High Speed low noise Instrumentation Amplifier with Disable?

Hello I need low noise high speed and high precise Instrumentation Amplifier, my first idea was any as

Figure 68. in datasheet  The 3-op-amp instrumentation amplifiers AD8620+AD8610.

Amplifier is used as differential amplifier with high resistance input. Input signal is sinus, amplitude about 1mV or 2V, frequency up to 1.2MHz Gain=1.

I thing AD8620+AD8610 would be a good choice, unfortunately, needs signals from several amps switch using something like ADG1404.

ADG1404 have crosstalk between chanmel  ~60dB. Consequently, the need to turn off the amplifier output, if not through for ADG1404.


AD8610 have not disable function, my first idea is replace AD8610 with AD8021, ot is good idea?

Any other solving?



  • Hi, wivern.

    In designing a three op amp in amp using discrete components, there are few things to consider.

    1. For a high impedance input, amplifiers with an extremely low input bias current are better. This is because the I*R drop created by the bias current and the high impedance, after multiplication of gain, will produce an almost negligible offset voltage at the output. So the actual output will be almost the same as the expected output. An option for this pre-amp is a JFET or FET input amplifiers such as ADA4627-1 and the one you're using AD8620. And for low input impedance, bipolar amplifiers are better.

    2. Low offset voltage, low offset voltage drift and low supply current are to be considered for the input stage amplifiers.

    3. Bypassing the inputs is essential for high (MHz range) gain bandwidth product.

    4. The headroom (input voltage range) must also be considered for proper operation of the amplifier.

    5. For microvolt input, a low supply such as +/-2.5V is better for the first stage in order to save power. 

    Also, can you please clarify the following to help you better?

    1. Are you setting a configuration for different gains or for only a gain=1? If you want different gains, you may use only one in amp (with three op amp topology) configurable with different gains via one external resistor such as AD8226. You may put this after ADG1404.

    2. Are you going to make 4 sets of three op amp in amp as inputs for ADG1404? I think this will cost you much and this design is not good to implement.

    3. Are you going to make one set of three op amp in amp after ADG1404? CN0122 might be of help. 

    4. If you may, can you post your schematic or even a block diagram?



  • Thanks you,

    One picture a thousand words

    The problem is with the output of both instrumentation amplifiers, dynamic voltage difference to their outputs can be up to60 db. When Channel-to-Channel Crosstalk at 60 db ADG1404.....

    I thought exactly the unmeasured amplifier somehow attenuate or disable.

  • Hi, wivern.

    I have few suggestions that you may consider for your design.

    1. Your 50ohms resistor is too low to be a load of your in amp. This will dissipate a lot of power. It would be better if you increase the resistance and reduce the 1uF capacitor in order to save power and reduce heat and thermal feedback. 

    2. May I know why do you need an in amp as a buffer?

         Since you configured the two sets of your three op amp in amp as buffer, you may try ADI's AD8066 FET input dual amps with a very high impedance and very low input bias current. You may also try AD8170 2:1 buffered multiplexer. The only available JFET op amp with disable that ADI has is ADA4817. It has 1GHz bandwidth which seems an overkill for your application. 

    3. Are you using the fourth channel of ADG1404? If not, is it grounded?

         Based on the posted schematic, you are only using 3channels of the multiplexer. The unused channel must be shorted to ground in order for it to avoid picking up stray signals and to provide DC path for the bias current. Also, floating the unused terminals of a CMOS multiplexer will degrade the performance of the ON-channels.  

    4. ADG1404's off-isolation at 1MHz is more or less 52dB which is high. You may want to replacement it with ADG1204 or ADG5204 which both have higher bandwith and, lower crosstalk and off-isolation (more or less 80dB) at 1MHz.

    5. You may also want to replace ADG1404 with two separate switches or mechanical relays which will not only fully eliminate crosstalk but also will fully turn-off your unused channels without affecting your on-channels.

    Hope this helps. If you still have other questions, just let us know.



  • Thank you for the information.

    You mentioned the size of the load that works in AD8610.

    What is the optimum load for the AD861i with  output sinus amplitude +-2.5V, powes supply -+12V?

    The datasheet is mostly spoken 1k ohm, but it seems too much for the + -2.5 V output, and in any case it is quite difficult to keep the line impedance 1k on FR4 PCB.






  • Hi, wivern.

    You don't have to worry about the matching or termination of your trace to the line impedance since you are only using 1MHz as frequency. This termination is more important in signals more than 100MHz. An unrealistic 700-inch is the minimum PCB track length you need for your signal of 1MHz so a load of more than 10kohms is reasonable enough for this trace. This high load is easier to drive than a 1kohm because of the low current requirement and also the power dissipation will be reduced.


    As additional information, the most common rule of thumb in termating the trace to line impedance is the rule of 2 inch PCB track length per nanosecond of rise or fall time. Once the trace exceeds this criterion, a termination must be implemented. The computation of the 700-inch trace of 1MHz is shown below as sample.

    For reference on terminating PCB traces with their characteristic impedance, please see MT-094.

    Let us know if you have other concerns.