Has anyone determined if the MUXOUT pin is being used?
Looking at the board photos recently posted in this thread this pin is not connected to anything.
where the "spreading" is coming from and it might be DSS.
There is no DDS. The 4360 PLL is referenced from a 25MHz oscillator
There is no DDS. The 4360 PLL is referenced from a 25MHz oscillator
DSS - Digital Spread Spectrum
So we don't have a lock pin, but we can look at the Charge Pump pin. Below is what it's doing. I would expect it to be more symmetric if it was locked.
If you have one there (and that wasn't an old picture), what does your CP pin look like?
I found my old proto board, this is what I have on the CP pin. The picture is scaled same as yours, 5uS/ and 10mV/.
There is no 100kHz or any other low frequency ripple whatsoever. The total noise includes noise from my on-board switching regulator and cross-coupled leak from the RF output.
EDIT: I re-posted your screenshot made on Agilent just for reader's convenience.
I found my old proto board, this is what I have on the CP pin. The picture is scaled same as yours, 5uS/ and 10mV/.
There is no 100kHz or any other low frequency ripple whatsoever. The total noise includes noise from my on-board switching regulator and cross-coupled leak from the RF output.
EDIT: I re-posted your screenshot made on Agilent just for reader's convenience.
Ah, that looks more like I would expect. Thanks for digging out your old board. Are you still running it at 193MHz?
As I recall there was some RFout coupled into my CP pin also. I did that capture in high-res mode, so it would have gotten filtered out. What I'm not recalling was the magnitude of the RF.
For interest's sake, I'll grab a CP pin capture in normal mode.
By dumping the 12M point memory and using a decent FFT this scope turns into a hobbyist-grade spectrum analyzer. I like.
I'm guessing the PLL's harmonics are those huge spikes... among all the internal CPU clocks and what not.
For interest's sake, I'll grab a CP pin capture in normal mode.
No real surprise. There is some leakage, but it's not a crazy amount. It looks to be higher than yours but that's probably because the higher frequency is coupling more.
Also below is an FFT of the same scope capture in the 1GHz region detailing the leakage. Again, no surprises that I can see, except of course this shouldn't be happening.
By dumping the 12M point memory and using a decent FFT this scope turns into a hobbyist-grade spectrum analyzer. I like.
I'm guessing the PLL's harmonics are those huge spikes... among all the internal CPU clocks and what not.
If FFT is your goal, this PLL problem is really going to make a mess out of it.
Feed the scope a pure tone, zoom in on some of those plots, and you should see your carrier modulated at 100kHz (similar to the one I just posted). And the FM deviation should vary from scope to scope because we see differences in the 5us jitter width.
Edit: And actually, the sample clock isn't even being modulated by a sine wave. It's a sawtooth with all of its lovely harmonics. Should create some interesting artifacts. Yummy.
MarkL,
Next time can you try a simple test - try to heat the 25MHz oscillator with hot air pencil or soldering gun , and cool with dust blower or a piece of metal cured in the freezer for a few min and watch the CP pin if it shows any sign of the PLL trying to lock, or if we are lucky it may even lock. That would indicate either the oscillator may be off of the spec frequency or the loop filter bandwidth may be too narrow.
A better test would be to unsolder the DC blocking cap between the 25MHz oscillator and PLL and connect an external signal gen there (DC blocked) and vary the frequency slightly and see if the PLL locks. That is only if you willing to touch the board with solder.
For interest's sake, I'll grab a CP pin capture in normal mode.
No real surprise. There is some leakage, but it's not a crazy amount. It looks to be higher than yours but that's probably because the higher frequency is coupling more.
Also below is an FFT of the same scope capture in the 1GHz region detailing the leakage. Again, no surprises that I can see, except of course this shouldn't be happening.
Something bugs me about this: for a 1GHz center, there is no actual "carrier" but rather a band of noise. If this thing is actually clocking there should be a significant tone, at least 125mVrms in Amplitude (or about 40dB larger than the -60dBV you're measuring right now).
You're probing the CP pin, so I'm thinking looking at the 1GHz center isn't really a good metric, but rather the DC-200kHz range. Garbage coming out of the VCO will be proportional to the noise at VTune/CP by way of the Kv of the VCO. In simplifier terms, if you had 1uVrms of noise at 1kHz on a 12MHz/V VCO, you would expect your skirt to distort at around 12Hz from carrier. 10uVrms and it goes to 120Hz, etc. Have a band of noise, and then the skirt gets ugly, quickly.
UPDATE FROM RIGOLShort and not explanatory, but:
1: We have reproduced the two issues in R&D side;
2: All issues can be fixed by firmware updating without deleting any feature;
3: The trail firmware form R&D will be released in early next week, that can be used for Dave or some urgent cases.
MarkL,
Next time can you try a simple test - try to heat the 25MHz oscillator with hot air pencil or soldering gun , and cool with dust blower or a piece of metal cured in the freezer for a few min and watch the CP pin if it shows any sign of the PLL trying to lock, or if we are lucky it may even lock. That would indicate either the oscillator may be off of the spec frequency or the loop filter bandwidth may be too narrow.
Great suggestion - I tried it.
There is sensitivity to temperature. The hotter it gets with an air pencil, the less deviation in the modulation. The colder with freeze spray, the wider the deviation. I was not able to get it hot enough to lock, but I didn't heat it beyond about 100C.
A better test would be to unsolder the DC blocking cap between the 25MHz oscillator and PLL and connect an external signal gen there (DC blocked) and vary the frequency slightly and see if the PLL locks. That is only if you willing to touch the board with solder.
If we were beyond the 30-day return window, I would go at it in an instant.
If this problem has no resolution within the return window, it's going back. I've learned the lesson that what you see in front of you is what you're buying, especially with this lower end stuff. I wouldn't blame Rigol in the least for not taking it back once the modding started.
That being said, I'm going to see if I can find a spec sheet for the oscillator on the off-chance that one of those pads is an output enable.
For interest's sake, I'll grab a CP pin capture in normal mode.
No real surprise. There is some leakage, but it's not a crazy amount. It looks to be higher than yours but that's probably because the higher frequency is coupling more.
Also below is an FFT of the same scope capture in the 1GHz region detailing the leakage. Again, no surprises that I can see, except of course this shouldn't be happening.
Something bugs me about this: for a 1GHz center, there is no actual "carrier" but rather a band of noise. If this thing is actually clocking there should be a significant tone, at least 125mVrms in Amplitude (or about 40dB larger than the -60dBV you're measuring right now).
You're probing the CP pin, so I'm thinking looking at the 1GHz center isn't really a good metric, but rather the DC-200kHz range. Garbage coming out of the VCO will be proportional to the noise at VTune/CP by way of the Kv of the VCO. In simplifier terms, if you had 1uVrms of noise at 1kHz on a 12MHz/V VCO, you would expect your skirt to distort at around 12Hz from carrier. 10uVrms and it goes to 120Hz, etc. Have a band of noise, and then the skirt gets ugly, quickly.
You're right, 1GHz is not the frequency of interest in the loop filter. The question was if there was any significant leakage from the RF out into the loop filter. I was just making sure since I had the scope in high-res mode when making the previous measurement.
It was just a verification. I thought it was interesting to include the FFT shot since measuring the leakage this way is more meaningful than eyeballing the fuzz in the scope screen shot.
But since there is a small amount of leakage, the 1GHz FFT of the leakage shows the same 100kHz modulation that was seen on the SA back on page 19. The 100kHz signal is also in the FFT, but I didn't post it. We can already see the 100kHz signal directly in the time domain.
Edit: Fix typo.
UPDATE FROM RIGOL
Short and not explanatory, but:
1: We have reproduced the two issues in R&D side;
2: All issues can be fixed by firmware updating without deleting any feature;
3: The trail firmware form R&D will be released in early next week, that can be used for Dave or some urgent cases.
It would be greatly appreciated if they or you would post both the trial firmware and the previous full release version as open downloads instead of having to play the email game to get them.
Short and not explanatory, but:
1: We have reproduced the two issues in R&D side;
2: All issues can be fixed by firmware updating without deleting any feature;
3: The trail firmware form R&D will be released in early next week, that can be used for Dave or some urgent cases.
Ok, great. Thanks for forwarding the info.
As a Rigol customer, I'm interested to see this echoed through my open case.
It's been a great discussion and investigation, but I hope every problem does not need the visibly of a 23 page, global pile-on to be addressed.
UPDATE FROM RIGOL
Without this Blog they would not bother
2: All issues can be fixed by firmware updating *without deleting any feature*
OK that is what we are hoping for
I get the impression shipping to dealers will continue as usual.
UPDATE FROM RIGOL
Without this Blog they would not bother
Well, if they want to play in the higher end market with their other gear, which is clearly their aim, they need to learn the customer support game. I've recently had a couple of open cases with Agilent/Keysight and I don't wonder at all where my extra money goes buying their gear.
A major part of evaluating a vendor is taking their technical support for a spin. That didn't really happen here. Maybe another bug will come up... Nah!
Maybe another bug will come up... Nah!
And they still have the third bug to address, I am dying to learn which one out of the 3 reported it is.
Maybe another bug will come up... Nah!
And they still have the third bug to address, I am dying to learn which one out of the 3 reported it is.
You have a point, but I'm assuming they're addressing the 5us jitter and the AC trigger coupling.
I thought the dual edge trigger "bug" was pilot error. The same edge happened to be synchronized to the sweep timing and could be corrected by increasing the holdoff.
Maybe another bug will come up... Nah!
And they still have the third bug to address, I am dying to learn which one out of the 3 reported it is.
1. Trigger jitter when using 5us horizontal offset
2. Trigger jitter when using AC trigger coupling.
Not sure what the third bug is? The "dual slope triggering not working bug" is not a bug at all, just correct behaviour of trigger holdoff:
link to earlier in this thread.
EDIT: MarkL, must have pressed "reply" at the same time! Great minds, right?
EDIT: MarkL, must have pressed "reply" at the same time! Great minds, right?
Doesn't hurt to repeat, especially in this thread!
I'm assuming they're addressing the 5us jitter and the AC trigger coupling.
I wonder if DS20xx AC trigger coupling will also be addressed, mine does not have the 5uS thing but does have that one.
2. Trigger jitter when using AC trigger coupling.
Not sure what the third bug is?
There is actualy 4th one.
I've showed it in my video: on DS2000A with EXT source it works only with DC coupled trigger. Menu is the same (AC/DC/rejects), but it doesn't affect anything, EXT trigger always work with DC coupling.
Despite the fact that this problem seems to be solved soon:
Yesterda my DS1074Z arrived, Software 00.04.00
I can't reproduce the 5us Jitter issue.
The AC trigger coupling is not relevant for me. Now I'll watch Dave's latest video to understand what it's about
Edit: Seen the video, now I'll check it out...
Well Rigol some time ago produced a kool paper signed by a sales rep that they have tested a few scopes from their stock and could not reproduce the problem. I guess you got one of those 'no-problem' units