What is the Design Process for Building an RF Mixer Circuit?

[perplexabot]

Hello all. So I have been trying to build a mixer lately, I have done my research on google but I am still kind of lost. I know now many mixers exist; single diode, multiple diodes, single balanced, double balanced, unbalanced and so on. I don't want to build a circuit blindly, so I am trying to understand what is going on. I stumbled upon this pdf during my search: http://www.carmarthenradioclub.org.uk/mixer design.pdf . Specifically, at the bottom of page 5 of the pdf is a nice little circuit. I thought this would be a nice place for me to start since I have no previous experience with mixers.

I have a couple of questions about this circuit, if I may ask...

1) The low pass filter (LPF) will work for a certain frequency of LO, not a range, am I right?

2) How is the LPF attenuating LO from the RF port? I thought freq. of LO < freq. of RF, no?
(EDIT: or is this because they are assuming low-side injection?)

3) I have no idea what the job of the IF tuned circuits is. I know if they are tuned to IF freq. then they will have least impedance for IF freq., but what does that do to the circuit? Is it supposed to "attract" the IF signal towards the diode?

Thank you.

PS: any links, pdfs or sources that you think are good/worthy reads on this topic will be greatly appreciated.

 

[davenn]

...

I have a couple of questions about this circuit, if I may ask...

1) The low pass filter (LPF) will work for a certain frequency of LO, not a range, am I right?

2) How is the LPF attenuating LO from the RF port? I thought freq. of LO < freq. of RF, no?
(EDIT: or is this because they are assuming low-side injection?)

3) I have no idea what the job of the IF tuned circuits is. I know if they are tuned to IF freq. then they will have least impedance for IF freq., but what does that do to the circuit? Is it supposed to "attract" the IF signal towards the diode?

Thank you.

1) ... No, a LPF is just that, it will pass everything below a given freq.

2) ... No, in this case it is a "high Side " local osc ... else that filter would have been a HPF

3) ... The 2 x IF tuned circuits are providing BPF, band pass filtering for the IF to cut out any "residual" products of the Mixing, the RF in and the LO signals

You can see they are 2 different styles the first one is series resonant and the second one ( after the diode ) is parallel resonant ... see points 2 and 3 in the comments directly above the schematicThe circuit in question...

cheers
Dave

 

[perplexabot]

First of all thank you for the much needed reply and for posting the circuit.

1) ... No, a LPF is just that, it will pass everything below a given freq.

2) ... No, in this case it is a "high Side " local osc ... else that filter would have been a HPF

3) ... The 2 x IF tuned circuits are providing BPF, band pass filtering for the IF to cut out any "residual" products of the Mixing, the RF in and the LO signals

You can see they are 2 different styles the first one is series resonant and the second one ( after the diode ) is parallel resonant ... see points 2 and 3 in the comments directly above the schematic

I have some questions/comments to your answers, if that is ok.
1) I understand what you mean, but say I have a variable LO, I would need to have a "variable" LPF that changes with the LO frequency, right?

2) Ok great, now I know how the circuit differs according to injection type (they should have stated this in the pdf).

3) Interesting.I think I now understand what the parallel LC circuit does, it has high impedance for IF so this allows IF to go to the output while removing all others (aka bandpass filter). I still don't understand the series LC circuit though. Here is my guess, can someone correct if needed please:

The series LC circuit has low impedance for IF, so this sets up a voltage at the '+' of the diode and at the same time removes current from entering the diode, right? Why remove the current?

 

[davenn]

1) I understand what you mean, but say I have a variable LO, I would need to have a "variable" LPF that changes with the LO frequency, right?

No because the LO freq, is not going to be the same as the RF freq in at any time
there will still be the difference of the IF freq ... 10.7 MHz or whatever
in some transceivers I have seen quite hi first IF freq's up ~ 40 - 50 MHz

2) Ok great, now I know how the circuit differs according to injection type (they should have stated this in the pdf).

yup, that would have helped the beginner
But as least you won't get caught out again, as now you recognise the configuration

I'm not sure of the reasons why they are using one series and one parallel resonant filter circuits
rather than just 2 of either ?? or why they are using the series one before the parallel one

Maybe its to do with impedance matching around the diode ?

Maybe some one more clued up than me can give some answers to that one Dave

 

[perplexabot]

Thank you davenn, you have cleared much of my confusion!

