4
\$\begingroup\$

I am working to understand this application's mixer design note. An image of the mixer is shown below. It uses two transistors and a hybrid ring coupler (rat-race coupler).

enter image description here

What I am having trouble understanding is the λ/4 RF chokes near the diodes. In the application note, the reason for the chokes is to prevent diode self-biasing. The reason the RFCs are needed because "At higher LO power levels, the diodes are self-biased and show undesired conversion loss and isolation values" (Page 5).

I am not sure I fully understand what self-biasing means regarding diodes. My current understanding is that if the voltage of the incoming waves exceeds the forward bias of the diode, then half of the waves are essentially clipped by the diode leading the wave to have a non-zero DC bias. See the LTspice schematic and output below.

enter image description here

enter image description here

Vout still looks somewhat like a sine wave, but now it has a DC bias around maybe 1.75 to 2 V. My questions are:

  1. Is this understanding of DC bias correct, and
  2. How could I add an RF choke to my LTspice circuit below to try and remove the DC component of my signal?

EDIT: To clarify, I am not trying to simulate the rat-race coupler above in LTspice, my main goal is to observe self-biasing in a Schottky diode in LTspice. It doesn't necessarily have to be the exact circuit of the rat-race coupler. If I need to have transmission lines in order to observe this effect, please let me know.

Below is the circuit with the inductor included. The signal is still shifted up so that the mean is nonzero

enter image description here

enter image description here

But if I include a 2 V DC component in my voltage source (Vin), that DC voltage is shorted by the inductor as expected and does not show up at Vout. So am I even observing a bias in my diode in my previous simulations?

enter image description here

enter image description here

\$\endgroup\$
12
  • \$\begingroup\$ In my LTspice circuit, I tried adding a 200uH inductor in parallel to my capacitor in the hopes of shorting any DC bias. But my results were essentially unchanged, so I am beginning to doubt if there is a DC bias \$\endgroup\$
    – CMH12
    Jan 20, 2023 at 13:52
  • 1
    \$\begingroup\$ Those are diodes and not transistors. \$\endgroup\$
    – Andy aka
    Jan 20, 2023 at 13:58
  • \$\begingroup\$ Thanks, I changed the label in the image \$\endgroup\$
    – CMH12
    Jan 20, 2023 at 14:06
  • \$\begingroup\$ And the text in your opening paragraph. They are not chokes BTW they are transmission lines. You can't model them with inductors. \$\endgroup\$
    – Andy aka
    Jan 20, 2023 at 14:24
  • 1
    \$\begingroup\$ Is RF input wideband? \$\endgroup\$
    – Antonio51
    Jan 20, 2023 at 14:52

3 Answers 3

2
\$\begingroup\$

Consider what may happen if the ring is not grounded via those \$\lambda/4\$ chokes...

Doppler radar requires I.F. output exdending to 0 Hz. Both R.F. input port and L.O. input port could easily be coupled with a capacitor, leaving the ring high-impedance DC. Where will I.F. output currents go when the ring is floating?.
I.F. output voltage will appear at RF or LO port, or both.

If either R.F. port or L.O. port had a DC path to ground, those chokes would not be needed. OP's drawing is incomplete - the block diagrams showing RF input and LO input do not clearly establish a path-to-GND.


edit:
The "DC bias" mentioned in the app note is not due to rectification. It is due to an imperfectly-balanced mixer. It is true that those diodes conduct for nearly half-a-cycle, but they should do so simultaneously. Non-simultaneous diode conduction causes a DC component at the I.F. port. It also causes some L.O. power to appear at the R.F. port too.

\$\endgroup\$
3
  • \$\begingroup\$ I didn't include the additional schematics you are referring to, but in the application note above, the chokes do go to ground as you expected. I believe you are saying that some of the byproducts of the mixing will be DC, and without a path to ground, the DC components could build up, leading to a bias. This is especially an issue when a capacitor is placed between the RF/LO inputs and the hybrid ring. Is this correct? So therefore unless I am performing the mixing operation I shouldn't expect to observe the bias \$\endgroup\$
    – CMH12
    Jan 22, 2023 at 23:55
  • 1
    \$\begingroup\$ A simulation done exactly won't produce DC bias - it is a balanced mixer. Consider this: no mixer has perfect balance...if diodes see slightly different L.O. amplitude or phase, I.F. port will see a DC component. Furthermore, some L,O. power will leak back to R.F. port...radiate...reflect...and cause a DC I.F. component too (which may further unbalance the mixer). Very difficult to simulate, and a very real problem for the Doppler radar example circuit where RF & LO frequency is the same. \$\endgroup\$
    – glen_geek
    Jan 23, 2023 at 0:37
  • \$\begingroup\$ Thanks for the clarification. I think this makes sense, as I look at the application note, it also points out that the biasing is only an issue with higher LO powers \$\endgroup\$
    – CMH12
    Jan 23, 2023 at 1:49
5
\$\begingroup\$

A 1/4\$\lambda\$ transmission line shorted at the end will act as a short circuit for DC, but as a high impedance at the frequency of interest. This is sometimes used as what's called a shorted stub for lightning protection and reducing static buildup on antennas.

In the case of the mixer it keeps DC from building up on the diode while not attenuating the signal.

To simulate it in LTspice you might be able to do it with a transmission line.

I haven't used transmission lines in LTspice much, but I think this might be close to what you need, the Td parameter changes the resonant frequency, divide frequency in GHz by 1/4 and use that in the formula (5.8/0.25 = 23.2). I just used 75 \$\Omega\$ as the characteristic impedance, so if you know what that should be you can change it.

transmission line stub

\$\endgroup\$
2
  • \$\begingroup\$ Hi @GodJihyo, thank you for your answer showing me ways to simulate the choke in LTspice. Can you maybe elaborate on why a DC bias builds up on the transistor? Based on the additional tests I did in my edit, I am not confident I have observed the diode biasing. I also tried using a 1H inductor as you suggested and observed vary similar behavior. The output signal was still biased its amplitude was just changed because my filter characteristics changed \$\endgroup\$
    – CMH12
    Jan 20, 2023 at 15:25
  • \$\begingroup\$ @CMH12 I think what you originally said about it is correct, rectification builds up a bias. I will have to play around with simulating it when I have more time. Also note that I change the schematic in my answer so it's at 5.8GHz. \$\endgroup\$
    – GodJihyo
    Jan 20, 2023 at 15:31
1
\$\begingroup\$

Here is the schematic of the mixer.
I used it at 3 MHz (LO) and 3.1MHz RF (for the "easiest" possible), it does really not matter, because the transmission lines are in fractions of a wavelength.
Your points 3 and 4 are wired with a length of "lambda/2" transmission line.
The composite gain of the mixer is ~ 6 dB. IF filter not optimized.

enter image description here

Waveforms across diodes and voltage at center point of diodes.

enter image description here

And a test with "low pass" filter at IF port.

enter image description here

\$\endgroup\$

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.