Re: FeelElec New Arrival FY-6900 Signal Generator

[StillTrying]

It looks like both attachments finally made it into this topic thread for all to enjoy (and perhaps be inspired by).

I could see them as soon as you posted them, 2 X Square is reference-Sine is 30mHz off.zip (2151.72 kB and 1X Sine is reference-Square is 30mHz off.zip (2135.43 kB

I know a bit about DDS and the 4ns problem, I surprised feelelec didn't make any attempt to show it, or lack of it, - well not surprised really.

[exe]

How does "magic pulse" work? (what a horrible name btw)

[StillTrying]

How does "magic pulse" work? (what a horrible name btw)

I think that for rectangular and triangular waves it chooses just the frequency steps that align with the 4ns intervals, so no jitter but much less frequency resolution, but that's just guesswork based on no info.

I can't play JBG's vids in VLC as .mkv or .avi, but that might just be my PC(s).

[Johnny B Good]

How does "magic pulse" work? (what a horrible name btw)

I think that for rectangular and triangular waves it chooses just the frequency steps that align with the 4ns intervals, so no jitter but much less frequency resolution, but that's just guesswork based on no info.

I can't play JBG's vids in VLC as .mkv or .avi, but that might just be my PC(s).

 Triangle waves (as opposed to sawtooth), like the sine wave, aren't afflicted by this 4ns jitter issue. As for the attached files, you'll need to extract each mkv file from within each zip file before you can play them.  I suppose I could have simply cheated my way past eevblog's file type restriction by simply renaming the mkv file extension to zip. I'll try that now with one of those files...

 I tried the new fangled "Drag 'n' Drop" option - it seems to have attached something although the uploading went rather swifter than I was expecting so I'm not sure whether it actually worked .

[EDIT #2]  It worked! I downloaded them and used "open with" VLC media player to play them. I'll keep that renaming to zip trick in mind for any future movie file uploads.

JBG

[chickenHeadKnob]

 
 Configure your function generator for 5MHz on both channels. Pick a convenient voltage level such as 5Vpp (2.5Vpp into 50 ohm load) and initially set both waveforms to square. Nominate one channel as your 'scope's trigger source and apply a 30mHz frequency offset[2] to the other channel and observe how it changes its phase displacement with respect to the triggering waveform (or slips past the triggering point if using a single channel 'scope triggered from an external trigger input signal set to precisely 5MHz).

 Now go through each waveform type in turn on the channel with the 30mHz offset and observe its side slipping behaviour. You might want to create a table of your observations against each waveform type but that's optional. It shouldn't take too long to determine which waveform types slide smoothly past the trigger point and which slip past it in 4ns increments.

BTW, this was the test I'd rather hoped chickenheadknob would be able to try out with his newly acquired FY6900. I guess he really mustn't have any spare time after all. It's a pity because it would have proved a lot more enlightening than that rather cryptic youtube video demo just posted by FeelElec, supposedly demonstrating their "MagicPulse" technology.

OK, I did your tests JBG. I figured you would not let go of this once learning I have a 6900 unit (sigh).  My problems at chez c-knob involve a lab in boxes and home renovation, but just for you I hauled out my o'scope and made a hole on my cramped and cluttered computer desk.

 I could not duplicate with sine,square,rectangle or triangle waves; ramps pos,neg, trapezoid, and CMOS and exponentials did however exhibit stepping. I don't want to spend any more time on further investigation right now as this "feature"  holds no obsession for me. I experimented with both sine and square wave as the source (edge) trigger channel. It did not change the cohort of wave shapes which exhibited stepping.

One probably irrelevant issue is that my cabling setup was haphazard and not properly terminated so there was a small amount of ringing with square waves.   The CMOS and trapezoid shapes were free of obvious ringing and they were stepping.

Testing at 5MHz with variable 30-70 mHz offsets

[StillTrying]

"I'll keep that renaming to zip trick in mind for any future movie file uploads."

Just adding something to the end usually works and leaves the real .ext visible   ....30mHz off.mkv.txt  or even
...30mHz off.mkv.remove

I can play them after downloading a new version of VLC.