I would still like to know more about the two tuned circuits. If anyone has input, please let me know.

 

[davenn]

I have asked for one of our forum Mentors to respond in the thread so you and I can both learn

cheers
Dave

 

[perplexabot]

I have asked for one of our forum Mentors to respond in the thread so you and I can both learn

cheers
Dave

Thank you good sir

 

[Baluncore]

L5 in series with C6; is a tuned circuit to peak the IF while strongly attenuating the RF and LO signals. That prevents saturation of the IF chain with unwanted RF that would cause all sorts of cross-modulation and desensitisation to the wanted signal.

L4 in series with C5; is a notch or trap to prevent backward traveling IF signal from passing through to the LPF and so radiating back out of the RF input.

L4-C5 can be considered to reflect the reverse traveling energy at the IF forward through the diode and so into the IF strip. It can also be seen as providing a short circuit to ground at the IF frequency, to give something solid for the IF voltage appearing across the diode to work against while driving the tuned L5-C6 at the IF.

 

[davenn]

L5 in series with C6; is a tuned circuit

Parallel I assume you meant to say

is a tuned circuit to peak the IF while strongly attenuating the RF and LO signals. That prevents saturation of the IF chain with unwanted RF that would cause all sorts of cross-modulation and desensitisation to the wanted signal.

Yes as I commented earlier ... I understand that

L4 in series with C5; is a notch or trap to prevent backward traveling IF signal from passing through to the LPF and so radiating back out of the RF input.

NO, I don't see how that can happen, since the injection point is BEFORE the filter
The whole idea of the LPF is to stop the LO from reaching the RF input point ... the text even states that


the question was ... I'm not sure of the reasons why they are using one series and one parallel resonant filter circuits rather than just 2 of either ?? or why they are using the series one before the parallel one ??


Still awaiting a visit to the thread by Berkeman

cheers
Dave

 

[perplexabot]

L5 in series with C6; is a tuned circuit to peak the IF while strongly attenuating the RF and LO signals. That prevents saturation of the IF chain with unwanted RF that would cause all sorts of cross-modulation and desensitisation to the wanted signal.

L4 in series with C5; is a notch or trap to prevent backward traveling IF signal from passing through to the LPF and so radiating back out of the RF input.

L4-C5 can be considered to reflect the reverse traveling energy at the IF forward through the diode and so into the IF strip. It can also be seen as providing a short circuit to ground at the IF frequency, to give something solid for the IF voltage appearing across the diode to work against while driving the tuned L5-C6 at the IF.

Hey, thanks for the reply. I don't exactly understand what you mean by "saturation of the IF chain."

I also have another question to anyone that can answer. By 50Ω (at the RF and LO input), do they mean that the source is assumed to have 50ohm output impedance? If yes, is that why they are designing the LPF to be 50ohm? Impedance matching?

 

[perplexabot]

NO, I don't see how that can happen, since the injection point is BEFORE the filter
The whole idea of the LPF is to stop the LO from reaching the RF input point ... the text even states that

Hmmmm. I think he stated what I was stating earlier. The IF signal will pass through the series L4 C5, due to the fact that Z->0 as ω->ω0 and will be a open circuit to all other frequencies, this sets up a voltage at the positive end of the diode... The left over unwanted frequencies that pass through the diode are then filtered out through the parallel C6 L5 BPF. What do you think?

 

[Baluncore]

davenn. Yes thanks, I did mean L5 in parallel with C6 is a tuned circuit.

The whole idea of the LPF is to stop the LO from reaching the RF input point ... the text even states that

Yes, that is why the LPF is there. But IF energy could pass backwards through the LPF to the RF input.

IF energy appears across the series detector diode, the IF energy could flow in either direction from the diode.
The C6-L5 IF tuned circuit, opposed by the C5-L4 IF trap, prevents IF energy appearing on the LPF side of the diode and so reaching the RF input. C5-L4 is a short circuit to ground for the IF.

Desensitisation… http://en.wikipedia.org/wiki/Desensitization_(telecommunications)
If the IF signal is smaller than the LO signal entering the IF amplifier then the IF amplifier could saturate with LO. The AGC would reduce the gain of the IF amplifier based on the nuisence LO signal. That would attenuate the IF signal wanted, the detector would see the LO not the IF signal.

