RIGOL DSA815-TG @ 10 MHz issue/bug

[Giorgio]

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.

They do monitor this forum for sure, there's evidences everywhere in other Rigol relate threads, at least the german support.

Happy to see see my problem with Rigol is getting everybody problem..... nice job guys.

We hope now they will take seriously my initial complain, the "just one guy rant.."

Please open tickets to Rigol support or we can stay here talking for the rest of our lives without a Rigol response.

Dave answered again he has no time for this so if  there's more people spreading videos about it would be great.

Use the filter DUT test not the external generator because Rigol already answered me it's not proving anything because the TG and SA are directly connected and there's always some kind of normalization on the signal to compensate known levels and frequencies anomalies.

So stick a crystal filter or a simple 10,7 MHz IF transformer and make tests and videos with it.

Ciao.




 

[H.O]

They do monitor this forum for sure, there's evidences everywhere in other Rigol relate threads, at least the german support.

Last time I was in contact with them (and it was the German office) they said they were not. But even if they drop by now and then it can not be expected from them to pick up anything and everything that gets written here - or on any forum.

They may or may not fix this issue and it and other issues should definitely be discussed here but the correct way to let the manufacturer know about it is thru their official channels. If they act on it based on a forum post that's great but they really can't be expected to. They CAN be expected to act on it, or at least tell you then won't act on it (which would suck), if it gets reported to them directly.

[N8AUM]

Jason from Rigol response:

"The tracking generator does effect the measurements when it is active. This is noted in the specifications:
From: User’s Guide for DSA800 Series section 5-2

But, I want to check and see if your reported shift is out of the ordinary."

I sent him a pic showing a 12.5Khz shift at 10.7Mhz
we will see what happens, probably nothing 
 

[wkb]

Jason from Rigol response:

"The tracking generator does effect the measurements when it is active. This is noted in the specifications:
From: User’s Guide for DSA800 Series section 5-2

But, I want to check and see if your reported shift is out of the ordinary."

I sent him a pic showing a 12.5Khz shift at 10.7Mhz
we will see what happens, probably nothing 

That remains to be seen of course.  Especially if 'all'  DSA815-TG owners file a problem report. Rigol have invested considerable
time and money into establishing themselves in the T&M marketplace, this is the sort of anti-marketing most companies are seriously unhappy about.

(/me wonders where to find the time to test my own DSA815-TG for this issue)

[Howardlong]

The reason I chose a crystal filter is that they tend to be narrow band and have very steep skirts, so showing frequency shifts like this is demonstrated much more clearly.

For a relatively gentle LC filter, even of several orders, you'd be hard pushed to show a 1.5kHz deviation at 10MHz. Indeed, for fiddling with LC filters at HF or above, I'm happy to be educated otherwise, but I wouldn't think the frequency shift thing is much of a problem practically speaking, a gentle breeze on the coil probably might have more effect.

For tuning a narrow band IF stage though, this is definitely going to be an issue, and I'd hazard a guess that at its price point, many of the purchasers of the DSA815-TG are ham radio operators who'd use it for just this kind of thing.

[ted572]

In case there are any doubts: Every Single DSA815-TG, NOT Just Those With the Newer Hardware (i.e. Main Board .07/8) have a problem with the SA Measured frequency using a Resolution Bandwidth (RBW) greater than 100 Hz or so, when the Tracking Generator is turned On.

This is NOT just a problem at 10 MHz, but over the whole frequency range of the DSA815-TG.  I typically see a +10 kHz offset using a 30kHz RBW.  Of course this frequency offset can be different depending on the Frequency Span/Scan Time used.

To see this, simply insert a CW Signal between 1 MHz and 1.5 GHz from a Signal Generator and use the SA Marker's Peak Search function to readout the frequency with various RBWs.  When you turn the TG from Off to On you will see the SA's peak maker frequency increase with the wider RBWs.  Note: TG Normalization does NOT help correct the SA frequency readout error at all! 

On the other hand the Glitch reported by 'Giorgio' while measuring a 10 MHz Filter seems to be only on DSA815-TGs with the Newer Hardware.  And as far as I know this is only at 10 MHz.   I don't have a unit with the Newer Hardware, so I can not verify this myself.  Although all the reports from other users so far would indicate that this is probably the case.

So I believe that we ALL have a problem with our DSA815-TG's (even with Main Board .04/5),.   Was it also there with older firmware?  I don't know, or think that it matters at this time.

Please re-frame from making negative comments about Regal.  As this won't get us anywhere, and I'm sure that they will correct this issue as they have with others in the past.  And we can thank ' Giorgio' for discovering it.  Hopefully it will be corrected soon, and how urgent can it be with it taking this long for us to uncover.  As long as it is corrected I will be happy.