[Johnny B Good]

"I'll keep that renaming to zip trick in mind for any future movie file uploads."

Just adding something to the end usually works and leaves the real .ext visible   ....30mHz off.mkv.txt  or even
...30mHz off.mkv.remove

I can play them after downloading a new version of VLC.

 Ok StillTrying,

 Thanks for the feedback. TBH, it did cross my mind to simply tack a .zip onto the end... after renaming them.

 Anyway, although it's nowhere near being past my normal bed time (it's only 2am), I'm off for an "early night".

JBG

[FeelElec]

Dear Customers,
We updated a video with subtitles about FY6900 "Magic Pulse" function in youtube.



FeelElec

[StillTrying]

"We updated a video with subtitles"

I think that's actually worse, it confirms that you're not showing any of the unique properties of "Magic Plus", just some 2.71MHz square/pulses.

[beanflying]

"We updated a video with subtitles"

I think that's actually worse, it confirms that you're not showing any of the properties of "Magic Plus".

Not to mention 'Magic' and Professional are just synonymous with kwality Test Gear 

[evava]

Men, enlighten me: it is jitter? Or what is it?
Shaking line is about 200ps ..

[Johnny B Good]

"We updated a video with subtitles"

I think that's actually worse, it confirms that you're not showing any of the properties of "Magic Plus".

 To be fair, it looks like they're trying to demonstrate their version of Siglent's "EasyPulse" technology... and failing badly for want of screen resolution in their demo video. 360p just doesn't cut it in this case. If they'd used 720p or (better yet) 1080p, it would have allowed their audience to see the actual 'scope settings they were using (and very likely the generator settings as well) which would have saved any need for sub-titles in the first place.

 If you're not familiar with Siglent's "EasyPulse" technology, here's a link to the manual for the SDG1062X Signal Generator which boasts this feature (along with, interestingly enough, a mention of 150Msps 14 bit sampling ).

http://www.labtronix.co.uk/drupal/sites/default/files/sdg1000x/SDG1000X_DataSheet_DS0201X_E01A.pdf

The relevant DSO trace screen captures are on the fourth page of that document where it looks like they've sacrificed a potential rise and fall time speed of circa 6ns with a 6.67ns jitter for an 18.2ns Tr and Tf with a residual noise jitter of just 200ps rms or so.

 My initial calculated potential improvement before I saw the actual figures had suggested 12ns Tr and Tf figures - the detail reveals they'd made an even greater sacrifice than just one extra clock cycle, it looks more like two extra clock cycles are being sacrificed on the altar of low jitteryness with their slower 150Msps sampling rate - Tr and Tf times as enduring as 18.2ns! 

  I wonder whether the technique inherently demands an extra two cycles to be sacrificed rather than just the one cycle it seems FeelTech had opted for in their own initial attempt to solve this issue on square waves (7.2ns versus the 3.8ns if they hadn't bothered at all - the other waveforms which contain such 'instant' transitions do so with rise and fall times of just 3.8ns).

 I noted this apposite factoid taken from the "Key Features" list on the 2nd page of that pdf

"Special circuit for Square wave function, can generate Square waves up to 60 MHz with jitter less than 300 ps+0.05 ppm of period"

 which I think answers the question "Only Square Waves?"  I mean, if it's good enough for Siglent (on a dual channel 60MHz AWG costing over five times the price of FeelElec's latest effort), then it ought to be just fine for Feeltech's customers too - just my attempt to keep things in perspective. 

[EDIT]
====================================================================================
 My bad! On taking another look at that "Key Features" list, it seems they're referring to an entirely separate feature aimed at only the square wave function (quite possibly just a squaring circuit driven from the sine wave output with it's own level adjustment circuit - the sort of add on device one could attach to any sine wave generator to achieve low jitter square waves - in this case, integrated into the SDG1062X Signal Generator itself).

