, the mode fails to simulate some property of the real circuit. The LTSpce model doesn't squeq with a bunch of inductances spread between 1mH and 33mH using more or less realistic series resistance values, and real circuits did.

The interesting question is why the real circuits squegged - Peter Baxandall never spelled that out, which does suggest that there's no easy explanation. He was using bipolar transistors driven from a separate winding on the transformer, where I've used croos-connected MOSFETs, but that shouldn't make any significant difference.

-- Bill Sloman, Nijmegen

made

ctors

of the

Did you see the squegging on the MOSFET version or did you do the bipolar?

The bias point of the transformer driven bipolars would not be controlled directly by the voltage at the center tap of the transformer. This would take away a lot of what I see as the damping effect. To increase the current taken from the center tap, the tuned circuit would have to ring up. This puts a lag into the effect.

I have seen the bipolar version used in a high voltage supply. The secondary side was a voltage multiplier. Regulation was done using the same inductor as the working inductor of a bucker.

e c --+-- -----+-----)))))-----+-- To transformer ! \\ / ! ! ! ---- --- ! ! ! ^ ! -/\\/+ ! ! ! GND ! ! ! --------- Sync ! ! PWM ! --

the

It's Slowman. What else do you need to know... jack of all trades, master of none.

...Jim Thompson

Squegging is usually caused by some bypassed biasing network that has a slow time constant, as the bipolar version may have had. I can't find any not-for-pay schematics of the original Baxandall oscillator.

John

s-D

u made

the

to

ductors

e of the

The circuit that I played with back in 1968 used a resistor to bias the centre tap of the winding that drove the switching transistors, which allowed me to make sure that they saturated, so they did exactly the same job as the MOSFETs in the current Spice model. The source inductor controls the "bias point" in both cases.

I don't think that this is a useful insight.

I don't think we are talking about the same circuit; the Baxendall class-D oscillator was invented to drive high ratio step-up transformers with lots of interwinding capacitance, often feeding Cockroft-Walton voltage multipliers, but that classic Baxandall circuit doesn't offer any built-in regulation point

-- Bill Sloman, Nijmegen

ade

tors

f the

I haven't seen the original Baxandall paper - Baxandall, P.J, Proc I.E.E 106, B, 748 (1959) - since 1966.

My appreciation of what was going on at the time was that the source inductor (L3 in my Spice model) was interacting with the capacitance of the tuned circuit, but - as I said - Peter Baxandall didn't say anything very specific about what was going on.

-- Bill Sloman, Nijmegen

"Jim Thompson" schreef in bericht news: snipped-for-privacy@4ax.com...

Bipolar

Jim-out-of-touch-with-reality-Thompson demonstrates his own mastery of the subject by posting ad hominem abuse.

I'm sure we are all ever so impressed by this exhibition of technical insight.

ass-D

you made

n the

n to

inductors

nce of the

ETs

ail

Did you try modeling it as it was?

One thing that may be at work is the storage time of the transistors. A transistor with the collector connected to an inductor and a highish impedance on the base can oscillate without any external base current (once it gets started). This is because at part of the swing the collector goes below the emitter and turns on the collector junction putting a large charge into the base. This means that there will be a change in performance if the tank rings up enough to make the collectors start going below ground. The frequency will shift down. This may take you below resonance. You need to model the circuit up before we can be sure that spice misses something that happens in real life. Spice doesn't do a good job on inverted transistors because the inverted mode can bring into action parts of the device that don't really matter in the forwards case and are thus left out of the model.

I think it is. It makes the transistors etc look like an inductance instead of a resistance to the unwanted oscillations.

Can you see the above ASCII art? The inductor I marked as "*** this one ***" is the inductor that is the one that feeds the transformer on the Baxendall. Because the circuit naturally has this inductor in its design, it can be shared with a simple bucker that regulates the supply voltage to the inverter. The working inductor of the bucker is also the input inductor of the Baxendall. The control circuit for the PNP is basically just a comparator and an op-amp. It makes a low parts count regulated HV supply.

- Did you try modeling it as it was?

Not until today

- One thing that may be at work is the storage time of the transistors.

- A transistor with the collector connected to an inductor and a highish

- impedance on the base can oscillate without any external base current

- (once it gets started). This is because at part of the swing the

- collector goes below the emitter and turns on the collector junction

- putting a large charge into the base. This means that there will be a

- change in performance if the tank rings up enough to make the

- collectors start going below ground. The frequency will shift down.

Something like this may well be going on. The model below does behave strangely for a while after the collector currents start going negative after 0.55msec of start-up.

- This may take you below resonance.