 

[perplexabot]

Hey, sorry for the redundant post, but can someone please answer my question in post 10?

 

[Baluncore]

You had several questions in post #10. I answered desensitisation ...

50 ohm, yes, it is for impedance matching ...
RF equipment is usually connected with standard 50 ohm coaxial cable. Most modules designed for experimentation are optimised for 50 ohm. It is only inside manufactured equipment where there is an advantage in using a different impedance.

A low pass filter can be designed to efficiently transform an impedance over a very wide bandwidth by using several stages. It was therefore necessary in the reference to specify that both ports of the LPF should be 50 ohm.

 

[perplexabot]

You had several questions in post #10. I answered desensitisation ...

You are right. I should have been more clear and I should have also thanked you for your previous answer (and also this one), so thank you very much!

50 ohm, yes, it is for impedance matching ...
RF equipment is usually connected with standard 50 ohm coaxial cable. Most modules designed for experimentation are optimised for 50 ohm. It is only inside manufactured equipment where there is an advantage in using a different impedance.

So if I wanted to connect an antenna to the RF input, I would have to check for the antenna's impedance and match the LPF to it?

A low pass filter can be designed to efficiently transform an impedance over a very wide bandwidth by using several stages. It was therefore necessary in the reference to specify that both ports of the LPF should be 50 ohm.

What if my LO and IF have different impedances, what then should be the matching impedance of the LPF (or should I simply add a resistance in [EDIT] [STRIKE]parallel[/STRIKE] series with either LO or IF to have them the same)?

I know I am asking too many questions, and they are all over the place. I am sorry for that, please be patient with me. I just need to understand this circuit and I don't see any other way

 

[Baluncore]

So if I wanted to connect an antenna to the RF input, I would have to check for the antenna's impedance and match the LPF to it?

Probably not, you would first match the antenna to a 50 ohm coaxial cable, then connect the other end of the coaxial cable to the 50 ohm RF input.

What if my LO and IF have different impedances,

The output impedance of the IF is 50 ohm. It resistively loads the C6-L5 tuned circuit and so sets the Q or IF bandwidth. If you change the IF Zout you should also change C6 and L5 reactance at the IF frequency to maintain the Q or BW.

The LPF only looks like 50 ohm over it's passband. At the LO frequency, the LPF will appear capacitive, as C4 is going to dominate the injection point. The C5-L4 IF trap is far enough away from the LO frequency to appear high impedance, and so is unimportant. The remaining effect will be the impedance of the mixer diode driving the C6-L5-50R. The diode is non-linear and so can be impossible to match perfectly across all signal strengths. There will be some mismatch at the LO injection point due to the LPF and the loaded detector diode. It is a compromise.

The output impedance of the LO is specified as 50 ohm. That means the LO is able to produce a signal with a 50V/1A ratio. There is no guarantee that the LO will look like a 50 ohm resistor at the RF input frequency. Some of the RF energy from the LPF may end up in the LO output.

By using a 3dB attenuator pad between the LO input and the injection point, a better match for the LO would be obtained, but more RF would be lost in the LO pad. More LO energy would also be required to overcome the 3dB pad.

 

[perplexabot]

Probably not, you would first match the antenna to a 50 ohm coaxial cable, then connect the other end of the coaxial cable to the 50 ohm RF input.

The antenna I got, has a 75Ω matching transformer, is that what you speak of (i believe it has a coaxial input and 2 wires as outputs)? The manual that comes with the antenna states that the matching transformer is to be used with FM. Can I use this (I am doing AM)? Or should I get a 50Ω one?

The output impedance of the IF is 50 ohm. It resistively loads the C6-L5 tuned circuit and so sets the Q or IF bandwidth. If you change the IF Zout you should also change C6 and L5 reactance at the IF frequency to maintain the Q or BW.

I will need to give this more thought. Thank you.

The LPF only looks like 50 ohm over it's passband. At the LO frequency, the LPF will appear capacitive, as C4 is going to dominate the injection point. The C5-L4 IF trap is far enough away from the LO frequency to appear high impedance, and so is unimportant. The remaining effect will be the impedance of the mixer diode driving the C6-L5-50R. The diode is non-linear and so can be impossible to match perfectly across all signal strengths. There will be some mismatch at the LO injection point due to the LPF and the loaded detector diode. It is a compromise.

