EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: entropi on March 07, 2021, 12:28:37 am
-
Either new or vintage, what are the most affordable scopes capable of a color graded eye pattern generation? This is a very cool feature and I’m just curious how attainable it is these days.
-
Its a shame this feature has been kept as a product differentiator, its useful beyond XXGS/s eye diagrams. I think Siglent are the cheapest new scopes that have this.
-
Anything that can do intensity grading and long persistence can do color grading with a modest software change in the graphics code, but yes, I think Siglent scopes are cheap end ones that have bothered to implement it. LeCroy scopes have had it since they went to color screens, so you may be able to find an LC series scope that was cheaper.
It's certainly an interesting feature, and I find it easier to glean data from than a regular intensity graded persisting screen, but it's really barely even a feature, basic implementations would only be a few lines of code, though it could be more with controls of it or with certain methods of drawing the display.
-
I think it's also important to have embedded clock recovery features to be able to leverage it well.
For general reference, this is what a color graded eye diagram looks like:
-
What I'm talking about with clock recovery...
Here is a Keysight MSO-X 6004A Showing the embedded clock recovery settings screen:
[attach=1]
And a Rohde & Schwarz showing similar:
[attach=2]
-
Also here's Dave back in 2012 when he had his hands on a $100k+ Agilent (Keysight) DSA91000 and was showing off its excellent eye diagram on a 5Gbps USB3.0 signal:
[attach=1]
https://www.youtube.com/watch?v=o8DPlqWVmzk (https://www.youtube.com/watch?v=o8DPlqWVmzk)
A good primer on eye diagram basics:
https://www.youtube.com/watch?v=tZiKRfH2yZ4 (https://www.youtube.com/watch?v=tZiKRfH2yZ4)
And more info:
https://www.youtube.com/watch?v=Nu6aXTMgksk (https://www.youtube.com/watch?v=Nu6aXTMgksk)
-
Either new or vintage, what are the most affordable scopes capable of a color graded eye pattern generation? This is a very cool feature and I’m just curious how attainable it is these days.
What kind of signals do you want to look at? If you want to look at high frequency stuff then the trigger-jitter and implementation of the trigger become important (some DSOs concentrate the trigger point digitally into a single point which causes the edges to smear out more than they actually are). The price of vintage depends on state & age. With some homework to avoid wrecks you can buy an older fixer-upper 1GHz+ oscilloscope for a couple of hundred US$. Usually there are a whole bunch of Lecroys with TFT screens for sale on Ebay.
-
I think it's also important to have embedded clock recovery features to be able to leverage it well.
Then you're stuck with the "high end" platforms that were targeting high speed serial links of the day. There are ways to do external clock recovery (or use the recovered clock of the DUT) so its not essential to have it built in, but by the time you are looking at something where that matters its a GHz scope and $$$
-
Either new or vintage, what are the most affordable scopes capable of a color graded eye pattern generation? This is a very cool feature and I’m just curious how attainable it is these days.
What kind of signals do you want to look at? If you want to look at high frequency stuff then the trigger-jitter and implementation of the trigger become important (some DSOs concentrate the trigger point digitally into a single point which causes the edges to smear out more than they actually are). The price of vintage depends on state & age. With some homework to avoid wrecks you can buy an older fixer-upper 1GHz+ oscilloscope for a couple of hundred US$. Usually there are a whole bunch of Lecroys with TFT screens for sale on Ebay.
Most of the jitter/eye analysis on deep memory scopes is not relying on trigger information at all. It recovers clock info from signal itself and does it's own edge and period extraction from the capture itself, and then overlaps such "chopped up" pieces to create eye diagram... Trigger simply serves for reasonable signal acquisition. Trigger jitter is important only if you create eye manually by using physical signal+clock method.
-
Or to answer shortly, your best bet is and old used Lecroy, Keysight or Tektronix with Eye/jitter analysis.
Or you could do what JoeQSmith did, sample with any scope that has Labview drivers, and do Eye/jitter analysis in LabView.
-
My Siglent SDS1104X-E does a acceptable job of color grading, but is a bit slow for eye diagrams. It was ~$500 and I hacked to to 200 MHz.
The best one for color grading I own is my 1 GHz Tektronix TDS784D. I paid $780 including shipping with a LCD screen. The color graded display on this one is beautiful.