I don't think so. The tank circuit dominates the behaviour of the circuit as a whole

- You need to model the circuit up

- before we can be sure that spice misses something that happens in real

- life. Spice doesn't do a good job on inverted transistors because the

- inverted mode can bring into action parts of the device that don't

- really matter in the forwards case and are thus left out of the model.

This might explain why we saw persistent squegging in real life.

- I think it is. It makes the transistors etc look like an inductance

- instead of a resistance to the unwanted oscillations.

The squegging didn't change the frequency of the basic oscillation; it just meant that voltage swing across the tank circuit never settled down - as it does with the Spice models - but kept on cycling up and down over periods of the order of some ten cycles of the basic oscillation

- Can you see the above ASCII art? The inductor I marked as "*** this

- one ***" is the inductor that is the one that feeds the transformer on

- the Baxendall. Because the circuit naturally has this inductor in its

- design, it can be shared with a simple bucker that regulates the

- supply voltage to the inverter. The working inductor of the bucker is

- also the input inductor of the Baxendall. The control circuit for the

- PNP is basically just a comparator and an op-amp. It makes a low

- parts count regulated HV supply.

I saw it - by pasting it into Notepad. You don't make it clear whether your PNP is acting as a saturating switch or a current source. Since you talked about PWM I'd imagine that it was acting as a saturating switch, which means that you've got to make clear whether the switching frequency is above the resonant frequency of the tank circuit, or below it. In any event you are talking about a circuit that is even less easy to simulate than the Baxandall circuit.

Here is a minimal model of a bipolar Baxandall oscillator - I've not done much more than replace the MOSFETs with 2N3904 transistors and fudge the transformer to drive the bases.

Version 4 SHEET 1 1316 776 WIRE 96 -160 -464 -160 WIRE 96 -128 96 -160 WIRE -208 96 -240 96 WIRE -112 96 -208 96 WIRE 96 96 96 -48 WIRE 96 96 -32 96 WIRE 256 96 96 96 WIRE 416 96 336 96 WIRE 448 96 416 96 WIRE -112 192 -160 192 WIRE 80 192 -32 192 WIRE 256 192 80 192 WIRE 368 192 336 192 WIRE -464 256 -464 -160 WIRE -368 256 -464 256 WIRE 16 256 -288 256 WIRE 80 256 80 192 WIRE 80 256 16 256 WIRE -464 336 -464 256 WIRE -160 400 -160 192 WIRE -80 400 -160 400 WIRE 0 400 -80 400 WIRE 288 400 64 400 WIRE 368 400 368 192 WIRE 368 400 288 400 WIRE -160 496 -160 400 WIRE 368 496 368 400 WIRE -240 544 -240 96 WIRE -224 544 -240 544 WIRE 448 544 448 96 WIRE 448 544 432 544 WIRE -464 624 -464 416 WIRE -160 624 -160 592 WIRE -160 624 -464 624 WIRE 368 624 368 592 WIRE 368 624 -160 624 WIRE -464 656 -464 624 FLAG -464 656 0 FLAG 16 256 Vct FLAG 416 96 Vout+ FLAG -208 96 Vout- FLAG -80 400 tank- FLAG 288 400 tank+ SYMBOL ind2 -128 208 R270 WINDOW 0 32 56 VTop 0 WINDOW 3 4 56 VBottom 0 SYMATTR InstName L1 SYMATTR Value 0.00025 SYMATTR Type ind SYMATTR SpiceLine Rser=0.022 SYMBOL ind2 240 208 R270 WINDOW 0 32 56 VTop 0 WINDOW 3 4 56 VBottom 0 SYMATTR InstName L2 SYMATTR Value 0.00025 SYMATTR Type ind SYMATTR SpiceLine Rser=0.022 SYMBOL cap 64 384 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 46 32 VTop 0 SYMATTR InstName C1 SYMATTR Value 100n SYMBOL ind -384 272 R270 WINDOW 0 32 56 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName L3 SYMATTR Value 0.033 SYMATTR SpiceLine Ipk=0.03 Rser=80 Cpar=8.5p SYMBOL voltage -464 320 R0 WINDOW 123 0 0 Left 0 WINDOW 39 24 132 Left 0 SYMATTR SpiceLine Rser=0.001 SYMATTR InstName V1 SYMATTR Value 5 SYMBOL ind2 -16 112 M270 WINDOW 0 44 45 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName L4 SYMATTR Value 0.0000278 SYMATTR Type ind SYMATTR SpiceLine Rser=0.004 Cpar=100pF SYMBOL ind2 352 112 M270 WINDOW 0 32 56 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName L5 SYMATTR Value 0.0000278 SYMATTR Type ind SYMATTR SpiceLine Rser=0.004 SYMBOL res 80 -144 R0 SYMATTR InstName R1 SYMATTR Value 47k SYMBOL npn 432 496 M0 SYMATTR InstName Q1 SYMATTR Value 2N3904 SYMBOL npn -224 496 R0 SYMATTR InstName Q2 SYMATTR Value 2N3904 TEXT -400 664 Left 0 !.tran 0 10m 0m 10n TEXT -448 8 Left 0 !K1 L1 L2 0.99 TEXT -448 40 Left 0 !K2 L1 L4 0.99 TEXT -448 72 Left 0 !K3 L1 L5 0.99 TEXT -448 104 Left 0 !K4 L2 L4 0.99 TEXT -448 136 Left 0 !K5 L2 L5 0.99 TEXT -448 168 Left 0 !K6 L4 L5 0.99 TEXT -472 704 Left 0 !.ic V(tank-)=0.0 V(Vct)=0.005 V(tank+)=0.001 V(Vout+)=0.0 V(Vout-)=-0.0 I(L3)=0.00 I(L1)=0 0 (L2)=0 00 I(L4)=-0.03 I(L5)=-0.0

