When I study the wireless transmitters , I find that these transmitters always generate digital IQ signal , and change the digital IQ into analog IQ signal. After this , there are two analog mixers using quadrature carrier to transform the IQ baseband signals into IQ RF signals . Next , combining the two quadrature signals into one signal and transmitting it. Based on my understanding , I think that I can use NCO to mix the digital IQ signal to a low frequency carrier in digital domain and combine the result info a digial bandpass signal , then change the digital bandpass signal info analog signal through ADC. Then we can transform the analog signal to a RF and transmit it . It seems this will be easier. Since my understanding is different from the practice , I think maybe I have made some mistakes. Can you help me about this? Thanks.
Why transmitters use analog IQ mixer instead of digital?
Started by ●January 5, 2012
Reply by ●January 5, 20122012-01-05
On 1/5/2012 7:04 PM, agump wrote:> When I study the wireless transmitters , I find that these transmitters > always generate digital IQ signal , and change the digital IQ into analog > IQ signal. After this , there are two analog mixers using quadrature > carrier to transform the IQ baseband signals into IQ RF signals . Next , > combining the two quadrature signals into one signal and transmitting it. > Based on my understanding , I think that I can use NCO to mix the digital > IQ signal to a low frequency carrier in digital domain and combine the > result info a digial bandpass signal , then change the digital bandpass > signal info analog signal through ADC. Then we can transform the analog > signal to a RF and transmit it . It seems this will be easier. Since my > understanding is different from the practice , I think maybe I have made > some mistakes. Can you help me about this? Thanks.I'll swing at this one - If you generate the digital signal at an Intermediate Frequency, bandpass filter the signal to remove the NCO spurs, and then up-convert to RF - you often get mixer products from the up-conversion that are unacceptable. If you generate the digital signal at the RF frequency, the NCO itself has spurious signals that are unacceptable. It is often simpler to generate a nice clean RF signal and use an analog QAM modulator. You can also get into transmitter "noise figure", or broadband noise issues if you have a lot of transmitter gain after a low power mixer. It's one of those practical problems. They all work great on paper... Rob.
Reply by ●January 5, 20122012-01-05
>When I study the wireless transmitters , I find that these transmitters >always generate digital IQ signal , and change the digital IQ into analog >IQ signal. After this , there are two analog mixers using quadrature >carrier to transform the IQ baseband signals into IQ RF signals . Next , >combining the two quadrature signals into one signal and transmitting it. >Based on my understanding , I think that I can use NCO to mix the digital >IQ signal to a low frequency carrier in digital domain and combine the >result info a digial bandpass signal , then change the digital bandpass >signal info analog signal through ADC. Then we can transform the analog >signal to a RF and transmit it . It seems this will be easier. Since my >understanding is different from the practice , I think maybe I have made >some mistakes. Can you help me about this? Thanks.Try studying some more wireless transmitters and you will find many strategies are used. Turning I and Q separately into analogue signals, and building the final transmitted signal in a largely analogue manner, used to be the norm, because most systems were pushing the bounds of what reasonably prices DACs could do. These days DACs are good enough and cheap enough that you have many more choices for building the transmitted signal in most applications. Of course, the fastest systems still push the bounds of what a DAC can do, and there will probably always be a place for systems that separately convert the I and Q signals to analogue. Steve
Reply by ●January 6, 20122012-01-06
On Thu, 05 Jan 2012 20:04:48 -0600, "agump" <aaronstempmail@n_o_s_p_a_m.gmail.com> wrote:>When I study the wireless transmitters , I find that these transmitters >always generate digital IQ signal , and change the digital IQ into analog >IQ signal. After this , there are two analog mixers using quadrature >carrier to transform the IQ baseband signals into IQ RF signals . Next , >combining the two quadrature signals into one signal and transmitting it. >Based on my understanding , I think that I can use NCO to mix the digital >IQ signal to a low frequency carrier in digital domain and combine the >result info a digial bandpass signal , then change the digital bandpass >signal info analog signal through ADC. Then we can transform the analog >signal to a RF and transmit it . It seems this will be easier. Since my >understanding is different from the practice , I think maybe I have made >some mistakes. Can you help me about this? Thanks.I think you didn't look at enough transmitters. It is not unusual for modern systems to employ a digital mix to an IF and then convert to a real analog signal. I've worked on many such systems, some deployed as long as fifteen years ago. There are a few tradeoffs to each method, but both are commonly used. For inexpensive systems that transmit at high frequencies (e.g., WiFi, BT, etc.), a direct conversion mixer that digitizes at baseband and mixes directly to the RF frequency in a single mix may be less expensive than an IF mix that requires image rejection processing, etc. The direct conversion tuner may also wind up with a greater tuning range, which can be important for many systems. For signal integrity, i.e., very high EVM performance, a digital IF system is almost always superior. Eric Jacobsen Anchor Hill Communications www.anchorhill.com
