Flex 5000 - Its ability to produce positive peaks cleanly

[W1AEX]

Being bored on a cold winter day, I decided to see for myself how hard the Flex can be pushed before strange stuff happens. Using a 10 watt carrier and a 60% carrier level setting in Power SDR, I injected a 1000 cycle tone into the rig and ran the level up to the point where it was modulating 100% and then drove it to the point where it nearly reached 100% in the negative direction. Things looked fine to that point.

I'm curious about what you would call what it does next. When pushed beyond roughly 135% it begins to form little "wavelets" that mimic the major lobes in the waveform. I'm assuming the proper name for this would simply be "distortion". At times I will hear a Flex on AM with a kind of grainy sound on peaks. Is it correct to assume that this crunching on peaks is what happens when it's driven to the point where these little envelopes are forming?

I typically run mine like I did when I measured it this morning, with an AM carrier setting of 60% in Power SDR. Lower settings than 60% will allow much more impressive positive peaks to be reached before the little "wavelets" appear. When I have run my Flex that way, I have received reports that it sounds very loud, but not clean, even though it looks fine on the scope and sounds fine with my "pick-off" monitor. I have heard that some receivers don't deal with heavy positive peaks very well. In fact, when looking at the AM broadcast stations in my area, they seem to run with around 125% positive peaks and rely on heavy compression to raise their average audio so they sound loud. What is the maximum level for positive peaks where the vast majority of old and new receivers are happy?

I'd be curious what others have seen for themselves or heard on the receiving end regarding the use of these rigs on AM.

Rob W1AEX

Note:  My little (low budget production) home movie can be viewed here if you are looking for some low-tech amusement:

http://www.youtube.com/watch?v=1gNsWl8BBD0

[K1JJ]

Rob,

The little wavelets filling in the TX waveform are normal for a balanced modulator. (like your Flex has) Those little peaks happen when 100% negative is exceeded and will grow right along with the positive peaks as the modulatiion % hits 200%, 300% and to infinity. The limiting factor is the headroom of the amplifier which will eventually clip the big positive peaks.

There would be no indication of distortion even at 500% positive if someone were using a sync detector.  But with a diode detector, depending on the receiver, you will hear complaints anywhere from 110% pos modulation on up.   A sync detector is the key to being able to run big audio on AM.

Personally, I turn my audio up until I naturally hit -100% negative and let the positive peaks fall where they may thru asymmetricity.  Then I add in some neg peak limiting or if a balanced modulator, do nothing further.  This seems to suit most receivers/listeners.

T

[KF1Z]

Overmodulation? I guess that's what I'd call it..

Causes a nice bunch of "splatter".

That's why most opt for either keeping the negative peaks under 100% by adjusting the audio, ar by clipping.


By rights, the main screen  RF (power) slider should be left at 100, and the carrier adjusted with the AM carrier control.

Bruce
kf1z

[Steve - K4HX]

It is not overmodulation and it won't cause splatter. The image below shows a typical scope display with sine wave modulation. The top is 100% modulation. The middle is overmodulation in a typical AM system (non-balanced modulator system). Notice the carrier cutoff. The bottom is roughly what the scope pattern looks like using the NE602 (or any other properly operating balanced modulator) with the same amount of audio as in the middle waveform. Notice there is now no carrier cutoff. As you increase the audio, the waveform will begin to look a lot like a DSB-SC waveform - and that's because it is. At some point, the audio to carrier ratio is so great, the signal is effectively DSB-SC.

With speech input, the scope waveform looks pretty normal, except you never see any "sparklies" or bright spots in the negative modulation troughs. Instead, the negative modulation appears to bottom out at about 95%. And the positive peaks can swing way up there if you choose to put the audio maul down.

[K1JJ]

(We posted at the same time, Steve)

Just to clarify...the waveform pattern in Flex 5000 picture three shows a normal balanced modulator on AM and cannot be "overmodulated" to cause splatter like a conventional plate modulated rig or equivalent... unless the system runs out of headroom. 500% modulation is just as clean as an SSB signal with suppressed carrier because they both use balanaced modulators.

On an AM diode detector RX, yes, big audio over ~140% will sound like hell, but on a sync detector, it will sound fine. A spec analyzer will not show splatter up the band on a heavily modulated balanced modulator rig unless the amplifier system is clipping, etc.  The limitations are usually in the amplifer stages following the balanced modulator and where the carrier level is set.

If the carrier is set at 1/4 watt and there is 1KW of headroom, then a huge double sideband signal can be generated cleanly, but is it still called "conventional AM?"  Probably not.

