Hi Aindrila,

with an analog signal, there seem to be two possibilities:

- Using a bucket chain, either with several outputs, one of which is selected by an analog multiplexer, and/or clocked by a variable frequency. Suitable ICs might be TDA1022, MN3005, MN3007, MN3102, MN3205, MN3207. However, all of these seem to be available exclusively from China nowadays; I have no experience with them.

- Digitizing the signal, delaying the digitized signal, and converting it back to analog by a DAC. This could be done with a microcontroller or a DSP. If the (input) signal had always the same parameters, it would suffice to recognize the start of the input burst, and to output (the digitally stored) burst after a delay.

In both cases, you cannot vary the delay from exactly 0 to 2 ms. If this is necessary, you have to delay the other signals by a constant time, say, 0.5 ms, and delay the burst by 0.5 ms to 2.5 ms, resulting in a relative delay of 0 to 2 ms.

Hope this helps.

Thank you sir Joerg Fricke . we are trying to do the second method as you mentioned by converting the signal by arduino uno. And we don't need to vary the delay.just we want to put a fixed delay of 2 ms to one signal. in that case, is there any chance of changing pulse shape? if pulse shape will change then we can't extract the micro volt signal from the pulse.

Hi Aindrila Sinha ,

how high a resolution in voltage and in time do you need? By digitizing, the continuous signal will be replaced by a "staircase" signal, and it is important to know the maximum permissible height (voltage) and length (time) of the steps.

In any case, it is convenient to have both the analog-to-digital converter (ADC) and the digital-to-analog converter (DAC) integrated in the microcontroller. So, you should not use the ATmega328P (used in Arduino) because it is lacking a DAC, and digital-to-analog converting by low pass filtering a PWM signal yields only good results at low signal frequencies (some 100 Hz at best).

If you prefer the AVR family, and don't want to develop an own board, I would recommend something like this one:

https://www.mouser.de/ProductDetail/Microchip-Technology-Atmel/ATXMEGAA3BU-XPLD?qs=950R68cwx9dz8sU4C7PasQ==&vip=1&gclid=EAIaIQobChMIsdmjxqfJ5AIVmeF3Ch0rNwU_EAYYASABEgJo1vD_BwE

It has a 12-bit ADC and a 12-bit DAC with a maximum of 1 MegaSamples / s, i.e., a time resolution of 1 microsecond. You can develop programs and download them to the board using the software "AVR Studio" which is free.

Hi Joseph Charles French

I'm not sure if I understand your term "information conservation" correctly. Certainly, some information is lost in the ADC but the crucial point seems to be whether the remaining information suffice in the framework of this application.

Joerg Fricke Sir the pulse we are using that is of 50-100 mV and the actual signal coming from sample is in that pulse which is order of 100 micro V. I want to extract that 100 micro V signal from that pulse by subtracting method in lock in amplifier. So this small signal can be preserved in ADC ? in the case of digitalization can we get a voltage resolution of few micro V?

Aindrila Sinha If I understand correctly, your "carrier" has an amplitude of 100 mV and is modulated with a signal whose resolution is about 100 microV. With a 12 bit ADC, there are 4096 different output values possible. Converting the input signal in the range -100mV to +100mV yields a resolution of 50 microV under ideal conditions.

In fact, the ADC will introduce some errors but it might be possible to improve the accuracy by oversampling. Here, the crucial question is:

What is the maximum frequency in your spectrum which is necessary to yield reliable results? Using a math program, I would generate samples of your typical signal, and apply FFT to them. By retransforming part of the FFT result to the time domain you can decide where to place the upper limit for the frequency domain without inacceptably distorting the retransformed signal. If necessary later on, you should insert a low pass filter between the signal source and the ADC. The minimum sampling frequency is twice the highest spectrum component.

Example: If you can cut off your spectrum above 50 kHz then the minimum sampling frequency were 100 kS/s (in reality, I rather would use 200 kS/s). Since the maximum sampling frequency of the ADC is 2.5 MS/s, there is enough room for oversampling in case the raw 12 bit-result is not quite sufficient.

Aindrila Sinha If it really turns out that you need a higher resolution than 12 bit, then you cannot rely on any microcontroller with integrated ADCs and DACs. Instead, you would need an ADC and a DAC plus a separated microcontroller or DSP. Please have a look at the attached file on ADCs. A Sigma-Delta ADC might fit your application.

A suitable device might be the SSM2603, available on a evaluation board, too (please see attachments 2 and 3).

But in order to exploit such a high resolution, great care must be given to keeping the supply voltage of ADC and DAC clean of spikes generated by the digital part (microcontroller) of the circuit. So, I would try the simple (and more inexpensive) way with 12 bit ADC and DAC integrated in a microcontroller first.

Thanks sir for those information.

If I understand correctly, you have a carrier of 100 mV and 1 kHz and a very small signal modulating the amplitude of this carrier. You do not say what is the frequency of the modulating signal, but it should be at a frequency much lower than that of the carrier, in order you can use the amplitude modulation properly.

If this is the situation, you can try to use a chain of all-pass filters. Despite mismatchings, the module of the transfer function is unitary regardless of the frequency. The phase can be considered linear up to a limit frequency, where the linearity of the phase starts to worse. The delay is given by the derivative of the phase in relation to the frequency. Since the phase is linear, the delay is constant.

The required bandwidth depends on the duration of the pulse. With the suitable number of all-pass blocks, you can define the delay to your signal with an acceptable deviation for a specified bandwidth, say 10 kHz. See e. g. minitutorial MT-202 about all-pass filters from Analog Devices. Hope I have helped.

Antonio Carlos da Costa Telles thank you sir for your suggestion.