Converting Collins 820D-2 AM broadcast transmitter to 4-400As.

[WB6QED]

That's pretty much what I said on the first page of this thread; apparently, you missed it.

Actually, I had tried adjusting the screen voltage previously as you had suggested and it didn't make a noticeable difference.  However this was before the installation of the 15H choke on the output of the screen voltage power supply and the realization that the 4-400As would not work well at low power which is the power level that I was doing most of my earlier measurements at.  After eliiminating other variables like the output network tuning, I recalled that I hadn't tried adjusting the PA screen voltage since I had installed the 15H choke in series with the screen voltage dropping resistor. 

Once I get the audio feedback issues sorted out, I anticipate that I will be adjusting all of the operating parameters for best performance.

I very much appreciate everyone's help and suggestions!

[WB6QED]

I pulled the modulator feedback networks and carefully reworked them.  I found one 100 pF silver mica capacitor that was leaky and changing capacitance.  I may have also found a cold solder joint.  I matched the resistors on both sides to be within 0.01% of each other and I replaced all of the 100 pF silver mica capacitors while I had it out.  I then reinstalled the feedback circuitry and found that the modulator balance pots now have range with the feedback connected!  I am now achieving 1% THD at 90% modulation from 50 Hz to 10 kHz.  4:1 IMD is about 2%.  Frequency response is -1 dB at 20 Hz and 10 kHz.  The transmitter is now meeting original specifications using 4-400As!  The 4-400As will not modulate much beyond 100% positive.  I believe that this is because of the reduced amount of voltage swing that is available when using the tetrodes instead of the pentodes.

I also fixed a leaky modulation transformer using a small amount Permatex around the insulators.

While there are still a few loose ends to tie up before I start the process of moving it up to 75 meters, I feel that I have pretty much tamed this beast!

To sum up everything that I have done to get it to work to factory specifications using 4-400As:

1.  Installed an additional 15H choke at the input to the screen voltage dropping resistor.
2.  Changed the screen voltage dropping resistor from a 3K fixed to a 5K variable and adjusted the PA screen voltage for exactly 500 VDC.
3.  Installed new carbon composition resistors on the audio driver card to replace those that had gone more than 5% out of tolerance (all of the 1/4 watt carbon composition resistors were out of tolerance!).
4.  Rebuilt the modulator feedback networks using OE components and matched each side to be within 0.01% of the other.
5.  Tuned the PA output network as specified by the original factory production test proceedure.

Brian, WB6QED

[WB6QED]

Perhaps someone knows the answer to one anomaly.  When the transmitter is right at 100% negative modulation, the positive modulation isn't quite 100%, but this is the point of optimum THD.  In the attached photo, the carrier reference is three divisions.

I would expect that optimum THD would occur with both the positive and negative modulation being perfectly symmetrical.

Any thoughts???

Brian, WB6QED

[WD5JKO]

Brian,

   When I do these kinds of tests, I overlay the plot of the modulated b+ over the modulated envelope while scaling the amplitudes to be equal. Use that 1000:1 precision divider probe again. This tells me real quick where the non linearity is. Using a triangle waveform is an option too but is hard to interpret since the both the modulator and the final RF linearity can change the slope of the ramp as seen on the RF envelope.

   Not sure but it looks to me that your non-linearity is near 90-100% negative modulation.

Jim
WD5JKO

[WB6QED]

Use that 1000:1 precision divider probe again. This tells me real quick where the non linearity is.

Jim,

Here you go.  What do you think?  The 1000:1 probe was DC coupled.  The RF sample is taken from the transmitter modulation montior sample port.

Brian, WB6QED

[WB6QED]

I finally managed to acquire enough 5-500A pentodes so I could restore my Collins 820D-2 AM broadcast transmitter back to its as delivered circuit configuration.  It is now 100% original.

The results with the 5-500As have been spectacular!

The biggest improvement is the performance at low power (250 watts).  It will now modulate like crazy at low power!

Restoring the screen voltage dropping resistor back to its original value of 3K brought the screen current down which is probably a good thing for overall reliability.  With the 5-500As, the audio performance was essentially unchanged after I removed the screen voltage power supply modifications that I had made for use with the 4-400As.  It will do +125% modulation at 1 kW output and more than +140% on low power!

90% modulation / 1 kW output:

100 Hz ==> 0.72% THD+N
1 kHz ==>  0.36% THD+N
10 kHz ==> 1.00% THD+N
60 Hz + 7 kHz ==> 2.00% 4:1 IMD
+0/-1 dB ==> 23 Hz-11 kHz
S/N ==> -61.5 dB


90% modulation / 0.25 kW output:

100 Hz ==> 0.40% THD+N
1 kHz ==>  0.30% THD+N
10 kHz ==> 0.70% THD+N
60 Hz + 7 kHz ==> 1.60% 4:1 IMD
+0/-1 dB ==> 21 Hz-12 kHz
S/N ==> -62.2 dB

I think this one is a done deal!  Viva La 5-500A!  The 820D-2 is nearly as good as a current production transmitter.  The only performance benefit that I can see to using one that is pulse width modulated is for improved square wave and transient response.  The engineers at Collins knew what they were doing forty years ago when they designed the 820D-2 with 5-500A pentodes in the PA and modulator stages!

Brian, WB6QED

[K5UJ]

Brian, congratulations!  Steve W9SWR up the road from me also runs an 820D.  Yes the 5-500s are hard to find--I guess that's the only downer about it but otherwise a great rig and beautifully built in my opinion.

Rob