T

[WA1GFZ]

Gee Rob, I'm running 25% on my rig. Output is 10 mW carrier from the exciter and hit over 150%. We need to do some more testing Sunday afternoon. I made a bunch of changes and it would be cool to have you take a look at my signal broadband. 

[W1AEX]

Frank - Sunday afternoon sounds like a good idea except it's my wife's birthday. Hmmm... I can picture how that would work out... I would like to mess around with the carrier % just to see how various settings affect the guy on the other end. I do have VAC working now with the video capture software, so I can do a 720p High Def screen recording that has the actual received audio instead of speaker-to-microphone. Send me an email or PM when you have time to mess around with this and we can check some of this stuff out.

Steve - Thanks for that picture, and the explanation. As you said, unlike my plate modulated stuff, I never see the scope light up with baseline sparklies under heavy modulation. I have heard what you described as DSB-suppressed carrier with some extremely heavily modulated Flex signals on AM. There is a kind of "sidebandy" quality to it that is not pleasant on the receive end. I want to stay away from that undesirable effect.

Bruce - As far as I have observed, the Flex behaves very well by using the "driver" slider to control power. I have not seen any ALC intervention when controlling output power that way as long as the levels through the PSDR audio chain are set to avoid crossing the 0 dB level at each stage. The leveler (limiter) actually does a very good job of holding the maximum audio threshold to stay below 0 dB with ALC. I will try your suggestion to control power with the carrier setting just to see how it works out.

Tom - I completely agree with the ability of the synch detector to handle big audio on receive. Synchronous detection makes a huge difference and never breaks off into "kazoo" mode as my cheap Grundig SWL receiver is prone to do with some of the rigs out there that run big audio. After lots of dummy-load transmissions while listening to the FRG-7700, TS-440, 756 Pro III, and my inline RF-detector, I have pretty much arrived at the point you described. I run the rig to the point where it approaches around 90% negative, and with my voice it seems to develop positive peaks up to around 120% with no sounds of roughness or tearing in any of my receivers. With the audio levels I am using, I ended up with the AM carrier setting of 60% to get to that point. Setting that to a lower value allows me to easily surpass 150% cleanly (no wavelets evident) on the scope, but the Grundig, FRG-7700, and TS-440 start to exhibit artifacts when I venture in that direction.

Thanks for the replies!

Rob

[w1vtp]

I've been playing with balanced mixers for many years.  HP used to put out an excellent booklett on balanced mixers and different application.  This URL is a nice primer on applications of a balanced mixer.  It's a very versitile gadget

http://www.emhiser.com/images/content/121/ETTI-_Mixer_Application_Guide.pdf

[steve_qix]

Hi Rob,

The only thing I'd be concerned about is the tops of the sine wave in the "overmodulation" example are starting to round off..

You have perfectly defined the so-called balanced modulator problem.  The INSTANT you try to "overmodulate" in the negative direction, you get the "reflection" of the modulating waveform, and this sounds seriously distorted using a standard AM detector.

If you could clip the waveform at, say, 90% negative - and prevent it from going further, while allowing the positives to have their own way, you would sound like a standard AM signal with lots of modulation.

I would think it would be a rather simple modification to the balanced modulator circuitry to cut one (or 2, depending on how the modulator is constructed) "keepalive" diodes into the circuit when using AM.. ... or if the preceeding amplifier stage is DC coupled, put the keepalive there.  A lot of rice boxes would sound a whole lot better when the modulation reaches 100% if this could happen 

[WA1GFZ]

Rob, VAC is on my around to it list. Actually want to integrate metis before I get into VAC. I should receive it soon. OK on XYL BD. Well some other time.
We are so close it is a great way to do testing during afternoon.

[W1UJR]

Great video Rob, excellent demo!


Was curious about this video I found on your site...http://www.youtube.com/watch?v=C-LrvGg1S9Q&feature=youtube_gdata_player

[Steve - K4HX]

Geez, there was a whole series of them about a year ago. Try to keep up OM.

http://amfone.net/Amforum/index.php?topic=23825.50

[KF1Z]

Software defined balanced modulator.

Never thought of it like that.


What's even better is when you realize that with SDR, the carrier IS modulation....sort of..
Well, the carrier is generated by an audio tone after all....

tis actually DSB with an inserted carrier.. the carrier generated via audio tones.

[W1AEX]

Was curious about this video I found on your site...

Yes indeed Bruce, that's definitely a curious video! What's hilarious (at least to me) is that the first segment http://www.youtube.com/watch?v=Sqn9b3bZSuY has been viewed over a thousand times. I can only conclude that our civilization must be hopelessly starved for entertainment and this little "reality" segment amused a few people!

