## down conversion vs de-modulation

Started by 7 years ago●14 replies●latest reply 7 years ago●2416 viewsI am a bit confused between digital down conversion and de-modulation (or up conversion and modulation). Let us assume I receive a modulated signal where a base band signal is modulated by a carrier. If I pass this through a DDC and de-modulation, would it be one and the same.

I assume de-modulation would involve some sort of filtering to remove carrier signal and get only base band. Down conversion would involve multiplication with complex exponential. I also understand that many times de-modulation needs to be carried out in analog domain due very high carrier frequencies.

Thanks a lot ...

Unless they've really gone downhill in the last ten years, you want to get an ARRL Handbook (or the GBRS Manual, if you're in England). Not just for this question -- for all of the ones you've been answering, and for the ones that you haven't thought to ask.

Down- (and up-) conversion is a specific operation, independent of the type of modulation. Demodulation is specific to the type of modulation used. Demodulating conventional broadcast AM is different than demodulating FM, and both are different from demodulating single sideband.

Single-sideband modulation is, essentially, the process of taking one half of a baseband signal (either the positive frequency components or the negative ones) and up-converting the result. The actual process doesn't necessarily happen that way, but the result is the same. So you can "demodulate" SSB by downconversion.

You cannot demodulate FM by downconversion, at all. Take FM, convert it to baseband, and the result is a nasty-sounding mess. You could down-convert to baseband and then demodulate there, but you'd still be demodulating.

You can, in theory, demodulate AM by downconversion, but you need to be synchronized to the carrier frequency to do it, or you get nasty fading. (It's done in practice in high-end AM receivers, by phase locking to the carrier and using the result for the local oscillator -- it's called exalted-carrier demodulation).

Hi Sharan123.

Your confusion is understandable. In the literature of signal processing some authors use the phrase "demodulation" when they should more correctly use the phrase "down-conversion." Likewise, some authors incorrectly refer to "up-conversion" using the misleading phrase "modulation."

Down- and up-conversion, to me, mean shift the center frequency of a signal either down or up in the frequency domain. Those operations are strictly frequency translation operations. The frequency-translated signals have the same bandwidth as the original untranslated signals!

Amplitude modulation (AM) and frequency modulation (FM) systems also generate signals whose center frequencies are higher than the center frequency of the original information-carrying baseband "modulated" signal. But those high frequency signals have wider bandwidths than the original baseband "modulated" signal!

I think it's useful for all signal processing folk to study and understand the processes of amplitude and frequency modulation as well as amplitude and frequency demodulation. Those processes are NOT difficult to understand and they're also easy to model (simulate) using software. (I teach a 3-day DSP class now and then and I'm regularly shocked by how many EE graduates do not know how the AM clock radio in their bedroom works.)

Hello Rick,

Thanks for answering. I just implemented a small Matlab code for DUC and then I got this doubt. I used a pure sine tone to do this. My next exercise would be to use a more irregular signal to study this and then I will work out some modulation examples.

Thanks again

Sharan123. After you implement frequency translation, by way of multiplication by a real-valued sinusoid or a complex-valued complex exponential, I hope you are then passing your translated signal through a lowpass filter.

Dear Rick,

Currently, this is more of an hands-on to get better understanding. I assume your reference to low pass filter is due to 2 frequency bands that result due to multiplication of signal with complex exponential. Is this correct?

Thanks again

Hi Sharan,

I think you already got nice answers. To give some extra flavor I will add one more. You can understand modulation as operation of translating your initial representation of information into certain new form. In case of AM and FM your modulating signal (which represents information that you want to transmit) would change some parameters of continuous carrier wave (usually sinusoidal waveform). For example, AM would change amplitude of carrier wave (it can be simply multiplied with carrier wave), while FM would change frequency of carrier wave (function that represents modulating signal would be part of sinusoidal carrier phase argument).

Now, we can go further to think about single carrier quadrature modulations such as BPSK (although one of quadrature components (usually quadrature one) is zero for the case of BPSK), QPSK, O-QPSK, MSK, QAM... For these modulations your information is usually first filtered in order to remove ISI (usually with RRC or SRRC filters), after that it is pushed into two orthogonal carrier waves (which have same frequency but, pi/2 phase shift -> simply sine and cosine). Orthogonality of carrier signals is important from many aspects, especially on receiver side in process of demodulation. Simply said, it enables you to achieve higher spectrum efficiency (higher level of information that you can push through certain frequency band) while still preserving ability to distinguish (demodulate correctly) that information on receiver side). In short modulation/demodulation assume more operations on signal compared to up/down conversion which assume frequency shifting (multiplication by sinusoidal wave form, and maybe some filtering).

