The AM Forum

Last night I was listening on 75 and overheard a QSO that K1KBW was having. He made a comment that caught my attention. He was saying that the Flex Radios weren’t capable of negative peaks beyond 100%. Any truth to this, especially with 6500’s? If so why? Since I don't have a mod monitor I couldn't verify it.

Any negative modulation over 100% is going to result in carrier cutoff and splatter, unless you have a rig with a balanced modulator, then the negative peaks are filled in with wavelets, which what I understand is the negative peak coming back around. Probably in the software the flex just doesn't allow the carrier to cut off, therefore not exceeding 100% negative.

No transmitter can modulate more than 100% negative. The comment makes no sense.

I believe that is in relation to the baseline problem plate modulated xmitters have.

A low level mod (balanced modulator unbalanced) that attempts to overmodulate makes 'wavelets' as they've been called on this site and others.

Due to the method of generation (mathematically making a balanced modulator), it is impossible to baseline the flex.

However, the 'wavelets' do add audible and visually verifiable distortion.

This is what is being described, I think.

--Shane
KD6VXI

Agreed. I thought the same thing. A cut-off carrier is a cut-off carrier once the modulating voltage equals the carrier voltage in the negative direction, IE, a 10 volt carrier is modulated by a 10 volt negative-going waveform resulting in zero volts, zero output.

Then I had a thought - what if a 10 volt carrier is modulated by a 20 volt negative-going audio signal? The result is still the same, a cut-off carrier, BUT the DURATION time of the cut-off carrier is going to increase due to being in the cut-off region longer, like a class C amplifier.   Is this true?  If so, could we say that a carrier is then "technically" modulated negative -200%++ or whatever?

Just curious.

T

It still is not modulating more than 100% negative, wavelets or not.

Tom:

Semantics. Since the RF output cannot be less than zero, I'd say negative modulation cannot be more than 100%.

That said, who cares? A properly set up Flex sounds as good as anything on the air.  :)

The key to removing the Flex wavelets is to use a QIX negative peak limiter or equivalent in the low level audio stages. That would produce unlimited positive peak upside while holding the negative peaks at 95% - no wavelets.

I once went thru the trouble of building a pair of 4-1000As in upside-down-tube config. It was actually a high level balanced modulator and produced wavelets. It used a pair of 3CX-2500s as modulators giving audio up the wazoo. But I was sorely disappointed when the reports came in that the wavelets produces a buzz-saw distortion. After a lot of fussing around I finally realized what was happening and went back to conventional AM limiting.

Again, all Flex owners wishing to run big audio need to use a standard AM limiter to inoculate against the wavelet flu.

BTW, surprise, surprise! The wavelet distortion is really a ghost - not real. Only those with diode detectors can hear it. Use a sync detector and the whirl is beautiful again.  I remember Willie / DUQ, with his tube sync detector, telling me my 4X1 rig sounded FB. But back in 1975, he was one of the 1%ers using one.

T

Thanks, because that was my initial reaction.
Thanks to all for the comments

A while back talking AM Flex to Flex, we both dropped the carrier level, and boosted the audio. Using the SAM detector, things sounded great, and the S/N ratio was vastly improved. Those listening with Model T receivers had to use their BFO to copy us, and some broke in and complained.

Darrin, KE9NS has just added a neat feature to AM for users of the Flex 1500, 3000, and 5000 series Legacy units:

06/06/17 v2.8.0.39: Add: H3E modulation (experimenting). Transmit Single Side Band AM. Toggle between DSB-SAM, SAM-L, and SAM-U modes. Also, Click on AM to toggle between DSB-AM, AM-L, and AM-U modes as well

Jim
Wd5JKO

He has added lots of neat stuff to the aging PowerSDR. The other night I spent hours diddling around with all the new features. I've talked with Darrin and since Flex probably doesn't have the time to update the Flex 5000, 3000, and 1500 manuals, Darrin gave me authorization to add his "features" document to the existing Flex manuals which are done in full color. It, at least, brings the documentation somewhat up to date even if the information is not integrated into the original specific chapters.