[K1JJ]

Hi Rob,

You have perfectly defined the so-called balanced modulator problem.  The INSTANT you try to "overmodulate" in the negative direction, you get the "reflection" of the modulating waveform, and this sounds seriously distorted using a standard AM detector.

If you could clip the waveform at, say, 90% negative - and prevent it from going further, while allowing the positives to have their own way, you would sound like a standard AM signal with lots of modulation.

That's an interesting point, Steve.   I second that.

Years ago I built up an upside-down plate modulated pair of 4X1's as a balanced modulator. It worked FB except when I tried to pour on the audio. Listening on ssb or a sync detector it sounded OK - but not on diode detectors. In that day no one had sync detectors, so I tore it down and went back to regular plate modulation.

My point is that since the E rigs (and other conventional AM systems) can get away with BIG audio using a (soft) negative peak clipper why can’t a standard SDR or ricebox  eliminate the balanced modulator generated wavelet and simulate standard AM with NP clipping ? – as you said.  

If the transmitter is run < natural 100%, no one could tell the difference as to what kind of modulation system is being run. But when going for big positive peaks, we can.

Interesting idea that should be looked into. I know it could be programmed into HPSDR / Flex software, but who would be knowledgeable and motivated enuff to do it, is the question.

T

[KF1Z]

I'll let you know Tom...
I know a Serbian working on it for me "when he gets time".

Problem is, the PSDR software he's writing it into won't run a Flex!   

[KF1Z]

By the way Rob....

Stu AB2EZ shows the same thing you see on your scope in this thread....

http://amfone.net/Amforum/index.php?topic=22728.0

[W1AEX]

My point is that since the E rigs (and other conventional AM systems) can get away with BIG audio using a (soft) negative peak clipper why can’t a standard SDR or ricebox  eliminate the balanced modulator generated wavelet and simulate standard AM with NP clipping ? – as you said.  

Tom - Well... unless I missed something, that capability is already present by adjusting the ratio of carrier to audio in Power SDR. With a 1000 cps tone, it's very easy to cleanly reach positive peaks well over 150% with no evidence of the little wavelets at all. I'll check it tomorrow and see just how far it can go with much lower carrier % values, but it would not surprise me if reaches close to 175% without generating the mutant wavelets.

The biggest limitation, as I see it, is that most (diode detector) communication receivers simply don't handle extreme positive peaks very well. The challenge for me has been to find a level that is demodulated cleanly on the other end by most receivers. I think the level used by most AM broadcasters (110% - 120%???) seems to work even with most cheap-o alarm clock radios and whatever else the typical consumer picks up at Walmart. Then again, in the amateur world, quite often the guy on the other end is listening to you with the 2 inch speaker mounted in the top of his metal box transceiver with a 4.8 kc AM filter as he complains that you sound muffled and boomy...

Bruce - Thanks for linking back to that thread. It kind of proves that everything old is new again since the old thread covers much of this discussion as it pertained to the earlier versions of Power SDR and the balanced modulator wavelet phenomenon.

[K1JJ]

Rob,

Interesting on the Flex software adjustment - making the % AM carrier higher to mask the BM wavelet. I have mine set at 25% now. I have the same software on my HPSDR and will try it tmw too. Hope that does it. If so, then we have eliminated just one more obstacle that's makes an SDR transparant on AM.  The big peaks will always be a problem, however, with any AM rig if the receivers are diode detectors.

Ya know, the time is getting closer when ALL AM receivers can hang a $20 Softrock off of it and upgrade to a sync detector. (needs a computer too)  Imagine if 99% of all AMers had sync detectors then we could run 200%, 300% audio and fidelity would be FB.

Anyway, tonight I got the TMC linear working with the HPSDR. 10mW in gives me a VERY clean 50W pep out. I'm seeing an easy -45 to -50db third IMD now. Really pleased.  The pair of 4CX-250's are running only 650 volts plate and 200V screen with 100ma idle each in class A.  I can barely hear the difference between the -55db HPSDR barefoot and the linear on.   Maybe we can get on over the next few days and give it an on-air check.

Later -

T

[W1AEX]

Interesting on the Flex software adjustment - making the % AM carrier higher to mask the BM wavelet.

