This is a problem that's always baffled me a bit. How does high frequency (more than 100MHz anyway) wireless communication actually work? I understand that it has an antennae and for receiving it amplifies it and checks for a logical 1 or 0 and reversed for transmitting.

What I do not understand is how an IC can communicate at such speeds? Take for example wifi, 2.4GHz. Is there a chip that is actually processing each bit 2.4 billion times per second? That seems impossible. Could someone explain how a transmitter and a receiver actually work electrically?

  • \$\begingroup\$ In your example 2.4GHz is the carrier frequency, not the data rate (which is measured in Baud not Hz). The max. data rate is roughly 100-1000 times lower than the carrier frequency (depending on many factors, e.g. type of modulation, SNR, etc.) \$\endgroup\$
    – Curd
    Commented Sep 10, 2013 at 13:01

4 Answers 4


The important thing to note here is carrier frequency and modulation.

2.4GHz is your carrier frequency, in modern modulation formats it is going to be in the air at all times. The transmitter radiates the entire time you are sending the signal.

How is the data actually sent?

Phase modulation is the most common method. You can think about what is happening very clearly, on a set timer you are going to either change phase or not. Wikipedia has a good graph of QPSK, where you are actually sending two signals at the same time out of phase and each one encodes a bit. http://upload.wikimedia.org/wikipedia/commons/b/be/QPSK_timing_diagram.png">

This may look a little confusing, but you see whenever they change what bit they are sending there is a sudden shift in the signal. PSK has the lowest bit-error rate of the different modulation techniques for the same baud rate. This means that for the same allowable bit-error rate you have the highest link speed with PSK.

I hope the image allows you to understand what is going on behind the scenes. Let me know if I can post more to help make this understandable.

What hardware does this?

This section I am keeping short because there are many different ways to approach this with hardware. The circuit that allows most ICs to do internal TX or RX comes from the gilbert cell.

When to do it?

If you modulate to the correct frequency directly before radiating and demodulate directly before receiving the signal your circuit deals with everywhere else is going to be a slower speed signal that is digital and your circuit can deal with.

  • \$\begingroup\$ Do chips actually use the Gilbert cell, though? Looks prone to noise, thermal variations, and high power consumption... \$\endgroup\$
    – tyblu
    Commented Dec 16, 2010 at 22:23
  • 1
    \$\begingroup\$ @tyblu, How many low power receivers have you seen. This is the foundation of IC modulation and demodulation. They have a lot of factors in their design. \$\endgroup\$
    – Kortuk
    Commented Dec 16, 2010 at 22:32
  • \$\begingroup\$ Gilbert cells are definitely legit. The major benefit of a Gilbert cell is that you can get gain out of it. \$\endgroup\$
    – W5VO
    Commented Dec 16, 2010 at 23:20
  • \$\begingroup\$ @tyblu, @w5vo, I believe you can cause a gilbert cell to be lower power but require a higher received power from the RX line. Also, most IC RX circuity can be placed in low-power mode. \$\endgroup\$
    – Kortuk
    Commented Dec 17, 2010 at 0:05

While I'll dodge the modulation question, I'm fairly well acquainted with the IC side of things.

"How can a IC communicate at speeds greater than 100MHz?"

I'll start with a simple case. Intel has designed a processor that operates at a clock frequency of 3.8GHz. This is performing multiple logic operations and storing the results each cycle. Thus, not only can signals be processed at 2.4GHz+, your computer probably does this already.

The reason for this is that transistors on a IC are FAST! In a 130nm SiGe BiCMOS process, the unity gain frequency is listed as 230GHz. I figure that I could make a circuit that works at least 5-10% of that value, and this isn't even a leading edge process.

If you wanted to maximize the serial clock rate, you could use a circuit called a De-Serializer, which is basically a high frequency shift register. You would need the very high frequency circuitry for the input, and then convert it into a parallel format at a lower data rate. This is commonly used in high speed protocols like HDMI.

