Hi all,
I am new to D class stuff, so forgive me if the question is stupid.
I have an application where I need to drive a speaker-similar load (1 Ohm ) with a sine wave of about 5-10 KHz constant frequency. The problem is that the mean value of the sine get shifted up or down so there is a good DC content in output. Peak to peak values never exceed 50 V either side.
Actually we are using straight class B amplifiers (something like 10 TO3 pairs). Would it be possible to do it more efficiently with a class D?
I am new to D class stuff, so forgive me if the question is stupid.
I have an application where I need to drive a speaker-similar load (1 Ohm ) with a sine wave of about 5-10 KHz constant frequency. The problem is that the mean value of the sine get shifted up or down so there is a good DC content in output. Peak to peak values never exceed 50 V either side.
Actually we are using straight class B amplifiers (something like 10 TO3 pairs). Would it be possible to do it more efficiently with a class D?
No problem for a class D amp to do that, and save a bit of power.
Got any specs for the application, like a maximum THD level of the sine wave?
Got any specs for the application, like a maximum THD level of the sine wave?
...means basically, yes. But take care with supply rail pumping.
Solution a) Go for bridged configuration or a full bridge design.
Solution b) If you need the load between amp out and GND then use an unbridged amp, but power the classD amp from an SMPS with synchronous rectifiers.
P.S.
Note that class D amps do have an superimposed HF ripple on the output.
Usually in the range between 300mVrms...1Vrms. Typical frequencies of such ripple are ranging between 150kHz...500kHz.
If this is Ok for your load, then a high current class D amp should work to power it.
Thanks to all for the kind answers. The thing starts to lok feasible.Then:
BTW, for me HF content is not a big deal, and THD = 1% is OK. As I said, we are using pure class B stuff, with some ugly crossover distortion.
Then, full bridge could be a problem, since the load is usually referred to ground (really the amplifier has a current feedback sensing a 0.1 ohm shunt to the ground) What do you mean with a sync. rectifier? That the clock of the SMPS should be in sync with the amplifier (and maybe pi out of phase)?
And ever seen a D class connected in current feedback?
Yeah, *very* low, in the sense that it can stay shifted (i.e. +10 V) for 1/2 hour. Prcatically a true RMS meter always reads a lot of DC on the output.
BTW, for me HF content is not a big deal, and THD = 1% is OK. As I said, we are using pure class B stuff, with some ugly crossover distortion.
Then, full bridge could be a problem, since the load is usually referred to ground (really the amplifier has a current feedback sensing a 0.1 ohm shunt to the ground) What do you mean with a sync. rectifier? That the clock of the SMPS should be in sync with the amplifier (and maybe pi out of phase)?
And ever seen a D class connected in current feedback?
If the SMPSU uses synchronous rectification (using MOSFETs) instead of diodes for the output voltage supply, then it may be able to absorb in an almost lossless manner the back-EMF caused by 'bus pumping' in the half-bridge class-D amplifier.
I guess by 'current feedback' you mean that the output of the amplifier is a regulated current into the load, rather than a regulated voltage across the load. The answer is yes, this is quite possible. I have made a class-D amplifier using a self-oscillating hysteretic output stage with the sensing signal for the hysteretic PWM modulator being the voltage developed across the current-sense resistor in series with the load.
If you don't need a very high frequency for the PWM oscillation, you could use a full-bridge output where the required current-sensed feedback is produced by a Hall-effect current sensor, such as the Allegro ACS712 series.
If you don't need a very high frequency for the PWM oscillation, you could use a full-bridge output where the required current-sensed feedback is produced by a Hall-effect current sensor, such as the Allegro ACS712 series.
Output of a bridged amp (and current sense resistor) can be grounded also, just the power supply have to be floated.
Hall-effect current sensors are very noisy (even the expensive LEM ones too), plus open loop ones like ACS712 are very inaccurate (and slow). Resistor is much better!
Hall-effect current sensors are very noisy (even the expensive LEM ones too), plus open loop ones like ACS712 are very inaccurate (and slow). Resistor is much better!
The only problem with a class d and DC is the residual carrier you get on the output.
My 2092 design has a few volts of residual carrier at the output.
My 2092 design has a few volts of residual carrier at the output.
Just as a sidenote: If an application requires better suppression of the carrier fundamental than the ordinary 2nd order lowpass offers then there are of course some options available.
Nosey question: Will the amp be used for some kind of positioning system ?
Regards
Charles
Nosey question: Will the amp be used for some kind of positioning system ?
Regards
Charles
Thanks all for the intersting suggestions. I believe I wouls stay half-bridged since to float the load or the PS needs a lot of work that I am not sure it is worth to do.
I am even thinking that I do not need an amplifier. It would probably make sense to just synthethize the (shifted) sinusoid just accordind to a digital clock in in input. The DC level would be the other input. AC amplitude is constant once choosen. Feasible?
I am even thinking that I do not need an amplifier. It would probably make sense to just synthethize the (shifted) sinusoid just accordind to a digital clock in in input. The DC level would be the other input. AC amplitude is constant once choosen. Feasible?
You are correct, Charles. The amplifier is meant to move an electromagnectic actuator, that is *very* linear with the amount of current supplied to it. The residuals at hf are not a problem since the actuator itself has a quite narrow band. Outside the current loop there will be a position loop, in a very similar way of how you control a DC motor.
I don't think that a proper synchro-rectifier would be simpler to design. It can be quite tricky! (Especially if this is the first time you heard about it.)
If an exact DC balance doesn't have to be maintained some charge-pump could be used for instance for the DC balancing act.
Regards
Charles
Regards
Charles
Then the questions then are:
Cannot the actual massive linear (transf-bridge-Capacitors) PS, about 2kW with 50 mF x rail, suffice, no way?
Would not some normal improvement (i.e. a PI filter) help, no way?
Must the eventual SMPS be specially bult in order to be synchronous with the amplifier?
What is a switch mode converter. IMHO 2 identical half bridge is simpler than 1 half bridge (twice as powerful as in the first case) + an other different switching regulator (charge pump or whatever).
If you want to use the original PSU, then Charles has a point, an additional charge pump or similar thing can help. Filter is not enough.
But the transformer may be configured to parallel secunders instead of the present series connection, and full bridge may be used. The PSU current capability gets higher this way!
Ok understood that full bridged has many advantages, that may overcome the cost of lifting the load.
But is it a bridged configuration really able to go down to 1 Ohm? Now rails are +-50V, even paralleling the transformer secondary coils an going for only 0-50, would it be feasible?
BTW I never saw yet a D class with parallel OPs , I belive because current sharing here can be a issue (the usual .22 Ohm source resistor will be way to big here I guess).
But is it a bridged configuration really able to go down to 1 Ohm? Now rails are +-50V, even paralleling the transformer secondary coils an going for only 0-50, would it be feasible?
BTW I never saw yet a D class with parallel OPs , I belive because current sharing here can be a issue (the usual .22 Ohm source resistor will be way to big here I guess).
- Status
- Not open for further replies.