Regarding the negative peak crossover wavelet phenomenon, with the ability to change the ratio of sidebands to carrier I don't ever see the cross-over wavelets with my voice no matter how much the asymmetry capability is pushed. To be honest, from what I have seen, most people who operate on the upper bands usually strive for symmetry and greater density in their transmitted signal as fades and large positive peaks without sufficient density don't really help.

At any rate, in OpenHPSDR, with an AM Carrier Level setting of 100, negative peak limiting in the ALC algorithms make it is impossible to cross through the baseline and create the dreaded wavelets, no matter how hard you drive the audio. It is also impossible to drive the DAC into clipping due to the look-ahead algorithm that soft limits the level just before the clipping threshold. Striving for Asymmetry with the SDR architecture actually makes no sense since you are not actually modulating a carrier as you do with a plate modulated transmitter. The software creates a DSB signal and then adds the carrier as an afterthought. Reducing the carrier to make it look asymmetrical on a scope will not change the DSB portion of the signal, so it has no effect on over-all loudness.

As Steve K4HX said, when properly set up an SDR will rival most other AM hardware that is out there. The v3.4.1 release of OpenHPSDR mRX PS includes major enhancements to the built-in transmit audio chain, with PRE-EQ, CFC (10 band continuous frequency compression), POST EQ, Phase Rotator, and adjustable ALC max gain to take advantage of the soft-limiting algorithm at the back-end of the TX audio chain. AM selectable sideband reception and transmit capabilities are there as well.

It just keeps getting better!

Rob W1AEX

Here is the negative peak limiter I use with my Flex 5000 (for those times when I actually use the Flex to generate AM).

(http://www.classeradio.com/negative_peak_limiter.jpg)

I've got 80% of it built. Just need to reallocate some time from the "honey-do" projects. Steve if you ever make a PCB board, I'm interested

Oh, those magnificent men in their flying machines!

With Pure Signal and now software control of asymmetry sideband ratios, the old schoolers are really feeling the bum's rush!   ;D

T

The sharp cutoff and cut-on of slamming into and leaving zero carrier is what causes the
 approximate square wave generation of infinite harmonics.

With decent AM modulation software from any manufacturer, mathematical algorithms will prevent the impressed, rapidly approaching audio from ever causing zero carrier.  Almost zero carrier in the calculus sense, but never quite reaching it.

Yes.  This.

If you have am being created mathematically, you don't need to worry about a lot of things.

Also, it's been reported predistortion is killing assymetry as well.

I prefer a nice, dense multiband compressed am signal fed to a 125 pct modulator.

--Shane
KD6VXI

Hi Shane,

Actually, predistortion doesn't kill asymmetry as it's implemented in any of the SDR platforms. The reality is that with SDR you are not modulating a carrier in the traditional sense as you do with plate modulation. Instead, the software mathematically generates a DSB signal and the carrier is added in at the end. Some software allows the user to adjust the AM Carrier Level of their signal and when the SDR users reduce the AM Carrier Level to make a signal with "pseudo" asymmetry, they are not increasing the peak power of the signal, all they are doing is changing the ratio of power between the carrier and the DSB signal. There is no increase in the peak power of the DSB signal when that adjustment is made, only the carrier power is reduced and so there is no increase in loudness. It just looks nicely asymmetrical on the scope!

What adaptive predistortion does after it analyzes the RF sample from the back-end of the transmit chain, is remove the illusion of asymmetry, because it looks like distortion to the algorithms involved. However, instead of limiting the positive peaks, APD raises the carrier level, and in the case of OpenHPSDR, you end up with 100% positive peaks (assuming you are driving the audio to that point) and 99% negative peaks (you cannot reach 100% negative with an AM Carrier Level of 100 in OpenHPSDR). There is no reduction in PEP and the positive peaks can be driven right up to just before the clipping point of the DAC, which is the absolute limit and is also the point where the ALC look-ahead algorithm will assert itself and soft limit them. The algorithms then predistort the signal to remove out-of-band IMD products, and as a bonus, they also clean up in-band distortion products, which improves the signal's THD from a nominal 2% or 3% depending on the audio source, to a very clean 0.2%.

If you listen to a station that toggles Pure Signal on and off while running AM you will not hear any difference in the loudness of their signal because the peak power output is unchanged. If you record them and then look at the amplitude of the signal's waveform, you will not see a significant difference in the level of the audio content. I have found when measuring this with my own signal that there is actually a very slight reduction of roughly 0.1dB with predistortion engaged while running AM. I suspect that this is probably due to the reduction in THD but I have no proof of that.