  • \$\begingroup\$ Interesting, I focused mostly on modulation. I enjoyed what you added. \$\endgroup\$
    – Kortuk
    Commented Dec 16, 2010 at 23:49
  • \$\begingroup\$ I don't think your answer actually gets at what he is asking. There is a difference between data rate and carrier frequency. The logic/transistors only need to be able to handle the data rate. Modulation and demodulation occur as the first last thing and very first thing in your system. \$\endgroup\$
    – Kellenjb
    Commented Dec 17, 2010 at 1:15
  • \$\begingroup\$ Also just because a computer is able to run at 2.4ghz doesn't mean it can handle 2.4ghz of bandwidth. This starts to come into play with sampling theory, size of each sample, etc. \$\endgroup\$
    – Kellenjb
    Commented Dec 17, 2010 at 1:17
  • \$\begingroup\$ @Kellenjb, I disagree with that statement. While there certainly is a difference between the data rate and the carrier frequency, it would still be OK if there wasn't a difference - i.e. if the data rate was 2.4GHz. I would argue that at the very minimum, a 2.4 GHz computer would have a bandwidth of 2.4GHz * number of data lines clocked at 2.4GHz. If the clock signal (which is generated on-chip) is 2.4 GHz, then the signal would have harmonics in the 7.2GHz range. I am addressing only one part of his question: : "how an IC can communicate at such speeds." \$\endgroup\$
    – W5VO
    Commented Dec 17, 2010 at 2:44
  • \$\begingroup\$ @kellenjb, @w5vo, you are both saying something different. Yes, modulation steps down the frequency to a frequency that is cheaper to operate at. Yes, circuits can go that fast, but it increases cost. I think both are important concepts. \$\endgroup\$
    – Kortuk
    Commented Dec 17, 2010 at 4:29

While there are some special exceptions, most radio communication is usually achieved with the aid of upconversion and downconversion.

Basically, a transmitter starts with a circuit for modulating information (be it voice or data) onto a signal of conveniently low frequency that is easy to work with - a few tens or hundreds of kilohertz for narrowband applications, often somewhere between 10 and 45 MHz for wider band ones. At these frequencies analog circuits work well, or one can actually use a D/A converter at the output of a DSP that does the modulation mathematically. (For data rates higher than a "DSP chip" can handle, parallel logic in an ASIC or FPGA is used, so each individual path may only have to calculate every 8th or 32nd or whatevereth sample needed by the DA).

The transmitter also contains an oscillator or synthesizer for generating a signal closer to the desired transmitter frequency, and a mixer which multiplies the two signals together, causing generation of sum and difference frequencies. Either the sum or the difference will be the desired transmission frequency, and is selected by a filter, amplified and sent to the antenna. (Occasionally multiple stages of conversion are needed)

The receiver works the same way, only in reverse. A local oscillator signal is subtracted from the amplified antenna signal (or the other way around), creating an intermediate difference frequency that is back down in the range more convenient to work with (In AM broadcast receivers, typically 455 KHz - for FM, traditionally 10.7 KHz and then converted down againto to 455 KHz, though today staying at 10.7 MHz works too). This intermediate frequency can be processed by a demodulator circuity, or digitized in a fast A/D converter and fed into a potentially parallel DSP to complete the process.

If the desired bandwidth of data to be transmitted is less than about 10 KHz, one can actually use a computer soundcard to make a high performance receiver or transmitter, by positioning the intermediate frequency at say 10 KHz and using software to process a bandwidth spanning 5-15 KHz.

Today a common technique is to exploit some properties of complex numbers, and do the modulation/demodulation balanced around a center frequency of 0, such that it contains both positive and negative frequencies. By using two phases of the oscillator and something called an image reject mixer, one of the two resulting frequencies cancels and the other reinforces. However, two D/A or A/D converters are needed - one for the "I" phase and the other for the "Q". You can sort of do this with a stereo soundcard, though the DC blocking caps will create a hole in the passband right in the middle, at what gets converted to 0 frequency.


100 MHz is the carrier frequency, not the data transmission rate. A modulation of the carrier frequency is what carries the data. AM radio varies the amplitude of the signal to modulate it. FM varies the frequency slightly from the carrier frequency. PSK is phase shift keying. It change the phase of the carrier signal.

A modulator tasks the data and applies modulation to the carrier to send it. A demodulator receives the carrier and separates the modulation from it extracting the data.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.