Edit: Underlined - when the Tracking Generator is turned On 

[Giorgio]

Ted572,
thank you for the credits about discovering the bug. I really hope they will fix it. Again Gents,  please fill your requests to your Rigol support, it's the only way to get it fixed.

And yes the shift in frequency readout is present from 10MHz to 1,5 GHz when TG is active.

As wrote before they were working on a new firmware they sent me in beta version that moved the problem from 10 to 8 and something MHz. When I spoke to the german support guy he said that the only way to solve it is to try to move the anomaly outside the display view because the problem is related to the hardware and at 10 MHz or 8 or outside the screen has to be there.

After that beta FW they stopped to give me any news. I think because I was the only one complaining.

So please send support requests and let's wait.

I am curious to know if maybe this problem is present also in the 3 GHz version, the DSA832-TG, anybody want to test it? I don't own one.

Edit: sorry I meant the 832, corrected.

[N8AUM]

The frequency shifting with TG on is present on ALL frequencies and the shift in frequency is dependent on SPAN and RBW combination.
As far as I can tell the worse case is a frequency shift of 5.833Khz with the 3Khz RBW
This amount of frequency shift might not even be noticed at high frequencies but it sure is noticeable on the low end, like at 455Khz !

I hope everyone ( all 10 of us LOL )  contacts Rigol about this bug.
Too bad all the schools that have thousands of these dont do the same, than for sure they will come up with a fix!

Its still a nice little toy for the money and I'v been playing with it almost every day for the past 8 months.

[pa3bca]

Well this is getting more and more disturbing. I can't understand why I (or anybody else) has not noticed this before.
I tried to measure a 475 KHz bandpass filter I built some time ago, with the TG.
Span 200 KHz (400-600 KHz), RBW 3 KHz (Yellow), 1 KHz (Purple) and 100 Hz (Magenta).
WTF?
1. The shift is there. The higher the RBW, the more the graph is shifted to the right. The filter is not steep enough to accurately measure the shift, but no doubt it is the same as measured by others and myself with other tests.
2. BUT with a RBW set to 3 KHz there is a 10 dB amplitude discontinuity at exactly 500 KHz!! RBW at 1 KHz also shows a small discontinuity at 500 KHz (hard to see in this screenshot).
Yes, I will definitely complain to Rigol. I urge others to do this too 
(oh and by the way, My DS815 does NOT show the 10 MHz discontinuity....)

[Giorgio]

(oh and by the way, My DS815 does NOT show the 10 MHz discontinuity....)

Please post your System Information, maybe you are the "lucky" one with old hardware not showing the 10 MHz glitch.

Maybe we discovered that the old hardware has the glitch at 500 KHz........ well, if this is the case ALL (old and new) 815-TG in the world have some kind of problem with the discontinuity using the TG.

Cool 

[pa3bca]

Please post your System Information, maybe you are the "lucky" one with old hardware not showing the 10 MHz glitch.

Maybe we discovered that the old hardware has the glitch at 500 KHz........ well, if this is the case ALL (old and new) 815-TG in the world have some kind of problem with the discontinuity using the TG.

Cool 

I have the 'older' hardware, already posted my system information, but for the sake of completeness here it is again:

[Orange]

If you increase the sweep time to 300 ms the freq shift is gone and also distorted filter curve

The Auto SWT setting can be set to normal and accy, whatever that means, probably accurate 

I have feeling that the sweep times are a little to high for the given RBW when the TG is on. The processor probably does not have enough time to deal with the SA and TG at the same time.

My analyser is an 'old' model revision 0.4, firmware 1.09

[pa3bca]

If you increase the sweep time to 300 ms the freq shift is gone and also distorted filter curve

The Auto SWT setting can be set to normal and accy, whatever that means, probably accurate 

I have feeling that the sweep times are a little to high for the given RBW when the TG is on. The processor probably does not have enough time to deal with the SA and TG at the same time.

My analyser is an 'old' model revision 0.4, firmware 1.09

YES!
I think you are on to something.
The default sweeptime (for my 400-600 KHz span and 3 KHz RBW) is 22.22 ms. Setting accy to on increases the sweeptime to 66.66 ms and the amp dip decreases.. around 200 ms sweeptime the dip is into the noise.
Yellow: sweeptime default 22 ms, purple sweeptime manually increased to 200 ms. Both frequency shift and amp dip are gone.
So, for now we have a workaround.
Thanks!
now: can anybody with a new board & the 10 MHz discontinuity try the same?

[orbiter]

I have just had a response from a contact at Rigol after pointing them in the direction of this thread and the SA issue. They do indeed know about the Issues we're having and R&D are now investigating.

