[minima] Re: J-KISS Version 3
- From: "Steve VK2SJA" <vk2sja@xxxxxxxxxxx>
- To: minima@xxxxxxxxxxxxx
- Date: Sat, 27 Sep 2014 00:00:21 +1100
The original circuit design I was sort of copying used two resistors
because they fed a bias voltage in at the centre of the two resistors. I
played around with this but couldn't get it to work as well as the
floating bias arrangement. So I left the two resistors in place and simply
connected the floating reference level to that point.
I think the resistors also serve to discharge the gate-source capacitance
allowing the FET to turn off more readily. At least that's what I've read.
But they definitely clean/square up the waveform as well. I note that the
Peak-to-Peak voltage across the secondary changes with the resistors
added. So the resistors may be doing a couple of things.
Most of my transformers are trifilar toroids. So removing a single turn
from the primary is a bit tricky. :-) The one pig-nose transformer I wound
recently was a bit if a disaster. I think the core was too large for the
number of turns I was winding. Maybe.
I did check VP and the LO voltage on the gate and I had room to spare.
I've also got lots of healthy harmonics above second order. So I'm fairly
certain that my mixer is switching. Mileage may vary with other FET's of
course.
> I reported several weeks ago that the J-KISS is LO-starved, at least with
> the pair of FETs I used. Depending on the FET lot, it might not swing to
> pinch-off. Taking a turn off the transformer primary helped greatly, as
> did terminating the transformer secondary with a resistive load. The
> latter cleans up the waveform and improves switching symmetry. I think
> this is what Steve's resistive termination scheme accomplishes, thought it
> can be done with a single resistor across the secondary winding.
>
> ----- Original Message -----
> From: Mvs Sarma
> To: minima@xxxxxxxxxxxxx
> Sent: Friday, September 26, 2014 4:01 AM
> Subject: [minima] Re: J-KISS Version 3
>
>
> Can we take a re-look into winding tolerances of the trifler would
> toroids, how ever confident we are. Perhaps, our seniors who have T&M
> instruments at hand, could try out few trafos for repeatability.
>
> I humbly feel that JFETs how ever best we think, we matched, need
> balance adjustment.
>
> We also need to see the tolerance limits of independent resistors used
> around the mixer.
>
> I am constrained having only an oscilloscope and DMM at hand. wondering
> whether elektor UK 2008 10 article "RF sweep frequency
> Generator/Spectrum Analyser" by Gert Baars could help making DIYing
> same.
>
>
>
> On Fri, Sep 26, 2014 at 9:11 AM, Steve VK2SJA <vk2sja@xxxxxxxxxxx>
> wrote:
>
> Hi Farhan,
>
> Schematic same as the last one I built. I will post it again later
> this
> afternoon. I just built another from scratch to make sure it was
> repeatable.
>
> No balance pot adjustment in this circuit. You manually match the
> FET's
> prior to construction. No harder than matching diodes for the product
> detector/BFO mixer IMHO.
>
> Two SA screen shots in previous post show the result of using matched
> and
> unmatched FET pairs.
>
> Interestingly CL seems about the same for both. Significant difference
> in
> unwanted LO and RF leakage.
>
> Like all versions of J-KISS posted to date the LO leak is not
> symmetrical.
> Inherent in the design. LO leak suppression is much greater at the
> mixer
> transformer centre-tap. So if the goal is to reduce LO leak level to
> the
> RF amp/antenna then this port is the one which should be connected to
> the
> LPF.
>
> I can get about 6 to 7dB CL from this mixer. 6.5dB in this latest one.
> All
> other J-KISS variants I've tried were at least 10dB CL or even much
> higher.
>
> This mixer is operating as a switching mixer. I checked. But voltage
> swing
> is not symmetrical around the VP cutoff point with this bias scheme.
> Which
> I guess is the desired target? Regardless this seems to work well and
> is
> the best I've managed building a switching mixer with two J-FET's.
>
> More later.
>
> Steve
>
>
> > steve,
> >
> > what's the schematic of this mixer? have you used any balancing pot
> with
> > these? i am at the pycon with my son. hence living off the tethered
> phone
> > for connectivity.
> >
> > - f
> >
> > On Fri, Sep 26, 2014 at 5:53 AM, Steve VK2SJA <vk2sja@xxxxxxxxxxx>
> wrote:
> >
> >> Hello All,
> >>
> >> Re-surfacing again after an illness and some "life" interruptions.
> >>
> >> I know many have now moved on to mixer experiments with high speed
> CMOS
> >> switches. But I'm still finalizing some JFET KISS experiments. Just
> to
> >> tidy things up before moving on.
> >>
> >> As promised I built another version of the original J-KISS version
> 1
> >> with
> >> both Joe's floating bias mod AND my added resistors. From now on I
> shall
> >> call this J-KISS version 3 as it saves typing. This was completely
> built
> >> from ground up with all new parts. It functions the same as the
> previous
> >> unit. ***Subject to the prior matching of the J310 FET's being
> used***.
> >> In
> >> fact performance is even better than my first version.
> >>
> >> A full report later but for now I need to make an urgent
> clarification,
> >> a
> >> correction and an apology to the group.
> >>
> >> Previously I had reported that I was not seeing any significant
> >> difference
> >> between a matched set of FET's and another set of deliberately
> unmatched
> >> FET's while experimenting. I reported what I was seeing. But
> somehow I
> >> had
> >> managed to mix my FET sets up!! In short - I had two sets of FET's
> that
> >> were very, VERY close to being universally matched. I have no idea
> how I
> >> managed to achieved this but I did. Suffice to say I didn't
> carefully
> >> mark
> >> the FET pairs with coloured fingernail polish. (Big red face! Smack
> head
> >> with hand!! - Sorry...).
> >>
> >> To try and make amends and set the record straight below are two
> >> spectrum
> >> analyzer screen shots showing my latest mixer with both a carefully
> >> matched and poorly matched FET pair (this time carefully marked
> with
> >> pretty coloured dots). The 1201.jpg is the good pair while 1204.jpg
> is
> >> the
> >> mis-matched pair. Exactly the same mixer.
> >>
> >> It is interesting to note that Conversion Loss seems to remain
> about the
> >> same. However both unwanted LO and RF leakage increase. The LO leak
> >> increases dramatically.
> >>
> >> More later.
> >>
> >> 73, Steve.
> >>
> >
>
>
>
>
>
>
>
> --
> Regards
> Sarma
>
Other related posts:
