SDS1104X-E Eye Pattern

[tomas1808]

I bought a Sigilent SDS1104X-E with the intention of using it for measuring the ethernet signal integrity of a device.

Does this scope have the function to do "Eye Pattern" test? The scope is hacked to 200MHz and have passive probes to match it.

I am new to digital scopes (been using a 7704A mainframe for a long time) so I still got to learn most of this scope's features.

Hope I didnt buy this scope in vain 

Thank you!

[tautech]

Very quick check with 10 MHz square wave. Scope settings visible in menus and OSD.

[tomas1808]

Nice! Never seen alternate triggering before! Ty

[bdunham7]

The SDS1104X-E really doesn't have an eye pattern function and even 200MHz BW is probably not enough for your application.  That is usually an extra-cost option on higher-end scopes and not something that can be completely implemented at this level.  You can get an eye-like display on the scope with alternate triggering, but I think you miss a lot of information that way since you only see the traces that are immediately after a transition.  Also, what you select as the trigger point may not be ideal for lining up rising or falling edges. The window trigger function where you set it to trigger anytime the signal goes outside the middle core is probably better.  And of course there is no clock recovery like a scope with a true eye pattern test would have.  You can trigger off of a separate clock if you have one, but I don't think that helps in your case.  You also need a differential probe to do it properly with ethernet AFAIK.

You can get eye signals even with old analog scopes, but it gets a bit harder at higher BW.  Diagnosing bad laser pickups on CD players was a classic example of this.

[Performa01]

The SDS1104X-E really doesn't have an eye pattern function and even 200MHz BW is probably not enough for your application.  That is usually an extra-cost option on higher-end scopes and not something that can be completely implemented at this level.  ...  And of course there is no clock recovery like a scope with a true eye pattern test would have.  ...  You also need a differential probe to do it properly with ethernet AFAIK.

Yes, having an Eye Diagram in a low bandwidth, low cost, entry level oscilloscope is not going to happen. Signal integrity checks on 100 Base T Ethernet require a minimum of 500 MHz bandwidth. For 1000 Base T, more than 1 GHz bandwidth is highly recommended.

Siglent decided to not offer this Application even on the SDS5000X, and this is already a 1 GHz midrange DSO.

Attached is a screenshot of the true Eye-Diagram application on a 2 GHz Siglent SDS6000.

[Fungus]

I bought a Sigilent SDS1104X-E with the intention of using it for measuring the ethernet signal integrity of a device.

Does this scope have the function to do "Eye Pattern" test? The scope is hacked to 200MHz and have passive probes to match it.

What speed Ethernet? You'll need a lot more bandwidth than that to look at anything above 10Mbits.

[tomas1808]

What speed Ethernet? You'll need a lot more bandwidth than that to look at anything above 10Mbits.

I can actually configure the DUT to work at 10BASE-T. Can the scope do eye pattern with for those signals?

Ty!

[tomas1808]

The SDS1104X-E really doesn't have an eye pattern function and even 200MHz BW is probably not enough for your application.  That is usually an extra-cost option on higher-end scopes and not something that can be completely implemented at this level.  You can get an eye-like display on the scope with alternate triggering, but I think you miss a lot of information that way since you only see the traces that are immediately after a transition.  Also, what you select as the trigger point may not be ideal for lining up rising or falling edges. The window trigger function where you set it to trigger anytime the signal goes outside the middle core is probably better.  And of course there is no clock recovery like a scope with a true eye pattern test would have.  You can trigger off of a separate clock if you have one, but I don't think that helps in your case.  You also need a differential probe to do it properly with ethernet AFAIK.

You can get eye signals even with old analog scopes, but it gets a bit harder at higher BW.  Diagnosing bad laser pickups on CD players was a classic example of this.

Do you think I can could get anything meaninful out of this scope (and non differential probes) if the signal was 10BASE-T?

[bdunham7]

Do you think I can could get anything meaninful out of this scope (and non differential probes) if the signal was 10BASE-T?

The threshold question IMO is whether you can connect up and get a triggered signal on the screen without interfering with the signal to the extent that the data is corrupted. 

The options are to use two 10X probes and MATH (CH1 - CH2) or to simply use one probe and ground reference one side of the twisted pair.  You'd need to make a breakout of some kind to do that as neatly as possible.  I think with 10BaseT you can probably get a usable signal either way, but I've never tried this.  If ground referencing one side of the ethernet doesn't kill it, that will probably be your best signal.  Whether all of that gives you a meaningful result or not is another question.  I'm guessing that even a basic 10X differential probe would be much better.

[tomas1808]

The threshold question IMO is whether you can connect up and get a triggered signal on the screen without interfering with the signal to the extent that the data is corrupted. 

The options are to use two 10X probes and MATH (CH1 - CH2) or to simply use one probe and ground reference one side of the twisted pair.  You'd need to make a breakout of some kind to do that as neatly as possible.  I think with 10BaseT you can probably get a usable signal either way, but I've never tried this.  If ground referencing one side of the ethernet doesn't kill it, that will probably be your best signal.  Whether all of that gives you a meaningful result or not is another question.  I'm guessing that even a basic 10X differential probe would be much better.

I will try it out using the MATH substract with the two 10X probes and see how it goes. Thank you!

[Anthocyanina]

I bought a Sigilent SDS1104X-E with the intention of using it for measuring the ethernet signal integrity of a device.

Does this scope have the function to do "Eye Pattern" test? The scope is hacked to 200MHz and have passive probes to match it.

What speed Ethernet? You'll need a lot more bandwidth than that to look at anything above 10Mbits.

triggering on clock i get this at 50Mbit on a 200 Mhz bandwidth oscilloscope, it still looks very usable

[2N3055]

I see lot of confusion here..
Like Performa01 said you need 1GHz+ (preferably 2GHz) scope with proper high speed active differential probe to actually extract the signal from the Ethernet, including a certified Ethernet T connector jig to not destroy signal integrity of the Ethernet line..

Then people start saying you could do it at 100 or 10 Mbits..
What is the point of doing eye diagram at 10 Mbits if you're interested at 1Gbit eye?

The whole idea of eye diagram is to be run at speed of interest.  Running it 100X slower will not give you any information... Except how it works when it is slow. And 10 Mbit is so slow it will run over water pipe.... No need for eye diagram..