# Question about simple schottky diode mixers

1. ### Mrcelloman99

7
I am new to mixer circuit design, but I need to build a simple mixer that will downconvert a 10 GHz signal to a DC to 3 kHz signal. I know I can do this with a schottky diode and lowpass filter to isolate the lower harmonic (f-IF = f-RF - f-LO). The problem is that the RF signal will be in the uV range. Will mixing still work with a weak signal or do I need to amplify it? I read somewhere that the RF signal should be much lower than the LO signal such that the diode transconductance is affected by the LO signal only, but how low is too low?

Thanks for your help!

2. ### vk6kro

4,059
Very difficult.

This would require a very elaborate receiver to convert this signal to a frequency where you could isolate one sideband.

You would have to buy most of it unless you have a lot of experience building equipment.

If you bought a receiver capable of receiving single sideband at 144 MHz, you could look for a converter that could mix your 10 GHz signal with a stable 10 GHz minus 144 MHz sine wave voltage.

You would be looking at at least $1000 to do that. Just using a Schottky diode wouldn't work at all. You would need a very stable oscillator on 10 GHz and a filter at the same frequency that just doesn't exist. And you would need a very high gain amplifier at 10 GHz. Also very difficult. Then you need an antenna. 3. ### meBigGuy The question of using an LO and schottky diodes to mix uV signals is something I have no experience with. Mixing 10GHz to 0 and using a quadrature detector isn't that difficult, conceptually. But there are practical issues that complicate it. You would need a 10GHz oscillator with very low phase noise in the +-3KHz band, which I think is a problem. Mixing to 144Mhz (as suggested, or some other frequency) means you need low phase noise out at 144Mhz, which is *maybe* more realistic. Then you can gain up and filter the 144Mhz before mixing to zero. Remember that the mixer output will contain the +-3KHz components from both the LO and the signal. Either you have to gain it up at 10GHz and then mix, or mix to an intermediate frequency, filter the image, gain it up, and mix to 0. 4. ### Mrcelloman99 7 I don't think I was specific enough. The entire setup would be composed of my ~10 GHz signal (which has already been received through a close range transmission from another circuit), the diode, and a 10GHz VCO. The intermediate frequency would be from DC to 3 kHz. The filter would then be applied at the output to get the lower sideband, which should be the easy part since the sidebands would be 20 GHz apart at the output. I'm not too concerned with output power as long as it's in the uV range at least. There are definitely diodes that move fast enough to achieve something like this. Heck, there are mixers that can do it that cost less than$40 or less. In fact, I have a microwave sensor that does the exact same thing I'm trying to do, and it costs \$8. I just can't use it because I need to move my frequency wirelessly to another point in my design and DC-3 kHz is impossible to transmit on a small scale.

This is a low power project so I don't need tons of gain either. And the antenna is not the issue; I'm pretty sure I have that part down. Is my proposal still unfeasible?

5. ### Mrcelloman99

7
meBigGuy:

Thanks for the insight; I'll consider your suggestion.

6. ### meBigGuy

Technically the intermediate frequency is -3Khz to +3Khz.

The mixed down signal you see in DC to 3Hz would contain the information from the signal's 10Ghz +- 3KHz plus the LO's 10GHz +- 3KHz. You have -3KHz to +3Khz summed in the DC to 3KHz signal. Does that make sense?

If you want to receive only 1 sideband of the +-3KHz signal, You need a quadrature mixer and detector (I and Q). If you really don't want 3KHz single-sideband, then your only issue is LO phase noise in the +- 3KHz band and whether you can pull your signal out of that.

7. ### Mrcelloman99

7
I see, I wasn't accounting for negative harmonics. Well I'd rather not use a quadrature mixer if I don't have to, but it depends. See the actual frequency I am trying to oscillate will be fed into a frequency counter of some type and information will be extracted from it. Will the -3 kHz image hurt me too much when trying to do that? Sorry for all the questions. I'm actually still a student.

8. ### vk6kro

4,059
Maybe you could clarify further.

What sort of modulation are you using?

How is your 10 GHz signal being generated? Is it a stable signal locked to a crystal oscillator?

Gain and selectivity are very difficult to obtain at microwave frequencies and not easy at 144 MHz.
So it is usual to use multiple conversion to get to a crystal filter at about 10 MHz which can deliver good selectivity.
In the conversion process, gain can be added to bring a 1 uV signal up to a volt or so where you can actually use it.

Most of this can be provided in the receiver (possibly 144 MHz because these are available) and a converter from 10 GHz to 144 MHz is very specialised but probably available.

A microwave detector for microwave ovens is a very different thing.
It operates on 2.4 GHz on millivolts or volts of signal.
It does not need to demodulate anything either.

9. ### meBigGuy

Whether the negative frequency aliasing helps or hurts depends on the nature of the signal you are trying to detect and the noisy environment it exists in.

What you are dealing with is not a "negative harmonic", but just an aliasing negative frequency. Think of a spectrum analyser view of the 10GHz signal. Replace the 10GHz with 0Hz. That is what you are creating with the mixer. Add LO feedthrough at 0 Hz, and LO phase noise. Then basically fold it over at zero. I expect LO noise and LO feedthrough will be the major issues with respect to the required signal level. You will have twice the noise bandwidth.

I have no idea how you will tune the LO to be right on to make this work. To the extent it drifts, the carrier drifts.

How fast are you trying to count? Is it just pulses (a signal level detector, basically)?

10. ### Mrcelloman99

7
I appreciate your replies.

I'm trying to build a speedometer for skis and longboards and the like that charts a user's speed by sending off off a signal and determining the velocity by reading the doppler-shifted return signal, like a radar gun only on a much smaller scale. I have a sensor that can do this; it transmits a ~10.525 GHz signal, receives the return signal, and mixes it with the same oscillator used for transmission (which is an integrated Dielectric resonator oscillator) to produce an IF signal that is at the doppler-shifted frequency, which should be DC to 6kHz.

The problem is, I need to output what the speed actually is on a wrist display, while the sensor will be mounted close to the ground, so I need to get the information wirelessly (hopefully) from the sensor to the display unit. To keep antennas small, I would like to transmit in the X-band, but the frequencies that contain the information will be DC-6kHz depending on how fast the user is going, hence, my downconversion predicament.

I'm also pretty inexperienced so forgive my naivety.

I think the sensor oscillators may be stable though and I may be able to access the onboard mixer in the device, which I could use to downconvert the signal for me. Otherwise, not sure how I would do it on my limited budget.

11. ### Mrcelloman99

7
Oh yeah, and I can't find VCos nearly stable enough to act as finely-tuned LO's so that's probably a no-go.

12. ### davenn

3,893
really wish people would describe their projects right from the start, it saves so so much guesswork

that's all cool and probably the easy part

that probably isn't a good idea. You should be using a lower freq that isn't so directional and also wont interfere with your Doppler shifted 10GHz received signal .... which is going to be really low level and easily swamped

use 2.4GHz or 5.7GHz for example

cheers
Dave

13. ### meBigGuy

Sure would have been easier if you had described the system initially.
If your DC to 6KHz is very noisy you will find it difficult to get robust and acurate results with a counter. You probably want a frequency to voltage converter like http://www.ti.com/lit/ds/symlink/lm2907-n.pdf

Bluetooth would be the method of choice for transmitting the signal to the watch.

Know someone interested in this topic? Share a link to this question via email, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted