Here are some more thoughts about the phase shifter idea. Use 5v VDD instead of the 15v shown in the diagram, as the circuit is shown, that voltage seemed to work. I tested out the circuit on the LTspice simulator. On the Yahoo Group pertaining to that simulator, I obtained the CD4000 CMOS library from the files section. I am not good with computers, so that says a lot for the user-friendliness of LTspice.
In my simulation, I also made the phase shift capacitor 100pF, and varied the resistor from a few hundred ohms for a maximum phase shift of about 180 degrees to 270k for the minimum of about 0 degrees. I used a square wave with a period of 40 microseconds because I hope to get the zero voltage switcher to use resonant capacitors in the range of 0.1uF on each output MOSFET totem pole. I really do not know if I can get it to work that way, but I expect a quiet running supply with little consideration needed for parts layout to be the result if successful. I would hope it to have usability in powering a class D amp. Yet, one caution about this ZVS approach which uses relatively large resonant capacitors is that if the circuit does not stay in ZVS mode, the MOSFETs may overheat from hard-switching against the capacitors, and the low EMI sought will disappear.
I may try to use phase modulation in a class D amp if I get around to it. This type of phase shifter may work in conjunction with a class D pulse width modulator to drive the output MOSFET totem poles out of phase, but, it really seems like a lot of work for an amp. A phase-modulated amp may also require a carrier-based approach. At least the frequency would need to be fairly constant, I'd say. I tend to think classd4sure was on the right track in regard to a good amp design along the lines of a UcD.
Many readers are aware that Texas Instruments has a line of low-power phase shift modulated class D chips, but alas, they are only available in surface mount. I think for that reason, they are out of the question for use in any class D amp to be potentially considered a circuit design reference.
I would not mind if a moderator placed my posts in this thread into a new one titled "Phase Shift Modulation" or something along those lines.
In my simulation, I also made the phase shift capacitor 100pF, and varied the resistor from a few hundred ohms for a maximum phase shift of about 180 degrees to 270k for the minimum of about 0 degrees. I used a square wave with a period of 40 microseconds because I hope to get the zero voltage switcher to use resonant capacitors in the range of 0.1uF on each output MOSFET totem pole. I really do not know if I can get it to work that way, but I expect a quiet running supply with little consideration needed for parts layout to be the result if successful. I would hope it to have usability in powering a class D amp. Yet, one caution about this ZVS approach which uses relatively large resonant capacitors is that if the circuit does not stay in ZVS mode, the MOSFETs may overheat from hard-switching against the capacitors, and the low EMI sought will disappear.
I may try to use phase modulation in a class D amp if I get around to it. This type of phase shifter may work in conjunction with a class D pulse width modulator to drive the output MOSFET totem poles out of phase, but, it really seems like a lot of work for an amp. A phase-modulated amp may also require a carrier-based approach. At least the frequency would need to be fairly constant, I'd say. I tend to think classd4sure was on the right track in regard to a good amp design along the lines of a UcD.
Many readers are aware that Texas Instruments has a line of low-power phase shift modulated class D chips, but alas, they are only available in surface mount. I think for that reason, they are out of the question for use in any class D amp to be potentially considered a circuit design reference.
I would not mind if a moderator placed my posts in this thread into a new one titled "Phase Shift Modulation" or something along those lines.
Ahh, it finally worked when I added a capacitor from the phase shift one to the power negative to conserve the original charge and to prevent it from shunting to the supply rails. One of the reasons it did better now is that I made it 50% larger than the phase shift cap instead of the wrong way I had tried it--50% smaller. Now it should not depend on the power supply. So if the phsh cap is 220 pF, the shunt cap value becomes 330pF. If the phsh cap is 100pF, the shunt cap becomes 150pF. The simulator may have a quirk when certain combinations are selected. So I had good test results with the 220/330 pair.
