Hi,

I want to make a pulse amplifier. I want to amplify a pulse of 0 to 5 Volts at a frequency of 100 kHz. The output of the cirquit should be +10 Volts top level and -10 Volt low level.

Now you might be thinking, that ain't too hard. Well it isn't!

I've found an opamp that does exactly what I want, downside is that the opamp can't handle a load. I want to add a capacitive load from let's say 1nF to 10 nF. The opamp just can't handle that.

I've tried to build a push-pull cirquit, but that didn't work. So I've decided to work with MOSFET's. I tried several things, but I can't seem to get it right.

Does anyone have an idea on how to fix this?

Kind regards,
ZweedNaat

 

It is true that many op-amps don't like capacitive loads, but you can "isolate" the capacitance with maybe 50 ohms if that doesn't mess up your slew rate.

math..math..math...Yep, that's going to mess with your speed.
So what are you working with for power voltages? What's the op-amp number? What's the allowable rise time?
Have you thought using a gate driver chip without the mosfet?

 

If this can not help you here it WILL come in handy for you sooner or later.

 

Hi,

I want to make a pulse amplifier. I want to amplify a pulse of 0 to 5 Volts at a frequency of 100 kHz. The output of the cirquit should be +10 Volts top level and -10 Volt low level.

Now you might be thinking, that ain't too hard. Well it isn't!

I've found an opamp that does exactly what I want, downside is that the opamp can't handle a load. I want to add a capacitive load from let's say 1nF to 10 nF. The opamp just can't handle that.

I've tried to build a push-pull cirquit, but that didn't work. So I've decided to work with MOSFET's. I tried several things, but I can't seem to get it right.

Does anyone have an idea on how to fix this?

Kind regards,
ZweedNaat

1. What are you going to be driving with this thing / how much current must it source/sink?
2. Will the +10 volt and - 10 volts times be equal?

 

Hi,

I want to make a pulse amplifier. I want to amplify a pulse of 0 to 5 Volts at a frequency of 100 kHz. The output of the cirquit should be +10 Volts top level and -10 Volt low level.

Now you might be thinking, that ain't too hard. Well it isn't!

I've found an opamp that does exactly what I want, downside is that the opamp can't handle a load. I want to add a capacitive load from let's say 1nF to 10 nF. The opamp just can't handle that.

I've tried to build a push-pull cirquit, but that didn't work. So I've decided to work with MOSFET's. I tried several things, but I can't seem to get it right.

Does anyone have an idea on how to fix this?

Kind regards,
ZweedNaat

Aa op-amp will usually have restrictive slew rate limitations - a fast comparator might be better.

Many comparators have open collector outputs and generally use an external pull up resistor. The first step is to casc-ode the output, which is easiest to do with a grounded gate MOSFET - but you can use a BJT if you can arrange a suitable base voltage. The casc-ode bit doesn't give you any current gain, but its damn fast and you won't have any difficulty getting voltage gain. By this time - you've got the same configuration as an average video amplifier. A high definition monitor just had a complementary pair emitter follower on the end to drive the capacitive load of a CRT cathode.

AFAIK: You have to put the complementary emitter follower outside the feedback loop, or you're back where you started.

 

It is true that many op-amps don't like capacitive loads, but you can "isolate" the capacitance with maybe 50 ohms if that doesn't mess up your slew rate.

math..math..math...Yep, that's going to mess with your speed.
So what are you working with for power voltages? What's the op-amp number? What's the allowable rise time?
Have you thought using a gate driver chip without the mosfet?

I am using a THS4081. The slew rate I want is 100V/us. I have tought about it and I have a few n-channel driver laying around, but you can't generate the negative voltage with that can I? I need to buy a p-channel driver extra to do that I reckon?

 

If this can not help you here it WILL come in handy for you sooner or later.

Read through it real fast. Might come in real handy indeed. It's nice to have so much good information in the same place! Much appreciated.

 

1. What are you going to be driving with this thing / how much current must it source/sink?
2. Will the +10 volt and - 10 volts times be equal?

1. I just want to put a capacitive load on the cirquit. Didn't really think about currents running through it.
2. Depends on the input pulse. I want it to follow the input pulse. When the input pulse high (5 Volt), I want the output to be +10. You probably guessed what I want when the input is low (0 Volt).

 

Aa op-amp will usually have restrictive slew rate limitations - a fast comparator might be better.

Many comparators have open collector outputs and generally use an external pull up resistor. The first step is to casc-ode the output, which is easiest to do with a grounded gate MOSFET - but you can use a BJT if you can arrange a suitable base voltage. The casc-ode bit doesn't give you any current gain, but its damn fast and you won't have any difficulty getting voltage gain. By this time - you've got the same configuration as an average video amplifier. A high definition monitor just had a complementary pair emitter follower on the end to drive the capacitive load of a CRT cathode.

AFAIK: You have to put the complementary emitter follower outside the feedback loop, or you're back where you started.