 However, since it's a technique which strongly resembles Siglent's (and no doubt every other brand's) solution to edge jitter on pulse waveforms, it probably applies to more than just the square wave example that FeelElec chose for their demo. No doubt, when FeelElec redo their demo video, they'll think to include another pulse waveform.
===================================================================================

 To Feeltech/FeelElec, may I suggest you redo that video in a high enough resolution to allow your potential customers to actually see the 'scope and generator settings? After all, it'll save you having to bother with an audio or subtitled commentary - "A picture (in this case a movie clip) can save a thousand words"... but only if it has sufficient resolution.

JBG

[Johnny B Good]

Men, enlighten me: it is jitter? Or what is it?
Shaking line is about 200ps ..

 Prey, do tell. How did you manage to discern that from such an ultra low res video? 

 JBG

[evava]

PrtScr 

Can be even bigger, if you like..

It is HD 1080p actually as you can see in the bottom right..


So, is it that jitter Magic Pulse technology is about?

[StillTrying]

Men, enlighten me: it is jitter? Or what is it?
Shaking line is about 200ps ..

I'm quite sure it's the scope's sinx distortion and nothing at all to do with the SG or jitter.

To be fair, it looks like they're trying to demonstrate their version of Siglent's "EasyPulse" technology... and failing badly for want of screen resolution in their demo video.

A complete fail on showing what Magic pulse does wouldn't matter, everybody makes mistakes, etc., but I'm including the other fails such as turning up after 18 months ? and saying sorry we didn't have the staff, were the staff numbers less than 1 ?

[exe]

afaik when new video is uploaded, it first only available in 360p. Only after a while higher resolutions are available (bc I've seen similar complains on last the signal path video).

[Johnny B Good]

PrtScr 

Can be even bigger, if you like..

It is HD 1080p actually as you can see in the bottom right..


So, is it that jitter Magic Pulse technology is about?

 Well, for some reason when I go to youtube to watch that video in a higher resolution, I still can't select any higher res than  the 360p initially offered. The missing options suggestions are no help and searching for a hi-res version fails to find any. I've just spent the past few hours watching 1080p youtube vids, mostly high tech electric aircraft vids, to verify that the problem isn't an Opera issue. Do you have a direct link to that 1080p version you found that you can post?

 As for the type of jitter, it can only be the 4ns clock jitter rather than the typical 200ps rms jitter that afflicts all waveforms regardless.

JBG

[StillTrying]

I can turn the quality up to 1080p on the latest video, the quality goes up but there's not much extra info to see!
I can't see a way to link directly to the 1080p version.

Code: [Select]

www.vdyoutube.com/watch?v=wxi_vpUhJwY

[Johnny B Good]

I can turn the quality up to 1080p on the latest video, the quality goes up but there's not much extra info to see!
I can't see a way to link directly to the 1080p version.

Code: [Select]

www.vdyoutube.com/watch?v=wxi_vpUhJwY

 Thanks for that YT d/l link, StillTrying.

 The best quality I could d/l was a 1280x720 version but that was sufficient to show what I needed to know. A 2.71MHz square wave on a 100ns per div X-axis scale isn't the most revealing of settings. Still, at least it wasn't a 2.5 or 5MHz setting to hide the 4ns jitter from view. The 4ns jitter is quite visible at 2.71MHz on an FY6600, modified or not.

 However, I did notice they'd taken care not to include the Tr and Tf figures in the statistics display. I guess we're just going to have to wait for a proper, independent review to materialise before we get to see just how much of a benefit this "Magic Pulse" business will prove to be.

JBG

[evava]

 
 Configure your function generator for 5MHz on both channels. Pick a convenient voltage level such as 5Vpp (2.5Vpp into 50 ohm load) and initially set both waveforms to square. Nominate one channel as your 'scope's trigger source and apply a 30mHz frequency offset[2] to the other channel and observe how it changes its phase displacement with respect to the triggering waveform (or slips past the triggering point if using a single channel 'scope triggered from an external trigger input signal set to precisely 5MHz).

