# High frequency, low noise clamp circuit

I have an analog circuit that outputs voltage between 0-5V. A simple representative circuit is given below.

The ADC can tolerate up to 4V and the interesting signal that I like to digitize is between 0-3.3V. In other words, it is ok to read all 1's at the ADC if the signal is more than 3.3V?

I am trying to find a circuit where it doesn't introduce:

• Doesn't disturb my signal if it is not above 4V
• fast enough to deal with my 1 MHz input signal (zeners are out at least I couldn't find one that is fast enough and small noise)

Any help would be appreciated.

You specify that it "should not introduce additional noise", which rules out any possible components in the signal path, at least in a strict reading.

So, okay, we could put an LDO regulator on the op-amp power supply rail(s) and reduce its supply voltage so that it is not 'railed' at 3.3V but does saturate under all conditions at 4V. If the LDO has the same or less noise than your current supply, there will be no additional noise, and if the op-amp functions well enough with a few hundred mV of headroom, the 1MHz signal may emerge unscathed. If the particular op-amp you have in mind does not quite allow that, it's often acceptable to reduce the allowable input swing to (say) 3.0V and lose a fraction of a bit of resolution.

A clamp diode as @user50443 suggests is possible- that means you use something like a zener (say a TL431) to create a stiff supply that can sink current and that's a few hundred mV below your desired clamp voltage and use a diode (conventional switching diode or a small signal Schottky) to clamp the op-amp output through a relatively low-value resistor such as 50 or 75$\Omega$. Of course the forward voltage of a diode changes with temperature, so unless you compensate the supply you'll have some change in the clamp voltage, and Schottky diodes tend to be leaky so you'll have to do the sums as to allowable error. If you're only concerned about high frequency noise, the Schottky (eg. BAT54) +TL431 should be a good choice.

There also "clamp" amplifier chips that are designed to clamp the output at a certain level. I've never found them adequate in performance, but if memory serves they might be okay for this kind of application if precision is not a big deal.

Edit: Here is a simple clamp diode setup that will not affect signals below 3.3V significantly (you could notice a slight DC shift at high temperatures) and does not require a minimum load on your 3.3V supply. If you're sure the signal can never exceed 5V and you're sure that the load on your 3.3V supply will never be less than about 15mA (maybe you put a resistor on there) you can lose the TL431 and resistors and just connect the BAT54 to the 3.3V supply.

simulate this circuit – Schematic created using CircuitLab

• I think my favorite article on this is B. Pease's "Bonding and clamping techniques". I found a copy here, edn.com/electronics-blogs/designing-ideas/4311721/… Commented Jul 31, 2014 at 13:10
• Could you please provide a reference schematic?
– Ktc
Commented Aug 5, 2014 at 15:05
• Which option appeals to you? Commented Aug 5, 2014 at 15:47
• The clamp diode option. I cannot reduce the supply on the opamp to 3V.
– Ktc
Commented Aug 6, 2014 at 2:31
• This circuit works well. I implemented this on the bench. I have one problem somehow circuit when it is not clamping it reduces my signal about 100mV. I am suspecting the leakage from the didode causes this. Any ideas?
– Ktc
Commented Aug 13, 2014 at 14:11

I don't think that a resistor divider at the input of the OA will introduce critical noise to your signal, have you seen http://www.ecircuitcenter.com/Circuits/Noise/Noise_Analysis/res_noise.htm ? also you could implement something like in the CN-0048 http://www.analog.com/static/imported-files/circuit_notes/CN0048.pdf and measure your signal differentially if noise is a mayor concern. You only have to change the component values in the schematic of the fig. 1 to get a full range use of your adc.