The output impedance of the LO is specified as 50 ohm. That means the LO is able to produce a signal with a 50V/1A ratio. There is no guarantee that the LO will look like a 50 ohm resistor at the RF input frequency. Some of the RF energy from the LPF may end up in the LO output.

WOW! That is crazy (in a good way of course), your saying the unmixed LO is forced to go through c4, right? So does this also mean the LPF can only be matched to the RF input (which is 50Ω, in this case) and NOT to both RF input AND LO input?

By using a 3dB attenuator pad between the LO input and the injection point, a better match for the LO would be obtained, but more RF would be lost in the LO pad. More LO energy would also be required to overcome the 3dB pad.

That is interesting. I am not familiar with attenuators, I will look into them. Thank you! So much info!

 

[Baluncore]

You should specify what your RF band, IF and LO frequencies will be.
That will make a big difference to antenna coupling and filter design.
What type of antenna are you using. What are you using as an IF.

The mixer design you are considering is not a current design, it is from the 1960s. There are now many balanced mixers available, (some later in chapter 7), that will prevent re-radiation of the LO and IF without any need to build and calibrate many tuned circuits. The last thing you want to do is re-radiate LO or IF, but there there are easier ways to avoid re-radiation.

The impedance standard for TV reception became 75 ohm because that is one quarter of 300 ohms, the impedance of a folded dipole. 75 ohm cable has slightly lower loss than 50 ohm cable.

The twin leads on your transformer are probably designed to be used as 300 ohm balanced parallel transmission line or a folded dipole. The coaxial connector is specified for unbalanced 75 ohm. It should work OK as a 200 ohm to 50 ohm transformer.

 

[perplexabot]

You should specify what your RF band, IF and LO frequencies will be.
That will make a big difference to antenna coupling and filter design.
What type of antenna are you using. What are you using as an IF.

• RF band: Medium wave for now.
• IF: If I am not mistaken this can be arbitrary, but there are standards for it. I have chosen 460KHz
• LO: I am using my frequency generator for this. It will be variable, since I am designing (or attempting to design I should say) a superheterodyne AM receiver. My frequency generator does have a 50 ohm output.

The mixer design you are considering is not a current design, it is from the 1960s. There are now many balanced mixers available, (some later in chapter 7), that will prevent re-radiation of the LO and IF without any need to build and calibrate many tuned circuits. The last thing you want to do is re-radiate LO or IF, but there there are easier ways to avoid re-radiation.

Yes, I understand I am using a "low quality" mixer design. I chose this circuit due to it's simplicity. Once again, I have no prior experience with RF circuitry. Do you still think I should start with something better as a beginner?

The impedance standard for TV reception became 75 ohm because that is one quarter of 300 ohms, the impedance of a folded dipole. 75 ohm cable has slightly lower loss than 50 ohm cable.

The twin leads on your transformer are probably designed to be used as 300 ohm balanced parallel transmission line or a folded dipole. The coaxial connector is specified for unbalanced 75 ohm. It should work OK as a 200 ohm to 50 ohm transformer.

I don't exactly understand what you mean here. You say it should work as a 200Ω to 50Ω transformer. Do you mean the antenna's output impedance is 200Ω and the transformer brings it down to 50Ω?

 

[Baluncore]

Let us assume that your IF is the traditional 455 kHz.
The MW band is from about 530 kHz to 1600 kHz.
Your mixer will generate IF = LO – RF, but there will also be an image at IM = LO + IF.

Consider the frequency ranges covered.
IF fixed at 455 kHz.
RF from 530 to 1600 kHz. MW band.
LO from 985 to 2055 kHz. = RF + IF
IM from 1440 to 2510 kHz. = LO + IF

See the attached diagram that shows clearly these relationships.
Horizontal axis is frequency. Vertical axis is effectively the tuning dial.

It shows that the RF overlaps the LO and demonstrates that a fixed LPF is not applicable if you want to cover the MW band.

The RF stage of MW superhets had a tuned RF stage that tracked the RF to eliminate the image and other strong MW signals. That is why the tuning capacitors used two or more gangs. One for the LO, another for the tracking RF BPF.

I don't exactly understand what you mean here. You say it should work as a 200Ω to 50Ω transformer. Do you mean the antenna's output impedance is 200Ω and the transformer brings it down to 50Ω?

