Signal mixing

Discussion in 'The Projects Forum' started by Jadams, Feb 8, 2016.

Jan 14, 2016
8
0
Hi, I am having some slight problems with signal mixing and I thought you guys might be able to help. My aim is to mix two frequencies together and then filter out the difference, (f1-f2) using a low pass RC filter. The circuit is comprised of two Colpitts oscillators, one of which oscillates at about 400khz and the other at 405khz, so I am expecting an output of about 5khz. Both these signals are fed into a mixer (the circuit of which can be found here http://electronicdesign.com/site-files/electronicdesign.com/files/archive/electronicdesign.com/files/29/11935/figure_01.gif)

The problem I'm having is that the output is a slightly distorted sine wave (I have attached an image of the waveform). I have played around with the RC component values and the attached waveform shows one of the least pronounced 'dips' in the sine wave I could get. I have tested the circuit with waves from a function generator and it worked nicely - I used different test frequencies for that though and should have obviously tried the actual ones. I have a few theories as to why: I thought it could be a result of the since waves from the oscillators not being 'perfect' sine waves and this has resulted in some distortion? I am not sure how large an affect this would have because while the waves are not perfect they are pretty good. I also thought maybe the highish frequencies compared to the gap I am trying to filter out may have some influence?

I hope some of you can shine a light on this for me.

File size:
409.3 KB
Views:
16
2. RichardO Well-Known Member

May 4, 2013
1,157
360
The question that first comes to mind is: "What are all of the differences between the function generator circuit and the final circuit?"

My first guess is the output impedances of the signal sources. The function generator low-- only 50 ohms. Your Colpitts is going to be much higher than that.

Some other thoughts... The gain/bandwidth product of the LM6134 is not very high compared to 400 KHZ. The waveform in your picture is pretty noisy. How is the circuit assembled and how well is the power supply bypassed?

3. crutschow Expert

Mar 14, 2008
12,511
3,062
If you want low distortion output, then you may need a balanced modulator/mixer such as the 1496 or the 1595, or an analog multiplier.

Jan 14, 2016
8
0
Hi guys, so basically that waveform in the first image should be ignored, I was doing some testing today and noticed that a ground connection had come loose! A very annoying error to make and I assume the ground was connected when testing with the function generator. However, re running the circuit gives the attached waveform, it is a bit more sine wave like to me but still not enough. Does this type of shape to an output remind anyone of any different problems ? Playing the the values of the RC filter changes it a bit but still maintains a similar overall shape. RichardO, I used LM741s in my circuit and I think the kind of frequencies I have used are fine with it? I'm not sure what you mean by 'how is the circuit assembled' exactly though? I am powering it from a power supply unit in a lab so its pretty reliable. I have never heard of the 1496 or 1595 mixers but I will do some research on them. Thanks for your help so far I feel I am close to breaking through!!

5. RichardO Well-Known Member

May 4, 2013
1,157
360
Happens to the best of us.

A 741 has no chance of working here. It has a gain/bandwidth product of about 1 MHz. This gives an open loop gain of about 2 at the frequencies you are using.

All the connections in the circuit need to be short since any stray capacitance or inductance can cause problems. The power supply pins must be bypased with a ceramic cap (about o.1 uf) with short leads from the power pins of the op amp to ground.

Jan 14, 2016
8
0
Ah okay, the current assemblage a rough and quick one, the wires are long and the power supply pins do not have bypass capacitors on them! That is something I can easily rectify though. I don't quite understand your point on the 741 gain/bandwidth though, are you saying the gain is too low at the frequencies I am using?

Jan 14, 2016
8
0
From my understanding the gain of the op amps are controlled through the input/feedback resistors. The two input stage op amps have 10k on both which is unity gain. Unity gain, which is low, means you get a large bandwidth?

8. RichardO Well-Known Member

May 4, 2013
1,157
360
Good to hear that you are adding proper bypass.
Yes, the open loop gain of the 741 is way too low at 400 KHz. The 741 is barely works at audio frequencies.

9. RichardO Well-Known Member

May 4, 2013
1,157
360
True for both statements, but... Since the gain/bandwidth product of the 741 is only 1 MHz the open loop gain is only about 2 at 400 KHz.

Jan 14, 2016
8
0
Well I placed the LM741s with TL081s and got the attached waveform, after some amplification, which I am pretty happy with. Thanks for your help.

11. slackguy Member

Feb 11, 2016
29
1
> Hi, I am having some slight problems with signal mixing and I thought you guys might be able to help. My aim is to mix two frequencies together and then filter out the difference

opamp does than indiginously. they are widely available and inexpensive.

12. atferrari AAC Fanatic!

Jan 6, 2004
2,608
738
If you manage to eliminate the clipping in the bottom (saturation?) you could feel even better.

13. slackguy Member

Feb 11, 2016
29
1
that scope has that same wave feature your mentioning built-in, most do actually. i'm assuming you know that.

are the two signals in phase throughout the process? if out of phase the result may be correct.

i dont think a "low pass rc" filter is for differentiating signals but for attenuation - it's a cheap way to "squash" unwanted (high,low) frequency, and (depending on design) a r series C parallel "filter" would add ALLOT of distortion to a wave

you have 2CH you could obviously measure them both at all points in the process

14. SLK001 Active Member

Nov 29, 2011
746
197
You are not measuring what you want (405k-400k). The signal you show is a 1.3k Hz signal. It's also way to large for a higher order intermodulation product at 8.2V p-p.

You should use a bandpass filter to grab your signal - at 5k Hz a simple opamp filter should suffice.