Re: uBitx Unfiltered

Jim Tibbits
 

So trying this from another path.   Did some poking and looking at the Q90-Q97 path today ..Noticed that Q911 thru Q98 all have 100 ohms from the base of the transistor to ground (biasing network) The problem is you only end up with 1 volt on the base ..And with a threshold of 0.7 volts Vbe, that doesn't give much room for any work to be done ...changed them all to 220 ohm ...some better...
pa to filters transformer mounted.jpg
10 meters - 1.8 watts.jpg
20 meters - 10.24 watts.jpg
40 meters - 10.88 watts.jpg
80 meters - 9.52 watts.jpg
L8-L9 substitute mounted.jpg
Photos show the gain now ..I will change them out to 330 ohm tomorrow and re-align the bias settings for maximum flatness again . Test 80,40,20 and 10 on the SA ...Photos also show the mod I made to replace L8 and L9 ...Those units have DC current flowing through them and that messes with the inductance ..Schematic shows them as 60 uH ..I measured mine when I removed them ..24 uH..Checked inductance while flowing DC current through them and at 10 mA inductance decreased to 14 uH....As those flow minimum 100 mA, I doubt there was much inductance left ...As these are supposed to keep all the nasty RF out of the power supply. I think its little wonder all the spurs created ...The band test I ran showed 9.5 watts on 80,... 10.8 watts on 40,... 10.24 watts on 20 and only 1.8 watts on 10 .. Likely the stacked core "binocular" is contributing th that as it isn't one lump of ferrite ..Losses between the pieces ...


On Sun, Sep 2, 2018 at 10:16 AM ajparent1/KB1GMX <kb1gmx@...> wrote:
On Sat, Sep 1, 2018 at 08:49 PM, jim wrote:

consider ..Ft of 2n3904 is about 250 Mhz ...If beta rolls off at 6db/octave, then =at 250 Mhz, beta = 1..125 Mhz beta 6db better  (3db doubles =beta 2, 3db doubles beta 4)..60 Mhz 12 db better (4 doubles  2,4,8,16)...30 Mhz 18 db better .(6 doubles 2,4,8,16,32,64)..
I finally found your text. You are writing inside the quoted section and the web version only shows a blue rectangle with three dots...  on an other wise blank page.  Be careful of the Emailer as its likely burying your answer as a the other guys quote.

Your supposition is however wrong.  For common emitter resistive feedback that is untrue.  The gain falloff is 6db octave
starting at some low frequency so by time you at 30mhz you have gone through a few octaves (3+ for 3-30 or -20db)
you are well down. Add feedback and you are way down as it does not decrease with frequency (in the ubitx case).
It is possible to have a feedback that decreases with frequency but you still run out down the gain.  Where does the
bend in the curve start?  Usually less that 3mhz and its downhill from there.  Also with wideband amsp the C-B feed
back capacitance and other parameters are not in that and they tend to be input and output impedance sensitive
as well as operating current influenced.

The usual calc for resisitive feedback the first parameter is FT/F=  for 3904 its about 300mhz/30 or a beta of 10.  With
that try to make a power amp stages of 18db, voltage gain maybe but not power.  Assume 50 ohm input and output,
but it doesn't happen even with 200 ohm output.  

That and the HFE as I've written many times before falls with increasing current over 10mA, its in the datasheet.  When
you combine that performance at even milliwatt power is a fail.  At 20Ma is on the down side and at 50mA its down
from the peak by 40%. That or unbounded optimism or a hot batch.  With no margin and a lot of luck how many out
of a thousand will perform well.

Reality is that the 3904s in enough units to be a valid confirm are not performing.  Dial in RF1 for 11W at 80M
the driver output is easily a half watt.  By 30mhz you are down 11db to maybe 45mw.  Where'd the gain go?

Same for the 45mhz amp.  In isolation at 10mhz its 18.9 db, at 45mhz its 13db for 50 ohm in and out as that is
what the filter matching network expects and the mixers.  Also around 15mhz the IF amp is showing noticeable
and increasing phase shift when that hits a high enough number instability is seen save for the projected path
is not enough gain at that point..  Now we add the darlington pair effects sine they are running at .99999 (alpha)
their voltage gain is essentailly one but power gain is the result of high input impedance (limited to the prior stage)
and output impedance or about 4 (6db) so the IF amp has an apparent gain even with the poor showing of the 
3904 of some gain just due to the output power gain of the compounded emitter follower.

Yes you can get 3904 to oscillate at 450mhz too.  Doesn't mean anything as not all do.  But then I got a uV201
to oscillate at UHF once.
 
Allison

Join BITX20@groups.io to automatically receive all group messages.