Baxandall converter strange output

[dreamth]

Coud somebody explain to me how that i get square wave signal out of a sinus input?
When i lower the resonance frequency i get that zigzagged signal...Also when i place my frequency meter on the output it get nuts as it sees ten times higher frequency...There i have two transformers in paralel on the same transistors, one outputing about 70v 62khz, the other 2.5v ac which then goes into a quadrupler and a doubler to get 390v dc and 6.3 v dc ,  but i repeated the measurements on single transformer systems and the results are identical just the resonance frequency is lower as expected.

[dreamth]

Well guys...you really are comunicative and helpful... i probably need a current probe to see if the transformer enters saturation which is almost certain, but i  have no idea how i should deal with it as i tried more than 20 types of wiring from 5 +5 to 12 +12 turns in the primary and very thick to very thin  litz wire  and variable gap and i get simillar results.the eficiency is about 80% though...i had different response before i used the actual constant current inductor which is using much thicker wire for the same inductance in a bigger pot so i migt just rework that current limiting inductor.

[johnf]

I have absolutely no idea what you are asking
what has sinus Sine wave i presume got to do with an inverter circuit.
Ans as to your instruments i see switching artifacts on your later pics so your counter is adding them in to.

Your are obviously trying to make an inverter to run a vacuum tube as a portable setup from a battery.
Note the Baxandall oscillator was developed to generate the high voltage needed for photomultiplier tubes that consume very little current.
Your use might be for a valve audio preamp and then the Baxandall circuit would be appropriate.
Basically it is a current fed current mode oscillator. There is always a series inductor from the supply (this makes it current fed).

I would totally expect square waves driving the transformer the secondary output will be more rounded due to secondary inductance
For us to help you have to give much more info a neatly drawn circuit diagram of what you have built and references to the components used especially the inductors

[dreamth]

I have absolutely no idea what you are asking
what has sinus Sine wave i presume got to do with an inverter circuit.
Ans as to your instruments i see switching artifacts on your later pics so your counter is adding them in to.

Your are obviously trying to make an inverter to run a vacuum tube as a portable setup from a battery.
Note the Baxandall oscillator was developed to generate the high voltage needed for photomultiplier tubes that consume very little current.
Your use might be for a valve audio preamp and then the Baxandall circuit would be appropriate.
Basically it is a current fed current mode oscillator. There is always a series inductor from the supply (this makes it current fed).

I would totally expect square waves driving the transformer the secondary output will be more rounded due to secondary inductance
For us to help you have to give much more info a neatly drawn circuit diagram of what you have built and references to the components used especially the inductors

Thanks for reply!
It's  a very classical Baxandall converter as described here: https://en.wikipedia.org/wiki/Royer_oscillator where the choke(current source) is about 350uH using 0.8  mm diameter   litz wire,almost same wire as used to wire the primaries of the transformers used in te test above .The load is purely resistive and quite tough. A class A tube preamp being a current sink might present a higher impedance than the load i am using now .I can't provide real data of these transformers as i got them from random smps, but at least one is comming from a high power ccfl circuit which had a very complet driving circuit so i'd expect at least one transformer to behave properly, but they are both quite high power gapped cores .I thought that you being more experienced in building such circuits could spot easily my problem without too much  calculation or measuring evidence, but it seems that i need to learn the whole body of theory on smps which i am programmed to do it anyway in a few months.

[dreamth]

there migt be some sense in this picture though as i'm using two transformers in parallel anyway:

[davekni]

Now that you've posted a circuit, it is possible to offer reasonable answers.  (It would further help if you add more information.  Label nodes on the circuit and label scope images with which node is being probed.  Label or list part values you are using when a specific scope image is captured.  List the time/division and volts/division of scope images.)

The quick answer is that the output bridge rectifier is clipping what would otherwise be a sine wave.  The load is forming almost a short-circuit to the transformer.  The oscillator is running with the intended inductors in parallel with the transformer's leakage inductance.  (Likely the transformer's turns-ratio is too high for the load resistance you are attempting to drive.)

[dreamth]

Now that you've posted a circuit, it is possible to offer reasonable answers.  (It would further help if you add more information.  Label nodes on the circuit and label scope images with which node is being probed.  Label or list part values you are using when a specific scope image is captured.  List the time/division and volts/division of scope images.)

The quick answer is that the output bridge rectifier is clipping what would otherwise be a sine wave.  The load is forming almost a short-circuit to the transformer.  The oscillator is running with the intended inductors in parallel with the transformer's leakage inductance.  (Likely the transformer's turns-ratio is too high for the load resistance you are attempting to drive.)