On Dec 27, 6:02=A0pm, "Bill Sloman" wrote: [.....]

ur

ans

More later but just on this bit:

e c --+-- -----+-----)))))-----+-- To transformer ! \\ / ! ! ! ---- --- ! ! ! ^ ! -/\\/+ ! ! ! GND ! ! ! --------- Sync ! ! PWM !

your

ked

means

the

re

Ouch. I should have been able to appreciate that, though it isn't the clearest explanation I've ever seen. There is still at least one interesting question, which is how fast the PWM reacted to a step change in output voltage - from the fiddling about I've done in the past, trying to use PWM to reduce the harmonic content of the current through the inductor, I'd imagine that you need quite a lot of low pass filtering to prevent the feedback via the PWM edges from messing up the sine wave oscillator at the output.

-- Bill Sloman, Nijmegen

On Dec 28, 10:48=C2=A0am, snipped-for-privacy@ieee.org wrote:

r your

alked

h means

e the

are

=C2=A0 =C2=A0 =C2=A0 =C2=A0 !

=C2=A0 =C2=A0 =C2=A0!

=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 !

=C2=A0 =C2=A0 =C2=A0 =C2=A0 !

=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0!

=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0!

=C2=A0 !

wice

ine

feed

e it

I freely admit to sometimes making explanations that leave the reader no better off for reading them. It is one of my talents for which there isn't much of a market.

If the intended load is a voltage multiplier, the waveform gets distorted anyway. The big issues are how strong the upper harmonics are and how quickly the back bias on the rectifier grows as it is turn off. The high Q tank still keeps both in line.

Here is a hacked together version of a 500V regulated supply based on what I remember of the circuit plus a little guessing and flinging down values. I used a couple of transistors as the comparator. I have spent 3.76 seconds optimizing the design.