Reply by ●January 6, 20122012-01-06
On 1/5/12 9:21 PM, Rob Doyle wrote:> On 1/5/2012 7:04 PM, agump wrote: >> When I study the wireless transmitters , I find that these transmitters >> always generate digital IQ signal , and change the digital IQ into analog >> IQ signal.you mean they output these two signals we call I and Q to a D/A converter?>> After this , there are two analog mixers using quadrature >> carrier to transform the IQ baseband signals into IQ RF signals . Next , >> combining the two quadrature signals into one signal and transmitting it. >> Based on my understanding , I think that I can use NCO to mix the digital >> IQ signal to a low frequency carrier in digital domain and combine the >> result info a digital bandpass signal ,that's how i would do it,...>> then change the digital bandpass >> signal info analog signal through ADC. Then we can transform the analog >> signal to a RF and transmit it .... but it depends on how clean the NCO is.>> It seems this will be easier. Since my >> understanding is different from the practice , I think maybe I have made >> some mistakes....> If you generate the digital signal at an Intermediate Frequency, > bandpass filter the signal to remove the NCO spurs, and then up-convert > to RF - you often get mixer products from the up-conversion that are > unacceptable.but say your NCO is just part of the DSP (say, it's a really fast Blackfin or something like that), the same DSP you would be using for the "mixing" (because of my audio sympathies, i would use the label "heterodyning"). now, with sufficient interpolation effort, you can make that NCO as clean as any analog oscillator. you could even generate a pair quadrature sinusoids from the real and imaginary parts of: u[n] + j*v[n] = e^(j*w0) * ( u[n-1] + j*v[n-1] ) u[0] = 1, v[0] = 0 . and with enough bits and MIPS, i would say cleaner than any physical analog oscillator. and the hetrodyning would be perfect, also. so i also don't get why, if the word width and sufficient instructions are there, why not do it all digitally and output to D/A a single real composite IF signal with quadrature components at some frequency high enough that it could be easily analogly bumped to any RF frequency with not much effort put into the anti-imaging filters. but, then again, i'm an audio geek and not a comm-systems engineer. hey Eric, yer one of them commies**, no? how would you do this? -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge." **1. Eric comes from SD 2. George McGovern comes from SD, 3. ergo, Eric is a commie.
Reply by ●January 6, 20122012-01-06
On Jan 6, 3:21�pm, Rob Doyle <radioe...@gmail.com> wrote:> On 1/5/2012 7:04 PM, agump wrote: > > > When I study the wireless transmitters , I find that these transmitters > > always generate digital IQ signal , and change the digital IQ into analog > > IQ signal. After this , there are two analog mixers using quadrature > > carrier to transform the IQ baseband signals into IQ RF signals . Next , > > combining the two quadrature signals into one signal and transmitting it. > > Based on my understanding , I think that I can use NCO to mix the digital > > IQ signal to a low frequency carrier in digital domain and combine the > > result info a digial bandpass signal , then change the digital bandpass > > signal info analog signal through ADC. �Then �we can transform the analog > > signal to a RF and transmit it . �It seems this will be easier. �Since my > > understanding is different from the practice , I think maybe I have made > > some mistakes. Can you help me about this? Thanks. > > I'll swing at this one - > > If you generate the digital signal at an Intermediate Frequency, > bandpass filter the signal to remove the NCO spurs, and then up-convert > to RF - you often get mixer products from the up-conversion that are > unacceptable. > > If you generate the digital signal at the RF frequency, the NCO itself > has spurious signals that are unacceptable. > > It is often simpler to generate a nice clean RF signal and use an analog > QAM modulator. > > You can also get into transmitter "noise figure", or broadband noise > issues if you have a lot of transmitter gain after a low power mixer. > > It's one of those practical problems. � They all work great on paper... > > Rob.I and Q weren't even invented when I studied wireless. What's wrong with food old fashion AM or FM. No buggering about with DSP crap there either. Cheap and it worked! Latest OFDM radios don't work well in a car (though they are supposed to). Lot of money spent on shit. Hardy
Reply by ●January 6, 20122012-01-06
HardySpicer wrote:> I and Q weren't even invented when I studied wireless.I and Q were invented by Armstrong and such. When you studied, the NTSC was already deployed.> What's wrong > with food old fashion AM or FM.No simple way to implement paid content with AM or FM.> No buggering about with DSP crap > there either.Indian programmers have something to do.> Cheap and it worked!That's the problem. Not helping the economy.> Latest OFDM radios don't work well > in a car (though they are supposed to). Lot of money spent on shit.Sure. Write a complaint to the world league for sexual reforms.