 Now go through each waveform type in turn on the channel with the 30mHz offset and observe its side slipping behaviour. You might want to create a table of your observations against each waveform type but that's optional. It shouldn't take too long to determine which waveform types slide smoothly past the trigger point and which slip past it in 4ns increments.

BTW, this was the test I'd rather hoped chickenheadknob would be able to try out with his newly acquired FY6900. I guess he really mustn't have any spare time after all. It's a pity because it would have proved a lot more enlightening than that rather cryptic youtube video demo just posted by FeelElec, supposedly demonstrating their "MagicPulse" technology.

OK, I did your tests JBG. I figured you would not let go of this once learning I have a 6900 unit (sigh).  My problems at chez c-knob involve a lab in boxes and home renovation, but just for you I hauled out my o'scope and made a hole on my cramped and cluttered computer desk.

 I could not duplicate with sine,square,rectangle or triangle waves; ramps pos,neg, trapezoid, and CMOS and exponentials did however exhibit stepping. I don't want to spend any more time on further investigation right now as this "feature"  holds no obsession for me. I experimented with both sine and square wave as the source (edge) trigger channel. It did not change the cohort of wave shapes which exhibited stepping.

One probably irrelevant issue is that my cabling setup was haphazard and not properly terminated so there was a small amount of ringing with square waves.   The CMOS and trapezoid shapes were free of obvious ringing and they were stepping.

Testing at 5MHz with variable 30-70 mHz offsets

Ok, I did your test JBG, too.
There's really something m a g i c about this FY6900 
See attachments, blue trace sine is trigger, yellow trace square is sliding fluently, just little bit twisting and curling  - don't know why!
From time to time there is skip like jitter but not very often.
Short videos show this curling/twisting of the yellow trace. 

[evava]

and another movie 

[evava]

and last one 

[Johnny B Good]

and last one 

 Ok, evava, and thanks for those interesting one second clips. 

 I had to set VLC to playback at 0.03 real time to see what was going on. I noticed a single brief 4ns ghost jitter in amongst the more continuous movements made up of tiny sub nanosecond 'jumps' in the first two videos but no sign of such a ghost jitter in the third.

 The 5MHz and 20MHz examples showed a curious 'double vision' effect whilst the 30MHz sample was showing a 'triple vision' effect. Most curious indeed! Just how do they do that!?

 As for the 'wiggly' effect, some of that might be the higher frequency components interacting with the sampling rate of the generator's DAC as they approach the HF end of the anti-aliasing filter's cut off frequency, supposedly somewhere between 75 and 100MHz.

 I've seen similar wrigglyness (but without any double and triple vision effects) on the amplitude and the flanks of the Sinc Pulse waveform at 10MHz as it slides past my homebrewed GPSDO's 10MHz square wave trace (the triggering signal in this case).

 At these frequencies, there may also be some cross-talk effect between the two channels so I wouldn't get too distracted by such wrigglyness. However, what can't be ignored is the double and triple vision effect which must be a side effect of whatever this "Magic Pulse" process is doing. I've not see such an effect on my more humble FY6600. That appears to be a new feature of the FY6900.

 Otherwise, when you're not using this 30mHz offset trick to reveal these signal generators' shortcomings, how does it perform with more normal settings? Can you still observe such effects on the square wave without resorting to the use of a 30mHz offset from a "Magic Frequency"?

 It does beg the question as to whether you'd see a similar effect if you run this 30mHz offset trick on a Siglent SDG1062X at its 'magic frequency' (based on a 150MHz sampling rate of course - 3MHz in this case).

 I think I may have a new motto: "You make 'em, I break 'em.". 

 Incidentally, just out of curiosity, I processed those clips with Handbrake to compress them into mkv container files and converted the 1st two 3MiB clips into 1.2MiB mkvs and the 3rd 4MiB clip into a 1.5MiB mkv. If you'd supported the camera on a tripod to steady the footage, I'd have probably been able to get a bit more compression.