Version 4 SHEET 1 2368 1204 WIRE 1344 -80 1264 -80 WIRE 1520 -80 1408 -80 WIRE 80 -32 48 -32 WIRE 176 -32 144 -32 WIRE 1088 -32 688 -32 WIRE 1152 -32 1088 -32 WIRE 1264 32 1264 -80 WIRE 1296 32 1264 32 WIRE 1376 32 1360 32 WIRE 1408 32 1376 32 WIRE 1520 32 1520 -80 WIRE 1520 32 1472 32 WIRE 1616 32 1584 32 WIRE 1728 32 1616 32 WIRE 2016 32 1728 32 WIRE 0 64 -176 64 WIRE 48 64 48 -32 WIRE 48 64 0 64 WIRE 176 64 176 -32 WIRE 208 64 176 64 WIRE 304 64 208 64 WIRE 464 64 400 64 WIRE 544 64 464 64 WIRE 752 64 624 64 WIRE 1152 64 1152 48 WIRE 1152 64 752 64 WIRE 48 80 48 64 WIRE 80 80 48 80 WIRE 176 80 176 64 WIRE 176 80 160 80 WIRE 464 80 464 64 WIRE 1152 80 1152 64 WIRE 1728 80 1728 32 WIRE 1088 96 1088 -32 WIRE 48 112 48 80 WIRE 464 160 464 144 WIRE 1264 160 1264 112 WIRE 1296 160 1264 160 WIRE 1376 160 1376 32 WIRE 1376 160 1360 160 WIRE 1408 160 1376 160 WIRE 1616 160 1616 32 WIRE 1616 160 1472 160 WIRE -176 176 -176 64 WIRE 752 176 752 64 WIRE 752 176 624 176 WIRE 352 192 352 128 WIRE 352 192 48 192 WIRE 1088 192 1088 160 WIRE 1152 192 1152 160 WIRE 1152 192 1088 192 WIRE 1264 208 1264 160 WIRE 2016 224 2016 32 WIRE 1328 256 480 256 WIRE -176 304 -176 256 WIRE 208 304 208 64 WIRE 688 304 688 -32 WIRE 1024 304 688 304 WIRE 1152 336 1152 192 WIRE 1152 336 800 336 WIRE 1328 336 1328 256 WIRE 1728 336 1728 160 WIRE 1728 336 1328 336 WIRE 2016 352 2016 304 WIRE 0 368 0 64 WIRE 160 368 0 368 WIRE 480 368 480 256 WIRE 480 368 256 368 WIRE 800 400 800 336 WIRE 1024 400 1024 304 WIRE 1728 416 1728 336 WIRE 1024 496 1024 480 WIRE 1024 496 960 496 WIRE 800 512 800 480 WIRE 864 512 800 512 WIRE 1024 512 1024 496 WIRE 864 528 864 512 WIRE 960 528 960 496 WIRE 1728 528 1728 496 WIRE 688 560 688 304 WIRE 1152 560 1152 336 WIRE 352 592 352 192 WIRE 864 592 800 592 WIRE 1024 592 960 592 WIRE 800 608 800 592 WIRE 800 608 752 608 WIRE 1024 608 1024 592 WIRE 1088 608 1024 608 WIRE 800 640 800 608 WIRE 1024 640 1024 608 WIRE 688 688 688 656 WIRE 1152 704 1152 656 WIRE 352 720 352 672 WIRE 384 720 352 720 WIRE 800 752 800 720 WIRE 1024 752 1024 720 WIRE 528 768 448 768 WIRE 1104 848 896 848 WIRE 1328 848 1328 336 WIRE 1328 848 1168 848 WIRE 528 880 528 768 WIRE 800 880 528 880 WIRE 896 880 896 848 WIRE 896 880 880 880 WIRE 1712 880 1232 880 WIRE 384 944 384 816 WIRE 1232 944 1232 880 WIRE 1328 960 1328 848 WIRE 1328 960 1264 960 WIRE 896 976 896 880 WIRE 1200 976 896 976 WIRE 480 992 448 992 WIRE 512 992 480 992 WIRE 624 992 624 176 WIRE 624 992 592 992 WIRE 1360 992 1264 992 WIRE 480 1008 480 992 WIRE 1232 1024 1232 1008 WIRE 1712 1024 1712 880 WIRE 1360 1056 1360 992 WIRE 384 1072 384 1040 WIRE 480 1120 480 1088 WIRE 1712 1152 1712 1104 WIRE 1360 1184 1360 1136 FLAG -176 304 0 FLAG 464 160 0 FLAG 1264 208 0 FLAG 688 688 0 FLAG 1152 704 0 FLAG 800 752 0 FLAG 1024 752 0 FLAG 480 1120 0 FLAG 384 1072 0 FLAG 1728 528 0 FLAG 1360 1184 0 FLAG 1712 1152 0 FLAG 1232 1024 0 FLAG 2016 352 0 SYMBOL voltage -176 160 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value 24 SYMBOL pnp 400 128 M270 SYMATTR InstName Q1 SYMATTR Value 2N4403 SYMBOL res 32 96 R0 SYMATTR InstName R1 SYMATTR Value 1k SYMBOL ind 528 80 R270 WINDOW 0 32 56 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName L1 SYMATTR Value 3.3e-3 SYMBOL diode 480 144 R180 WINDOW 0 24 72 Left 0 WINDOW 3 24 0 Left 0 SYMATTR InstName D1 SYMATTR Value MURS120 SYMBOL res 176 64 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R2 SYMATTR Value 4.7 SYMBOL cap 144 -48 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C1 SYMATTR Value 0.1=EF=BF=BD SYMBOL ind2 1136 -48 R0 SYMATTR InstName L2 SYMATTR Value 100=EF=BF=BD SYMATTR Type ind SYMBOL ind2 1136 64 R0 SYMATTR InstName L3 SYMATTR Value 100=EF=BF=BD SYMATTR Type ind SYMBOL ind2 1280 128 R180 WINDOW 0 36 80 Left 0 WINDOW 3 36 40 Left 0 SYMATTR InstName L4 SYMATTR Value 0.01 SYMATTR Type ind SYMBOL cap 1072 96 R0 SYMATTR InstName C2 SYMATTR Value 3.3n SYMBOL diode 1296 48 R270 WINDOW 0 32 32 VTop 0 WINDOW 3 0 32 VBottom 0 SYMATTR InstName D2 SYMATTR Value MUR460 SYMBOL cap 1408 -96 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C3 SYMATTR Value 1000p SYMBOL cap 1360 144 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C4 SYMATTR Value 1000p SYMBOL cap 1472 144 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C5 SYMATTR Value 1000p SYMBOL npn 1088 560 R0 SYMATTR InstName Q3 SYMBOL npn 752 560 M0 SYMATTR InstName Q4 SYMBOL res 784 624 R0 SYMATTR InstName R3 SYMATTR Value 330 SYMBOL res 1008 624 R0 SYMATTR InstName R4 SYMATTR Value 330 SYMBOL res 784 496 R0 SYMATTR InstName R5 SYMATTR Value 470 SYMBOL res 1008 496 R0 SYMATTR InstName R6 SYMATTR Value 470 SYMBOL res 784 384 R0 SYMATTR InstName R7 SYMATTR Value 2k SYMBOL res 1008 384 R0 SYMATTR InstName R8 SYMATTR Value 2k SYMBOL cap 848 528 R0 SYMATTR InstName C6 SYMATTR Value 0.1=EF=BF=BD SYMBOL cap 944 528 R0 SYMATTR InstName C7 SYMATTR Value 0.1=EF=BF=BD SYMBOL npn 448 720 M0 SYMATTR InstName Q2 SYMATTR Value 2N4401 SYMBOL res 464 992 R0 SYMATTR InstName R9 SYMATTR Value 1k SYMBOL res 608 976 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R11 SYMATTR Value 4.7k SYMBOL diode 1408 48 R270 WINDOW 0 32 32 VTop 0 WINDOW 3 0 32 VBottom 0 SYMATTR InstName D3 SYMATTR Value MUR460 SYMBOL diode 1520 48 R270 WINDOW 0 32 32 VTop 0 WINDOW 3 0 32 VBottom 0 SYMATTR InstName D4 SYMATTR Value MUR460 SYMBOL res 1712 64 R0 SYMATTR InstName R14 SYMATTR Value 2e6 SYMBOL res 336 576 R0 SYMATTR InstName R15 SYMATTR Value 2k SYMBOL pnp 448 1040 R180 SYMATTR InstName Q5 SYMATTR Value 2N4403 SYMBOL res 1712 400 R0 SYMATTR InstName R10 SYMATTR Value 5k SYMBOL voltage 1360 1040 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V3 SYMATTR Value 1.23v SYMBOL voltage 1712 1008 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V4 SYMATTR Value 5v SYMBOL cap 1168 832 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C8 SYMATTR Value 0.1=EF=BF=BD SYMBOL Opamps\\\\LT1012 1232 912 M0 SYMATTR InstName U1 SYMBOL res 896 864 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R12 SYMATTR Value 1k SYMBOL pnp 256 304 R90 SYMATTR InstName Q6 SYMATTR Value 2N4403 SYMBOL current 2016 224 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName I1 SYMATTR Value PULSE(0 100e-6 30e-3 1u) TEXT 1136 -224 Left 0 !k1 l2 l3 l4 0.8 TEXT -208 1144 Left 0 !.tran 0.1 startup

