I have a simplistic power stage which currently delivers 21W RMS to a load. I use a transformer at the output so I estimate an 80% efficiency and therefore the output stage must be delivering close to 27W on the transformer primary.
My calculations show that the power spent on the output stage is around 30W (15W on the push side and 15W on the pull side).
So for 27W on load we spend 30W on the push-pull output stage (transistors/drivers/stabilising resistors).
Because my device is portable and battery powered, this creates two problems, first there is a lot of heat that needs to get rid off, secondly current consumption is great and the batteries drain quickly.
Is there a way to have an output stage more efficient than this?
The signal is a pure sinewave.
My calculations show that the power spent on the output stage is around 30W (15W on the push side and 15W on the pull side).
So for 27W on load we spend 30W on the push-pull output stage (transistors/drivers/stabilising resistors).
Because my device is portable and battery powered, this creates two problems, first there is a lot of heat that needs to get rid off, secondly current consumption is great and the batteries drain quickly.
Is there a way to have an output stage more efficient than this?
The signal is a pure sinewave.
I have absolutely 0 experience in building a class D amplifier. Also, my signal frequency is around 200KHz. Would that mean that I would need switching frequencies much higher than that? Does anyone here know how to design/build a class D? 🙂
200kHz?? What are you using your amplifier for?
That frequency will be too high for class D. If it is a single frequency, what about a tuned amplifier at class c?
That frequency will be too high for class D. If it is a single frequency, what about a tuned amplifier at class c?
Unfortunately I have 0 experience with class C... Just reading on wikipedia.
My application currently is:
1) Oscillator producing around 8V pk at 200KHz (16V p2p)
2) current power amplifier consisting of 3 pairs of drivers/power transistors
3) output transformer 1:8
4) load attached to the secondary of the transformer. Load is around 100R.
Not only I am spending 30W as heat on the current amplifier, in addition I need to up the power, so the transformer will need to become 1:9 or even 1:10. That means even more heat at the output stage...
Would a Class C be able to fit in the picture I have described above?
My application currently is:
1) Oscillator producing around 8V pk at 200KHz (16V p2p)
2) current power amplifier consisting of 3 pairs of drivers/power transistors
3) output transformer 1:8
4) load attached to the secondary of the transformer. Load is around 100R.
Not only I am spending 30W as heat on the current amplifier, in addition I need to up the power, so the transformer will need to become 1:9 or even 1:10. That means even more heat at the output stage...
Would a Class C be able to fit in the picture I have described above?
Induction heater? Is that what you are building? Ohh, I should have full read what you wrote... Sorry about that.
I can best describe the device I have built as a rather large resistor/capacitor on/through which I need to apply about 200KHz and around 170V p2p. Currently I am giving it 135V p2p, but it is not enough, and already the device is hot and needs two internal fans, as well as the battery packs draining quickly.
Yes it must be as pure as possible. The possibility of higher harmonics entering might create damage as the "load" may get damaged.
Change the oscillator to produce a square wave, then use a bandpass filter at the amplifier's output to turn it back into a sine wave. That is then effectively a tuned class D amplifier.
I do not think the current power amp would be able to cope with the frequencies involved in a square wave at 200KHz (actually the rise and fall times as well). On simulation at least, there is ringing and overshoot, and DOUBLE the amount of power expended on the transistors. I do not profess to understand everything that is going on, but from similar experiments with power square waves and a push-pull arrangement, I remember that one main reason is cross conduction, highly exaggerated with square waves. I would need a properly designed class D stage rather than an class AB stage.
Square wave cross conduction can be minimized at the expense of idle current. Using low value base-emitter resistors on the output helps. But current in the driver stage goes up of course. You may need a complementary pair of source follower mosfet, driven by an EF2 (with 10 ohm Rbe) that's in itself capable of driving an 8 ohm load. That'll suck the charge out of them sumbiches.
Investigate Class E and Class F. At 200kHz these will give you almost zero dissipation in the transistor/switch - but you will still have tuned circuit losses.
If you just need a low distortion fixed amplitude 200KHz sinewave it may be the way to go, but if you need anything else it won't be. In order to get zero dissipation in the transistor in class E or F, it is not an amplifier at all. It must be run at full output. As an amplifier, it is slightly higher efficiency than class B, with it reaching full efficiency at max output. And if you need any sort of linearity at all it just won't work.
Class AB is about 60% efficient with 40% of power lost in heat to the heat sink.
Class D is about 90% efficient.
Class D is about 90% efficient.
Hi, we swapped a few PMs on this previously.
What you want is basically a fixed frequency inverter.
To try to build it as a linear (all-purpose) power amplifier will be inefficient.
I would use a three step stair wave approximation to a sine wave, explained >HERE<.
Then put an inductor capacitor resonant filter to clean it up.
With a 3 step approximation you don't have any even harmonics or multiples of the 3rd.
So you only have to filter the 5th, 7th, 11th etc. Pretty easy.
The inductance could be part of your step-up transformer so losses are minimal.
Efficient, simple and there is a lot of literature and circuits for inverters.
200 kHz is in the order of what is used inside DC to DC converters.
Best wishes
David
Followed some references in the link above.
A resonant inverter would be even simpler. Here the resonant filter is part of the inverter switch mechanism.
I haven't seen a three step resonant inverter but it should be possible, and if it's not already done it would be a nice improvement.
Last edited:
What I need is a (very) pure sine at, say, 8V peak on a load of 0.6R-1.5R. The "load" is the primary of a step-up transformer of relatively small size (RM12 or RM14).
The voltages aren't high, but the currents are.
If anyone has a schematic for the resonant inverter please let me know.
The voltages aren't high, but the currents are.
If anyone has a schematic for the resonant inverter please let me know.
You would probably be better to pursue this in a switch mode power supply forum, it's a reasonably specialized topic but there is a lot of expertise.
I have only done some basic transformer power supply models in Spice and I learned this is not trivial to do accurately.
But I do have a Spice reference that covers this in some detail, as well as my reference books from university days and a few memories so I will do my best.
How "pure"? i.e. What is the distortion specification?
Not much use to have a very pure sine wave in the primary if the core non-linearities are substantial.
What core material do you plan to use, number of primary turns? and turns ratio?
Is that load the reflected load or the primary coil resistance?
Stray capacitance, leak inductance and similar second order effects actually become the dominant factors in a resonant circuit, can't just assume an ideal transformer, so we need quite a lot of information.
Best wishes
David
I thought the transformer was not an issue at all, OK, I lose about 4% voltage amplitude at full load, but that is OK. Two primary turns, 20 secondary turns, therefore 1:10. The transofrmer is an RM12, and could be the RM14 for less loss. The power amp is isolated from the load though a 20uF cap.
At full load the secondary of the transformer sees around 100R, so the primary sees around 1R.
At full load (and with the necessary bias) I get a very pure sine on the scope, both on the primary and on the secondary, I do not have a distortion analyser but it looks very clean. The simulation shows around 0.15% THD.
200KHz on a class AB power amp is a pain due to cross conduction. It resembles thermal runaway but it is not. It actually causes thermal runaway just to confuse you. There is a very large thread where we examined this problem in detail and also got a neat solution (use very low collector resistors on the complementary pairs).
A more efficient method will require a bit of a jump, in the meantime perhaps it is possible to improve your current solution.
What output transistors do you use?
Could you post the ASC if you use Spice?
Perhaps it is possible to reduce the cross conduction.
Best wishes
David
What output transistors do you use?
Could you post the ASC if you use Spice?
Perhaps it is possible to reduce the cross conduction.
Best wishes
David
- Status
- Not open for further replies.