A transformer has an impedance ratio = turns ratio squared.
The coaxial connector design determines the optimum impedance of the transformer, but below 100 MHz it is not that critical. When a 50 ohm coaxial cable is viewed through your impedance transformer, the antenna will see 200 ohms. You have not described your antenna, but it is safe to say that over the MW band, your antenna is unlikely to have a fixed impedance.

 

[perplexabot]

Let us assume that your IF is the traditional 455 kHz.
The MW band is from about 530 kHz to 1600 kHz.
Your mixer will generate IF = LO – RF, but there will also be an image at IM = LO + IF.

Consider the frequency ranges covered.
IF fixed at 455 kHz.
RF from 530 to 1600 kHz. MW band.
LO from 985 to 2055 kHz. = RF + IF
IM from 1440 to 2510 kHz. = LO + IF

See the attached diagram that shows clearly these relationships.
Horizontal axis is frequency. Vertical axis is effectively the tuning dial.

It shows that the RF overlaps the LO and demonstrates that a fixed LPF is not applicable if you want to cover the MW band.

That was exactly what I was speculating at the beginning of this thread (post #2, point 1). So I would have to a variable LPF for this particular mixer. Would a better mixer void this requirement (as in not require a variable LPF)?

The RF stage of MW superhets had a tuned RF stage that tracked the RF to eliminate the image and other strong MW signals. That is why the tuning capacitors used two or more gangs. One for the LO, another for the tracking RF BPF.

I was going to skip the RF stage as I assumed it was there to only remove the image station. I also assumed that image stations can be disregarded when trying to receive major stations. I was going to add this stage once I had a simple working receiver.

A transformer has an impedance ratio = turns ratio squared.
The coaxial connector design determines the optimum impedance of the transformer, but below 100 MHz it is not that critical. When a 50 ohm coaxial cable is viewed through your impedance transformer, the antenna will see 200 ohms. You have not described your antenna, but it is safe to say that over the MW band, your antenna is unlikely to have a fixed impedance.

I have the following antenna: http://www.terk.com/radio-antennas/?sku=AM-FM+
I am still not sure how I should connect my antenna to the mixer.

For now I need to know what mixer I should start with and how I can interface my antenna with my IF mixer. Once again, I am assuming I can disregard image stations for now (not sure if this is correct).

THANK YOU!

 

[AlephZero]

So if I wanted to connect an antenna to the RF input, I would have to check for the antenna's impedance and match the LPF to it?

Probably not, you would first match the antenna to a 50 ohm coaxial cable, then connect the other end of the coaxial cable to the 50 ohm RF input.

RF band: Medium wave for now.
Do you mean the antenna's output impedance is 200Ω and the transformer brings it down to 50Ω?

I was going to skip the RF stage as I assumed it was there to only remove the image station. I also assumed that image stations can be disregarded when trying to receive major stations. I was going to add this stage once I had a simple working receiver.

I have the following antenna: http://www.terk.com/radio-antennas/?sku=AM-FM+
I am still not sure how I should connect my antenna to the mixer.

The web page says your antenna has 300Ω impedance at AM.

But if you are starting with a simple AM receiver for strong local stations, the antenna design probably doesn't matter much. Even a random length of wire would be pretty effective - assuming you don't live hundreds of miles from the nearest transmitter, of course.

Maybe two comments will cut through some of the confusion here:
1. To design any type of filter or mixer circuit, you need to know its input and output impedance, even if you choose them for no particular reason.
2. Aside from the image rejection of a tuned RF stage, even an untuned RF stage will have a well-defined output impedance, which is the input impedance to your mixer. So the RF stage separates the rest of the receiver from the vagaries of whatever aerial happens to be in use.

I would be inclined to have a simple, low-gain, untuned RF stage in your first design because of (2). You can always improve on it later.

 

[perplexabot]

The web page says your antenna has 300Ω impedance at AM.

But if you are starting with a simple AM receiver for strong local stations, the antenna design probably doesn't matter much. Even a random length of wire would be pretty effective - assuming you don't live hundreds of miles from the nearest transmitter, of course.

That is interesting.

I would be inclined to have a simple, low-gain, untuned RF stage in your first design because of (2). You can always improve on it later.

