I am using this as a single thread for my questions relating to a current driver I am making - I'm trying to design a simple class G output stage for driving current into a coil*, this circuit is used inside a larger circuit. I am trying to use an op-amp to feed the output stage directly as I don't need any signal gain.
Having amended the circuit as per stigigemla's advice I now have a working circuit but I have a couple more problems -
* I have added amended the circuit pics to reflect the changes I have made and to clean up the confusing power supply labelling *
Having amended the circuit as per stigigemla's advice I now have a working circuit but I have a couple more problems -
- The response is much lower bandwidth than I was expecting - the bandwidth of the output devices is listed as 1MHz, but I get roll off around 10KHz - what dictates the roll off of this amplifier and how can I choose transistors to push the roll off up to around 150KHz (if possible?)
- When testing the real circuit in the flesh there is an oscillation of a couple of hundred mV overlaid on the output. It doesnt seem to be feedback in the global feedback loop, because changing C1 and C2 dont seem to affect it, but when I cut the global feedback it no longer appears. The real circuit does not yet have the 0.22R emitter resistors which I will add in the next few days when they arrive.
- There is a large spike in the output when the outer transistors turn on, what is causing this?
* I have added amended the circuit pics to reflect the changes I have made and to clean up the confusing power supply labelling *
//Original question//
As far as I can see the circuit looks OK but in simulation the outer transistors never turn on and the output is clipped at the inner voltage rails - could someone help me understand why, and how I rearrange my circuit to give the desired operation?
any help much appreciated!
* I realize that a lot of designs use PWM for this, but I need to use direct current here.
Last edited:
Thanks! Silly of me not to notice that. If I raise R26 / R30 to 2.2K, does this seem like an acceptable value?
I am concerned about the correct biasing of the transistors - although the simulation shows around +/- 630mV on the base of Q3 / Q16 with 0 input voltage, the current through the emitter resistors R13 / R24 is only about 7nA - do I need to change this?
I am concerned about the correct biasing of the transistors - although the simulation shows around +/- 630mV on the base of Q3 / Q16 with 0 input voltage, the current through the emitter resistors R13 / R24 is only about 7nA - do I need to change this?
Where does V and V+ top left go?
The opamp power supply is going to the outer transistors. Is that right?
The opamp power supply is going to the outer transistors. Is that right?
Not sure if this was the reason / motivation for a Class-G stage but PWM could also be used to inject direct current (DC) into a coil the same way it's used for AC.
I cut this down from a larger design to make it clear - the unused power supply can be ignored. The rails are +/-10V and +/-20 . The opamp label is a bit confusing but its correct
- and yes I realise about PWM but we cant use it here
I followed the link to the one you cut down.
Did you cut the zena diode voltage down in the same proportion?
Did you cut the zena diode voltage down in the same proportion?
The hi voltage transistors should be mosfets now as they are easy to drive off the speaker output. A 5.6v zener and small resistor to high rails and cap on zener would do it.
stocktrader200 could you elaborate a little on that - you think I should replace the TIP35 / TIP36 outer pair with mosfets? and then bias them with the arrangement you describe - what would this achieve?
I have changed the bias resistors and now the circuit is no longer clipped. I now have a couple more problems -
The linked file should now be exactly the same as the picture, I have amended them. I did not change the zener voltage, do I need to? In what proportion?
I have changed the bias resistors and now the circuit is no longer clipped. I now have a couple more problems -
- The response is much lower bandwidth than I was expecting - the bandwidth of the output devices is listed as 1MHz, but I get roll off around 10KHz - what dictates the roll off of this amplifier and how can I choose transistors to push the roll off up to around 150KHz (if possible?)
- When testing the real circuit in the flesh there is an oscillation of a couple of hundred mV overlaid on the output. It doesnt seem to be feedback in the global feedback loop, because changing C1 and C2 dont seem to affect it, but when I cut the global feedback it no longer appears. The real circuit does not yet have the 0.22R emitter resistors which I will add in the next few days when they arrive.
- There is a large spike in the output when the outer transistors turn on, what is causing this?
I have looked at the amended circuit and looks like the drive will be clipped to 2XVBE of the inner transistors.
The op-amp will try to drive the load via the BE junctions of the inner transistors.
The outer transistors need to kick in when the drive is over 3XVBE above the inner transistor power supply as there is an extra diode in the base.
The op-amp will try to drive the load via the BE junctions of the inner transistors.
The outer transistors need to kick in when the drive is over 3XVBE above the inner transistor power supply as there is an extra diode in the base.
If you tell us more about the coil and how it should be driven i guess we will give you a better schematic.
Sine ẃave, pulse or other waveforms? Continous or at intervals? Current and voltage?
Sine ẃave, pulse or other waveforms? Continous or at intervals? Current and voltage?
That would be much appreciated, Im fairly experienced with electronics but I'm completely new to these circuits.
I am trying to drive DC current round a coil of around 10uH, the circuit has to be a transconductance stage - for a voltage V of 0-5V there is current of 0-10A in the coil. The circuit will hold state for an indeterminate amount of time (1mS second, 1 hour, two days?) then change to a new value, but must change within 20uS, or as fast as possible.
My intention is to make a class G stage with proportional - derivative feedback which overshoots the voltage and makes the current through the coil reach maximum value faster. The class G part is useful as the voltage can swing high when the current changes but in its steady state the voltage rails are lower so there is less heat being generated in the output devices.
Clearly im getting a muddled with all the different possibilities of the circuit, it feels like there is some small change I can make so I can drive this with an opamp without getting the input voltage clamped? I can see now how the output load is referred to the input of the darlington (thanks Refugee1). It feels like there's some magic combination just with just a couple of extra components (im hoping)
I am trying to drive DC current round a coil of around 10uH, the circuit has to be a transconductance stage - for a voltage V of 0-5V there is current of 0-10A in the coil. The circuit will hold state for an indeterminate amount of time (1mS second, 1 hour, two days?) then change to a new value, but must change within 20uS, or as fast as possible.
