Reducing Power Toob Rigs

[K1JJ]

Yes, the "wavelets" generated by a balanced modulator are a real thang.  It seems those with diode detectors are more bothered by them. Those with sync detectors hear no problem.

Back in the late 1980's I built a 4-1000A "upside-down tube" rig. Essentially it was a high level balanced modulator using two 4X1s in class C.  I used a pair of 3CX-2500F3s as modulators for huge audio capability. What a rig that was.  It swept cleanly and had the balanced modulator wavelets. But I could never understand why some reported grunge on the audio when I hit it hard. This, of course, was the wavelets problem. Guys like Bill/W3DUQ who used an early tube sync detector said it sounded fine when running +180% positive.

Since the majority used diode detectors  back then, I converted it to a conventional plate modulated rig and all was FB.

So, I think the external negative peak limiter that Steve/QIX is talking about will be a nice option for low level balanced modulator/ linear amplifier combos.  There are many of these rigs out there out there that could find them very useful for AM.

T

[Pete, WA2CWA]

In a software defined rig such as the Flex series, is there really a "balanced modulator" as we traditionally are familiar with. I can find no reference to a "balanced modulator" in any of the Flex documents, specs, or blurbs. I did find, "The FLEX-5000™ family comes with FlexRadio PowerSDR software that performs all DSP, modulation, demodulation and control functions for the radio." I assume the hardware being software controlled for modulation would be the A/D, D/A converters. Even my Icom 756 PRO II doesn't have anything called a "balanced modulator" in it. So, are you guys just generalizing using the phrase "balanced modulator"?

[WD5JKO]

Yes, the "wavelets" generated by a balanced modulator are a real thang.  It seems those with diode detectors are more bothered by them. Those with sync detectors hear no problem.

So, I think the external negative peak limiter that Steve/QIX is talking about will be a nice option for low level balanced modulator/ linear amplifier combos.  There are many of these rigs out there out there that could find them very useful for AM.

  I played with this many years ago with one of my QRO modified Central electronics 20A's. Those wavelets on AM were very bothersome, as they would appear with any attempt to get the positive peaks close to 100% modulation. More recently I was in QSO with Sammy, W5DPP using his Flex 5000, and Sammy had the wavelets going big time as he had the Flex miss-adjusted for AM. I was on my 20A, so I did the same thing. He was copying me on the Flex with SAM, and I was copying him on my Icom R75 with SAM. Essentially we we running DSB with reduced carrier. Sounded great, and had very punchy signals with good S/N ratio. Some folks then broke in, running normal AM gear (diode detector receivers), and they wondered what the hell we were doing, since to them we sounded awful.

  What I did on my 20A audio chain was to make a circuit to eliminate those wavelets. I chose to make a soft clipper with adjustable threshold since different amount of RF out needed a proportional change in the peak audio required. I called it PNCL, progressive negative cycle loading. When used lightly to catch the occasional peak, the system sounds just fine. When used to really soft clip the negative cycle, the 2nd harmonic distortion is high. Still, with a heavily soft clipped signal, on a diode detector, it sounds better then you'd think...but with those big wavelets, a diode detector's output would be horrid sounding.

  I am not sure yet on the Flex, if we need an operator adjustable threshold control. My Flex 3000 makes wavelets just like my old 20A.  

  I attach some pictures, and a schematic. The carrier was at 2 cm, so the modulation positive peaks are at 150%.

Jim
WD5JKO

[N2DTS]

I think we went through the flex radio stuff before. From what I remember, the sdr stuff gets down to zero then starts going up again instead of staying at zero.

But the flex software has an adjustment that will allow 150% positive without going down to zero, its called the carrier control.
If you adjust it to get say 15 watts carrier out, you should still be able to get 100 watts pep.
At least, that is the way it used to work.

[Opcom]

I agree with the poster that said reducing the plate voltage. At least it works for me without upsetting the rest of the set. variable auotransformers are your friends,

[WD5JKO]

Using a ferrite core (or several ferrite cores) one can make a power divider that will send a portion of a transmitter's power to an ordinary dummy load, and the remainder to an amplifier's input.
Stu

   Stu, This is a very interesting concept. I find that when running a linear on AM, especially with enhanced positive peaks, getting the drive to the amp at the correct level is crucial. Lowering the output of the exciter can work, but that can create problems as well. Having a 100 watt rated switch attenuator uses some pretty big resistors (wattage) wise). I'd love to have a 100 watt rated 1, 2, and 4 db switch attenuator such that I can vary the drive in 1 db steps from 0 to 7 db attenuation.

   I am wondering if the transformer concept could be modified with added taps to get multiple output levels as described above?

I think we went through the flex radio stuff before. From what I remember, the sdr stuff gets down to zero then starts going up again instead of staying at zero.

But the flex software has an adjustment that will allow 150% positive without going down to zero, its called the carrier control.
If you adjust it to get say 15 watts carrier out, you should still be able to get 100 watts pep.
At least, that is the way it used to work.

   Brett, well yes and no. The carrier level on the transmit tab sets the maximum carrier for AM when the front panel drive is at 100%. Setting the carrier at 15 watts does allow about 150% positive peaks as you describe. It also allows those pesky wavelets just like the images I posted of my 20A with PNCL OFF. So my position is that a Flex on AM running high percentage AM % must have external processing to limit the negative peak. If this is not done, the thing makes wavelets, or stated another way, DSB with reduced carrier.

Jim
Wd5JKO

[AB2EZ]

Jim
The simple transformer method I described does not provide the precision that you are looking for (1dB steps).

