On 7/28/16 7:29 AM, Joe Rocci wrote:
AllisonWhile we seem to disagree I try to get past device limitations and proceed. Varicaps are at best
Sorry to belabor an academic point, but it seems to me that ‘distributed currents’ are an issue when looking for component heating (especially important in oscillators), but it doesn’t affect Q. The only way to improve unloaded Q is to somehow change the resistive or reactive component independent of the other. I just can’t see how paralleling does any good.
I guess one could argue that, if the varactor capacitance is only a small percentage of the overall capacitance of a resonator in the filter, then the affect of it’s poor Q would be lessened, but then the tuning range of the filter would be correspondingly small, and rather useless to accomplish the goals of a wide-range input pre-selector. Aside from going to a high-IF double-conversion design, I think filter switching of some sort is unavoidableI agree, switching is a better way to go and the circuits used are simple though repeated. The reason I go on this is
*From:* allison <mailto:ajp166@xxxxxxxxxxx>
*Sent:* Wednesday, July 27, 2016 10:09 PM
*To:* minima@xxxxxxxxxxxxx <mailto:minima@xxxxxxxxxxxxx>
*Subject:* [minima] Re: for minima project
On 07/27/2016 06:31 PM, Joe Rocci wrote:
Qu of the varicap equals Rp/Xp. Paralleling varicaps halves both, so I can't see how it improves the unloaded Q. Paralleling coils doesn't help for the same reason.Measured says no. The distributed currents are important. However the Q doesn't double
either. Paralleld Xp is for the unit (pair) but each X and each R are getting a portion of the
total current. The coil case to get more inductance you use series not parallel.
Even if we accept the the marginal Q its a means to tune a double or a triple tuned band pass.
The losses at HF are less problematic as we are not working with a exotic high performance
receiver or transmitter. Even with loss and low Q it may beat trying to low pass and trap
the IF. If a mechanical cap (still not the highest Q) were available it would be better but
then you trade electronic problems with availability and mechanical ones. No matter
what you vary its going to be a compromise over the active range. I was shooting for
something that can work (least for one case I actually tried) and can be built without
space, tools, and treasure that may be unavailable.
For myself I'd grab a bag of latching relays and build band specific filters. That we know
can work but while its repetitive circuits is not always simple to do well (or tune up).
One thing with any lossy system, gain is cheap. That and for minima we are talking about
better not perfect. I do plan to build it and put it on Genesys and PNA and see,
once I get a moment.
Allison/KB1GMX
Joe
Sent from my Verizon Wireless 4G LTE Tablet
-------- Original message --------
From: allison mailto:ajp166@xxxxxxxxxxx
Date: 07/27/2016 6:01 PM (GMT-05:00)
To: minima@xxxxxxxxxxxxx
Subject: [minima] Re: for minima project
On 07/27/2016 01:43 PM, Joe Rocci wrote:
> I agree in concept, except that varactor Q's are generally not good
> enough to get low loss in a hi-Q, narrow band filter. Making the
> coupling track over several octaves is also tricky. And then there's
> the intermod potential, since the voltage on the varactors will be Q
> times the input voltage. Problematic for me, but an academic exercise
> since I've never actually tried to do it.
>
Joe,
that is true the Q's are not so great. However parallel diodes can
improve the Q (their internal resistances
are in parallel).
IMD is a matter of signal levels and managing them. Also choice of
inductance.
I've modelled it and you need to parallel coils when you reach about
10mhz to keep the coil Qs
up but that's a relay (or two). I did try it on a KNQ7A on 40M.
The story there is they use a
VXO so it only tunes maybe 25khz. I dropped in a DDS to cover the band
and found the
double tuned front end (antenna switch>>DTBP>>SA612) was rather
narrow. So I used
a pair of 1n4007 (20PF as varicap) on each coil and tuned it with 1.5V
bias for the low end
and at 6V bias it hit the high end. Instant preselector. Didn't test
the bandwidth but the
RX sensitivity was unchanged. More importantly the killer BC stations
above .200 didn't
show as intermod. I call than simple and good enough.
Matching varicaps is a matter of fixed bias and a AADE LC-II though I've
tested a bunch of
MV109s and they seems to be very close to each other at several
voltages. Also the filter
can be designed to tolerate some mistuning by way of bandwidth at center
frequency.
It can be an approach. Also loss is tolerable if one uses a small
amount of gain.
Allison/Kb1GMX
> Joe
> W3JDR
>
> -----Original Message----- From: allison
> Sent: Wednesday, July 27, 2016 1:11 PM
> To: minima@xxxxxxxxxxxxx
> Subject: [minima] Re: for minima project
>
> On 7/27/16 9:19 AM, Joe Rocci wrote:
>> Allison
>>
>> As I recall, the problem with the original Minima was that the 20 Mhz
>> IF leakage through the mixer (IF-RF) wasn't sufficient to prevent
>> unwanted transmit spurious at the IF frequency, and that the simple
>> 2-band lowpass filter at the input didn't have adequate performance
>> to improve it enough. I don't see any specification in the ADE-1 or
>> SBL-1 datasheets for IF-RF rejection, but my recollection is that
>> numerous packaged and home-brew mixer designs were evaluated with no
>> definitive solution.
>>
> Yes that was it single balanced mixed had blow through and inadequate
> balance. The ADE/SBL rejection was typically
> around 40db.
>
> If you don't filter and rely on balance then you need at least 50db to
> be "legal" and 56db would be suggested for
> If rejection on transmit. The secondary problem is even using high q
> coils and caps a notch filter is wide enough that
> 15m abd maybe even 17M performance is questionable.
>
>> I think the proper solution for the original minima is a lower IF
>> frequency (maybe 9 MHz?) and switched bandpass filters at the
>> front-end. One could use 2 CMOS bus switches and up to eight 2-pole
>> bandpass filters to accomplish this (I've done it and it works very
>> nicely). The final product isn't much more complicated than the
>> 2-band minima lowpass approach and can be implemented incrementally,
>> adding one band filter at a time as desired. I think one could also
>> consider this as a souped-up BIT-X.
>>
> Right again. At 9mhz a notch filter has a much narrower and lower
> loss. Mixer performance is
> NO worse and likely better and the only near band is 10.1 (30M). so a
> filter in the 8.5mhz range
> works well or move up to 11.5mhz.
>
> The problem of bandpass filters is the implied spec of minima, the
> ability to tune a wide segment of HF with
> one filter. For receiving it may fly generally for transmitting band
> pass filters per band are the safest way to
> meet most PTT/FCC specs for spurs and harmonics. It is possible to
> cover 1.5-30mhz suing half octave low
> pass and combined high pass filters but t will still be 5 or 6 filters
> needing to be switched in and out.
>
> An alternate for that is tunable preselector/passband filters. In the
> past that meant a big multigang cap or
> a rack of moveable slugs in coils with mechanical headaches. I'd trying
> to do it using Varicap doides
> to see if that can fly (allowing for tracking and IMD needs) as variable
> DC is easier than a rack of gears.
> Old school idea, newer parts.
>
>
> Allison/KB1GMX
>
>> FWIW, my 2 cents and all that.
>
>>
>> Joe
>> W3JDR
>>
>> -----Original Message----- From: allison
>> Sent: Wednesday, July 27, 2016 8:57 AM
>> To: minima@xxxxxxxxxxxxx
>> Subject: [minima] Re: for minima project
>>
>> On 7/26/16 1:54 AM, Ashhar Farhan wrote:
>>> There are many ways to continue with the Minima:
>>>
>>> First:
>>> If one crosses over to the fulll H-mode mixer, it might solve the
>>> leakage problem. I have been trying to keep the design simple with
>>> the KISS mixer. Our (Joe and Mine) experiments wiht using MOS
>>> switches were encouraging.
>>>
>> MOs switches or a fully balanged mixer works.
>>
>>> Second:
>>> The other option is to just go with a diode mixer. I already have a
>>> transceiver like that in the works with Si5351. It is a low-cost
>>> design that works upto to 21 MHz and uses two oscillators for the
>>> BFO and the VFO. A a post-mix amp had to be added between the diode
>>> mixer and the crystal filter to provide decent termination to both
>>> sides as well as overcome the filter loss. This is a pretty simple
>>> design. The BFO and VFO are entirely software controlled. A single
>>> LPF that cuts off at 21 Mhz is all that it takes. I will soon post
>>> the details.
>>>
>> DBM is a easy way to get a balanced mixer. RF/IF gain is easy and
>> cheap.
>>
>> Filters are both easy and hard, easy to suggest and without instruments
>> or a procedure that
>> uses easy to acquire gear hard to assure they work as expected. This is
>> hampered by fear of
>> making coils.
>>
>> Also the IF is too magic as crystal filters are scary for most and high
>> HF filter being more so.
>> As it works the 20mhz is not much better for images or wide tuning than
>> 9, 10 or 11mhz
>> because there will be some frequency or range that will be inaccessible
>> as its the IF. Also
>> the lower frequencies its easier to obtain quality crystals.
>>
>> One solution is a preselection system. The tube radios did this and at
>> least one vendor
>> (Tentec) did that on their early transceivers. Ganged L or ganged C are
>> a mechanical problem
>> but with Varicaps I have prototyped this and considered it a possible
>> approach. Its another
>> knob but it is less a problem if its a simple pot and can provide a
>> higher selectivity front end
>> which helpful in Eu and many places with high power SW broadcast
>> stations.
>>
>>> Third:
>>> The JFETs as mixers could be trouble as their gate-source and
>>> gate-drain can go into conduction easily. Instead, if we use BS170
>>> or 2N7000 kind of switching MOSFETS it might just work.
>>>
>> capacitance from gate to drain is first order issue. Then the drain
>> source parasitic diode is the next problem.
>> I did some work trying them using some TI parts (lower gate to dran
>> capacitance) and it was not as good as
>> Jfets properly biased. The trick with Jfets is matched for Idss and
>> gate cutoff (dual parts are good for this
>> but scarce) and then raise to source to the gate cutoff value. The
>> problem is its still a single balanced mixer.
>>
>>> The interesting thing is, though Minima itself was not a very
>>> replicable design, it did teach us quite a lot. It is a robust
>>> receiver and it has had a large number of spin-offs.
>>
>> Its trouble spots were the 20mhz filter, and the first mixer. Both are
>> easily solved.
>>
>> I know from prior work I've done that a 16mhz filter is easier to
>> realize and crystals
>> had better temperature stability vs band width. Also 16mhz is far
>> enough from
>> 14 and 18mhz to avoid spurs in ham bands and images.
>>
>> At HF a wide band RX or TRX is easier with a VHF if (45mhz or higher).
>> But for monoband or
>> a limited number of bands single conversion and a if at 6-12mhz makes
>> for a simple design
>> with fewer issues with retaining good performance.
>>
>> Having built many monoband transceivers (160 through 2M SSB) and a few
>> multiband plus
>> a large number of receivers for various uses and tuning ranges there is
>> experience. There are
>> few magic circuits (thought the DDS and SI570 may qualify) the basic
>> radio is more a matter
>> of trade offs and choices. This has not changed since the 40s though
>> the devices to attain
>> the needed gain have.
>>
>> Allison/KB1GMX
>>
>>
>>
>>
>>
>
>
>
>
