RIGOL DSA815-TG @ 10 MHz issue/bug

[Giorgio]

Howardlong,

Finally a video on youtube, I didn't because my english is not so fluent.

That's the test I suggested to do too early in this thread if someone has no narrow filter on 10 MHz as DUT.

It's the test I sent 1 year ago to Rigol to have evidence there's something strange happening at 10 MHZ, their reply was they were not interested in such test because bla bla bla... you need to read the how to use TG on the Rigol site etc etc..... and this test proves nothing because it's not the way the instrument has to be used and so on.

For me it was the simplest way to check if the problem is present in a unit. Period.

It's clear that the problem is within the SA when the TG part is activated, beside the different output level of the TG that one can normalize before to measure, the SA has a hop exactly at 10.000.000 and the hop value varies with the sweep and bandwidth values settings. This suggest me there's a bunch of missing bits lost during the pass through one stage to another along the digital circuits. Maybe cut on purpose to mask some other anomalies at 10 MHz.

The german support guy I was in touch via email and I met at Friedrichshafen hamfest said that they can't solve the issue because exactly at that frequency in this hardware version of the instrument there's a switch between circuits (if I understood well, sorry if I am wrong) and that they just can try to shift the problem on a different frequency (this was made with the beta FW they sent me shifting the problem at 8 MHz and something) and that the old hardware version has not this problem.

Ciao

[Gertjan]

I found it very easy to reproduce Howardlong's video.
The phenomenon is there, at all signal levels and frequency / span / RBW / variations.
I used a Rigol DG1032z as a source.  Gen and  DSA815 were both locked to the same 10MHz reference.

I must say, in practice I don't have much problems with this phenomenon.
Recently I did a lot of measuring frequency curves of 10MHz transformers.
I noticed the amplitude discontinuity at 10 MHz, but with a new normalise it would disappear.
The problem was worst at a TG level of 0dBm. Taking the TG level a few dBm down made the problem much less bothersome.

Here is an example, comparing frequency curves of three transformers:


Regards, Gertjan.

[pa3bca]

I tried to replicate Howardlong's findings, exactly following his test setup.
On my DSA815-TG I cannot reproduce the amplitude discontinuity. The 156 Hz frequency upshift of the displayed spectrum however does appear as soon as I activate the TG.
Yellow: 9.9999 MHz (from a DG1032Z) without TG active,
Purple: with TG activated (level of TG does not seem to have any influence), note 156 Hz shift.
Also notice the 1.5 dBm lower peak level... what gives?

[Gertjan]

I reproduced exactly the same screen shots as pa3bca:





It appears the difference is in the version of the Main Board.....
pa3bca has an older Main Board, version 00.04 (with boot 01.03)
I have a newer Main Board, version 00.08 (with boot 01.04)

So it seems that the problem was not in the original design of the DSA815, but introduced in a later revision......

regards, Gertjan.

[mcinque]

Thanks for the video! 

Interesting test, so it's not an issue in the signal that the TG produces! It's something that happens inside (mixer?) just when it's been activated! Really weird.

Thanks again for the video.

[N8AUM]

Well this is really starting to suck now! I applied a 10.8Mhz sig as if I was messing around with a typical IF frequency and found that the SA shifts ONLY 6Khz when I have the RBW set at 3K and the span at 100Khz  !!!!!  I used an external 10Mhz GPSDO to eliminate if the internal clock  farts or not when the TG is on.
Being off 156Hz is not a big deal but 6K ??   
I also noticed the screen grab program ONLY works when I have the SA setup in a certain way. If I try screen grab with the RBW set at 1K the LAN is AFU but it works fine if I change the RBW to 300 or 100Hz
Can someone with the hardware mod try and duplicate this ?
Also, when the screen grab is working how long does it take using LAN ?

 

[johnny_canuck]

I tried to replicate Howardlong's findings, exactly following his test setup.
On my DSA815-TG I cannot reproduce the amplitude discontinuity. The 156 Hz frequency upshift of the displayed spectrum however does appear as soon as I activate the TG.
Yellow: 9.9999 MHz (from a DG1032Z) without TG active,
Purple: with TG activated (level of TG does not seem to have any influence), note 156 Hz shift.
Also notice the 1.5 dBm lower peak level... what gives?

I have a 00.03 main board. Otherwise, the same configuration as PA3BCA, with the same results.
Ken, VE3FIT

[pa3bca]

Well this is really starting to suck now! I applied a 10.8Mhz sig as if I was messing around with a typical IF frequency and found that the SA shifts ONLY 6Khz when I have the RBW set at 3K and the span at 100Khz  !!!!!  I used an external 10Mhz GPSDO to eliminate if the internal clock  farts or not when the TG is on.
Being off 156Hz is not a big deal but 6K ??   
I also noticed the screen grab program ONLY works when I have the SA setup in a certain way. If I try screen grab with the RBW set at 1K the LAN is AFU but it works fine if I change the RBW to 300 or 100Hz
Can someone with the hardware mod try and duplicate this ?
Also, when the screen grab is working how long does it take using LAN ?

I can replicate these (disturbing) findings.
There always seems to be a frequency shift  when the TG is activated. The shift depends on the RBW and the SPAN. for instance if I decrease the span in abovementioned test from 100 KHz to 50 KHz, the shift halves to 3 KHz. The shift also decreases if the RBW is decreased. But there is always a shift...
Looks like a simple to fix SW bug. Let's see how long it takes Rigol to notice and (after realizing what a serious issue this is) to fix it.

[H.O]

Let's see how long it takes Rigol to notice and (after realizing what a serious issue this is) to fix it.

