EMRFD Message Archive 8364
Message Date From Subject 8364 2013-03-19 13:19:48 Bob NE602 Conversion Gain Experiment Hi All -
While working on a sweep frequency RF source I stumbled upon an interesting and I think generally unknown behavior of the popular NE602 IC. This has to do with how, and how much, LO signal is injected.
The app notes for this IC describe using an external injection level of 200 mV. As best I can tell this means 200 mV p-p. I've also seen reference to using up to 300 mV p-p. However, I observed that when using the internal LO oscillator the effective conversion gain is noticibly higher than with the recommended level.
I was unable to measure the internal LO oscillator amplitude (probing affects the result) so I resorted to measuring the gain while varying the external drive and plotting the results. These results are given in the Files folder "K3NHI-NE602".
I suspect the QRP Tech Group may also find this information useful so I will repeat the Post there as well. All comments are welcome.
8365 2013-03-19 15:00:46 Roelof Bakker Re: NE602 Conversion Gain Experiment Hello Bob,
Thank you for your research.
I have always assumed that the recommended LO level is 200 mV RMS or about
I am happy to see that this agrees very well with your findings for
maximum conversion gain.
8366 2013-03-19 19:15:57 Nick Kennedy Re: NE602 Conversion Gain Experiment Hi Bob,
I went through similar considerations a few years ago building a 60M SSB
TX. My notes say:
"The data sheet says use 200 mV p-p minimum. EMRFD suggests that 0 dBm
into 50 ohms, or 630 mV p-p produces good results. No conflict there."
Well, wait a minute ... maybe there is a conflict. This is directly from
the Philips SA612 data sheet:
"External L.O. should be 200mVP-P minimum to 300mVP-P maximum."
Anyway, I used about 600 mVp-p.
8367 2013-03-19 19:44:27 Ashhar Farhan Re: NE602 Conversion Gain Experiment bob,
I tried the NE602 in the arduino sweeper that I built last year. As we
had discussed in our personal correspondence, using it to sweep LC
filters produced spurious response below the pass-band frequency. A
closer examination revealed that these were due to the harmonics
generated by distortion within the NE602. When I replaced the NE602
with an ADE-1, the results were much better.
I suppose we can reduce the IMD products by reducing RF input drive to
NE602. I did try that, however, I had to dip down to -40dbm (I don't
have the lab notes available right now to confirm this) before the
spurious responses were reduced. That meant that extra amplification
was required to get the signal up to 0dbm level.
8372 2013-03-19 21:51:56 Gene Dorcas Re: NE602 Conversion Gain Experiment Valuable info since I'm adding a DDS LO to my receiver and was wondering
what the optimum LO input level should be.
Thanks for your good work,
8375 2013-03-19 21:51:56 Leonard Re: NE602 Conversion Gain Experiment Here's a rig that shows 200-300mvpk-pk
8378 2013-03-20 06:08:00 victor Re: NE602 Conversion Gain Experiment Bob, your results are somewhat pessimistic compared to theory.
The mixing operation in NE602 or any mixer with Gilbert cell topology is done by two differential pairs. Each differential pair receives the LO signal differentially between the base of the transistors. As I remember from theory, 90% the total current of a differential pair will be directed to one transistor for 50mV input differential voltage, or 98% for 100mV. 100mV input differential voltage is really 200mV Peak to peak, so this is the reas
8379 2013-03-20 07:32:18 Bob Re: NE602 Conversion Gain Experiment Hi Victor et al -
I fully agree - it surely seems like the 200 mV p-p should be plenty - BUT - measured results are at odds with this, thus:
I was experimenting with a chain of functional blocks one of which was a 602 mixer. The gain of the overall chain came up short about 5 dB. I traced this back to the mixer block which utilized an external LO drive. I decided that particular 602 must be defective so I replaced it - but got the same result.
I then shifted over to a different 602 assembly that happened to use the IC LO and got the "right" outcome. This led me to question "external" vs "internal" LO drive and do the experiment I have described herein.
In summary, I used 3 different (mfgr / lots) of 602's in two different circuit assemblies which led to the results presented. At this point I have no choice but to once again acknowledge the dominance of the "To Measure Is To Know" creed. However, at the same time, I do welcome any data to the contrary (or in concert). This is why I made this Post in the first place - i.e. it just didn't seem "right", but after all, TMITK.
Cordially and Thank You,