EMRFD Message Archive 4275

Message Date From Subject
4275 2010-02-18 00:49:32 iami78 Double balanced mixer response to LO * 3
Solder smoky regards fellows!

I've search this data really hard, but haven't found it. EMRFD also doesn't cover this question: How much (dB) is the mix of LO*3 * RF/IF below the 1*LO*RF/IF.

I know that double balanced mixer responses quite well to the local oscillator's 3rd harmonic. I also noticed, that driving the local port will create harmonics, even when driving with very spectrally pure signal and thus the mixer will always response to the 3rd harmonic of the LO.

The problem occured in the 4 meter transverter I built. The local FM BC station leaks through from LO*3. For example with IF of 28.2 MHz (would mean I'm monitorin 70.2 MHz), the FM BC station might come through the the 28.2 MHz IF from 42*3-28.2 = 97.8 MHz.

I thought I was driving the LO port too hard, but when I tried to atteanuate the LO signal, the relation between 1*LO and 3*LO didn't change.

All ports are terminated with 3 dB attenuators.

73 de Janne OH1SDR
4276 2010-02-18 01:34:01 sm5glc Re: Double balanced mixer response to LO * 3
Hi Janne
I took a quick look in my ageing WJ catalogue "RF Signal Processing Components" and found the table I was looking for.
They suggests that 3*fLO+/-fRF is -13 dB using a 7dBm level mixer.

You will have a hard time to reduce this, as the LO is really acting as a square wave, even if you feed the mixer a sine wave, and thus will have a high spectral component at 3fLO, a perfect square wave has third overtone at -8 dB!

Using a square wave as LO input may improve the dynamic range vs. sine wave. This is related to the faster switching with the high-slew rate of a square wave.

May I suggest a low-pass filter with notches, i.e. elliptical (Cauer), that might help you with attenuating the FM-BC band, to compliment not replace your front-end filter.


Lasse SM5GLC


>
> The problem occured in the 4 meter transverter I built. The local FM BC station leaks through from LO*3. For example with IF of 28.2 MHz (would mean I'm monitorin 70.2 MHz), the FM BC station might come through the the 28.2 MHz IF from 42*3-28.2 = 97.8 MHz.
>
> I thought I was driving the LO port too hard, but when I tried to atteanuate the LO signal, the relation between 1*LO and 3*LO didn't change.
>
> All ports are terminated with 3 dB attenuators.
>
> 73 de Janne OH1SDR
>
4277 2010-02-18 02:00:06 Johan H. Bodin Re: Double balanced mixer response to LO * 3
Hi Janne,

I would guess that the difference in conversion gain between 1*LO and
3*LO is around 10dB if the mixer is driven hard. My guess is based on
the fact that the amplitude of the 3*f harmonic of a perfect square wave
is exactly 1/3 of the fundamental (1*f) amplitude.

I have been experimenting with mixers using CMOS analog switches on low
frequencies and these mixers were, as usual, RF polarity reversing
switches operating at the LO rate (like hard driven diode ring DBMs). In
other words, the RF signal was multiplied by alternating +/-1 which in
turn is equal to multiplying by a square wave. 1/3 amplitude = -9.54dB.

I may be wrong, I didn't do the trig math this time, but I said it was a
guess...

BTW, it is easier in the DSP domain where it is easy to make a clean
sinusoidal LO and a perfect mixer consists of nothing more than a
multiplication: *

;-)

73 de
Johan SM6LKM

----

iami78 skrev:
> Solder smoky regards fellows!
>
> I've search this data really hard, but haven't found it. EMRFD also doesn't cover this question: How much (dB) is the mix of LO*3 * RF/IF below the 1*LO*RF/IF.
>
> I know that double balanced mixer responses quite well to the local oscillator's 3rd harmonic. I also noticed, that driving the local port will create harmonics, even when driving with very spectrally pure signal and thus the mixer will always response to the 3rd harmonic of the LO.
>
> The problem occured in the 4 meter transverter I built. The local FM BC station leaks through from LO*3. For example with IF of 28.2 MHz (would mean I'm monitorin 70.2 MHz), the FM BC station might come through the the 28.2 MHz IF from 42*3-28.2 = 97.8 MHz.
>
> I thought I was driving the LO port too hard, but when I tried to atteanuate the LO signal, the relation between 1*LO and 3*LO didn't change.
>
> All ports are terminated with 3 dB attenuators.
>
> 73 de Janne OH1SDR
4278 2010-02-18 03:22:18 iami78 Re: Double balanced mixer response to LO * 3
Hi Lasse!

Thank you for the invaluable info Lasse!

-13 dB is quite much compared to the fundamental mixing product.
I think that the 28 MHz IF in 70 MHz transverter was a bad choise, but I didn't think that of much when I started the design progress, because 100% of the commercial 70 MHz transverters has the same IF.

At the presence of strong FM BC stations there will be problems, that's for sure. I used a quarter wave wire to measure the field strenght of the station that broke through, it was -40 dBm. Huge signal!

The present bandpass in the front end is triple top coupled resonator with quite low coupling - higher losses, but better out side of band attenuation. The elliptical with bandstop close to the BC station would definitely be the best opti
4279 2010-02-18 05:00:50 sm5glc Re: Double balanced mixer response to LO * 3
Aha!
This important information!!! A top-coupled filter has poorer high side rejection vs. lower side! The filter sort of "degenerates" to a high-pass filter look-alike! This is easy to see if you run the circuit in any simulator (I really reccomend the FREE RFsim99 which is super-easy).