They are not monitoring or searching forums for potential issues to be reported so you need to actually tell them directly.
I don't have a SA and I don't know much about them but this, unlike some other obscure "bugs" that gets totaly blown out of proportion, looks like a pretty serious issue which I'd be surprised (and seriously pissed of if it was my SA) if they didn't look into properly but again, you need to tell them directly.

And even though it has been reported it's probably better that everyone who does have an issue with this does so that they're not thinking it's just "one guy".... The more people complainging about it the larger the chanse of it getting fixed, I would think.

[KD0RC]

It seems like the severity of the issue needs to be determined.  The normal use for the TG is to sweep filters and that kind of thing.  It would be interesting to compare with a high end SA to see if the shift is intentional.  In other words, is the SA display intentionally shifted to properly display the TG freq as it sweeps?  If someone could verify that one way or the other, it would help the discussion.

Len

[Howardlong]

It seems like the severity of the issue needs to be determined.  The normal use for the TG is to sweep filters and that kind of thing.  It would be interesting to compare with a high end SA to see if the shift is intentional.  In other words, is the SA display intentionally shifted to properly display the TG freq as it sweeps?  If someone could verify that one way or the other, it would help the discussion.

Len

I have started doing that. I have an HP 8753A here and some narrow 10.7MHz crystal filters that I'm matching up to 50 ohms for some tests. It will be interesting to see what happens.

[Howardlong]

...and here is the result:

[georges80]

Thanks Howard for the video. I presume the frequency 'offset' changes depending on the RBW of the DSA ??

I notice the frequency shift changes if you feed say 10MHz into the DSA from an external signal generator and you vary the RBW of the DSA with the TG turned on. With the TG off, the peak is spot on at 10MHz.

cheers,
george.

edit: dumb typo.

[N8AUM]

Nice job Howard 

I found a few crystals in my scrap box and trying to find one that works !
Do you have couple of 3 or 6dB pads that you can stick on the the input and outputs so the crystal sees the same regardless of which instrument your hooked up to ?
I see the 1.6Khz shift with my Rigol and will try to see what my old HP 8591E comes up with.

Thanks for posting video, reading results are ok but seeing is much better LOL !

[Howardlong]

Three more vidja's for you.

The first does the same as before but with 6dB pads. The result is the same: I spent some time matching the filter for the previous video as it was, if it had been significantly different I would have had some explaining to do!

The second video follows on showing the impact of RBW. Eeeewww, not nice.

The final video is more educational, and shows how I came up with the matching network. This kind of thing is all too often shrouded in mystery and the black magic voodoo of Smith Charts, but it's also the bread and butter of RF engineers. In this video I show how to calculate your matching network without any of that scary nonsense. While I am not dissing traditional methods, you'd be nuts to want to do this kind of thing on a Smith Chart these days.

[N8AUM]

EXCELLENT videos Howard, many tnx !!!

[georges80]

More fun.... Feed in your signal (10MHz etc), turn on the TG, set the RBW and note the frequency offset (from 10MHz).

Now, push the Sweep/Trig button and adjust the "Time" parameter. Note how the center frequency moves closer to the 10MHz (or whatever you set) as you increase the Sweep Time.

So, there is definitely something weird going on across the board when the TG is turned on.

The real question becomes "are these effects due to some purposeful software that compensates/aligns for the TG as a source"? i.e. if the TG is actually feeding a circuit that the DSA then gets input from is the frequency aligned correctly? And we are only seeing frequency shift anomalies due to feeding externally a frequency source that is not related to the internal TG timing.

Of course the 'discontinuity' in the newer mother boards adds in yet another quirk (a more irritating one).

cheers,
george.

[Howardlong]

The more I look at it the more I believe this is fixable in firmware, particularly the frequency offset thing.

[KD0RC]

Nicely done Howard!  Thanks for the effort on this.  I concur - it looks like a bug, not an intentional shift.  For my hobbyist purposes, this is not terrible, but good to know so that I don't try to adjust a filter that is actually already in spec...  I hope Rigol addresses this, but I am not holding my breath waiting for them...

Len

[AF6LJ]

The more I look at it the more I believe this is fixable in firmware, particularly the frequency offset thing.

I am sure you are right.
My question is why hasn't Rigol addressed this?

[AF6LJ]

Nicely done Howard!  Thanks for the effort on this.  I concur - it looks like a bug, not an intentional shift.  For my hobbyist purposes, this is not terrible, but good to know so that I don't try to adjust a filter that is actually already in spec...  I hope Rigol addresses this, but I am not holding my breath waiting for them...

Len

You do have to be careful; your test equipment will lie to you from time to time and often at the most critical of times.

I still think for the money I am better off with a well cared for HP-141T or an HP-8569.
Back in the old days we used seperate sweepers and diode detectors, they seemed to work just fine.

[KD0RC]

The more I look at it the more I believe this is fixable in firmware, particularly the frequency offset thing.

I am sure you are right.
My question is why hasn't Rigol addressed this?

My guess is that they either don't know about it, or do know about it but have not heard enough complaints to do anything about it yet.

[AF6LJ]

The more I look at it the more I believe this is fixable in firmware, particularly the frequency offset thing.

I am sure you are right.
My question is why hasn't Rigol addressed this?

My guess is that they either don't know about it, or do know about it but have not heard enough complaints to do anything about it yet.

The third and fourth possibilities;

They don't feel it is worth spending ham-hours to resolve a firmware issue that is in their mind a minor issue.   

They have in some way painted themselves into some kind of engineering corner and cannot band-aid or patch their way out.

[Bud]

And the most obvious possibilty that everyone have missed is they are simply incompetent .

[AF6LJ]

And the most obvious possibilty that everyone have missed is they are simply incompetent .

I wasn't going to say that considering they seem to make things that work more often than not.