Well Tom, as it turns out, it would have been nice if it worked out like I thought it would! I ran 7 different tests with the goal of finding a combination of carrier percentage set point, drive level, and power level that would produce the highest modulation percentage without the appearance of wavelets. I used the same 1000cps tone during the tests. Here's what I found:

CARRIER SET%     CARRIER PWR      PEAK PWR     ALC LEVEL    MOD %    
        25%                      5W                    23W          -15.5 dB       110%
        60%                      5W                    28W            -7.5 dB       135%
        60%                      7W                    34W            -7.5 dB       130%
        60%                      9W                    47W            -7.5 dB       130%
        60%                    10W                    51W            -7.7 dB       130%
       100%                   10W                    50W            -0.5 dB       125%

In the above tests, I set the power output with the front panel drive control in the PSDR GUI. In the test below, I took Bruce's suggestion and set the drive control to maximum and used the "Carrier Set" percentage control to set the carrier power to 5 watts.

        15%                      5W                     23W           -18.5 dB      130%
        
From the data above, I can only assume the following:

1. It looks like the Flex can reach roughly 130% - 135% maximum before the little wavelets appear.

2. The Carrier Set value does have some influence on the maximum positive peaks that can be reached before the wavelets appear.

3. There seems to be a sweet spot for "Carrier Setting" that may change depending on the amount of carrier power used.

4. It does not appear that ALC is much of a factor, unless you set the "Carrier Set" value to 100%. The ALC will begin to intervene right at the point where the wavelets appear when it's set up in that way. The result of this intervention is that the envelope will begin to flatten out on the peaks.

5.  Setting the front panel drive control to the maximum and using the "Carrier Set" value to determine the carrier output power doesn't seem to offer any advantage.

6.  Steve WA1QIX is right and the wavelets are a limitation of the balanced modulator scheme that is employed.

7.  Assuming that Steve K4HX and Tom K1JJ are correct, and the wavelets aren't a bad thing, as soon as you reach the point where the ALC intervenes, the positive peaks start to flatten off. At my usual Carrier Setting of 60% this happens at around 175% positive peaks. The first week that I owned my Flex I pounded the audio quite hard and ran the "Carrier Setting" at a value of 100%. I assume this is why I received some reports initially that it sounded like something was "flat topping" and I thank Dave K3ZRF for bringing that to my attention. With a Carrier Setting of 25% I reached 300% positive peaks, with big wavelets present, but with no flattening of the envelope evident. The ALC was reading -4.5% under those conditions.

8. When it's a voice rather than a 1000 cps tone modulating the rig, a lot of asymmetry is evident, and the scope will show extreme positive peaks without coming close to the baseline. I suppose your mileage may vary with that depending on what receiver is being used on the other end of the conversation.

That's about all I can see from this. I think some of the roughness I hear at times with these rigs is caused by guys who are hitting the point where ALC is intervening, and the peaks are getting squashed. I think I will be content to run mine in that sweet spot where I see positive peaks between 110% - 125% unless someone can convince me otherwise!

Thanks for the ideas and explanations in the thread!

Rob

[w3jn]

All of this is immaterial if the guy on the other end is running a diode detector, Rob.  Some receivers particularly 50s-60s Hallicrafters with ANL - start objectionable distortion at 70% modulation.  I seldom exceed 90% positive and, with the CRL audio chain, everyone tells me my audio is LOUD.  I think the key is more in the audio processing, than focusing on modulation percentage.

[w1vtp]

<snip>
<snip>
The biggest limitation, as I see it, is that most (diode detector) communication receivers simply don't handle extreme positive peaks very well. <snip>
<another snip>

I think it's time for the AM community to start thinking of distortion that is being introduced by our AM detectors regardless of type.  Jay, W1VD has done a great service to the AM community by testng some of our beloved BA receivers. Note that as modulation levels approach 100% the distortion level rises. To observe distortion levels produced by 100%+ modulation is going to result in higher distortions levels.

W1VD BA receiver tests 

It seems to me that as modulation levels exceeding 100% becomes more and more of an experience, the discerning listener is going to need to consider the fact that higher distortion levels that traditional AM detectors produce will make for distortion reports that are not accurate. I have heard on several occasions these distortion reports but was not at all aware of any such thing on my system - that is, until I went from Sync AM to AM (make that traditional AM) detection.  The difference between the two modes is quite remarkable when the source of the distortion is produced by receiver system rather than the transmitter system because of high positive modulation peaks.  That is not to say Sync AM will remove distortion which emanates from the transmitted source - not at all.  Transmitted distortion is just that -- transmitted distortion and no demodulating system will resolve that.

I would really like to see an inexpensive, easy to build Sync AM detector board that could be installed on the average BA receiver.  The audio of this detector should be brought out and amplified by a quality AF Amplifier which in turn could drive a good speaker (not a matching yellowey speaker that looks pretty and goes with that BA receiver).  While the Sync detector manufactured by Sherwood Engineering is a great piece of equipment, most of us cannot or will not invest this much money in the single function.  It is a happy condition that the Flex radio system offers it.