I may have tried this idea already, not sure. The thing is, I think that the circuit I have now with a opamp (THS4081) just can't deliver enough power. It is fast enough. I wanted 100V/us and right now I have 100v/0.7us, but I cant add a load. That's the whole problem really. I just want to add a capacitive load.

Thanks for thinking with me!

 

I am using a THS4081. The slew rate I want is 100V/us. I have tought about it and I have a few n-channel driver laying around, but you can't generate the negative voltage with that can I? I need to buy a p-channel driver extra to do that I reckon?

Your circuit configuration might allow a bootstrap rail for voltages outside the main rails.

Application notes for MOSFET drivers like the range from IR are probably as good a starting place as any.

 

1. I just want to put a capacitive load on the cirquit. Didn't really think about currents running through it.
2. Depends on the input pulse. I want it to follow the input pulse. When the input pulse high (5 Volt), I want the output to be +10. You probably guessed what I want when the input is low (0 Volt).

1. It's not a "cirquit", it's a circuit.
2. How large will the capacitance be, and is there series or parallel capacitance associated with the load?
3. OK, then, what's the duty cycle of the input pulse train and how fast do you want the capacitive load to charge and discharge?
4. What's up with that attitude?

 

I may have tried this idea already, not sure. The thing is, I think that the circuit I have now with a opamp (THS4081) just can't deliver enough power. It is fast enough. I wanted 100V/us and right now I have 100v/0.7us, but I cant add a load. That's the whole problem really. I just want to add a capacitive load.

Thanks for thinking with me!

Complementary emitter follower will give a current boost - but slow and you lose a couple of Vbe out of the headroom.

Insert a casc-ode stage to compensate for the slow follower and you can run it off a bootstrap rail to get the pulse up to Vcc - but it still won't go completely down to 0V unless you use "pulse shaping" tricks of the trade.

 

Does it have to be a linear amplifier? A pulse driver circuit with saturating output transistors can be very fast, but variations in the input amplitude and edge times are lost.

ak

 

Does it have to be a linear amplifier? A pulse driver circuit with saturating output transistors can be very fast, but variations in the input amplitude and edge times are lost.

ak

I vaguely remember a CMOS chip containing 6 MOSFETS, 4007 I think - there was an RCA application note using this as a power buffer for one of their CMOS op-amps.

Whether the parts can still be got is another matter......

 

Does it have to be a linear amplifier? A pulse driver circuit with saturating output transistors can be very fast, but variations in the input amplitude and edge times are lost.

ak

No it doesnt have to be a linear amplifier. I don't want to lose any time on the rising and falling edge. I've build bootstrap thingy like Ian suggested with MOSFET's but without drivers. I think what happens here is that some times the MOSFETS are open simultaneously, which basically means the circuit doesn't work.

 

Complementary emitter follower will give a current boost - but slow and you lose a couple of Vbe out of the headroom.

Insert a casc-ode stage to compensate for the slow follower and you can run it off a bootstrap rail to get the pulse up to Vcc - but it still won't go completely down to 0V unless you use "pulse shaping" tricks of the trade.

Sorry for replying so late, I've been very busy lately. I've tried the emitter follower principal, but it is as you say. Also I've taken a look at the IC you are talking about in your other reply. Looks try-able. I'll give you a heads-up when I've found something.

 

I am using a THS4081. The slew rate I want is 100V/us. I have tought about it and I have a few n-channel driver laying around, but you can't generate the negative voltage with that can I? I need to buy a p-channel driver extra to do that I reckon?

Look at the National LH0063 buffer. Also, the ap notes in the National linear handbook shows an application that uses that part to do what you are looking for. Be sure to follow the current limit advice given in the ap note. That buffer can put out amps of current at tremendous slew rates!

 

Hi,
Below I have pasted the screendump of a circuit which I've simulated in LTSpice. It looks as though it will do what you want. I am assuming your input signal is bipolar since you want a bipolar output. The first stage is a differential amplifier and the second a non-inverting booster. As you can see from the screendump the output voltage into the capacitor ranges fron +9.5 to -11.3. There is a phase difference between input & output but that is unavoidable. You didn't specify zero plase difference in your original spec. Phase compensation would add more complexity to the circuit. I have omitted bypass capacitors in the simulation - you will have to put those into the physical circuit to remove power line noise, as the amplifier gain is very high. As you can see, the CA3140 opamp is sufficiently fast for this application and is much cheaper than using the much faster slew rate op-amps which you mentioned in post #9. I have used the largest load capacitor which you specified (10n) to confirm that the output voltage meets your spec.

I also tried using comparators because they are much faster than op amps as other posters have mentioned. But their open collectors make getting a bipolar output difficult.

 

Hi,
Just for comparison I removed the differential amp and duplicated the non-inverting anp, giving two cascaded non-inverting amps. There is some loss in signal fidelity but since this is a pulse amp, that may not be that important:

 

See