My intention is to make a class G stage with proportional - derivative feedback which overshoots the voltage and makes the current through the coil reach maximum value faster. The class G part is useful as the voltage can swing high when the current changes but in its steady state the voltage rails are lower so there is less heat being generated in the output devices.
Clearly im getting a muddled with all the different possibilities of the circuit, it feels like there is some small change I can make so I can drive this with an opamp without getting the input voltage clamped? I can see now how the output load is referred to the input of the darlington (thanks Refugee1). It feels like there's some magic combination just with just a couple of extra components (im hoping)
Do you need positive and negative current or are you driving the coil in one polarity only?
Do you have need for back induction voltage if it is going in both or one direction.
(10A in 10 uH gives an induction of 100v x microsecond)
Do you have need for back induction voltage if it is going in both or one direction.
(10A in 10 uH gives an induction of 100v x microsecond)
Yes positive and negative is preferred as it needs to both charge and discharge. Back EMF will not be an issue here, at least to start with.
I forgot to say before, the load is a 0.5R resistor in series with the 10uH inductor
I forgot to say before, the load is a 0.5R resistor in series with the 10uH inductor
If i have guessed right of what your are doing this will do. It is not hifi but i think the precision will be more than enough.
I would use high power surface mounted output transistors on a small aluminium board glued to a CPU cooler.
I dont know of your demands but ordinary VAPO PC CPU coolers last about 3 years in a dusty industrial environment.
If You need less service you ought to use ball bearing fans and air filters.
For hole mounted transistors they ought to be parallel connected but then we loose a few volts in the high freqency area. It is possible that you need to raise R4 to about 9 k ohm to get it to work in a simulation but in the real world 4,7k is the correct value. With 0,5 ohms as sense resistor R13 it needs to be 50 watt but you can use a smaller resistor and divide down the input voltage. The OPA 2156 has very low offset voltage but if you use soft ware adjusting you can use any cheaper OP that manages 30v power supply.
It is double and if you have no use of the other section i is just to connect it as a voltage follower with the input to ground.
I think you may need an overheating control but i guess you have that in the software.
I would use high power surface mounted output transistors on a small aluminium board glued to a CPU cooler.
I dont know of your demands but ordinary VAPO PC CPU coolers last about 3 years in a dusty industrial environment.
If You need less service you ought to use ball bearing fans and air filters.
For hole mounted transistors they ought to be parallel connected but then we loose a few volts in the high freqency area. It is possible that you need to raise R4 to about 9 k ohm to get it to work in a simulation but in the real world 4,7k is the correct value. With 0,5 ohms as sense resistor R13 it needs to be 50 watt but you can use a smaller resistor and divide down the input voltage. The OPA 2156 has very low offset voltage but if you use soft ware adjusting you can use any cheaper OP that manages 30v power supply.
It is double and if you have no use of the other section i is just to connect it as a voltage follower with the input to ground.
I think you may need an overheating control but i guess you have that in the software.
Last edited:
thanks, that's interesting but it is too much heat for what I am designing - I have a current design which works but produces way too much heat. I was particularly trying a class G however because I want to drop the power during the steady state. Is such a design possible to implement using FETs? I havent seen any designs from searching the web..
Class G with fets and so low voltage is difficult.
I believe Class D would be the heat solution. I think to mod Hypex or similar modules is impossible because they have no second source and will be reolaced to new models almostevery year.
Then it is just to start reading. TI has some interesting files. One example.
https://www.ti.com/video/series/power-solutions-for-class-d-audio-amplifiers.html
I believe Class D would be the heat solution. I think to mod Hypex or similar modules is impossible because they have no second source and will be reolaced to new models almostevery year.
Then it is just to start reading. TI has some interesting files. One example.
https://www.ti.com/video/series/power-solutions-for-class-d-audio-amplifiers.html
OK, interesting. What as you see it is the limiting problem with the original design i posted (apart from my mistakes in the schematic) ? Is it too low voltage to make class G work? Is the load too small? Is it incompatible with using an opamp to replace the first stages? As far as I could see that design was very close to working in exactly the way I wanted.
The problem with mosfets is that the VGS is 7 to 10 volts for 10 A with low VDS.
That means we need extra amplifiers for the Gate voltages and maybe extra voltage source.
The switching point would theoretically be 12,5v with 20v full voltage and 10 A max current.
Max power 75w in one or the other transistor. I think you need fans.
The rise time will be 6us with 10uH.
With 15 volts you have a switching point of 10v for 10A max.
Max power 50w in one or the other transistor. I think you still need fans.
The rise time will be 10us with 10uH.
With 10 volts the switching point will be 7,5 volt for 10A.
Max power 25w in one or the other transistor.
The rise time will be 20us with 10uH.
Maybe you can work with lower voltage?
The resistance for full conducting mosfets will be under 10 milliohm so i dont count with that.
That means we need extra amplifiers for the Gate voltages and maybe extra voltage source.
The switching point would theoretically be 12,5v with 20v full voltage and 10 A max current.
Max power 75w in one or the other transistor. I think you need fans.
The rise time will be 6us with 10uH.
With 15 volts you have a switching point of 10v for 10A max.
Max power 50w in one or the other transistor. I think you still need fans.
The rise time will be 10us with 10uH.
With 10 volts the switching point will be 7,5 volt for 10A.
Max power 25w in one or the other transistor.
The rise time will be 20us with 10uH.
Maybe you can work with lower voltage?
The resistance for full conducting mosfets will be under 10 milliohm so i dont count with that.
Last edited: