EMRFD Message Archive 14604

Message Date From Subject
14604 2018-04-01 07:04:30 radiog3xaq TR switch diode limiter

I am making a 50W PA to add to a 5W monoband QRP rig.

I was intrigued by the simple limiter circuit in fig 6.126 of EMRFD and the extended writeup in the QEX article, but there is a dire warning that the simple circuit is only suitable for low power transmitters and 100W can only be handled "with circuit modification".

I modelled this with LTspice and the simple circuit with four 1N4148 Silicon signal diodes looked OK at 100W. Although the peak current is 228mA, the average is 35mA (113mA RMS). The dissipation is only 35mW. I thik all these values are well within the Vishay datasheet ratings? Receive insertion loss and transmit feed-through power in the model are as described in the book.

I thought perhaps the diode model was too good to be true so I built the circuit for 14MHz and tested it at 100W. The results are in fair agreement with the model: Receive insertion loss is 0.8dB and feedthrough is -2.5dBm. My toroid coils probably aren't quite as good as the Qu=200 I used in the model.

What came as a pleasant surprise is that the diodes show no signs of distress at the 100W level. Nothing gets warm to the touch. Maybe long term reliability is suspect? I don't know, but if the diodes are each really only dissipating 35mW then I don't see them failing prematurely.

For a single band application this circuit looks simpler than one using high voltage DC and PIN diodes.

73, Alan G3XAQ


14605 2018-04-01 11:01:05 w7zoi Re: TR switch diode limiter

It sounds like you are on the right track with this T/R scheme.   The major issues include both the stress on the diodes, which you have addressed, and the IMD introduced in the receiver path.   It will take some careful measurements with a spectrum analyzer to determine the intercepts for this.    This measurement does not require that the transmitter be operating.

Good luck with your experiments.

73, Wes

14606 2018-04-02 06:28:19 radiog3xaq Re: TR switch diode limiter
Hello Wes, and thanks for taking an interest in this. I've made some IMD measurements with home-lab accuracy using a pair of signal generators and my Elecraft K3 as detector.

The headline result is the four-diode limiter shows a third order input intercept of +19dBm. Increasing the number of diodes to six (three per leg) makes no obvious improvement.

Here is some background detail on the measurements. I used two HP8640B signal generators. They are old boat anchors and need regular repairs but they didn't cost much from the salvage dealers. The generators each go into a 6dB hybrid and I added a 20dB fixed attenuator between the hybrid and the T/R switch+K3.

I first checked without the diode TR switch in circuit. With the generators spaced 10KHz apart and each set to 0dBm I could hear the third order product at the same level as an on-tune signal from one generator when set to -105dBm. Via the hybrid and 20dB pad this is -131dBm into the K3. If I have the arithmetic correct this corresponds to an IIP3 = +26.5dBm. So this is the limit of my measurement setup.

The ARRL review for the K3 puts IIP3 at +28dBm so I am probably being limited by only having 20dB on the generators' internal attenuator when set for 0dBm output (it goes up to +20dBm). The hybrid with the pad on its output into the K3 shows a bit over 40dB return loss on my ancient spectrum analyser and tracking generator, so there is only about 80dB isolation between the two generators. Also, because I don't have the ability to use high signal levels, I am using the Standard British Ear to estimate the signal in the K3; it is well below the AGC threshold and I can easily hear a 6dB change in level and fairly reliably a 3dB change.

Inserting the diode T/R limiter required me to reduce the generator outputs from 0dBm to -5dBm for the same -intermod level. Again, someone check my arithmetic but I think after the 26dB loss through the hybrid and pad it means IIP3 = +19dBm.

I tried adding another diode to each leg of the TR limiter but couldn't detect any difference.

As an aside, the target use for this TR limiter is with an add-on PA for a  QCX QRP transceiver

When I checked its performance with this test setup I was surprised to find the bare QCX gives IIP3 = +23.5dBm, only 3dB worse than the K3. However it is a bit deaf and mine can just hear -125dBm; others have measured the MDS as -121dBm (in 250Hz). This 96.3dB dynamic range is pretty good for a $49 kit! Adding a 10dB preamp would move the MDS/IIP3 closer to the ideal for normal 20m band conditions.

My conclusion is that this simple diode TR switch has merit for single band use even with pretty good receivers such as a QCX+preamp. It is not suitable for use with top of the range receivers in the K3 class and beyond.

73, Alan G3XAQ

14607 2018-04-02 07:48:31 Mike Re: TR switch diode limiter

Thank you for providing all the details of how you measured IMD for your
diode limiter/TR switch setup. For many years I have struggled to grasp
how the different parameters of receiver overload interact; I have been
somewhat constrained by limited test equipment availability, and maybe
also by limited brain power.

But I'm saving a copy of your email for future reference. You obviously
spent some time doing the measurements, and comprehensively documenting
(and sharing) exactly what you did- great job! Along with emrfd and
SSDftRA, I should be able to put it all together somewhat!

Mike Dinolfo N4MWP

14764 2018-05-08 10:51:13 Alan Ibbetson Re: TR switch diode limiter
When I reported on the EMRFD fig 6.126 diode limiter I claimed the third
order input intercept (IP3) was +19dBm. This was somewhat optimistic.
I've improved my IMD measurement setup to include a cascode amplifier
between each signal generator and the 6dB hybrid. Each 2-stage amplifier
has 30dB gain and can deliver up to 100mW into 50 ohms. The isolation
between the generators now exceeds 120dB when the hybrid summing port is
terminated in 50 ohms. I have replaced my Standard British Ear with a PC
soundcard based RMS audio voltmeter program, available here


It is in good agreement with an HP3400A thermal type meter. As a sanity
check I've used my Elecraft K3 as the Device Under Test and the IMD
signals do not change as I add and remove 10dB pads around the hybrid
and adjust the drive level from the HP8640B sources.

More careful measurements on the 1N4148 4-diode limiter gives IP3 just
over +14dBm. My Elecraft K2 (level-7 DBM front end) has IP3 +19dBm so
the limiter would compromise performance. The diode limiter is OK with
the RF amplifier in the K2 turned on (IP3 +6dBm).

A simple modification to the original limiter circuit improves IMD
performance substantially. Applying 12 volts reverse bias across the
four diodes makes the limiter stronger than my Elecraft K3, which itself
has IP3 +25dBm when using the 8-pole 400Hz CW filter. Insertion loss
remains well below 1dB. I did some measurements with 10dB attenuation
between the limiter and K3. IMD from the reverse biased limiter is not
well behaved, in that the slope of IM3/Pin is 2.5dB/dB instead of the
theortetical 3.0, so the IP3 appears lower at low signal levels.
Nevertheless the "arithmetic" IP3 exceeds +29dBm all the way down to the
K3's noise floor, so the limiter will not compromise the performance of
receivers in the K3 class. Transmit isolation is not affected but of
course the 12V bias must be shorted out on transmit.

The limiter perhaps could be integrated into a multiband transceiver by
adding an instance of the circuit to each section of a per-band switched
preselector. If better IMD performance is needed, for example with a
PA3AKE Holy Grail front end, a lower loaded Q series tuned circuit could
be used to reduce the signal voltage across the diodes, although on
transmit the resulting increased diode dissipation and transmitter
feedthrough needs investigating. If the IMD comes from signal modulation
of the diode capacitance then a lower capacitance diode such as the
MMSD4148 might perform better than the 1N4148.

Throughout these experiments I have been helped and advised by
Roelof/PA0RDT. He has been making IMD measurements for a long time and
with better test equipment than me and he took the trouble to build the
reverse biased limiter and confirm my observations.

73, Alan G3XAQ