Discrete chopper switch

Hi CVictor914,

It looks like you want to create a discrete chopper amplifier, which was typical in the last century, because there were no such integrated circuits. I'm correct?

Regards,

Kirill

Hi KirV,

Thank you for your reply.

Acually, chopper amplifier is not my target. I just want to use a chopper to realize the function of multiplication. For clarification,  I have an analog input signal and I want to multiply it with a sequence of 1 and 0. And the output is connected to a differential amplifier.

I know it's hard to find a discrete chopper switch. Do you have any other suggestions?

Thanks.

Victor

Hi Victor,

Yes I noticed almost the same thing I had already read, but dated December 9, which confused me a bit. But it's great that everything worked!

These devices are specifically designed for precision analog devices, and are used, in particular, in the most precision DMM in the world. They can be controlled by an external signal

It seems to me that when using a non-periodic signal, there should be no big problems if the durations of your signal allow the internal circuit to switch reliably

Hi KirV,

You are totally right!

I missed something before and now I know how to use them as a chopper. They are perfect choices.

By the way, did you use some single transistor switches? I know it will be crazy if I use 32 single switches on PCB to make 8 choppers I need. But if you know some high-performance analog switches, it would be better. I can use them as alternatives or use them to compare with LTC1043 and 6943.

Thanks a lot for your help!

Victor

Hi Victor,

Note how LTC1043 and 6943 are used in conjunction with capacitors - only the difference of the injected charges can create a voltage on the capacitor. Even a significant but identical injected charge applied to both capacitor plates does not affect performance.

The same unique property has a differential integrator - if the same charge is injected into both inputs, it will cause a common-mode shift of the output voltages, which is not a useful signal. Only the difference of the injected charges matters.

I find it difficult to advise you any switch. The difference of the injected charges is important. You can make individual selection of components or add a final adjustment of this parameter using external components. But it's not the best choice:)

Even more difficult is that the integrator has a small input resistance and the internal resistance of the switch can make an error, in addition, this resistance is nonlinear. What is the value of the input resistors and input currents of your integrator? This is important for selection

Regards,

Kirill

Hi Victor,

The product tables are sorted by parameters, and I tried to select the parameters important to you here

www.analog.com/.../10624

It is a simple interface, suitable contact configuration and minimal injected charge

I think I know why your message was marked as spam - you gave links to third-party resources and products that are prohibited by the forum rules.

Regards,

Kirill

Hi Kirill,

Thanks for your really helpful advice!

This is my first thread in this forum and I will obey the forum rules. Thanks for noticing me. And so lucky to disscuss with you and learn many stuff from you.

This chopper is for a function prototype PCB. Optimization will be at next stage. I think I have already known which switch I should use.

Thank you.

Victor

Thank you Victor!

By the way, which signal goes through the chopper-is it current or voltage? Does the signal source have a high or low internal resistance?

Regards,

Kirill

Hi Kirill,

In my case, signal goes through the chopper is the voltage from a single-to-differential amplifier. Here, I plan to use AD8138. There is no illustration of output resistance in AD8138 datasheet. But from the output voltage swing and output current, it should be in the order of several tens ohm. Could I estimate the ouput resistance by this way?

For the integrator, I plan to use LTC 1992 you recommended in another thread as the fully differential amplifier. Its input resistance is 500MΩ. For input resistor I will use, its resistance can be tuned from several hundreds to several thousands ohm. Input current for integrator is 10-100μA.

Do you have some comments for my choices?

Thanks a lot.

Victor

And if I use LTC6943 as one chopper, from my understanding of the datasheet, https://www.analog.com/en/products/ltc6943.html, I need to short the terminals, for example, 9 and 10. But in the datasheet, they are usually connected with a capacitor. So I am not very sure. Could I short them?

Hi Victor,

If you don't mind, I'll study the LTC6943 datasheet myself, it might take a while, okay?

You mean this table?

It is not necessary to determine the output resistance with it, because the short-circuit current and the output voltage in the absence of a load are indicated here. You can refer to this graph

Apparently, the output resistance in differential mode will be equal to twice the value on the graph. By the way, why do you use this driver? Should it be high frequency for your application?

Regards,

Kirill

Hi Kirill,

Of course, I'd like to know your opinions after you studying the LTC6943 datasheet.

