Wideband Amp Tests
Steve Ratzlaff
Design article that I somehow overlooked all these years. The article is
from the February 1989 issue of RF Design (an industry trade journal,
free if you work in electronics), titled "A New Negative Feedback
Amplifier" by Victor Koren (of Tadiran in Israel). I was sent a copy of
the article recently, but it looked vaguely familiar, and sure enough, I
had the original article torn out and saved (in my vast unsorted pile of
paper--which I waded through looking for the article, and found it).
Re-reading the article, it didn't look all that promising, but the
actual prototype circuit looked easy enough to duplicate.
Instead of classic negative feedback amps using transformers, this
article claims use of a directional coupler to improve reverse
isolation. The original prototype used an MRF587 bipolar transistor. I
didn't have one of those, but I did find an old MRF586--the same die as
the 587, in a TO-39 case. I used FT37-75 cores for the input directional
coupler and the output bifilar transformer. A 1:10 turns ratio for the
coupler was used in the article's circuit. I used the same turns ratio,
using an actual 2:20 turns to get a hoped-for lower frequency response.
The bifilar output transformer used 16 turns for each winding.
I used test equipment at work to measure gain, response, IMD; and a
calibrated signal generator with communications receiver and audio
voltmeter at home to measure MDS and derive noise figure.
The circuit worked immediately and was unconditionally stable (input or
output or both open circuit or terminated), using a 1.5GHZ spectrum
analyzer to check for instability products. Bias was set for 35mA with a
12.8volt supply. Performance of the single-ended circuit turned out to
be quite good.
Gain was about +10.5dB, being very flat over a wide bandwidth, with a
slight peaking of +11.0dB from 58-104 MHz, with +11.6dB peak at 82MHz
using my cores and suspended-over-a-groundplane circuit construction. At
the low end, response was down -1dB at 75kHz; -2dB at 48kHz; -3dB at
35kHz--pretty good LF response for only 2 turns on the directional
coupler primary.
Reverse Isolation was very good-- 38dB from 225kHz to 30MHz, dropping to
30dB at 95MHz and 29dB at 110MHz.
Return Loss, both input and output, were both very good; in fact, input
RL got slightly better when the output termination was removed; output
RL got poorer when input termination was removed, which is what's
normally expected.
Input Return Loss--250kHz 15.4dB; 500kHz 21.8dB; 1MHz 25.4dB; 15MHz
24.5dB; 30MHz 20.4dB 50MHz 14.2dB.
Output Return Loss--250kHz 24.2dB; 500kHz 24.5dB; 1MHz 24.8dB; 15MHz
22.4dB; 30MHz 18.6dB; 50MHz 15.0dB. (Recall that 14dB RL = 1.5:1
VSWR.)
IMD--tones of 3 and 4 MHz were used, set to 0dBm each at output of amp,
to spectrum analyzer.
3IMD product at -77dBm = +38 OIP3 (products at 10,11,5,2 MHz)
2IMD product at -50dBm = +50 OIP2 (products at 7,1 MHz)
Max output power--amp was still linear at +20dBm single-tone output, at
3MHz.
Noise Figure tests--using information from various sources, notably
Hayward's book "Introduction to RF Design", by measuring the MDS of a
communications receiver, then determining its filter's noise bandwidth,
the noise figure of the receiver alone can be determined. Then by
cascading the amp with the receiver in series, the overall noise figure
can be determined, and then the noise figure of the amp itself can be
determined.
Hayward discusses noise bandwidth for a communications receiver using
steep-skirted filters to be equivalent to the -6dB bandwidth of the
filter. This is easily measured with a calibrated signal generator and
an audio rms voltmeter connected to the receiver's audio output. Then
the MDS (3dB increase in audio output) of the receiver and amp/receiver
combination is measured, and the values plugged into the formulas to
derive the various noise figures.
My receiver, an AR-7030, with internal preamp off, in CW mode with
narrowest filter selected (500Hz--measured -6dB bandwidth of 570Hz) had
an MDS of -127dBm, to give a noise figure of 19.4dB.
Amp and receiver together had an MDS of about -136.7 dBm, for an overall
noise figure of 9.7dB.