Keep in mind, that some times up conversion can be part of modulation (or it can be done in that way) and that down conversion can be part of demodulation. Lets take as example QPSK modulation and usage of direct conversion architecture with dual channel DAC/ADC (popular architecture for SDR platforms - Analog devices FMCOMMS1/FMCOMMS2/FMCOMMS3....) you can create RRC filtered waveform which represents your information in digital domain, up sample it, convert to analog domain, and then in same time complete quadrature modulation and perform frequency shifting (up conversion) by multiplying it with sine and cosine carrier wave forms.

Kind Regards,

Tarik

Dear Tarik,

Thanks a lot such detailed explanation.

you can create RRC filtered waveform which represents your information in digital domain, up sample it, convert to analog domain, and then in same time complete quadrature modulation and perform frequency shifting (up conversion) by multiplying it with sine and cosine carrier wave forms.

I need a few clarifications with respect to above scheme.

If I have understood correctly then modulation in the scheme above happens in the base band and the up converted by complex multiplication. Is this correct?

Probably, a very basic question. After quadrature modulation, can we not do AM/FM modulation?

Thanks a lot

Hello Saharan,

Sorry for late reply.

In that example both mapping to quadrature components and frequency shift to RF happens in same time. You would need to use dual channel DAC to convert both I and Q components of your signal to analog domain. After that you would multiply those signals with carrier which will be pi/2 phase shifted for one branch. An then after this you can do summation and get your RF signal.

Maybe someone more experienced can help us here. I might be wrong. I think in case of FM/AM modulation in order to use direct conversion receivers (like ANALOG DEVICES FMCOMMS1/FMCOMMS2), you would need to complete modulation in digital domain first and then do up conversion.

Kind Regards,

Tarik

Thanks a lot, Tarik

Hi Sharan,

Modulation is how you encode the data into the carrier. The most common modulation schemes are AM, FM, and PM. For the amplitude modulation (AM) you alter the carrier amplitude by the data/information to be sent. While for the frequency modulation (FM) the carrier frequency is being altered by the data. In other words, the data/information are forming the shape of the carrier in either the amplitude, or frequency, or phase.

On the other hand, up/down conversion is shifting the carrier and/or data within the frequency domain. If we consider a simple FM receiver, the antenna receives the RF signal (around 400MHz) then it is mixed-down to an intermediate frequency (multiplication like) and even after that it most probably down-mixed again into the base-band. If the carrier frequency fc=400MHz, and local oscillator frequency fo=300MHz. fc x fo = (fc + fo) + (fc - fo) = 700MHz | 100MHz (refer to sinusoidal format of the signals). As you see the signal has been down converted to an intermediate frequency of 100MHz instead of 400MHz.

Regards.

Hi Ahmed,

Thanks a lot for the detailed explanation. One more question - what is the practical application of DDC or DUC?

Thanks,

"One more question - what is the practical application of DDC or DUC?"

Purely digital up-conversion is useful in the transmit chain. Consider a system where you have a complex signal at baseband, and for some reason (perhaps for really good carrier suppression, or for compatibility with a legacy system) you don't want to do an analog up-conversion from baseband. So, you digitally up-convert, then generate your signal at RF, then either amplify it and put it on an antenna, or you or up- (or down-) convert it by analog means.

Purely digital down-conversion has similar utility. If you've got a receive chain that has an IF signal, you can sample that at RF, downconvert in digital-land, very possibly filter it and then decimate it, and then do your processing (possibly at a much lower sampling rate than you used at RF) at baseband.

Thanks a lot. I think I get it now ...

I believe that in the advent of fast ADCs, down and conversion became an easier and cheaper method to perform the frequency conversion, which is traditionally done by mixers, i.e., multiplying by a sinusoidal function. The up/down conversion just discards samples or add zero samples to perform the operation, in lieu of multiplication.

I suggest that you look into the free ebook on digital communication using software defined radios (SDRs) that uses such operations naturally. Also you can buy a $20 SDR (the ebook uses such SDR for practical exercises) and play with it. Here is the link:

go to the download section and download the book and the MATLAB code to explore many concepts in digital communications.