Attachments
Hi,
I've been watching this forum and have decided that I would like to make a UcD clone. Upon successfully simulating the amplifier I decided that I would construct a PCB. The board is stuffed but the circuit fails to oscillate. I've tried several things but the circuit still fails to oscillate. I must admit that I'm no controls expert.
My main objective with this project is to learn about self oscillating amplifiers. It doesn't matter if all I can get with my board layout is 5 watts I just want to see it working and if possible I would like to do that while keeping things as simple as possible.
Thanks,
Randy Knutson
Mankato MN
I've been watching this forum and have decided that I would like to make a UcD clone. Upon successfully simulating the amplifier I decided that I would construct a PCB. The board is stuffed but the circuit fails to oscillate. I've tried several things but the circuit still fails to oscillate. I must admit that I'm no controls expert.
My main objective with this project is to learn about self oscillating amplifiers. It doesn't matter if all I can get with my board layout is 5 watts I just want to see it working and if possible I would like to do that while keeping things as simple as possible.
Thanks,
Randy Knutson
Mankato MN
Hi Randy,
One thing I think it needs is power to the upper driver circuit before the bootstrapping takes over. To ease that requirement, you may wish to increase the value of R10, and probably R16 as well, by a considerable amount. Then a resistor connected from the upper power rail to the positive power node of the upper driver can dissipate less power.
One thing I think it needs is power to the upper driver circuit before the bootstrapping takes over. To ease that requirement, you may wish to increase the value of R10, and probably R16 as well, by a considerable amount. Then a resistor connected from the upper power rail to the positive power node of the upper driver can dissipate less power.
subwo1 said:
Sub is right, the simulation schematic omitted all startup circuitry since it would just have slowed down the simulation and wasn't pertinent to the normal up-and-oscillating operation of the amp. However, before changing the value of the pull downs resistors, I'd recommend you try precharging the bootstrap capacitor prior to enabling the amplifier.
Put a 12 volt Zener across C6 and connect a 100k from its cathode to Vdd. Use the Vee connection to R2 as an enable switch (lift to disable). Apply power and wait about 10 seconds for the bootstrap supply across C6 to come up, then reconnect R2 to Vee. This is a kludge only good for testing, but it should get the amp to start up.
Good luck -- analogspiceman
Hi Randy,
You should do what I did and use a protoboard to get it working first, save you the cost of PCB's that don't work, and you can experiment far more simply. It can and will work on a protoboard.
Subwo1's suggestions are on target, but aren't the complete solution, which you'll find within and throughout this thread, since you said you wanted to learn I'll leave it at that and let you read it, but as a hint it has to do with a zener and another resistor from someplace to the negative power rail.
Keep us posted with your progress.
Regards
Chris
You should do what I did and use a protoboard to get it working first, save you the cost of PCB's that don't work, and you can experiment far more simply. It can and will work on a protoboard.
Subwo1's suggestions are on target, but aren't the complete solution, which you'll find within and throughout this thread, since you said you wanted to learn I'll leave it at that and let you read it, but as a hint it has to do with a zener and another resistor from someplace to the negative power rail.
Keep us posted with your progress.
Regards
Chris
analogspiceman said:
I forgot to mention that you must provide a return path for the bootstrap charging current. This can either be through some sort of load, if connected, or through another 100k resistor from the junction of the output mosfets to the existing 12 volt Zener you have referenced to Vee.
Regards -- analogspiceman
BTW Microsemi is making some class D phase-shift ("filterless") amps now with somewhat higher power, with the modulation scheme allowing "filterless" operation, and has some modulators/drivers (sans output FETs) that will operate filterless as well. Unfortunately like so many things these days they are strictly SMD packages.
Crown's amplifier patent includes a PWM scheme that is the same as phase-shift regular class D. I believe at one point they were looking at the feasibility of using one of the TI parts as a cheap front end for a bigger Crown-style output stage.