-
Not sure if this is done right. :-//
SDS1104X-E
CAN H and L from a STB3
Trigger set to one or the the other and H Pos moved to center the packets on the display. Min Holdoff.
Screenshots should say the rest...........
-
Seems like with a CAN signal there might be no need for a clock signal to trigger on.
The $399 SDS1104X-U
-
Either new or vintage, what are the most affordable scopes capable of a color graded eye pattern generation? This is a very cool feature and I’m just curious how attainable it is these days.
What kind of signals do you want to look at? If you want to look at high frequency stuff then the trigger-jitter and implementation of the trigger become important (some DSOs concentrate the trigger point digitally into a single point which causes the edges to smear out more than they actually are). The price of vintage depends on state & age. With some homework to avoid wrecks you can buy an older fixer-upper 1GHz+ oscilloscope for a couple of hundred US$. Usually there are a whole bunch of Lecroys with TFT screens for sale on Ebay.
Most of the jitter/eye analysis on deep memory scopes is not relying on trigger information at all. It recovers clock info from signal itself and does it's own edge and period extraction from the capture itself, and then overlaps such "chopped up" pieces to create eye diagram... Trigger simply serves for reasonable signal acquisition. Trigger jitter is important only if you create eye manually by using physical signal+clock method.
With eye pattern analysis you go one step further and reduce the number of availble oscilloscopes to choose from. It is also a rather specialised purpose which gets outdated quickly with ever increasing signalling speeds. If you want to see how stable an oscillator is, compare 1PPS pulses or how much jitter comes from a DDS then using the trigger is a good enough method (even better with some added statistics) for showing the time difference between the edges.
-
Either new or vintage, what are the most affordable scopes capable of a color graded eye pattern generation? This is a very cool feature and I’m just curious how attainable it is these days.
What kind of signals do you want to look at? If you want to look at high frequency stuff then the trigger-jitter and implementation of the trigger become important (some DSOs concentrate the trigger point digitally into a single point which causes the edges to smear out more than they actually are). The price of vintage depends on state & age. With some homework to avoid wrecks you can buy an older fixer-upper 1GHz+ oscilloscope for a couple of hundred US$. Usually there are a whole bunch of Lecroys with TFT screens for sale on Ebay.
Most of the jitter/eye analysis on deep memory scopes is not relying on trigger information at all. It recovers clock info from signal itself and does it's own edge and period extraction from the capture itself, and then overlaps such "chopped up" pieces to create eye diagram... Trigger simply serves for reasonable signal acquisition. Trigger jitter is important only if you create eye manually by using physical signal+clock method.
With eye pattern analysis you go one step further and reduce the number of availble oscilloscopes to choose from. It is also a rather specialised purpose which gets outdated quickly with ever increasing signalling speeds. If you want to see how stable an oscillator is, compare 1PPS pulses or how much jitter comes from a DDS then using the trigger is a good enough method (even better with some added statistics) for showing the time difference between the edges.
Well this is literally what OP asked about.. Creation of color graded eye diagrams... Not phase noise analysis.
Unless you have a source clock at twice the frequency of your watched signal, a clock recovery needs to done, and for that he can either create/acquire a hardware based clock recovery device to trigger scope or a scope (device) that takes your signal and does clock recovery internally. In scopes clock recovery is exclusively done by software extraction (software PLL), and those are available inside various Jitter/eye diagram packages, usually combined together..
OP didn't really specify anything, or hinted what would that be for, except stating that eye diagrams look cool..
So unless he does, we cannot give any real advice here, methink....
Regards,
-
My Micsig can do it for about $600:
(https://www.eevblog.com/forum/testgear/most-affordable-oscilloscope-capable-of-color-graded-eye-diagram/?action=dlattach;attach=1188954;image)
-
That is not a proper eye diagram.
https://www.eevblog.com/forum/testgear/most-affordable-oscilloscope-capable-of-color-graded-eye-diagram/msg3498324/#msg3498324 (https://www.eevblog.com/forum/testgear/most-affordable-oscilloscope-capable-of-color-graded-eye-diagram/msg3498324/#msg3498324)
Eye diagram has capture of each pulse that contributes to diagram, centered around center of pulse period.
So you can see both left and the right side jitter.
Without software to perform analysis and extracting data from data-stream you cannot do it right.