You can re-design and use inductors as coupling devices, but it often becomes unpractically large values, or you could try some other variants having better high-side rejection

Just for fun, I simulated a 69,5 to 71,5 MHz filter using top-C and it gives ~68dB attenuation on 93,4 MHz, swapping to top-L improves to ~88 dB! BUT the coupling inductor shows at 27 microH, not something I would have on 70 MHz :) so I would rather try a shunt-C which gives fair component selection and 88 dB attenuation!
My qick-and-dirty design using 270 nH inductors shows largest capacitor 813 pF and smallest 19 pF. I'm sure you can tweak to suit your needs.

Still a 3 pole Cauer would do nicely as a notch, and should provide 35 dB attenutation on 2MHz spot around 95 MHz...

/Lasse

> The present bandpass in the front end is triple top coupled resonator with quite low coupling - higher losses, but better out side of band attenuation.

>
4283 2010-02-18 07:32:33 Tim Re: Double balanced mixer response to LO * 3
Not just 3*LO, but also 5*LO, 7*LO, etc.

7*LO works out to the middle of FM broadcast band for 14MHz.

The resulting sound when I hear this, is a kinda scratchy broad noise.

Trying to cure this problem with LO filtering is IMHO hopeless.

For 14MHz this completely goes away with a 3-element CLC lowpass filter.

4285 2010-02-18 07:55:16 Tayloe Dan-P26412 Re: Double balanced mixer response to LO * 3
I myself have not worried a lot about the high frequency response of a
top coupled bandpass filter because this normally sits behind the TX low
pass filter, which adds quite a bit of rejection to the third harmonic.

- Dan, N7VE

________________________________

4286 2010-02-18 09:02:37 Chris Trask Re: Double balanced mixer response to LO * 3
>
> I've search this data really hard, but haven't found it. EMRFD
> also doesn't cover this question: How much (dB) is the mix of
> LO*3 * RF/IF below the 1*LO*RF/IF.
>
> I know that double balanced mixer responses quite well to the
> local oscillator's 3rd harmonic. I also noticed, that driving
> the local port will create harmonics, even when driving with
> very spectrally pure signal and thus the mixer will always
> response to the 3rd harmonic of the LO.
>
> The problem occured in the 4 meter transverter I built. The
> local FM BC station leaks through from LO*3. For example with
> IF of 28.2 MHz (would mean I'm monitorin 70.2 MHz), the FM BC
> station might come through the the 28.2 MHz IF from
> 42*3-28.2 = 97.8 MHz.
>
> I thought I was driving the LO port too hard, but when I tried
> to atteanuate the LO signal, the relation between 1*LO and 3*LO
> didn't change.
>
> All ports are terminated with 3 dB attenuators.
>

Sounds like you are in need of a linear (meaning multiplying) mixer rather than a switching mixer. You can roughly approximate on by way of an LM1496/MC1496 or a suitable equivalent by driving the lower transistor pair with a sine wave LO, keeping the LO current in the mixer much less than the bias current but at least ten times greater than the RF signal current. The signal is applied to the upper quad of transistors.

A more ambitious approach is to apply the RF to the lower transistor pair and then feed the LO sine wave to the upper quad, but also add a pair of diodes across the LO line so as to approximate a translinear function that will better linearize this form of analogue multiplier.



Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
4291 2010-02-18 22:18:09 iami78 Re: Double balanced mixer response to LO * 3
Hi Lasse!

Your approximation/simulation didn't fall far from the throught - I have measured only around 65 dB attenuation for the BC FM station at the moment. If I would've been smart enough, I would haved used something like 50 MHz or 20 MHz IF, hihi. I already have RFsim99 in use, but tnx for suggestion.

By the way, I haven't found any software that would design shunt-c topology. What did you use? I only know one web page where you can design such filters.

Like Dan N7VE said (Tnx Dan for the comment!), I might just wait until I finnish the PA and use it's LPF (like you normally would do) to attenuate the FM band some more. The particual frequency 70.200 MHz isn't even in use here
4300 2010-02-19 14:21:21 Lasse Re: Double balanced mixer response to LO * 3
Janne,
I am fortunate enough to have access to Agilent Genesys at the office,
and it is a matter of minutes to try out various filter designs and
topologies :)... Not sure what software that would allow to design
filters automatically, and beeing free for hobbyists.

As for the WJ data, I found almost the same stuff published in their
Tech Journal, i.e.
http://www.triquint.com/prodserv/tech_info/docs/WJ_classics/vol10_n4.pdf
There are several intesting topics, so just d/l all of them!
http://www.triquint.com/prodserv/tech_info/WJ_tech_publications.cfm

Cheers
Lasse SM5GLC

iami78 skrev:
> Hi Lasse!
>
> Your approximation/simulation didn't fall far from the throught - I have measured only around 65 dB attenuation for the BC FM station at the moment. If I would've been smart enough, I would haved used something like 50 MHz or 20 MHz IF, hihi. I already have RFsim99 in use, but tnx for suggestion.
>
> By the way, I haven't found any software that would design shunt-c topology. What did you use? I only know one web page where you can design such filters.
>
> Like Dan N7VE said (Tnx Dan for the comment!), I might just wait until I finnish the PA and use it's LPF (like you normally would do) to attenuate the FM band some more. The particual frequency 70.200 MHz isn't even in use here on the bandplan, but of course the FM signal is very broad.
>
> We gave this a lot of thought at yesterday's club evening (OH1AA). It would be a lot easier just to change the IF frequency.
>
> Lasse, does the WJ catalogue provide additional info about other harmonic results, like LO*5 and LO*7 ? That would be really interesting to know!
>
> TU agn Lasse!
>
> 73 de Janne OH1SDR
>
4301 2010-02-19 14:34:51 w4zcb Re: Double balanced mixer response to LO * 3
There is the filter program called ELSIE done by ex WB6BLD ne W4ENE, Jim Tonne. A free student version of it is available from his website, and it does several different topologies.Might give it a look.

W4ZCB
----- Original Message -----