I'm trully thankful for your input Sir and i read some of your topics with great interest!
My true circuit is actually in this small sketch i attached below  .That schematic i have shown prevously was taken from a public document found Here: https://core.ac.uk/download/pdf/19136619.pdf and made me think of a possible circuit that might also raise the current regulation and possibly make the noise smaller in the anode circuit, but that remains to be verified .

I had somewhat simillar thoughts on the rectifiers and that is why i used a doubler and a quadrupler wich made the output higer impedance and make te inverter work a little bit easier and also charge the phono preamp circuit somewhhat slower ,especially the fillament which needs to be protected from inrush startup currents, but it looks like it's not very helpful on heeping the output very low noise although i got better output power using them.On the other and the clipping looks softer than usual hard switch smps.

And the circuit is the same i tried  to improve in these circuits i described and showed here.
https://highvoltageforum.net/index.php?topic=1416.0
When coosing to go with two transformers in parallel i also started to use 100w 150mhz low sat transistors (not as low sat as te ztx1055a but better heatsink...as they get quite hot,about 6...8watts of dissipated heat) :

[davekni]

I don't see any resonant capacitor in your circuit.  Do you get sine waves based on just winding capacitance?

The 350uH current-feed inductor is usually on the transformer center-tap to + supply, not emitter to - supply.  Inductor in emitter circuit will make odd base current waveforms.

[dreamth]

I don't see any resonant capacitor in your circuit.  Do you get sine waves based on just winding capacitance?

The 350uH current-feed inductor is usually on the transformer center-tap to + supply, not emitter to - supply.  Inductor in emitter circuit will make odd base current waveforms.

i forgot the capacitor...220nf...
I just saw that variation in some document and while measuring the voltage i didn't see anything suspicious when comparing the two variations with inductor in emiter and center tap .I actually added even a split 0.1ohm resistor too to linearise the emiter current as i saw some professional royers doing it, but i clearly need to make a current probe and see the currents too.

[johnf]

Dave right on the inductor has to go in series with the positive supply.
This inductor must not saturate so this almost rules out ferrite cores you need either iron powder core or air core.
If this inductor saturates the inductance disappears and there is no limiting of current to the inverter so there is a good chance you will make a silicon sacrifice to the gods who control inverters

[dreamth]

Dave right on the inductor has to go in series with the positive supply.
This inductor must not saturate so this almost rules out ferrite cores you need either iron powder core or air core.
If this inductor saturates the inductance disappears and there is no limiting of current to the inverter so there is a good chance you will make a silicon sacrifice to the gods who control inverters

I have a number of professional baxandals using   inductors wich look like normal ferrites, but i'm not a metal expert .One is made with uc3872 driving two fet transistors and it looks like a a gapped pot ferrite with te indicative ASW W74...
All Jim Williams app notes on royers(baxandalls in fact ) include something that looks like  ferrite pot inductors and they also provides specific codes  like this one:
 https://ro.mouser.com/ProductDetail/Coiltronics-Eaton/CTX300-4-R/?qs=yzwxPInThYQ4vFyw6jLRIQ== 
which is described here:
https://ro.mouser.com/datasheet/2/87/eaton-econo-octa-pac-plus-toroid-power-inductors-t-1608358.pdf

Jim Williams app notes AN55, AN49 and a few others  indicate the inductor used in the emitter as well...just that the operation is common base so faster than the usual common emiter Baxandall, but i'm using 120Mhz FT  switching transistors and 100...220 ohms base resistor  so switching speed shouldn't be a really big problem.His configuration isn't clearly faster than mine anyway as the inductor is interfering with the LT1172 switching anyway...so it might very well be the opposite.

My gapped ferrite pot inductor is from a 200watts old  smps where it was used as inductor ...At least for the moment i can see it's doing its job.I measure  370mv dc fall on it when operating at full load.As i rebuilt the thing i also introduced 0.1ohm emitter resistors .I can't measure anything abnormal for the moment.
Although some theory papers i found on baxandall converter specify that i don't need any antiparallel diodes, i have it implemented for protecting both the CB and CE junction for reverse voltages as one of my commercial royers(true inductorless royer using 1.2 ohm linearizing emitter resistors) does it with 4 x 10 ohms resistors and 4 ultra fast diodes.

[davekni]

Yes, gapped ferrite cores work fine for inductors.  DC fall cannot be used to detect core saturation, however.  Scoped waveforms can show effects of saturation, or knowing the inductor's current rating from specifications, or measuring.  Your inductor may be fine.

The inductor on the negative (emitter) side appears to be for cost savings by it's double-use as the ZVS oscillator inductor and buck-converter inductor.  Current through the 1k base resistor will be zero at the ZVS zero-crossings.   Base junction stored charge from the transistor that is turning off must be providing the charge to turn on the opposite transistor.  That adds an extra requirement (extra complication) in picking transistors - making sure there is sufficient stored base charge.  Low parts cost, but at added design work.  (If I am mis-analyzing this circuit, someone please chime in.)

