Frequeny shifting using SSB Modulation

[extcjoey]

Hi All,

I'm working on a project using the SSB method to shift a band of frequencies. I'm considering both the "Phasing" and "Weaver" methods. For the phasing method, I'm using an HEF4013 dual D flip-flop and the Hc4066 analog switch to modulate an incoming sine wave. I'm totally confused about the Vdd supply which can from 5 - 15 V and gives better mark-to-space ratios for higher supply. If i do use 15V , how can i connect it to the 4066 which only accepts max +5 V? I am using a Pico ADC 212 spectrum analyzer also, and on mixing a 40 khz sq signal from the hef4013, with a 50khz sine signal, i get significant signals at 20khz,60khz, 70khz which i cannot explain (and i have ruled out aliasing). Is this some second harmonic distortion from the flip flop square wave? Is there any better quadrature circuit or switch i can use to improve this? I had the entire phasing circuit set up sometime ago, using an opamp (LM318) as the final summing stage, but could not adequately reject the unwanted sideband. I'm really at my wits end as how to proceed from here Please please help!!

 

[Nigel Goodwin]

It's rather vague what you're trying to do?, but it's obviously essential that you use sinewaves and NOT squarewaves (which will have masses of high level harmonics).

The filter method for SSB requires a VERY sharp filter, which is why it's normally only used for voice bandwidth - and the carrier frequency is many times higher than the audio - your frequencies seem MUCH too close to each other?.

 

[extcjoey]

Hey thanks for replying. I guess what im trying to do here is step down a band of frequencies (40khz-500khz) to 0 to 40 khz frequencies. I had thought of using sine waves as this is how the AM and hence SSB is typically produced. However, the problem with sine waves is that exact quadrature synthesis is not always possible, and i need to produce several local osc signals in quadrature and maintaining required amplitude constant. I have seen a few circuits where the phasing method is carried out using square waves in the mixer. I figured i'd keep switching the sq wave frequency like- for the band 40khz - 80 khz, i'd use a 40 khz or 80 khz sq wave, for 80 - 120khz use 80 or 120khz etc. I have tried the cd4013 and hef4013 dual d flip-flops , but i think that theyre the cause of my second harmonics showing up. According to the data sheets, increasing the supply voltage can reduce the rise/fall times, but in that case, my o/p will be of far greater amplitude than my hc4066 bilateral switch can handle (datasheet max Vdd is 5V), i was hoping if i could somehow find a solution for this?

 

[JimB]

I think the problem is that if you put frequencies F1 and F2 into a mixer, you dont just get F1+F2 and F1-F2 (and F1 and F2) at the output, you also get 2F1-F2 and 2F2-F1 plus lots of other stuff.

Can you tell us what you are trying to achieve?
There may be a more effective way to do it.

JimB

 

[Analog]

Why SSB? Why not just use a mixer with standard AM and then a LPF?

 

[Nigel Goodwin]

Yes, what has SSB got to do with it?, and why a flip flop?, or a 4066? - all you need is a double balanced mixer and a filter - and it's essential that you use sinewave carrier oscillators, you can't use squarewaves to do this.

You might also consider why spectrum analysers commonly use very high IF frequencies? - there are good reasons for this!. There are a number of homebrew analyser designs on the net, you might find it useful studying them!.

 

[k7elp60]

I agree with Nigel and Analog. Use a SA602 mixer and a active low pass filter.
I once did something very similar. I mixed 25Khz with 21Khz and used a active lowpass filter to get 4Khz. Like Nigel says using sine waves is essential.

 

[stevez]

My understanding of a pure sine wave is that all of the energy is at the fundamental frequency and there is virtually no energy at other frequencies. Generating a pure sine wave with no other noise, harmonics, etc is a challenge all by itself. A square wave, as already implied here, is quite the extreme departure from a pure sine wave and contains lots of energy elsewhere in the spectrum. A theoretical breakdown of the square wave would yeild energy at predictable frequencies. All of that assumes a perfect square wave.

If you apply the math that JimB suggested, to anything other than a pure sine wave, I would think you'd then generate an awful lot of stuff with much of it in the spectrum you are trying to analyze. The less than perfect sine wave may be some of the cause of the "other stuff" that JimB mentions. Now, the energy levels would be way down but they might be sufficient to cause problems.

 

[extcjoey]

One more thing - i was talking to someone, and they hinted that a balanced mixer can cancel out my unwanted signal? I dont see how to go about building one using the HC4066 though, but the AN1981 article seems to use a balanced setup...?

 

[kchriste]

Your project using the 4066 sounds like it uses technology simular to the old frequency domain scrambers used for analog voice encryption. It's been 20 years since I've worked on one those buggers so I'm spotty on the details. Anyway, here's a link to a chip that may (Then again maybe not) provide some insight:

https://www.electro-tech-online.com/custompdfs/2006/07/cmx264ds.pdf

To use a higher voltage with the HC4066, you need to get rid of the HC and go with a plain old 4066. Then you can power it from 16-18V depending on the manufacturer.

 

[JimB]

OK lets try and summarise:

You have an existing spectrum analyser which covers 0 to 30khz.
You want to make a "frequency extender" so that it can be used in the range 30 to 500(?) khz.
You appear to be talking about generating SSB signals.
Your terminology is confusing the hell out of everybody including me!


