In some of the small radios I've worked with such as the Nordic Semiconductor 24LE line, there is a significant (100us+) delay between transmitting data and being able to receive a response. By my understanding, a big factor in this delay revolves around the fact that a transmitter requires a PLL to be tuned to the exact desired transmission frequency, while a receiver requires a PLL tuned to a frequency which differs from the desired one by an Intermediate Frequency. Thus, switching between transmission and reception requires retuning the PLL, and the need to retune the PLL represents a significant portion of the turnaround time.

Would it be practical to construct a radio such that the receiver could be configured to receive a frequency either above or below the PLL frequency, and then have a handshake sequence operate something like:

  • Transmitter sets PLL to 2315Hz, sets receiver mode to "100Mhz above"
  • Expectant receiver sets PLL to 2415Hz and receiver mode to "100Mhz below"
  • Transmitter listens for a little while to let its AGC settle.
  • Transmitter sends out packet at 2315Hz, while keeping PLL constant
  • Receiver hears packet which was sent 100Mhz below its PLL frequency
  • Transmitter switches receiver mode to "100Mhz above" while keeping PLL constant
  • Receiver sends reply at 2415Hz, while keeping PLL constant.
  • Original transmitter receives reply which was sent 100Mhz above its PLL frequency

When using a 100Mhz Intermediate Frequency (which I understand is normal for 2.4GHz communication) this approach would unfortunately be limited to using the upper and lower 50Hz of the 2.4GHz band, but it would seem that a reduction in transmit/receive turnaround time would be helpful. Alternatively, use of a lower Intermediate Frequency would reduce the required separation between transmit and receive frequencies.

Are such approaches used at all, whether in 2.4Ghz radios or in other applications? Do they have any particular problems?


When a transceiver is transmitting the PLL frequency is set to Fs and when it is receiving the PLL frequency is set to either a frequency greater than Fs by the intermediate frequency or one lower than Fs by the intermediate frequency i.e

When transmitting:

$$ F_{\text{PLL}} = F_{\text{S}} $$

and when receiving:

$$ F_{\text{PLL}} = \begin{cases} F_{S} - F_{IF} & \text{for low side Injection} \\ F_{S} + F_{IF} & \text{for high side Injection}\end{cases} $$

Remember that we are only permitted to transmit over a very low frequency range e.g 2.39GHz to 2.42GHz so we can not use any of the approaches you propose above because Fs has to be constant.

Further notice that for use the only way to avoid retuning the PLL is to set the intermediate frequency to zero, which will essentially now turn our superhet receiver to a one stage TRF receiver (You can read the the disadvantages of TRF receivers in the link provided).

As for the solution you propose for reducing the IF frequency to a very low frequency so that we can reduce the amount of tuning to be done, you have to remember that the lower the value of our IF frequency the poorer the image channel rejection our system will have so there is a limit to how low our IF can be.We can't just set IF to a very low value.

  • \$\begingroup\$ What would you suggest as a practical way of making a low-cost ratio of the style used for things like Bluetooth Low-Energy which can manage efficient packet turnaround? \$\endgroup\$ – supercat Feb 23 '15 at 20:41

This sounds like overkill but couldn't you use two transceivers - one permanently set to receive and one permanently set to transmit then neither have to shift their respective PLLs. There may be issues connecting them to the same antenna but something in the back of my mind tells me that a few of these "types" of transceivers have separate receive and transmit output lines and you use external components to join them to one common antenna. It might be worth a try just to see what response you get.

I'll also add (and I haven't looked at the tech specs for these devices in ages) that there may be a "planned" delay to allow for ringing to die down (following a RF transmission) on the components that connect rx and tx to the antenna and you may in fact gain nothing.

  • \$\begingroup\$ The objective would be to allow devices to handshake efficiently without too much additional hardware cost. Using separate tuning oscillators would work, but I would expect it would be more expensive than using one oscillator and selecting high-side or low-side injection. \$\endgroup\$ – supercat Aug 3 '14 at 21:19

The Nordic Radios are, I suspect zero IF radios, where the signal is down converted to 0Hz in quadrature, so there isn't an IF frequency. This lets you integrate the receiver into a single chip. The retuning is to allow for half duplex operation, so that different units are transmitting and receiving on different frequencies. 100uS retuning speed isn't too bad, you've still got synchronization issues between different radios to allow for. Even if you could cut that in half, how many bits would you be able to transmit in that time?

  • \$\begingroup\$ The issue isn't just how many bits one could transmit. If a radio doesn't know whether there's anyone out there, the amount of time it has to stay awake listening for a response after a ping could be made a lot shorter if turnaround time were fast and predictable. Given that listening for a response after a ping takes more energy than sending the ping, cutting that time would allow major energy savings. \$\endgroup\$ – supercat Oct 3 '14 at 12:27
  • \$\begingroup\$ What factors, other than retuning, would prevent a radio from receiving a response 5us after transmitting? Freeze the AGC state before transmission (since it would otherwise get swamped) and it should still be at a good level when transmission is done. \$\endgroup\$ – supercat Oct 3 '14 at 12:30
  • \$\begingroup\$ No, the issue is entirely the number of bits you can transmit across a huge population of radios. The whole reason you have a radio network it to move bits back and forth, there's no other reason for it to exist. As far at other factors, Possibly a bit one is getting data through filters, both digital and analog, as well as compensating for a DC offset. Switching times for gain stages are in the 10's of nSec range, so filters, DC offsets would be the first two things I would worry about \$\endgroup\$ – rfdave Oct 4 '14 at 1:35
  • \$\begingroup\$ There are many applications where a radio may be expected to go for days without successfully exchanging any information with anyone, and use very little energy, and yet still be able to honor an outside wakeup request within a few seconds. One way to handle this is to have a radio send a "ping" periodically at somewhat-randomized intervals of a few seconds and listen for a response. If nobody responds to the ping, the radio can go back to sleep. The more quickly a another unit would be able to respond to a ping, the less time and energy the waiting unit would have to spend listening. \$\endgroup\$ – supercat Oct 4 '14 at 20:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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