Official Position on Adaptive Predistortion

Rob N4GA

https://community.flexradio.com/discussion/comment/20607015#Comment_20607015

$$$$

Rob N4GA

Good conversation guys. Very civil. :)

Rob

Mark K1LSB

Here are two screenshots taken this evening of me speaking into the mic (transmitting into a dummy load) with APD both enabled and disabled.

With APD disabled, the splatter is at best in the high 30's dB down from passband peaks, and is clearly visible in the waterfall:

APD_disabled.jpg

With APD enabled, the splatter is at least 60 dB down from passband peaks, and is not visible in the waterfall:

APD_enabled.jpg

Also notice that the energy density of the passband in the waterfall is somewhat greater with APD enabled. Not sure why that is but my own suspicion is that the energy that was being lost in splatter has been re-captured into actual passband talk-power.

Mark

Rob N4GA

Good morning Mark.

You would need a spectrum analyzer looking at the output of your amplifier.

Also, the IMD test is a two tone signal, not voice.

I don't say APD does not work, feedback schemes like these are used all throughout commercial communications, especially where channels are packed in by the hundreds, and the signal is digital. (think cellular) And they go by many other names.

My stance is for Amateur SSB, it's way overkill. And, if an OP's signal is really bad, say 6kHz wide, APD is not going to fix that.

If you did want to use your screenshots, with APD off, I see -25db peak down to, call it -70db on the right side peak. This is -45db. If this is 1kW, then that shoulder on the right is 0.032W

So you already don't have a problem, and then you're running APD.

Pretty clean man!

Rob N4GA

Mark K1LSB

Rob,

1) The waveform in the screenshots is a snapshot of the RF output of my AL-80B as captured by and displayed in Thetis. That is the exact same waveform you'd see if you piped the same signal directly into any spectrum analyzer.

2) IMD is a distortion product resulting from the amplifying of a mix of any two or more frequencies in a nonlinear amplifier. Voice is just one example of multiple (actually numerous) frequencies, and provides a plenty satisfactory signal for visualizing IMD distortion on a spectrum scope. Two tone is simply a convenient (and steady) source of multiple signals useful for observing static displays on a scope, but an instantaneous snapshot of voice will show the same relative dB IMD levels in the skirts as a steady state two-tone signal would show.

3) The intent of APD isn't to improve the quality of the TX audio, rather it's to clean up the splatter caused by nonlinearity in the amplifier. I agree that no amount of linearization correction will reduce the distortion originating from a nasty audio source.

4) Personally, I take a rather conservative view of the screenshots (I'm always looking for the poorest results, lol) -- i.e., looking at the disabled APD screenshot I see a delta of only -35dB (a -25dB peak signal and -60dB to the edge of the splatter, so a 35dB difference), which is actually kinda dirty IMO. But that's okay, because I always run APD -- in fact, that same observation (a rather dirty uncorrected IMD) was brought to the attention of Apache Labs' engineers and their response was that they were actually aware of that situation but the presumption was that the end user would be running APD anyway so the matter was officially moot, lol!

Mark

Rob N4GA

1) not sure about that one. You'd be surprised what a real spectrum analyzer shows, vs a software version of one...called "Thetis" haha ;) Also, where exactly you probe the Analyzer also makes a difference, as well as what kind of sampler you are using, what dummy load, on and on. But nothing to discuss here really.

2) Yeah but - you can't measure what you can't measure and taking IMD figures using the human voice as the input is extremely approximate. Its all over the place. This is why the 2-tone is the standard. But I think you said pretty much so.

Also, all this will vary wildy with power level and frequency (band). So when you speak, the amp goes from 0W to 1kW on the "puh" of the letter P. The IMD is great at say 10W and really bad at 1kW. You just can't gleen much from using speech as an input. And Mr. Thetis is not fast enough to show it.

3) QSL

4) That's smart to do.

Rob

Mark K1LSB

Rob,

1) You're laughing at a software package you know absolutely nothing about. Thetis is the authoritative example of the true power of APD. I've shown you screenshots of how it cleans up IMD. Ridicule it at your own expense to your credibility.

2) You're consistently avoiding acknowledging the unarguable benefit of APD. Everyone would do well to recognize that it's coming. Those who ridicule it and ignore it do so at the cost to everyone else who shares the HF bands with them. Every one of us has experienced numerous occasions where some clueless splatter-meister has wreaked havoc on everyone within 10k of his transmissions. APD has the ability to clean up the spectrum for everyone who has to share the bands. Those who refuse to get on board with the technology (including the manufacturers) are only perpetuating the problem.

Mark

Rob N4GA

Well now you've gone and gotten all heated, which is ridiculous.

I laughed at the name Thetis. And assuming I know nothing about it is pretty inaccurate. Assuming anything about me at all would be pretty inaccurate at this point.