Reply by ●January 6, 20122012-01-06
Hi ,Eric . I have some knowledge about the direct conversion architecture. Maybe I haven't express myself clearly. Based on my understanding , I think the direct conversion architecture is super. It just need one RF channel. We needn't care about the imbalance problem which exists in the analog I/Q channels if we just process I/Q signal in digital domain. But in practice , I find a lot of high performace transmitter using analog I/Q.>If you generate the digital signal at an Intermediate Frequency, >bandpass filter the signal to remove the NCO spurs, and then up-convert >to RF - you often get mixer products from the up-conversion that are >unacceptable. > >If you generate the digital signal at the RF frequency, the NCO itself >has spurious signals that are unacceptable. > >It is often simpler to generate a nice clean RF signal and use an analog >QAM modulator.Do you mean the NCO's SFDR? I have developed some NCOs and think we can realize a NCO with superios spur performance such as 130dB.>You can also get into transmitter "noise figure", or broadband noise >issues if you have a lot of transmitter gain after a low power mixer. >I think these problem should exists in the two architecture, are there some differences?
Reply by ●January 6, 20122012-01-06
>Hi ,Eric . I have some knowledge about the direct conversionarchitecture.>Maybe I haven't express myself clearly. >Based on my understanding , I think the direct conversion architecture is >super. It just need one RF channel. We needn't care about the imbalance >problem which exists in the analog I/Q channels if we just process I/Q >signal in digital domain. But in practice , I find a lot of highperformace>transmitter using analog I/Q. > > >>If you generate the digital signal at an Intermediate Frequency, >>bandpass filter the signal to remove the NCO spurs, and then up-convert >>to RF - you often get mixer products from the up-conversion that are >>unacceptable. >> >>If you generate the digital signal at the RF frequency, the NCO itself >>has spurious signals that are unacceptable. >> >>It is often simpler to generate a nice clean RF signal and use an analog >>QAM modulator. >Do you mean the NCO's SFDR? I have developed some NCOs and think we can >realize a NCO with superios spur performance such as 130dB.The NCO isn't where you hit hard to cirumvent spur issues. The DACs are.>>You can also get into transmitter "noise figure", or broadband noise >>issues if you have a lot of transmitter gain after a low power mixer. >> >I think these problem should exists in the two architecture, are theresome>differences?Steve
Reply by ●January 6, 20122012-01-06
On Jan 6, 8:32�pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote:> HardySpicer wrote: > > I and Q weren't even invented when I studied wireless. > > I and Q were invented by Armstrong and such. > When you studied, the NTSC was already deployed. > > > What's wrong > > with food old fashion AM or FM. > > No simple way to implement paid content with AM or FM. > > > No buggering �about with DSP crap > > there either. > > Indian programmers have something to do. > > > Cheap and it worked! > > That's the problem. Not helping the economy. > > > Latest OFDM radios don't work well > > in a car (though they are supposed to). Lot of money spent on shit. > > Sure. Write a complaint to the world league for sexual reforms.Ha ha good