At any rate, it's easy enough to see the effect that predistortion has on "pseudo-asymmetry" when running AM and shortly after I acquired my ANAN-200D I made a screen capture recording with my SDRPlay to get a look at it. I posted the result on YouTube yesterday. The level of the VU meter that is displaying the amplitude of the recovered audio is pretty consistent as predistortion is toggled on and off. Unfortunately, due to the latency of the USB connection used by the SDRPlay the synch between the VU meter and the panadapter is slightly off, but you can still get the idea. Also, keep an eye in the reduction of the out-of-band IMD products in the waterfall when predistortion is running.

https://www.youtube.com/watch?v=0_92HNhpfnU (https://www.youtube.com/watch?v=0_92HNhpfnU)

To my eyes, as previously mentioned above, the waveform of the audio track from the video shows a very slight reduction of recovered audio with predistortion engaged in some parts of the video. A screenshot of that is shown below.

As you and Rick both mentioned, the magic is in the math! At any rate, the only way to increase loudness on AM with an SDR is to build up the average audio content with density. The software's internal 10-band continuous frequency compressor does that quite nicely!

73,

Rob W1AEX

Rick,

Thanks for the explanation.  That makes total sense.

--Shane
KD6VXI

Thanks for posting that Rob, I was trying to find a way so say it politely but you nailed it.  No point in audio negative peak limiting ahead of an SDR, at least an ANAN SDR, because it does it for you.  The carrier ratio control will give whatever positive peak to carrier ratio desired right down to double sideband with no carrier.

73s  Nigel

Shane and Nigel,

All the SDR platforms do a pretty spectacular job with AM, but I have not used anything that comes close to the algorithms that Warren NR0V has come up with. His implementation of soft limiting to prevent clipping at the DAC and to restrain the negative peaks with AM works really well.

Another way of seeing it is to watch the scope as an SDR running AM is fully modulated with a 1000 cps tone while pre-distortion is toggled on and off, and then, as the AM Carrier Level is run back and forth through values less than 100 and greater than 100. Keep an eye on the positive peaks, which have been driven to the maximum. They don't flinch at all.

https://www.youtube.com/watch?v=069_uUSz3X4 (https://www.youtube.com/watch?v=069_uUSz3X4)

You'll never catch me running AM with an AM Carrier Level less than 100 with OpenHPSDR!

73,

Rob W1AEX

Rob, you and Stu ought to write a book.    Well actually you have, entire manuals no less.
Succinct, reasonably downward pitched (pun intended for AM topic) and with illustrations no less. I've learned a lot; too bad this type of explanation seems to elude the usual machinations on the Anan sites. Don't mean to be too patronizing, but you've sure helped me.  ;D. This is what makes viewing several open forums so productive.
Bravo.

Hi Rick,

Thank you the kind comments! Stu (AB2EZ) could definitely write the complete college textbook explanation of everything that's going on under the hood of the OpenHPSDR software. My best effort would be more like a child's coloring book version! Fortunately, neither of us has to do that as Warren NR0V has documented everything he has done for the open source framework of the project. If you are interested, at the link below he goes through the basic equations behind the algorithms he implemented in OpenHPSDR mRX PS for the transmission of AM and he also offers some best practices for running it.

Note that the TX processing chain is now quite a bit more evolved with the implementation of the CFC Audio Tools (Pre-EQ, Continuous Frequency Compressor, Post-EQ, Phase Rotator, and Adjustable Soft-Limiting ALC Compression) in OpenHPSDR mRX PS version 3.4.1 so some of the information is now a bit out of date. Additionally, with the implementation of Warren's WDSP and the UDP communications protocol, the sampling rate buffer size is now completely independent of the DSP buffer sizes, the DSP TX/RX filter sizes and your choice of either linear phase or low latency TX and RX filters. This allows you to create razor sharp filters or very relaxed filters for TX and RX without disturbing the performance of your interface buffer and sampling rate selections.

73,

Rob W1AEX

http://anan-100d.wikidot.com/100d-am (http://anan-100d.wikidot.com/100d-am)