 I reckon you'd have been able to offer 5 second or longer clips within the 4MB limit if you'd chosen a lower resolution such as 1280x720 of footage shot from a locked off tripod supported camera and compressed them with Handbrake or a similar video compression tool. Even 720x576 would likely have shown sufficient detail and allowed you to offer even longer clips within the 5MB limit, just a suggestion to get the most out your movie files if you want to avoid hosting them on youtube just so you can post youtube links instead of 4 and 5 MB attachments.

 At the time I did my own low res movie clips, I'd had a 2MB limit in mind. Knowing now that I can post 4MB clips, I'll try using 720x576 to create 20 to 30 second clips next time. You don't always need full HD to show short movie clips of unusual 'scope traces. 

JBG

[Johnny B Good]

 
 Configure your function generator for 5MHz on both channels. Pick a convenient voltage level such as 5Vpp (2.5Vpp into 50 ohm load) and initially set both waveforms to square. Nominate one channel as your 'scope's trigger source and apply a 30mHz frequency offset[2] to the other channel and observe how it changes its phase displacement with respect to the triggering waveform (or slips past the triggering point if using a single channel 'scope triggered from an external trigger input signal set to precisely 5MHz).

 Now go through each waveform type in turn on the channel with the 30mHz offset and observe its side slipping behaviour. You might want to create a table of your observations against each waveform type but that's optional. It shouldn't take too long to determine which waveform types slide smoothly past the trigger point and which slip past it in 4ns increments.

BTW, this was the test I'd rather hoped chickenheadknob would be able to try out with his newly acquired FY6900. I guess he really mustn't have any spare time after all. It's a pity because it would have proved a lot more enlightening than that rather cryptic youtube video demo just posted by FeelElec, supposedly demonstrating their "MagicPulse" technology.

OK, I did your tests JBG. I figured you would not let go of this once learning I have a 6900 unit (sigh).  My problems at chez c-knob involve a lab in boxes and home renovation, but just for you I hauled out my o'scope and made a hole on my cramped and cluttered computer desk.

 I could not duplicate with sine,square,rectangle or triangle waves; ramps pos,neg, trapezoid, and CMOS and exponentials did however exhibit stepping. I don't want to spend any more time on further investigation right now as this "feature"  holds no obsession for me. I experimented with both sine and square wave as the source (edge) trigger channel. It did not change the cohort of wave shapes which exhibited stepping.

One probably irrelevant issue is that my cabling setup was haphazard and not properly terminated so there was a small amount of ringing with square waves.   The CMOS and trapezoid shapes were free of obvious ringing and they were stepping.

Testing at 5MHz with variable 30-70 mHz offsets

 Apologies for the tardy response. 

 I'd gotten a little distracted by the low res youtube video issue I'd been experiencing so your reply got pushed so far onto the back burner, it took evava's reply to remind me about it and the fact that I hadn't actually made a reply.

 Anyway, thanks for taking the time to run those tests which seem to indicate that this Magic Pulse is only being applied to square waves. Sine and triangle waves didn't suffer from this 4ns clock jitter effect anyway, just square waves and any other waves with an instant transition between voltage levels such as ramps and sawtooth waveforms.

 It rather looks like they've moved the square wave out of the jittered waveform category and into the jitterless category which includes sine and triangle waveforms. I'll have to test the trapezoidal wave since I'd have expected it to be free of this jitter just like the triangle wave which I'd swear was a (surprisingly) jitterless example when I first checked it out.

 Anyway, that's more food for thought. I'll have another look at what is and what isn't free of this 4ns jitter on my FY6600 when I get some free time. I'm currently trying to complete a GPSDO project I've just started transferring from a prototyping breadboard lashup onto a piece of veroboard that I've prepped up to slide into a small extruded aluminium project enclosure, so I'm a little preoccupied right now - I do appreciate were you're coming from 

JBG

[evava]

Thank you JBG for your answer.
Indeed it must have been some crosstalk or crossmodulation between channels or external modulation  of a special kind.
If I set both channels to square wawe this triple curling appears in both channels!

(you can set your player to continuously replaying this short clip, so you can better see what is going on)

If I push Single on scope, there are always just thin lines (yellow and blue ones), but always they have another shape.
Photos in next post.