Home made HF transceiver improvements

I made some improvements.

1. 1) A preamplifier was added. Gain is 12 db. The signals were just to weak on higher bands. It can be turned on and off using a switch on the front panel. I had some space left for it. Now it is getting really crowded inside.
2. Filter for band pass tuning was replaced.The old one was too narrow.  I bought 15 xtals for 6 MHz. It took some fiddling to get a filter response that was wide enough and was flat enough.
3. The pushbuttons that I bought on ebay were awful. They are break when pushed. However when push in to the max, the switch closed again. I have replaced them all by normal on/off switches.
4. when switching from TX to RX the AGC attenuates maximal. It takes several seconds before audio is heard again. This was fixed by adding a transistor that empties a capacitor in the  AGC circuit when switching to RX.   
5. 160 meter power output was just 250 milli Watt. I have change a capacitor in the driver. Now I get 3.5 Watts output.
6. No more software crashes were seen after improving the software.

AGC improvement and more

The AGC is too limited for HF. Strong signals are distorted.
I use the 1350p IF amplifier. It has 60db dynamic range that is not enough. So I added a voltage controlled attenuator to the input of the IF unit.
It is just a npn transistor. Collector is connected to the input. Emitter to ground .Voltage over the transistor is low so it works as a resistor.
By changing base current I can change the attenuation. I can reduce S9 +10db signals to S1signals.

Now when the AGC voltage for the 1350p exceeds 5.5 V the attenuator starts to work. It helps a lot.
Most distortion is gone. However I found the attack time of the AGC could be faster . That would help to increase audio quality.

It seems I found the problem in the software. I use the PIC 16F877.  PORTB is configured as input. It is configured to generate an interrupt when one of the pins changes. Now according to the PIC datasheet I should not poll the port while it uses the interrupt. So I disabled interrupts while polling. This seems to help.

on air results

The set is not ready. However I have tested it on the antenna. Monitoring the signal on the spectrum analyzer.

I made a short QSO on 40 meter with a Russian station. Power was 2 Watts. Not bad. The same day I tried CW also on 40. Someone came back to my cq but the station faded away.

The next day I tried the reverse beacons. The signals were heard in several countries.
That is a good motivation to continue.

Now I have increased power a bit. It is still not optimal.
80 meter 3 Watts
40 meter 4 Watt
30 meter 3.5 Watt
20 meter 4 Watt
17 meter 3.5 Watt
15 Meter 5.5 Watt
12 Meter 4 Watt
10 Meter 2.5 Watt

Today I made 160 Meter work. TX Power needs to be improved. It is only 0.250  Watt.

Transmitter improvements

I spend some time to improve the transmitter.

Low pass filter for14 MHz and 18 MHz caused too much attenuation of the signal. So I rebuild I. Using a different design and component values. Now it works fine.

CW while key was up resulted in a signals at -40 dbc. This was cause by the sideband oscillator.
I have changes the software and hardware to turn it off this oscillator during CW TX.
Now It works fine.

SSB signal sounds clean although I think it can be improved by optimizing the compression level in the speech processor.

Output for 80 and 40 meter can be improved. I may have to rebuild the low pass filters. But also the driver can be improved. At keydow it produces 3 watt on 80 meter. After a second it drops to 2 Watts. Same thing happens on 40 meter. This must be investigated.

It looks the microphone wiring picks up too much signal when an antenna is connected.
This  signal is rectified and is audio then, that causes oscillation. Microphone input circuit must be adapted.  adding some filters may do the trick. I have to test

 

Receiver Improvenents.

All modules are build in. Now it is debugging time.

First I made the receiver operational.
Some problems that I had:

• Audio distorted.
• Banpass tuning not enough range to be useful for USB
• Attenuation of first xtal filer  too high.
• First xtal filter in not wide enough for SSB.
• Oscillation in audio path.

These problems were solved but there is room for improvement.

The first SSB xtal filter at 9 MHz contains just 4 xtals. The shape factor can be improved.
The second  xtal filter is at 6MHz that is used for passband tuning. It also contains 4 xtalsThe bandwidth is not optimal for SSB. It is usable but it should be wider. Now it is 2 KHz wide at -6 db. It is usable but not optimal. Shape factor is not really good so SSB signals can be heard.
In the future I will rebuild these xtal filters to optimize performance.

A problem that needs to be solved is the software. Sometimes the display or the microcontroller crashes/hangs. It seems to be related with the new pushbuttons that I use. During development I was using on/off switches .No crash was seen back then. The pushbuttons suffer from bouncing.

