precision clamping circuits exhibit awful spiking

I wish to design a circuit that may clamp an input arbitrary waveform voltage with bandwidth less than MHz. The possibility to tune the clamping voltage is a requirement, and I also require to be able to clamp from a reference voltage. I tried two designs :

• A) one called "precision clamping" with a simple diode as feedback on an op-amp
• B) the other one consists in short-circuiting the input waveform with the clamping voltage value by means of a FET

Both designs are shown in the following picture.



A simulation of each circuit exhibit awful spiking if the input signal is a square wave. I tried various op amps, with various slew rates and bandwidth, this does not change a lot. Is there something to do - better than a simple R-C filter that dramatically reduces the bandwidth - to avoid these spikings ? Some other kind of design ?

Best Regards,
Mike

There are special opAmps dedicated to limiting/clamping, s. p. 46 of this TI handbook below.

P. 47 shows a circuit similar to your set-up A .

I am not sure that your simulation circuit is correct, the signal source needs an internal impedance and must be AC coupled to the clamp, so the charge on the coupling cap carries the clamping voltage through to the next clamp pulse.
Frank

Thanks for your two answers.

eriki : I hoped there was something more "home-made" circuit than integrated solutions. I will try this and give some feedback.

Chuckey : I do not accurately understand this answer. I think you speak about a quick return path for the spike to the source, that's it ? so I should put a capacitor somewhere ?

The overshoot of clamped output voltage is mainly caused by the delayed action of the OP slewing out of saturation. That's an intrinsic disadvantage of this cicruit, when operated with fast input signals. The FET clamp circuit suffers from the same problem and most likely additionally from charge injection, particularly when using a power FET like SI2305.

Driving a high performance analog switch by a fast comparator is the concept of the Analog Devices clamping amplifier, by the way.

There's no easy solution to your problem. The dedicated clamping/limiting amplifiers from TI (or Analog) have been mentioned. My personal favourite for high-speed are non-feedback clamping circuits with voltage compensated diodes. By nature, they have a finite clamping impedance, but no overshoot.

Thanks a lot for this clear answer. I tried non feedback clamping circuits (for example replacing the Fet + op amp driver by a simple diode in my design B).
I did not know analog device's solution. I am a little bit suspicious about the quality of this solution, because if I'm not wrong analog switch are based on FETs and may exhibit spiking too, is that true ? Moreover, it's very hard to find an accurate spice model for these switches isn't it ?

I agree with you concerning non-feedback clamping circuit. However in my application (a laser diode driver) the user of the circuit should tune the limit as a "readable" parameter as preliminary set-up (he sets-up the maximum current that may go through the laser diode). I think it's difficult to display an accurate limit with clamping diode isn't it ? If you have a solution for that I'd like to know it.

To finish : do you have an idea of how the voltage limiting is achieved in the voltage-limiting op amps ?

Best regards,
Mike

In your OPA350 circuit, you have already reduced the OP output swing by choosing a high reference voltage. I guess you know, that the behaviour will be worse if you reduce the reference voltage level. You can design different feedback clamp circuits, that limit the OP output swing independent of the reference voltage, but a minimal swing of at least several 100 mV can't be avoided.

A better feedback clamp circuit should be possible with OTA, allowing faster slewing while the feedback path is open. But there aren't much available general purpose OTA devices, so this isn't a promising option for a discrete circuit design.

I assume that you mean the TI "voltage limiting" amplifiers, which perform output related clamping in contrast to the input related operation of TI devices. Unfortunately the datasheet doesn't give any details about the implementation.

FvM said:

In your OPA350 circuit, you have already reduced the OP output swing by choosing a high reference voltage. I guess you know, that the behaviour will be worse if you reduce the reference voltage level. You can design different feedback clamp circuits, that limit the OP output swing independent of the reference voltage, but a minimal swing of at least several 100 mV can't be avoided.

Click to expand...

ok I have seen this by performing various transient simulations.

FvM said:

A better feedback clamp circuit should be possible with OTA, allowing faster slewing while the feedback path is open. But there aren't much available general purpose OTA devices, so this isn't a promising option for a discrete circuit design.

Click to expand...

Yes, some years ago OTA had a great development for high frequency "op-amp-style" circuits, it was very nice. Unfortunately most high speed OTA are no more supported ...

FvM said:

I assume that you mean the TI "voltage limiting" amplifiers, which perform output related clamping in contrast to the input related operation of TI devices. Unfortunately the datasheet doesn't give any details about the implementation.

Click to expand...

Yes I thought about TI.

Yes, some years ago OTA had a great development for high frequency "op-amp-style" circuits, it was very nice. Unfortunately most high speed OTA are no more supported ...

Click to expand...

Linear LT1228 is still available. I have used it in a previous design (for a programmable slew rate generator).