I'm predicting that we are going to hear more and more distortion reports as modern AM transmitters produce higher and higher positive levels of modulation necessary for the assymetrical nature of the human voice.

Best regards, Al

[Steve - K4HX]

Tom - Well... unless I missed something, that capability is already present by adjusting the ratio of carrier to audio in Power SDR. With a 1000 cps tone, it's very easy to cleanly reach positive peaks well over 150% with no evidence of the little wavelets at all. I'll check it tomorrow and see just how far it can go with much lower carrier % values, but it would not surprise me if reaches close to 175% without generating the mutant wavelets.

I don't see how this is possible with a sinewave input. The negative percentage of modulation should always equal the positive percentage of modulation. With a BM, you should see the wavelets on the negative as soon as you go over 100% positive.


Al, your request for an inexpensive snc detector for BA receivers already exists - it's called a SoftRock.   

JN makes a good point. The average level of modulation is more important than the peak level. Add a big carrier to the mix and you are loud!

[KF1Z]

Tom - Well... unless I missed something, that capability is already present by adjusting the ratio of carrier to audio in Power SDR. With a 1000 cps tone, it's very easy to cleanly reach positive peaks well over 150% with no evidence of the little wavelets at all. I'll check it tomorrow and see just how far it can go with much lower carrier % values, but it would not surprise me if reaches close to 175% without generating the mutant wavelets

I don't see how this is possible with a sinewave input. The negative percentage of modulation should always equal the positive percentage of modulation. With a BM, you should see the wavelets on the negative as soon as you go over 100% positive.

I wonder if this is due to the audio output of the software, or the DAC onboard the Flex being unbalanced?
Or maybe the the hardware (QSE) is unbalanced, and the audio offset to obtain balance in software is causing a phase "shift" that shows up as higher positive peaks than negative...?

This maybe could be tested if one were to reverse the I/Q signal lines, and see if then the negative peaks were higher than the positive..

Actually, I think a DC offset in the audio path more likely?

Rob, were you using the tone generator built into the Flex software? or an external generator?

[w1vtp]

Tom - Well... unless I missed something, that capability is already present by adjusting the ratio of carrier to audio in Power SDR. With a 1000 cps tone, it's very easy to cleanly reach positive peaks well over 150% with no evidence of the little wavelets at all. I'll check it tomorrow and see just how far it can go with much lower carrier % values, but it would not surprise me if reaches close to 175% without generating the mutant wavelets.

I don't see how this is possible with a sinewave input. The negative percentage of modulation should always equal the positive percentage of modulation. With a BM, you should see the wavelets on the negative as soon as you go over 100% positive.


Al, your request for an inexpensive snc detector for BA receivers already exists - it's called a SoftRock.   

JN makes a good point. The average level of modulation is more important than the peak level. Add a big carrier to the mix and you are loud!

"I don't see how this is possible with a sinewave input."  Absolutely!  I didn't want to make any reference to what appears to me to be an inordinate emphasis on excessive positive modulation. As you have correctly pointed out if an AM transmitter is faithfully reproducing a sinewave the positive peaks will be 100%.  Period. 

But to make a point: with the human voice that is quite a different matter. If our audio is "in phase" there will likely be greater than 100% positive modulation on a properly adjusted AM transmitter. Now add to that some negative null limiting and there is the possibility of additional positive modulation.  I think the key here is our careful monitoring of our modulation levels and proper adjustments that will result in high talk power but all the while minimizing "splatter."  In that regard, I never cease to be amazed at Gary, WA1OXT's spectrum - he typically has the high end brickwalled at around 5 KC and it is SO evident.  Yet his "talk power" and quality is excellent.

"Al, your request for an inexpensive snc detector..."  True, and I am aware of that option and have actually used that general approach in conjunction with my Yaesu FT-301 -- I do, however,  wish we could get away from the term "softrock."  We really are talking about a very nice SDR program that is free, combined with a mixing scheme that allows one to use the IF output of our BA receiver and insert the mixing product (generally 11 KC) into the soundcard on a computer to access, among other neat features, SAM (Sync detection AM).  How one gets to that function is up to  application of a signal generator and a mixer along with some other components such as an attenuator for proper level adjustment.  My point was that I would like to see a board that would allow one the option of NOT using a computer in the shack -- something similar to what Sherwood offers but much cheaper and perhaps what would make a nice winter's project.

Finally, who can argue with lots of fire in the wire.    In that regard, I'm getting there - one step at a time.

Al