[Orange]

Hello,

these steps are the tracking generator calibration data.
If you turn off the tracking generator calibration in service menu, these steps disappear.
I have documented the format of the calibration data.
There are more frequencies with these errors.
e.g. 500 kHz, 600 kHz, 700 kHz, 1.1 MHz, 1.3 MHz, 2 MHz, ...
There are 21 calibration groups with my DSA.

Peter

There are definitely anomalies around the calibration points. Some of them cause really bad curves the 500kHz one is a good example. I'm not going to mess with the service menu, as some of us have had bad experience with this.

This issue should be fixable for RIGOL. It is not a discontinuity in the HW I think

[cncjerry]

I'm trying to figure out what is worse, the RBW shift or that TG glitch.  I think the RBW is a problem in their FFT algorithm dropping the results in the wrong bins or something.  That would drive me crazy tuning a filter and I recently did a lot of people or around 10Mhz.  I was planning to purchase this setup as a backup to my Advantest with TG but now will keep looking for an older HP.

[pa3bca]

I'm trying to figure out what is worse, the RBW shift or that TG glitch.  I think the RBW is a problem in their FFT algorithm dropping the results in the wrong bins or something.  That would drive me crazy tuning a filter and I recently did a lot of people or around 10Mhz.  I was planning to purchase this setup as a backup to my Advantest with TG but now will keep looking for an older HP.

The freq shift and the amplitude glitch have the same root cause. I think Orange https://www.eevblog.com/forum/testgear/rigol-dsa815-tg-10-mhz-issue/msg749828/#msg749828 nailed it. It should be fairly easy for Rigol to fix.
Until then the work-around would be to check if the sweep time is 200-300 ms or more. Adjust manually if required.
The default 22.22 ms sweep time resulting from the settings I used (200 KHz span, default 3 KHz RBW) does not result in a significant faster update rate than a 200 ms sweeptime anyway so not much is lost. No Idea why they would default to this 22.22 ms sweeptime as the screen update rate does not seem to be more than 5 times/second anyway 

Sanity check:
1: Yellow trace: Sweeping my 475 KHz filter+amp with the TG and sweeptime set manually to 200 ms
2: Purple trace: TG off, and using my DG1032Z on the input, generating a 1 sec sweep from 400 to 600 KHz @ -20dBm and the DSA815 on max hold. wait a few seconds.
Perfect match. So reliable enough.

[Giorgio]

Until then the work-around would be to check if the sweep time is 200-300 ms or more. Adjust manually if required.

I think this is valid for the old hardware with the 500 KHz glitch not for the new one. I already tried to slow down the sweep time but the 10 MHz glitch is always there and if it would be so easy for Rigol to Fix it should be done 1 year ago when I let them know about the problem.

73

[Giorgio]

Hi,

Any news from people who opened a ticket with the Rigol support, if any did?

[N8AUM]

I submitted a "ticket" one over a month ago, received a reply several weeks ago to supply them with my hardware and firmware revs WHICH I DID THE 1st time !
Anybody having better luck ?

[orbiter]

No reply here from Rigol either, other than the confirmation and agreement of the issues, although I'm still hopeful they'll sort this out.

Sent from my HTC One using Tapatalk

[mcinque]

received a reply several weeks ago to supply them with my hardware and firmware revs WHICH I DID THE 1st time !

They don't know how to reply and they are taking time...

[Pinkus]

the latest firmware of Rigol at http://int.rigol.com/File/ModelSoftWare/20150929/DSA815%20Software%20solidification.rar
is showing as the latest version: 1.14. Unfortunately the change log is not included to the download too, so I am attaching it here.
However, the 10Mhz issue still exists.

[Gertjan]

firmware version 1.14 was released on the 19th of June, 2015.........
I installed it months ago...

Regards, Gertjan.

[cncjerry]

wrt the glitch:  Do you have a switchable frequency counter on the SA?  The reason I ask is that I've been playing around with my HP 8568b and I noticed a similar glitch when the frequency counter is turned on and it makes sense on this older unit.  For instance, as the trace reaches the peak (or wherever you have the marker) the counter is triggered and counts the frequency.  During that period of time (at least on the 8568b), the counter is running and the trace is still being drawn.  So the trace is normal, then the processor gets busy at the peak and when the trace starts writing again the time base is further along the X axis so you get an 'L' right at the top.  The size of the 'L' is related to RBW, VBW, etc.  Since the processor on the 8568 is relatively slow, I would expect this from a box that old.  This all makes sense to me and I started wondering if it also had something to do with the Rigol's glitch.

So other than turning off the TG, can you turn off other functions?  Is the glitch the same?

Also, on the frequency shift problem, both my analyzers experience this as you change the RBW unless I have signal tracking turned on.  It's like the analyzer doesn't center the signal in the RBW bandwidth perfectly.  It is a simple task to hit marker->peak and then marker->cf.

Jerry