I told you I was not against APD. Why would I be against the concept of a form of feedback in an amplifier, dubbed APD? I also conceded to you many good points made by you.

APD can't clean up 10kHz of IMD splatter. We already went over that.

Not sure why you're looking for a constant argument here.

But I do know.

Anyway, you can write more but I'm at the end of the road here.

73

Mark K1LSB

1) "...vs a software version of one...called "Thetis" haha"..."Mr. Thetis". The only times you mentioned Thetis was when you were laughing at it or mocking it.

2) "You just can't gleen much from using speech as an input". That's a flat untrue statement. Thetis is perfectly capable of generating fully formed correction curves (for both amplitude and phase errors) from the user simply speaking into the mic. I use that feature all the time, in fact I haven't used the two-tone feature in ages.

3) You've repeatedly let us all know that "nobody needs APD". You mock the users of Apache Labs radios as a "small group of users [that] has taken on APD like a religion...Mostly because they were the only ones to have this so-called holy grail and nobody else did...Usually those people also have what they think are studio quality audio setups - all for really bad SSB audio quality...Pretty entertaining".

4) "APD can't clean up 10kHz of IMD splatter...We already went over that". There are a number of possible causes of splatter. When splatter is caused by nonlinearity in an amplifier, APD can clean it up.

And I'm done here too.

73,

Mark

P.S. I was never "all heated"...tonal inflections can't accurately come across in written words. That's another assumption you made.

Rob N4GA

Wow. You really got me.

let me ask you this basic question why is a anon fan boy trolling around on the flex radio message boards?

Imaflex radio user that's why I'm here. Why are you here?

David Decoons, wo2x

Guys. Let’s drop it down a notch or we’ll lock the thread.

It is OK to discuss various competitor’s equipment and compare pros and cons. That is how people can learn what is right for their station. I find the old adage “If I own it, then it is good and other manufacturers are bad”. I am a member of a lot of different forums.

73

Dave wo2x

Rob N4GA

Hi David. Please do lock the thread.

We have a Anan fan here with exactly that attitude "if I own it...." as you stated.

73

Mark K1LSB

Dave, please let me answer Rob's accusations...

Rob,

1) You accuse me of being a "fan boy" of Anan, but I'm not. I'm just a proponent of APD. So is John K3MA, per his October 30th comment. That's what the most recent subject of this thread is all about, and that's the only reason I'm here. I automatically get notified when anyone posts in any thread I've ever posted in. I'd made several posts in this thread several years ago so I got a notification when John posted his October 30th comment on APD.

2) You accuse me of trolling, but I haven't made a single disparaging remark about Flex, nor have I mentioned Apache Labs at all (except to agree that their transmitters are dirty). I've mentioned Thetis several times, but Apache Labs didn't write Thetis and doesn't support it.

Dave, thank you for your indulgence.

Mark

G2YTpeter

This is my first venture into the subject of APD. I’m not much of a ‘forum’ guy, but I got interested in the subject when realising how astonishingly clean and tight the output of several friends Anan radios were, despite running high power. Having read and understood (most) of what people have been trying to put across, I’d point out that this thread was started in January 2015 as a statement from Flex as to where they stood on the subject. It seems strange to me that having stated so long ago that the 6000 series was designed with implementing APD in mind that Flex has not, in the intervening NINE years since then actually got around to actually doing it. There seems to have been a lot of ‘firefighting’ with supply issues, designing military equipment, not to mention a couple of near debacles with SmartLink, and probably other issues to which us owners/users are not privy to, but even so… nine years is an extraordinarily long time for it to be on the ‘back burner’. I would suggest that now might be the time to ramp the subject up the priority ladder if it is really possible. Either DO it, or make a statement that Flex doesn’t consider it worthwhile or necessary.

Sure, having or not having APD is unlikely to be a ‘deal breaker’ as to whether or not someone would invest in the Flex route, and I’m definitely not going to change to Anan just because of Flex not having APD, but as has been pointed out by others, its starting to be implemented in other radios and it would be good for Flex not to get left behind… plus I’d like a super-clean signal like that too - and yes, I KNOW that my Flex isn’t a dirty transmitter as it stands (and cleaner than most).

In other areas Flex are considered a leader… but evidently not in this subject. I guess that most people interested in APD are now aware that Icom have made a trial implementation for the IC7610. At the moment it’s only on the Icom JA site, but you have to imagine that it won’t be long before it’s available for the IC7610 worldwide. Others will surely follow. There are already several YouTube videos on the subject of it being implemented on the IC7610… here is a link to one of them…

https://www.youtube.com/live/sRFvfvW4l80?si=4P9NsADS2DZu5714

Regards,

Peter