OK, what I think you need to consider is what is known as a "direct conversion receiver".
In its usual form, this takes an SSB signal and converts it directly to baseband (audio).
What you need is direct conversion receiver with a wide baseband (0 to 30khz).
From the datasheet you posted, look at figure 13, and start from there.
The baseband phase shift circuits (fig 14) may be a bit of a challenge.

Does that help?

JimB

 

[Nigel Goodwin]

One more thing - i was talking to someone, and they hinted that a balanced mixer can cancel out my unwanted signal? I dont see how to go about building one using the HC4066 though, but the AN1981 article seems to use a balanced setup...?

What is the relevence of that application note to what you want to do?, it's for SSB - you're not doing SSB, so why are you following it?. All you need is a simple down converter, as done in any superhet radio (or spectrum analyser).

1) Input band-pass filter.
2) Double balanced mixer (SA602 would be good).
3) Output band-pass filter (could be low-pass for your purposes).
4) Carrier oscillator (sine wave, preferably crystal controlled).

 

[extcjoey]

OK lets try and summarise:

You have an existing spectrum analyser which covers 0 to 30khz.
You want to make a "frequency extender" so that it can be used in the range 30 to 500(?) khz.
You appear to be talking about generating SSB signals.
Your terminology is confusing the hell out of everybody including me!


OK, what I think you need to consider is what is known as a "direct conversion receiver".
In its usual form, this takes an SSB signal and converts it directly to baseband (audio).
What you need is direct conversion receiver with a wide baseband (0 to 30khz).
From the datasheet you posted, look at figure 13, and start from there.
The baseband phase shift circuits (fig 14) may be a bit of a challenge.

Does that help?

JimB

Hey Jim and the rest of u guys. This actually is my first electronics project and I'm not familiar with all the conventions yet, but im learnin' ....Anyways, a "direct conversion receiver" is exactly what i'm looking for and i did come across that term in my research. The method used in the AN1981 figure 13 is the "third method" for SSB, and I'm trying something similar only using the "phasing method" see? I'm using the HC4066 because it seems to be adequate for use upto 500khz and its cheap. I have the whole thing setup at a single test frequency (50khz sine) but still getting a lot of the upper sideband as well as other problems i have mentioned in previous posts. Now i have just a couple of weeks to figure out the best way to do this thing and i could really really use all the help i can get

 

[Nigel Goodwin]

Now i have just a couple of weeks to figure out the best way to do this thing and i could really really use all the help i can get

Read my previous post! - that's EXACTLY what you need.

You're just confusing yourself further, first with SSB, and now with 'direct conversion' - you're NOT building a radio, so why start with a radio?.

'Direct conversion' is actually the same as my post, but a specific limited application of it - it's really a VERY, VERY simple, standard, common, radio technique - any radio amateur will now exactly how to do it.

 

[JimB]

Nigel

I think a simple converter (like a 2 metre converter from day gone by) would be fine but for the frequencies involved in this case.
The big problem as I see it is image rejection. because the "IF" is 0 to 30 khz, the image rejection is going to be a BIG problem.
Either, it will need some really good filter at the measuring frequency, or the images can be rejected by using a phasing technique, as has been used in some direct conversion receivers. (The simple type of DC receivers have zero image rejection).

One thing I have to ask is:
1 Is there a further goal to this project, ie is it to be used as a measuring tool as part of some other activity.
or,
2 Is it just a project to learn more about various aspects of radio/electronics.

If the answer is 1, I think the requirements need to be defined more precisely.
If the answer is 2, pick a range of frequencies to convert, make a converter, try to understand its good points and its bad points.

Somewhere extcjoey commented that this was his first project, sounds rather ambitious to me.

JimB

 

[Nigel Goodwin]

Nigel

I think a simple converter (like a 2 metre converter from day gone by) would be fine but for the frequencies involved in this case.
The big problem as I see it is image rejection. because the "IF" is 0 to 30 khz, the image rejection is going to be a BIG problem.
Either, it will need some really good filter at the measuring frequency, or the images can be rejected by using a phasing technique, as has been used in some direct conversion receivers. (The simple type of DC receivers have zero image rejection).

That's one reason why spectrum analysers up-convert, to a very high IF, so you get good image rejection.

Presumably the reason for the 0-30KHz range is because it's an audio analyser?, and not a real spectrum analyser?.

Somewhere extcjoey commented that this was his first project, sounds rather ambitious to me.

I agree!.

 

[extcjoey]

I'm thinking of using a double balanced mixer and I need to have my input signal in quadrature as well as inverted. Could i use the TL084 jfet opamp for this? will it give me uniform output for all input frequencies (i.e 40khz to 500khz)?

 

[Nigel Goodwin]

I'm thinking of using a double balanced mixer and I need to have my input signal in quadrature as well as inverted. Could i use the TL084 jfet opamp for this? will it give me uniform output for all input frequencies (i.e 40khz to 500khz)?

No, it's fairly unlikely - for RF use you need RF techniques, and not audio ones. But why do you want inversion and quadrature?, use a double balanced mixer that just needs single ended inputs.