[dreamth]

Base junction stored charge from the transistor that is turning off must be providing the charge to turn on the opposite transistor.  That adds an extra requirement (extra complication) in picking transistors - making sure there is sufficient stored base charge.  Low parts cost, but at added design work.  (If I am mis-analyzing this circuit, someone please chime in.)

So the very low CE sat and very hig Hfe  ztx 1055 works better also because of its huge capacitance? Never thought about it...Unfortunately i would have needed 3...4 ztx 1055 in parallel for the whole 30...40watts job and also resin potted for dissipating the heat so i opted for the higher CE sat and much lower hfe  2sd1213 recovered from a sony car amp inverter....maybe not the best choice but at least i could use a heatsink with them easily ... I didn't have the ztx849 at hand unfortunately.That's a hell of a transistor...but its VCEmax at 30v isn't good enough for a Baxandall converter working at 12v supply .Even ztx1051 with its max VCE at 40v dc performed worse than xtz1055 althoug its Vce sat is lower than the one of ztx1055 and it seemd that 1055 was superior in real world than 1051 maybe due to its max vce at 150v or  its SOA .

[davekni]

Transistor selection should be less critical once you move the supply inductor to the + side (to transformer center-tap).  Just make sure the base pull-up resistors feed enough current to keep one transistor saturated at ~5A collector current.  (40W out, 80% efficiency for 50W in at 12V = 4.2A.  Add a little inductor ripple current, so ~5A for each transistor when on.)

[dreamth]

Transistor selection should be less critical once you move the supply inductor to the + side (to transformer center-tap).  Just make sure the base pull-up resistors feed enough current to keep one transistor saturated at ~5A collector current.  (40W out, 80% efficiency for 50W in at 12V = 4.2A.  Add a little inductor ripple current, so ~5A for each transistor when on.)

Wouldn't a darlington transistor work better?

[davekni]

No, darlingtons are generally too slow and have higher saturation voltage.

For best efficiency, I recommend using FETs.  Feedback needs to change for FETs since they are controlled by voltage rather than current.  Here's my variation on the typical ZVS oscillator circuit to avoid startup issues:

[dreamth]

No, darlingtons are generally too slow and have higher saturation voltage.

For best efficiency, I recommend using FETs.  Feedback needs to change for FETs since they are controlled by voltage rather than current.  Here's my variation on the typical ZVS oscillator circuit to avoid startup issues:

Good to know that cause i already tried the mazzilli converter and was dissapointed by some startup failures when trying to supply the gate voltage from the 5v regulated supply whose output capacitor(220uf) probably charged too slow.LM340-5 would have been better tere as it doesn't need an output cap at all. I may have tried with a very low output cap though... I had no problem supplying te bjt base from the same 5 v reg.

[davekni]

The standard ZVS circuit can work too, but the gate supply needs to start before the main inductor supply, not after.  Also, for most FETs, 5V is too low.  12 is good for supplying gate pull-up resistors in the normal circuit.  It is just a nuisance to require gate supply to switch on first.  That is why I suggested the alternative with two small FETs for gate drive instead of pull-up resistors and diodes.

I've built several variations of the circuit as shown above.  Different FET part numbers, though.  Above were just some parts that happen to have models already within LTSpice.  Was playing with slow supply ramp-up simulation this morning, finding that adding resistors across the small FETs reduces peak startup current:

[Twospoons]

Those little gate drive fets are a nice addition! Neatly gets over the issue of overvolting the main fet gates, allows the gates to be pulled to 0V (no diode drop), and provides more gate drive capability than resistive pull ups.

Do you have favored devices that you've found work particularly well?   I'm still trying to work out what device parameters would be optimal.

[dreamth]

The standard ZVS circuit can work too, but the gate supply needs to start before the main inductor supply, not after.  Also, for most FETs, 5V is too low.  12 is good for supplying gate pull-up resistors in the normal circuit.  It is just a nuisance to require gate supply to switch on first.  That is why I suggested the alternative with two small FETs for gate drive instead of pull-up resistors and diodes.

I've built several variations of the circuit as shown above.  Different FET part numbers, though.  Above were just some parts that happen to have models already within LTSpice.  Was playing with slow supply ramp-up simulation this morning, finding that adding resistors across the small FETs reduces peak startup current:

Sorry for my  late comming back!
I saw that your fet drvers are very low power.What's the heat dissipation on them? If i use resistors i get them really hot for a useful value working over a wide frequency area...
The only low power fet transistor i have at hand is bss295 and 2sk941, not as fast as yours...I'm gonna try your circuit  today though cause i need to finisk my phono preamp supply this week..