The input impedance, to the power divider, is not exactly 50 ohms; but you could use a tuner between the transmitter and the power divider to make the input impedance, looking into the tuner, 50 ohms... and thus keep the load on the transmitter constant.

The power splitting ratio is best suited to specific (not integer multiples of 1dB) selections in a switched version. For example: 50% out, 36% out, 20% out, and 5.9% out; using turns ratios of: 4:4, 3:4, 2:4, and 1:4, respectively.

If you swap the positions of the dummy load, and the amplifier's input (assumed to be a 50 ohm input impedance), then the same switched power divider will give the following percentage of the total power to the amplifier: 94.1% out, 80% out, 64% out, 50% out.

Therefore, with the four position switch and the swap to employ... the input-to-amplifier attenuations are: 0.26dB (94.1%), 0.97dB (80%), 1.9dB (64%), 3dB (50%), 4.4dB (36%), 7dB (20%), and 12.3dB (5.9%)


The output power percentage (of the input power) is 100% x (k x k) / [1 + (k x k)]; where k is the turns ratio.

For example, if the turns ratio is 3:4, i.e. 0.75, then the output power percentage is 100% x (.75 x .75) / [1 + (.75 x .75)] = 36%

Note:

When using only part of the primary turns, the unused turns should be left open (not shorted). There will be a voltage produced across the unused primary turns.

In the above example, if we use the configuration where the dummy load is on the primary side, and we only use 2 of the 4 turns on the primary side (the secondary side has 4 turns, with the 50 ohm amplifier input across it)... then the percentage of the transmitter output power going into the 50 ohm input of the amplifier is 20%. The rms voltage across the amplifier's  input will be: the square root of {0.2 x the transmitter's output power x 50 ohms}. The rms voltage across the unused, open, 2 primary turns will be approximately 0.5 the rms voltage across the amplifier's input. If the RF output power of the transmitter is 100 watts, then the rms voltage across the input to the amplifier will be approximately 32 volts, and the rms voltage across the unused, open, 2-turn portion of the primary winding will be approximately 16 volts.

Stu

Using a ferrite core (or several ferrite cores) one can make a power divider that will send a portion of a transmitter's power to an ordinary dummy load, and the remainder to an amplifier's input.
Stu

   Stu, This is a very interesting concept. I find that when running a linear on AM, especially with enhanced positive peaks, getting the drive to the amp at the correct level is crucial. Lowering the output of the exciter can work, but that can create problems as well. Having a 100 watt rated switch attenuator uses some pretty big resistors (wattage) wise). I'd love to have a 100 watt rated 1, 2, and 4 db switch attenuator such that I can vary the drive in 1 db steps from 0 to 7 db attenuation.

   I am wondering if the transformer concept could be modified with added taps to get multiple output levels as described above?


Jim
Wd5JKO

[W1AEX]

In a software defined rig such as the Flex series, is there really a "balanced modulator" as we traditionally are familiar with. I can find no reference to a "balanced modulator" in any of the Flex documents, specs, or blurbs. So, are you guys just generalizing using the phrase "balanced modulator"?

Pete,

Back in 2011 I had a nice chat with Bob McGwier who worked on the implementation of the AM mode in PowerSDR and he described the approach they used as a software derived balanced modulator. Instead of a hardware balanced modulator they generate a double sideband signal with PowerSDR and give you complete control over the ratio of carrier power to sideband energy with the carrier level setting. They describe it as coherent AM. Part of that discussion is in the Flex Edge Reflector here:

http://www.mail-archive.com/flexedge@flex-radio.biz/msg05980.html

Just a few comments about other items in the thread. It seems that many SDR rigs exhibit an anomaly with AM. If you feed them with a sine wave, they may exhibit some unexpected asymmetry. Is it distortion? Is it PowerSDR acting like a negative peak limiter? Is it caused by a slight imbalance between the I/Q streams? I really don't know but as can be seen in the pictures below at 100% positive, it's not even close to reaching the baseline in the negative direction. The screenshots come from a video I made while messing around with this a few years ago:

http://www.youtube.com/watch?v=Ebkmja3xDPE

If you have an asymmetrical voice I suppose this will exaggerate your positive peaks somewhat. At any rate, for what it's worth, with my 5000A I have no problem creating positive peaks to 150% and beyond with normal speech, without crossing the baseline and getting into the wavelet phenomenon, by simply working with the AM carrier level setting. That being said, I usually set things so the positive peaks reach between 110% to 115% and work instead to bring the average audio level up as high as I can.

After 3 years I finally gave up trying to get by with only the software voice processing offered within PowerSDR. It always sounded nice, but the leveler never quite behaved as Flex dreamed it would. It simply isn't up to the task of acting like an effective audio limiter. I found it impossible to bring up the average audio level to the point where the audio density was high enough. This was especially noticeable on sideband where I would frequently get comments like "Your audio sounds nice but the loudness factor is not where it should be for the amount of signal you're putting in here." What everyone runs into is that if you push the audio, you will incur the wrath of the ALC and the IMD will fall apart.

I ended up with a simple external audio chain with a compressor limiter at the end just before feeding the audio into the balanced input. I can crank up the audio level in the external chain and use the compressor/limiter to hold the line on the ALC. This creates audio density that is far higher than before and the external compressor/limiter catches the peaks very effectively to keep the ALC from getting involved.

After running an external hardware audio chain I'm never going back to running naked audio again. Keep in mind with the SDR platform that an "external audio chain" can also mean using software voice processing and coupling to PowerSDR (or the MRX version for HPSDR) through VAC.

Rob W1AEX