Crown's amplifier patent includes a PWM scheme that is the same as phase-shift regular class D. I believe at one point they were looking at the feasibility of using one of the TI parts as a cheap front end for a bigger Crown-style output stage.
for testing the scheme -- why not do it on low power levels (several watts) with a voltage comparator chip plus some medium power transistors (or use a OPAMP without internal phase compensation such as LM301 )?
start-up circuit? add a little hysteresis / insert a schmitt trigger before the power stage can do this. It can prevent the MOSFETs from running into linear operating zone.
I have built two self-OSC classD: a BJT one and a MOSFET one. Both of them have a slow switching stage, either shoot-through or have a big dead time. The big dead-time caused a crossover distortion visible on the oscilloscope.
Many feedback scheme has been tried:
pure hysteresis / SODA / UcD / Kenshin Simpw. But there isn't much difference on distortion levels.
The main difficulty for me is to build a fast switching stage out of general purpose transistors such as 5401/5551. Intergrated gate drivers are expensive and rare here. Maybe I'll try a low side driver & capacitor coupled N+P switch soon.
start-up circuit? add a little hysteresis / insert a schmitt trigger before the power stage can do this. It can prevent the MOSFETs from running into linear operating zone.
I have built two self-OSC classD: a BJT one and a MOSFET one. Both of them have a slow switching stage, either shoot-through or have a big dead time. The big dead-time caused a crossover distortion visible on the oscilloscope.
Many feedback scheme has been tried:
pure hysteresis / SODA / UcD / Kenshin Simpw. But there isn't much difference on distortion levels.
The main difficulty for me is to build a fast switching stage out of general purpose transistors such as 5401/5551. Intergrated gate drivers are expensive and rare here. Maybe I'll try a low side driver & capacitor coupled N+P switch soon.
classd4sure said:
analogspiceman,
I did some more messing around last night but no luck getting it oscillating I'm really quite unsure what I'm doing wrong. I placed a zener diode across C6 [100uF] and a 100K from the cathode to VDD. [I had a 100K load resistor for the return path] waited 10 seconds and connected R2 back into VEE and no oscillations.
My supply voltage was +/- 25 Volts and the input signal was grounded.
Is there an easy way to "divide and conquer" this circuit? I mean is there something I could do to say the feedback loop or an input which if didnt do as expected could help me isolate the problem to either the comparator or the mosfet driver? This could save me lots of time.
Otherwise I'll go through all the nodes and record the voltages and post them here later tonight.
Thanks!
Randy Knutson
I did some more messing around last night but no luck getting it oscillating I'm really quite unsure what I'm doing wrong. I placed a zener diode across C6 [100uF] and a 100K from the cathode to VDD. [I had a 100K load resistor for the return path] waited 10 seconds and connected R2 back into VEE and no oscillations.
My supply voltage was +/- 25 Volts and the input signal was grounded.
Is there an easy way to "divide and conquer" this circuit? I mean is there something I could do to say the feedback loop or an input which if didnt do as expected could help me isolate the problem to either the comparator or the mosfet driver? This could save me lots of time.
Otherwise I'll go through all the nodes and record the voltages and post them here later tonight.
Thanks!
Randy Knutson
Randy Knutson said:Attached is a schematic of the circuit I am testing. All voltages shown are referenced to GND. [-25 gnd +25] Per Bruno's advice I still did not manage to see anything on the output... This circuit is starting to frustrate me!
Does anyone spot anything wrong
Thanks,
Randy
Yes, the input current source (associated with Q1) should go up to the positive supply, not ground.
Please fix and report back.
-- analogspicemanYes, that would explain a lot. You should see of order +0.65V d.c. average with a zero average d.c. source (that is, your 200kHz generator) at the emitters of Q4-Q5 when there is actual collector current out of Q1.
The measurement of -0.79V at the base of Q1 indicates that all the current from R2 is flowing through R1 and the base-emitter junction of Q1---was this a sim result or a measurement?
The measurement of -0.79V at the base of Q1 indicates that all the current from R2 is flowing through R1 and the base-emitter junction of Q1---was this a sim result or a measurement?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.