I didn't find this graph of output resistance. Thank you for reminding. "Ctrl+F" is a good shortcut that I often forget to use...

The reason why I plan to use AD8138 is that I need to convert an analog voltage to differential voltage. You can refer to the figure below which is my basic idea. I want to drive each capacitor with a random bit stream (about 5MHz). Then the top-plate voltage is converted and amplified to a differential voltage.

Actually I think lower input impedance of single-to-differential amplifier is better considering the settling error (total capacitance of capacitor array is about 2μF). But the single-to-differential amplifiers I find in ADI product list https://www.analog.com/en/products/amplifiers/adc-drivers/single-ended-differential-amplifiers.html are with high impedance.

Does my point of view make sense? What do you think about the parameters I should care about S2D amplifier? I'd like to hear from you.

Thank you so much Kirill.

Victor

Hi Kirill,

Of course, I'd like to know your opinions after you studying the LTC6943 datasheet.

I didn't find this graph of output resistance. Thank you for reminding. "Ctrl+F" is a good shortcut that I often forget to use...

The reason why I plan to use AD8138 is that I need to convert an analog voltage to differential voltage. You can refer to the figure below which is my basic idea. I want to drive each capacitor with a random bit stream (about 5MHz). Then the top-plate voltage is converted and amplified to a differential voltage.

Actually I think lower input impedance of single-to-differential amplifier is better considering the settling error (total capacitance of capacitor array is about 2μF). But the single-to-differential amplifiers I find in ADI product list https://www.analog.com/en/products/amplifiers/adc-drivers/single-ended-differential-amplifiers.html are with high impedance.

Does my point of view make sense? What do you think about the parameters I should care about S2D amplifier? I'd like to hear from you.

Thank you so much Kirill.

Victor

Hi Victor,

Unfortunately, I attached the wrong link to the parametric search for switches, note this, there are 25 parts, I have already fixed this. 

Do you want to get pulse differentiation or do you want to transmit the tops of the pulses lossless? If you want differentiation, it makes sense to have low resistance on the input side of the amplifier to get a small time constant. If your goal is the opposite, all reasoning is reversed.

It's not hard to get a low input impedance just by adding an external resistor. Getting high resistance from an unsuitable amplifier is much more difficult.

Regards,

Kirill

Hi Kirill,

Thanks for being so meticulous. I have opened the link that you re-edited.

Could you explain more about the reason why it should have low resistance for getting pulse differentiation. What did you mean by differentiation? Actually, the top plate voltage will change after the PRBS is input to bottom plate. I just want this signal converted to differential signal. I must misunderstand something...

Victor

Hi Victor,

I found on page 4 an interesting note

You can select multiple contact configurations that meet or do not meet this requirement. In any case, both sections have the same clock signal and you will have to apply one LTC6943 to each integrator. And in any case, it goes beyond typical applications and will be very cool if you get great results. I can suggest such a configuration as an example, but it is not the only way

I saw analog switches that have four keys and four separate control contacts, and at the same logical level on all these contacts, two keys are closed while the other two keys are open. It's like it's made for chopper:)

There are a few remarks about the injected charge. For LTC6943, the condition of proximity of the input signal to half of the supply voltage must be fulfilled - only in this case the charge is minimal. For analog switches it is necessary to study each datasheet of the device of interest. As a typical example:

You want to apply to the integrator 100 μA, please note:

Regards,

Kirill

Hi Victor,

The capacitor together with the input impedance of the amplifier form a differentiating RC circuit. If the circuit time constant is small, only short pulses are input to the amplifier.

To transmit the pulse without distortion, the system must have infinite bandwidth. But here the high-pass filter is formed and it is capable to transfer only edges of impulses.

For un-distorted transmission of pulse tops and bases, the ability of the system to transmit very low frequencies, up to DC.

Depending on the time constant of the input circuit, you will get the desired pulse shape at the amplifier output.

But note that it is necessary to provide a path for the bias current to flow, otherwise it will not work. Your simplified diagram does not show this.

Regards,

Kirill

Hi Kirill,

Thank you so much for so detailed explaination.

May I ask what simulation tool you used?

Victor

Hi Victor,

This is LTSpice

https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html

And I would like to contact you via private messages. Can you activate this option in your profile?

Regards,

Kirill

Hi Kirill,

It's done.

Thanks.

Victor