Now the receiver works but it needs improvement.
But first I will make the transmitter work. Then the transceiver is usable.
After that the improvements will be made one by one.

All modules build in

All modules are build in the transceiver. It is complete now.
Now I need to make them all work together.

Outside

Inside

Driver

I have build the driver. It seems to work fine.

I used a 1350p as variable gain amplifier. The signal out of the 1350p is amplified by a MAR2.
Finally the signal is boosted to 10 milli Watt by a vhf power transistor that I happened to have. A 2n2219 would do as well. It works in class A
The signal is detected at the collector of the vhf transistor. This signal is fed back to the 11350p input.
After some fiddling with components values the output was constant between 1.8 and 30 MHz.

I have applied a 2 tone signal. to test the linearity. It looks good on the scope. I have repeated this test on all HF bands an 160 meter. It looks good at all those bands.

I have to test how the signal sounds when real speech is used. This concept works a speech compressor. So I have to use real speech to test it. I don't expect problems. I may have to play a bit with the attack and decay times. Now it has fast attack and slow decay.

The amplification is quite high. I only need a few millivolts to get 10 milli Watt. The signal out of the bandfilters is about 10 times stronger so I need some attenuation.

Current state

Al but one modules are ready. Only the amplifier driver for the PA is not build.
Most other modules are functional but need some work.

PA
This module was a kit from box73.  It seems to work fine.

LPF module
The band filers for all bands but one (28 MHz) are working the way they should. I may be able to increase harmonic suppression but they harmonics are below the legal limit. So no immediate need.
The 28 MHz filter is not OK.  The harmonic suppression is not good enough. On 56 MHz only 30 db suppression was measured. As a temporary fix I have added a series LC circuit that short circuits the 56 MHZ signal. It works but it reduces power to 3watt.

IF strip
I use two xtal filters. One of them is used for passband tuning. Both filteres were not OK.
The one that is used to make the ssb signbal out of a dsb signal was very bad. It was very wide and it has a high attenuation. It turned out that one of the xtals was too far off frequency. This filter is now working Ok. It was needed tom measure all xtals and select the ones that have the smallest  frequency difference.
The other xtal filter is a bit too narrow. I have bypassed it temporary.

TX/RX mixer and noise blanker.
This module contains the following items:

• The TX/RX mixer . This is a HPF505 diode mixer similar to the SBL1
• Amplifier for the DDS LO. It should have the same ampliufication from 1.6+9 to 29+9 MHz. It took some fiddling with capacitor get it right but it works ok now.
• Noiseblanker. This concept was 'borrowed' from the elecraft  K2 design. It worked badly but I used different transitors and some other components. After some adaptions the noise detection works. I still have to see how it performs in practice.

Bandpass filter:
I did not use the filter from the Ukraine after all. I made a new one.  It works Ok but at some bands the bandwidth  is a bit too small. For receiving this is no big deal. I suppose A few db's  will not be noticable.  For transmitting it is not acceptable. Power output should be the same over the whole band. The amplifier/driver should solve this. This has to be done.

Speech compressor/clipper.
This is base on a design of PA0FRI. It is a double sideband clipper. I had some problems making it work. First off all the oscillator did not start. I fixed it by changing capacitor values.
Then I had to fiddle with the clipping. It worked but the output was not symmetrical.
After I got all issues fixed I was quite happy with the performance. The signal sounded OK when I monitored it using a PC speaker. Not really HIFI but usable. When I used it as audio source for the IF strip the resulting ssb sounded pretty good. 

Driver for the PA
This module has yet to be build.
It is complicated. As the signal levels out of the band filters can vary I have to use variable gain. The output level should be the same for all bands.  I think I can do it by using a detector to measure the output signal level and feed back this DC signal to a amplifier that has variable gain. This way the input level can vary while the output level remains at a fixed level.
I will use the 1350p IF amplifier for this.








  

Building it all together

Most modules are ready. Now it is time for the boring stuff like drilling holes.
Picture below shows the frontpanel.

Next picture shows most modules mounted. Wiring is also shown
On the right is the IF strip
The middle one is the bandfilter module.
On top is the TX/RX mixer and the noiseblanker.
PCB on the bottom is the relay module that switches TX/RX ,LSB/USB .
The other PCB contains the CW sidetone generator.

Next picture shows the backside. It contains the PA an the Lowwpass filters.