(All noise figure tests were done at 450kHz, mainly because I'm
interested in a low-NF LF amplifier.)
Substituting values in the cascade noise figure equation to solve for
the amp's noise figure, using Fnet = 9.7dB; F2 = 19.4dB; G1 = 10.5dB,
gives a value for F1, the amp, as 2.2dB noise figure. (I have the
hardest time remembering that all values in the cascade NF equation are
not dB's but algebraic numbers, and then converting the final result
back to dB for the final NF value.)
So, the amp has a respectable low noise figure at 450kHz.
Overall, this is a pretty decent amp. I should substitute a common
2N5109 and redo the noise figure measurement, but I don't expect the
results to change much. The datasheet for the MRF586 doesn't indicate a
very low NF for this device. Perhaps the best features of the amp are
its very good input and output return loss (VSWR) values and its good
reverse isolation. Having a low noise figure and good low frequency
response, along with decent IMD performance, doesn't hurt, either! And
no additional "bandaids", such as ferrite beads, were needed to tame any
instabilities.
If you're looking for a decent-performing amp, you might want to get a
copy of the article and breadboard the circuit, and give it a try.
(If you're unable to find this article, I could be persuaded to mail a
copy, for postage and copying costs. Note that 4 patents are listed,
pertaining to this amp, if you have thoughts of making this amp
available for sale commercially.)
Steve AA7U
Ralph Mowery
compair to one of the MMICS ? Something like a mar6.
Most of the work has been done internal so all you need is a couple of
isolating capacitors and one resistor.
Steve Ratzlaff
There is little comparison between a MMIC and a discrete amp. Only noise
figure might be similar. The discrete amp will almost always be superior
in terms of overload resistance, and most of the other parameters I
tested.
Makers of amps for the hobby market, such as Kiwa, Wellbrook, Advanced
Receiver Research continue to sell their products, when MMICs are
readily available.
If your only interest in a preamp is sensitivity, and you don't care
about any of the other parameters, and you plan to use the amp in a
very-rural area, far from any sort of potential overload problem
(transmitters), then perhaps a simple MMIC circuit might be the
best/easiest solution.
Steve
JD
"Steve Ratzlaff" <srat...@flash.net> wrote in message news:3C9A8656...@flash.net...
> Hi Ralph,
> There is little comparison between a MMIC and a discrete amp. Only noise
> figure might be similar. The discrete amp will almost always be superior
> in terms of overload resistance, and most of the other parameters I
> tested.
> Makers of amps for the hobby market, such as Kiwa, Wellbrook, Advanced
> Receiver Research continue to sell their products, when MMICs are
> readily available.
> If your only interest in a preamp is sensitivity, and you don't care
> about any of the other parameters, and you plan to use the amp in a
> very-rural area, far from any sort of potential overload problem
> (transmitters), then perhaps a simple MMIC circuit might be the
> best/easiest solution.
> Steve
>
but there are some damned good mmics out there. For example, take
a look at the Stanford SGA-4586 or the SGA-6X89... They can put out
between 10 and 100mw (depending on unit), and have somewhere between
1.8 and 3dB NF. They are also made up from SiGe, therefore doing
better on the low end (NF) than PHEMT fets.
There is indeed an issue of input/output isolation when using most
mmics, but there are some parts that have lesser specs in some areas,
but also provide much better input/output isolation.
John
Paul Keinanen
<rmo...@dialpoint.net> wrote:
>While it is nice to play with circuits do you have any idea how it will
>compair to one of the MMICS ? Something like a mar6.
MAR-6 is the worst device for HF preamplification, unless you have a
very narrow front end. The gain is very high and output very low, so
it will saturate at quite low input levels. It is however, very usable
for LF/MF indoor loops, which are quite selective but produce a very
low output.
However, MMICs like MAV-11 have output power and output IP3 points a
few dBs below those of the 1989 RF Design specifications with similar
gain figures.
If you are going to wind transformers, why not put two MAV-11s in push
pull and get similar specifications as the 1989 design.
Paul OH3LWR
carltons
<kein...@sci.fi> wrote:
This is very good suggestion from Paul. You would also reduce even order
effects.
Steve WB4CZR