her your

talked

ich means

ove the

ou are

ice

ne

it

The circuit looks feasible. L1 looks reasonable at 3.3mH. L2 and L3 are both shown as 100H each. I shifted them to 100uH and made a desultory attempt to run a transient analysis - I didn't go as far as specifying start-up conditions, so it isn't surprising that the simulation crashed.

-- Bill Sloman, Nijmegen

your

talked

means

the

are

And you believe climate models! Now THAT is hilarious.

John

I think the mu's got corrupted, so check the 0.1uF capacitors too.

Also I had to add some parallel capacitance to L1 to stop the simulation crashing (I used 5pF). Spice often seems to dislike perfect parts!

Seems to work OK then, ~500V regulated output.

I made a backlight inverter like this, sort of, (i.e. buck regulator with the buck inductor feeding current to the "Baxendall" part).

Is this close?

It looks to me like an astable MV inductively coupled to the output.

Cheers! Rich

ether your

ou talked

which means

above the

you are

he

he

ng

twice

line

ed

pe it

Ooops. For some reason the "mu" of the uH got lost in the copy and past. This is the second time it has happened on posting schematics.

My simulation ran fine.

Yes, the op-amp would not move very fast with a 0.1F capacitor on it. The bases of the transistors wouldn't care much 0.1F and 0.1uF would work about the same.