EMRFD Message Archive 1687

Message Date From Subject
1687 2008-05-30 12:34:29 bkopski EMRFD FEEDBACK AMPLIFIERS
Hi All,

I've been pursuing subject amplifiers particularly for a wideband
application i.e. beyond HF. As part of this effort I have been
trying different transistors in the same circuit. One such circuit
is the 200 Ohm, 10.3 dB version listed in Table 2.2.

My test setup consists of an old Wavetek 2001 Sweeper set up for ~1
to 200 mHz sweep. The detector is my variant of the QST Power Meter
which has the option of connecting a scaled meter drive output to
a `scope. The source cable and the power meter cable have 6 dB pads
at the amplifier connectors. The nominal amplifier input power is ~
0 dBm.

I have an inexpensive DSO which can store 4 waveforms at a time. The
associated software can port the scope display into to my pc (via
USB) where I can capture and annotate it in PAINT. I can input to the
scope the output of my pwr mtr and trigger the scope at the
initiation of a sweep. Thus I can acquire and display the results of
4 sweeps over the frequency range in amplitude – here done at the
display scales of 2 dB / grid and 20 mHz / grid.

I am including 7 photos or .jpg's of the test results. In each
picture one of the four traces is given as "reference". This is the
amplitude response of my test setup with a "pad-to-pad" connection.
The other 3 "gain" traces are for various transistors tested in the
amplifier which is connected "between pads". Thus, the "real"
response of the amplifier would be the displayed gain trace minus the
reference trace.

In some cases I have more than one transistor of the same part # from
the same manufacturer but taken from different lots and / or quite
different dates of purchase. These are distinguished by a numeric

What amazes me no-end is the remarkable consistency of performance of
3904 devices. Those characterized here represent a VERY WIDE variety
of purchase over many years. If I had not done this work myself I'd
not believe what is shown here.

BTW – the DSO is OWON model PDS 5022S recently purchased for $325 new
from Salig. I have used it to capture and display other things like
filter sweeps and SA displays. It is a wonderful adjunct to my
variation of the QST power meter and my variant of the QST (W7ZOI) SA
because it permits seeing very slow sweeps (such as needed for very
narrow filters) that is not possible to do with a classic analog

(It's not clear to me how many photos I can post at the same time so
if all seven don't show up right away check back later – I'll do my

1690 2008-06-01 08:23:22 bkopski Re: EMRFD FEEDBACK AMPLIFIERS
Hi All -

Well folks - I have made an error of omission in my related posts and
this needs fixin'.

Yes - I did use EMRFD Table 2.2 values as described BUT I failed to
mention that my circuit was not exactly the one associated with Table
2.2 - i.e. I did not use Fig. 2.57. Rather, I've been working with a
SIMILAR circuit from SSD - namely Fig 12, page 147 for a few weeks
and then began applying Table 2.2 values to it.

I regret this oversight and any issues it may have caused anyone.
What happened was I'd been corresponding with Wes, W7ZOI re: the SSD
circuit which I've been working with for several weeks and then stuck
with it as I began playing with values that affect the AC operation.
This was an easy temptation since I had several built to play with.

RE: EMRFD Fig 2.57 - this "same but different" circuit is biased
differently and no doubt this accounts for the current drain
differences and would also have a small gain affect as well. The
lighter-biased circuit I used typically drew 18 - 19 mA with some
slight variation with transistors due DC Beta variation no doubt.

I'm not set up to do IMD but rather have been keeping watch on
harmonics with signal level changes
1691 2008-06-04 17:51:51 ve7ca2 Re: EMRFD FEEDBACK AMPLIFIERS
Hi Bob. Interesting notes re your measurements of the EMRFD Feedback
Amplifier. I would be interested in seeing a gain flat response curve
from 1.8 to 50 MHz.

Many months ago I requested further information about the biasing
method you used for your MOSFET Power amp.

In February I finally got around to building a new power chain for my
homebrew transceiver, (HBR2000) based on the G6ALU's design.

I only need 5 watts to drive my 100 watt amp and I used a big heat
sink so the MOSFET's run so cool that I didn't bother to implement
your biasing cct. (Maybe at a later date)

With the drive level available in my HBR2000, I got 8 watts output
power from 3.5 to 28 mhz but the output was only 4.5 watts on 1.8 and
2 watts on 50 MHz.

Since the gain flatness was not as reported I decided to change the
input cct as it's gain flatness was far from being flat. I used the
standard W7ZOI EMRFD Feedback Amp. using a 2N5179 biasing it for 20
ma's and putting a RF chock in series with the feed back resistor from
the collector to base to increase the gain at 50 MHz.

Next I spent hours optimizing the coupling xmfm T2. I checked the
input SWR of T2 and it was not even close to 50 ohms at 50 MHz. I
rewound the 2nd coupling xmfr T2, several times using different wire
sizes and techniques to achieve a flat SWR (high return loss from 1 to
50 Mhz.) Using the largest wire size I could fit into the core, I
was able to obtain a SWR that is 1.1:1 from 1 to 50 MHz and an
impedance of 50 ohm within a needle width. I found an easy way to
check the SWR. T2 is a 1:4 impedance ratio design. I terminated the
output with a 200 ohm resistor and connect the input to my MFJ antenna
analyzer which reads SWR and Impedance.

Since the output on 1.8 MHz is only 4.5 watts, I think the output
transformer inductive reactance is a little low however I have been
reluctant to play with it fearing that I may upset the gain at 50 MHz.

All in all I am pleased. I have enough gain proceeding this module
that I can increase the power output control to increase the final
output on 1.8 and 50 MHz to be able to drive my 100 W. Amp to full

I used the PC board layout indicated in G6ALU's site. I photocopied
the pc board layout onto a special clear plastic sheet that I found at
a local electronics supply store. I then ironed it onto the PC board
which left the traces on the PC board. I etched the board and it came
out like a professionally manufactured board. Real slick way to make
a single board.

I could upload a few pictures and a copy of the new input stage if
anyone was interested.


Markus VE7CA