JohnSwenson Posted March 19, 2021 Share Posted March 19, 2021 3 hours ago, The Computer Audiophile said: There must be something in the SFP+ telling the switch that it handles 10Gb, else the switch would just try to send 10Gb in a 1Gb SFP. See the post Alex shared. It talks about the pin on the connector that is grounded on the SFP, an SFP+ port can apply a pull up on that port, when SFP is inserted it will be pulled down, an SFP+ will not pull it down so the switch can if it wants to, switch the speed. A switch with SFP+ port is free to either run at just 10Gb, have an external mechanism to change the speed (physical switch, web page configuration etc) or use the above property of SFP modules. John S. Link to comment
Popular Post JohnSwenson Posted November 6, 2021 Popular Post Share Posted November 6, 2021 30 minutes ago, R1200CL said: I don’t think so. There is said a chain of dual good switches (meaning EtherRegen and/or the opticalModule) have a good effect in some systems. Adding an external clock even better. (At least in a A >B configuration). This has to be done in a way that power supplies and ethernet cables doesn’t create loops and bypass the moat created by EtherRegen. Remember the EtherRegen was developed towards RJ45 interfaces from it's B side. But it’s been said using it in a B to A may be equally good. Well the “moat” is the same to my understanding, but there is for sure a minor reduction in phase noise since the clock signal also have to cross the isolation (the moat). It would be nice if Uptone could enlighten us with numbers if ever measured. Assuming you now have an opticalRendu, and lack a FMC, the Mikrotik’s may be a good option to the opticalModule, as this is a 10GB switch and we expect due to the requirements of 10GB, (eye pattern), it ought to perform very well. (I have one, actually 2, myself). Howerver it’s a mystery to me how “those bad clocks” in 10GB can deliver better (or equal) quality than what the clocks used in Johns design. Some has asked for an eye pattern test of the EtherRegen. The white paper has promised some measurements, but so far we haven’t seen any. Since you also mentioned the SE version as an option, there is a small project ongoing testing the opticalRendu with external clock done by sine wave and the circuitry EtherRegen lacks to turn sine to square 😀 Just a few words on testing these things. Eye patterns can only show gross levels jitter, no where near what we are talking about in high end audio. They DO show the slope at intercept and AM noise on the signal, which is important because that affects how much jitter is added to the signal in the receiver. So you really need to use a real jitter analyzer that goes down to very small levels of jitter. Fortunately I recently acquired one that can go down to 1ps resolution (way better than you can do with an eye pattern), BUT using it with an Ethernet signal is very difficult. The problem is that what goes out over the wires in RJ-45 Ethernet is NOT binary! 100Mb has three voltage levels, 1000Mb has 5 voltage levels and 10G has 16 voltage levels! All jitter analyzers are designed to work with binary signals, high and low, with the RJ-45 signals voltage levels are all over the place. Have you ever seen an eye pattern on a 1Gb signal, it is VERY weird looking! I haven't yet figured out how to do a proper setup to measure these signals. Since the B output of an ER is 100Mb (only three voltages) I MAY be able to get decent results from this, but I have not tried it yet. On the other hand using the jitter analyzer on a Gb optical signal (well actually the electrical signal going into or out of the module) is a piece of cake. I can easily do this. So what would you like measured? It has to be an electrical signal that is binary (such as going in or out of an SFP module). I can't measure 10Gb SFP+ signals, they are too fast for this analyzer (if someones wants to send me $100K I can get one that DOES). I MAY be able to make measurements of the B side of the ER. I also have an extremely good phase noise analyzer that can measure phase noise of clocks up to 200MHz. So all of you, come up with a set of one or two tests of what you want measured. Be warned it can take time. I will almost always have to make test fixtures to to get the test signals into the equipment, which is not easy to do since almost anything you do degrades the signals. This means it can take quite awhile. If this entails using devices (switches, DACs whatever) that I do not have, please ship said devices to me. John S. MarkoL, Superdad and R1200CL 1 1 1 Link to comment
Popular Post JohnSwenson Posted October 22, 2022 Popular Post Share Posted October 22, 2022 On 10/14/2022 at 12:06 PM, R1200CL said: @agillis or @vortecjr or @JohnSwenson Please tell us about what’s new with the V2 opticalModule released today. There are no new features, it does exactly the same thing as the previous oM Deluxe. The reason for this version is that all the main chips in the previous version have become completely unavailable. This is a re-design to use chips that ARE available. The search took a very long time, almost every chip in this space had zero stock in the world. What I finally found was a single chip that does everything in one chip and was designed by a group of engineers that do a VERY good job, I have worked on chips with them before. The result is a chip that will have significantly lower power and ground noise than the previous version. I have no idea exactly what this does to the sound, I have not had time to spend quality time with it to find out. John S. R1200CL, Markus8, kennyb123 and 2 others 1 3 1 Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now