EMRFD Message Archive 1196

Message Date From Subject
1196 2007-11-28 00:46:42 Heikki Ahola (OH... Class-A driver/PA in EMRFD and SSD for 1.8 - 30 MHz
Dear group and Wes in particular,

I have spent some nights with this two-stage circuit intended as a
class-A driver for SSB or other linear applications. It consists of
one 2N3553 feeedback amp as the first stage and two 2N3553's in
parallel as the second stage. When biased for 50 mA and 250 mA
correspondingly, it is claimed to deliver 1 W into 50 ohms.It appeared
already in SSD and have been duplicated in EMRFD as such.

Personally I have used this circuit for several years already in my
transverter but never been able to extract more than 6 V peak/50 Ohms
(0.36W) undistorted and when simulating with LTSpice the same results
are obtained. Have I missed something or is this the case ? Currently
I am using 2N4427 devices which, in my opinion, are close enough to
the 2N3553's. The simulations were run with 2N2219 available in LTSpice.

BR

Heikki (OH2LZI)


P.S. Hopefully we can carry on with this group in the furure also ! I
have found it exremenly interesting and useful for my experimentations.
1207 2007-12-02 15:15:11 Wes Hayward Re: Class-A driver/PA in EMRFD and SSD for 1.8 - 30 MHz
Hi Heikki and group,

Yes, that amplifier (Fig 39, p206, SSD, and Fig 6.98 in EMRFD) did
exactly what we said it would do. I used the amplifier in a dual
band (40/20) 1 Watt CW transmitter that was the basis of my station
for about 15 years. It also got used in a couple of other rigs that
did not last as long. The thing is still kicking around in my junk
box somewhere. I used it for some experiments when I needed a bit
more RF at HF for one application or another.

These were CW applications. 1 watt output is 10 volts peak, which
is about what you would expect for a circuit like this when loaded
with 50 Ohms and using Vcc=12.0. The effective supply is closer to
10 volts than 12 owing to the emitter degeneration and decoupling
resistor. Hence, you would expect a collector voltage of 2 x Vcc
and that comes out to 1 W.

The numbers quoted in SSD were measured. That is, two quarter watt
(+24 dBm) tones were extracted from the output while looking at the
IMD. This output is +30 dBm PEP, 6 dB above each tone. The IMD
was down by 39 dB from each tone. That's reasonable linearity. I
would expect it to degrade quickly though as you drove it further.
Also, the current may well have been higher than 250 mA in the PA
during the measurement. As I recall, the amplifier current does
increase with applied drive. It is not pure Class A, although it is
close. The heat sinks shown in the photo on P207 of SSD are not
adequate for continuous service. They were fine when the power
supply was keyed, as it was in the previously mentioned CW
transmitter. However, more heat sink is in order for typical phone
applications.

You can't really say that you get no distortion up to a given power,
and then have distortion. An amplifier like this one that is
essentially a Class A design will have a fairly well defined output
intercept that is controlled by the standing current. With less
drive, the distortion will be less, but it will still be there.
Even gain compression sets in slowly. So you can't really say that
below 0.36 W there is not distortion.

There is no need to stick to the transistors that are available in
the canned library of LT SPICE when doing a simulation. The SPICE
models for things like the 2N3866 and similar parts are readily
available on line. I usually take the model parameters and put them
into the library which can be edited with any text editor such as
Notepad. If you can't find a SPICE model for your 2N4427, I would
try a 2N5109 model, which is available. The 2N2219A is an OK
transistor, but not in the same F-t league as the others mentioned.

If I was going to build a sideband rig these days and needed a 1 Watt
Class A driver, I would probably get a more robust TO-220 device and
drop it into that same circuit. I bet the 2SC1969 would be great in
this application, but there are numerous other parts that would do as
well. Allis
1210 2007-12-04 00:12:47 Heikki Ahola (OH... Re: Class-A driver/PA in EMRFD and SSD for 1.8 - 30 MHz
Wes and the group,

Mni thanks for the reply ! I understand that you have been quite busy
with the cascode if amp issue and still have time to respond to my
problems which are related to an aging circuit. I built mine some
three or four year ago as the output stage of my transverter and in
spite of the reduced output it has served well in continuous (RTTY)
service. The reason I brought this thing up is that I saw the circuit
kinda revived in EMRFD and decided to carry on with experimenting in
order to extract the missing 5 dB or so. After studying other options
I agree that a more robust design might be in order, I am currently
working on another SSB transmitter project where this will be applied.

The QST December issue arrived yesterday, the hybrid cascode looks
very interesting, since I have almost spent my stock of double gate
MOSFETs during my latest receiver projects. Thay are also easy to
destroy since all modes do not have the protecti