What you show will align all left sides to perfect timing with trigger, so you cannot see if that edge came late or early, you see all variations as basicaly PWM, eg. as a variation of eye width, and only right side variable. While that is not useless (if it looks perfect than you're probably good as far jitter and amplitude variation) it is not showing real eye opening as it should.
-
Yep eye pattern with clock recovery is a whole different can of worms compared to just rainbow intensity gradation.
Triggering is indeed not used for clock recovery. Instead the scope tends to capture a long capture and then perform clock recovery on the whole thing in software. This actually requires a fair bit of processing power to do and some of the algorithms that do this can be pretty complicated. Then once clock is recovered the long waveform is chopped up into sections according to the clock and overlayed over each other to produce the image.
But what would you use an eye diagram for? Typically it is to verify the signal integrity of high speed links, be it USB3.0 over a cable or signals between a CPU and RAM. These are things that run at very high speeds and as a result not only need a very high end scope but also the appropriate low capacitance active probes that cost as much as a couple of those Siglent scopes put together. So your typical 100MHz entry level scope only really has the bandwidth to produce a usable eye diagram for data rates up to about 20Mbit at best. At these low speeds you rarely even need a eye diagram. Its easy enough to produce cables and board interconnects that support these speeds with so much headroom that the signal looks like a nice pretty square wave. So as a result it is easy to tell the signal is good just by looking at it. While at 5Gbit cables become very imperfect so these high speed serial protocols ride on the edge of still working so signals always look pretty rounded and wavey, this is where an eye diagram helps to quantitatively measure how good the signal integrity is.
Even tho i have a scope capable of generating real eye diagrams i very rarely make use of it. When im dealing with more lowish speed signals in the 10s of MHz i will just go for the "poor mans eye diagram". This is where you simply put your trigger in the middle and set the trigger on both edges. Modern scopes are very fast to capture a lot of edges, while you can crank up the waveform intensity to be able to see any lone single runt pulses going trough the middle of the eye. With nice clean sharp fast rise time signals the "clock recovery" by using a edge trigger is close enough.
-
That is not a proper eye diagram.
What you show will align all left sides to perfect timing with trigger, so you cannot see if that edge came late or early, you see all variations as basicaly PWM, eg. as a variation of eye width, and only right side variable. While that is not useless (if it looks perfect than you're probably good as far jitter and amplitude variation) it is not showing real eye opening as it should.
An easy solution to that is to move the trigger point off screen as far as possible. But this requires a scope with a stable enough clock source so the jitter from oscilloscope itself is significantly smaller compared to that of the signal. In the end any clock recovery algorithm will choose a point somewhere in the signal as t=0 and goes from there (and the clock source from the oscilliscope will still be a factor).
-
If you are probing a CAN bus you can use the PHY TX line to trigger off. According to this post:
If you have access to one of the devices, trigger on the signal going into CANbus phy (TX line)
Leo
I would also like to find an affordable way to produce eye diagrams to test physical networks. I was hoping there was a device similar to NanoVNA (@joeqsmith) that could just be designed for doing eye diagrams. This would be great to probe CAN bus and RS485 if you have a wiring setup that is acting up. Does anyone know if that would be even possible with a VNA? Also is there a PC software that could do it for a cheap USB oscilloscope?
-
A VNA is the wrong tool. And for RS485 / CAN you don't really need color grading. Just infinite persistence will do. But even then you'll be much better off with a DSO which can trigger on a bus fault occuring.
-
If you are probing a CAN bus you can use the PHY TX line to trigger off. According to this post: If you have access to one of the devices, trigger on the signal going into CANbus phy (TX line)
Leo
I would also like to find an affordable way to produce eye diagrams to test physical networks. I was hoping there was a device similar to NanoVNA (@joeqsmith) that could just be designed for doing eye diagrams. This would be great to probe CAN bus and RS485 if you have a wiring setup that is acting up. Does anyone know if that would be even possible with a VNA? Also is there a PC software that could do it for a cheap USB oscilloscope?
Sorry it took so long to get around to it. Post showing a comparison with my old LeCroy DSO and the LiteVNA. Previous post has a link to a recent video showing how it works.
I have not yet released it to the public. Because of the limited frequency range of the V2Plus4 (along with other limitations) I have only been testing with the LiteVNA. Demonstrations are using the 9.3GHz harmonic mode. If there is something beyond that demo that you would like to see, feel free to ask.
https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg4489321/#msg4489321 (https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg4489321/#msg4489321)