Ok, I guess I will add this to my design then. This will set the impedance to one of the mixer inputs (as you were saying). So for an untuned RF stage, would a buffer do (simple source follower)? Or what exactly should I be looking for? I quickly googled "untuned RF stage" with no luck.

Maybe two comments will cut through some of the confusion here:
1. To design any type of filter or mixer circuit, you need to know its input and output impedance, even if you choose them for no particular reason.
2. Aside from the image rejection of a tuned RF stage, even an untuned RF stage will have a well-defined output impedance, which is the input impedance to your mixer. So the RF stage separates the rest of the receiver from the vagaries of whatever aerial happens to be in use.

Mixer1 will see 2 input impedances: (the now added) RF stage output impedance and the frequency generator output impedance. I don't know how to treat 2 different input impedances to the mixer. So for the output impedance seen by the mixer, will I have to calculate (or measure) the rest of the circuit?

One more thing, from what I concluded from point 1, designing a mixer that works as a modular component is not possible due to the fact that different circuits will have different i/o impedances?

Thank you so much for the help.

 

[Baluncore]

Because the tuned RF stage of a superhet tracked the RF, it also prevented re-radiation of the LO and IF. No LPF was needed where a narrow BPF was used.


Aircraft and controllers use the band from 108 to 138 MHz for AM voice communication in and around airports. The AM channels are separated by 100 kHz. The international emergency channel is on 121.5 MHz.
If you built a mixer to down-convert aircraft band signals to about 1.5 MHz you could use an AM radio as your IF, detector and audio stages. Your FM antenna should work OK to receive the VHF. A LO range over the same band would be needed, and the mixer would need to be balanced so as not to re-radiate the LO.

Take a look at this NXP mixer chip SA612A. Download the data sheet, see the examples. http://www.nxp.com/products/rf/fm_if_system_ics/SA612AD.html

 

[perplexabot]

Because the tuned RF stage of a superhet tracked the RF, it also prevented re-radiation of the LO and IF. No LPF was needed where a narrow BPF was used.


Aircraft and controllers use the band from 108 to 138 MHz for AM voice communication in and around airports. The AM channels are separated by 100 kHz. The international emergency channel is on 121.5 MHz.
If you built a mixer to down-convert aircraft band signals to about 1.5 MHz you could use an AM radio as your IF, detector and audio stages. Your FM antenna should work OK to receive the VHF. A LO range over the same band would be needed, and the mixer would need to be balanced so as not to re-radiate the LO.

Take a look at this NXP mixer chip SA612A. Download the data sheet, see the examples. http://www.nxp.com/products/rf/fm_if_system_ics/SA612AD.html

Thank you!

Amazing! Truly amazing. I think you may have fixed all my RF mixer issues with that circuit. I should have gone active from the beginning I guess. A quick inspection of the circuit, I am wondering 2 things:

• What are those boxes with "Bias"?
• Do I still have to worry about impedance matching (connecting my RF stage to mixer input) with such a circuit?

EDIT:
I have gained a great deal of knowledge from this thread, but I am at a point now where I am completely clueless.

I think I will look into gilbert cells. I need to figure out passive or active. I also need to figure out impedance matching... I obviously don't have enough knowledge about this topic, I need to read a bit more.

I want to thank everyone for helping. I seriously don't know what I would do without you guys. Thank you.

 

[Baluncore]

Note that the SA612AD is in a small surface mount package. The SA612AN is the bigger original package with pins, which makes experimentation easier.

What are those boxes with "Bias"?

“Bias” is inside the chip. It provides the DC reference voltages and currents needed to accurately position and balance the circuit at the point where it works best. Internal bias saves you having to use external components. For example, the signal inputs In_A and In_B should be floating or capacitor coupled as signal input DC bias is provided by the chip.

Do I still have to worry about impedance matching …

Not really. For early experimentation, accurate impedance matching is probably a distraction. Once you have the circuit working you can improve the performance by sorting out the matching.

EDIT: I have gained a great deal of knowledge from this thread, but I am at a point now where I am completely clueless.

That indicates you have begun to understand the design problem.
http://en.wikipedia.org/wiki/Dunning–Kruger_effect

I want to thank everyone for helping. I seriously don't know what I would do without you guys. Thank you.

Thank you. That is what we are here for.
Without us you would have to get an old copy of the ARRL Handbook. Get one anyway.