Howdy Y'all!

First I'd like to say, this forum has me super excited and my curiosity cranked up. The sheer amount of information here is astounding. But let's get to the point!

I've worked in Data Centers for the last 8 years of my life, going on 9 years. Every job I've been at has always had a high overturn of optics. I've always enjoyed finding a solution to a problem, and one of my fellow colleagues dumped a project on my lap, quite literally. He handed me a 40G QSFP Transceiver Tester and told me "You look like you know what you're doing, so here's a project for you." The issue is, they no longer have the software for it, and want me to re-engineer it. Ok, I'm smart, but I have no clue where to even begin with this.

If I'm going to take the time to develop an interface for a single 40G optic, why shouldn't I include all of the optic speeds?

I need to develop a board with multiple SFP / QSFP ports on it, interfacing it with software via USB cable to test if Optic Receivers are truly bad or not.

What I'd like is some directions, a good starting point. This goes way beyond my skills of an IT Engineer, but I'm willing to put in the work, learn, and have fun. Electrical & Mechanic Engineering has been my second choice for a career path if I ended up hating IT (Which I do right now). I'm 110% a complete novice, I fully understand this is a massive undertaking, and I totally understand it'll probably cost me a pretty penny. However, if I can pull it off, and if it works, it makes me look amazing to my work and it's something I've wanted to do for a long time. I'm very familiar with Digi-Key.

Edit:
I forgot to mention that, I wasn't sure which section this goes under. HW Design sounded like the right section.

 

I would not know where to begin either, but I found this https://www.gigalight.com/40g-100g-qsfp-checker.html tester.
So, I'd look at testers to see what sort of information they give you.

 

Howdy Y'all!

First I'd like to say, this forum has me super excited and my curiosity cranked up. The sheer amount of information here is astounding. But let's get to the point!

I've worked in Data Centers for the last 8 years of my life, going on 9 years. Every job I've been at has always had a high overturn of optics. I've always enjoyed finding a solution to a problem, and one of my fellow colleagues dumped a project on my lap, quite literally. He handed me a 40G QSFP Transceiver Tester and told me "You look like you know what you're doing, so here's a project for you." The issue is, they no longer have the software for it, and want me to re-engineer it. Ok, I'm smart, but I have no clue where to even begin with this.

If I'm going to take the time to develop an interface for a single 40G optic, why shouldn't I include all of the optic speeds?

I need to develop a board with multiple SFP / QSFP ports on it, interfacing it with software via USB cable to test if Optic Receivers are truly bad or not.

What I'd like is some directions, a good starting point. This goes way beyond my skills of an IT Engineer, but I'm willing to put in the work, learn, and have fun. Electrical & Mechanic Engineering has been my second choice for a career path if I ended up hating IT (Which I do right now). I'm 110% a complete novice, I fully understand this is a massive undertaking, and I totally understand it'll probably cost me a pretty penny. However, if I can pull it off, and if it works, it makes me look amazing to my work and it's something I've wanted to do for a long time. I'm very familiar with Digi-Key.

Edit:
I forgot to mention that, I wasn't sure which section this goes under. HW Design sounded like the right section.

I hate to throw cold water on the enthusiasm, but are you sure you're not being setup for a colossal failure?
I do know that people and their time are far more expensive than a piece of test equipment.

 

I would not know where to begin either, but I found this https://www.gigalight.com/40g-100g-qsfp-checker.html tester.
So, I'd look at testers to see what sort of information they give you.

Thanks for that link, those tests it runs are about what I'm looking for. I just need to replicate it to handle more than just 1 or 2 at a time. We swap anywhere from 5 to 100 optics a day, per site.

I hate to throw cold water on the enthusiasm, but are you sure you're not being setup for a colossal failure?
I do know that people and their time are far more expensive than a piece of test equipment.

I thought about this quite a bit before responding. You're probably right, I'm setting myself up for a colossal failure. I did make it clear to people that this is far outside of my skill set, and that I promise nothing. One of the quotes I remember is "Nothing ventured, nothing gained." I'm hoping to explore and find out if I like Electrical & Mechanic Engineering more than I like IT right now. I've lost my drive with IT and my current job as made IT work toxic by crappy business practices and "cost saving" choices.

 

Thanks for that link, those tests it runs are about what I'm looking for. I just need to replicate it to handle more than just 1 or 2 at a time. We swap anywhere from 5 to 100 optics a day, per site.


I thought about this quite a bit before responding. You're probably right, I'm setting myself up for a colossal failure. I did make it clear to people that this is far outside of my skill set, and that I promise nothing. One of the quotes I remember is "Nothing ventured, nothing gained." I'm hoping to explore and find out if I like Electrical & Mechanic Engineering more than I like IT right now. I've lost my drive with IT and my current job as made IT work toxic by crappy business practices and "cost saving" choices.

I'm not trying to be critical of your choices in this matter, but it seems to me there are less stressful ways of making a career change than the path you have chosen. I wish you luck but I'm not sanguine about your prospects. Please come back to this thread and tell me I'm all wet if it works out for you.

 

And this is exactly why I avoid places like here.

Thank you, KeepItSimpleStupid for your help and suggestions. I'll find other sources than this place.

 

Thanks for that link, those tests it runs are about what I'm looking for. I just need to replicate it to handle more than just 1 or 2 at a time. We swap anywhere from 5 to 100 optics a day, per site.


I thought about this quite a bit before responding. You're probably right, I'm setting myself up for a colossal failure. I did make it clear to people that this is far outside of my skill set, and that I promise nothing. One of the quotes I remember is "Nothing ventured, nothing gained." I'm hoping to explore and find out if I like Electrical & Mechanic Engineering more than I like IT right now. I've lost my drive with IT and my current job as made IT work toxic by crappy business practices and "cost saving" choices.

Lofty goal- unlikely success for many, many reasons- but that doesn't mean you can't try.

1. You don't know enough about ethernet or fiber to handle the signaling/emf issues. Do you know what an 'eye' is?
2. You can't make something like this from scratch without an oscilloscope and or other test-set capable of testing against- a reference tester (which could cost $200K)
3. You probably can't even purchase the necessary semiconductors because they are only sold in monthly quantities to large entities under contract (100's of thousands to millions per month)

So do what you *CAN* do-- Divide and conquer- pay for what you can't do:



This solves you signaling issue and connectivity, and all the design work is done. All you have to do is build a board that this can mount on, that can drive it at the 40G- that is a far more achievable task.

Then, if you can get one working reliably, you can then figure out how to scale it-

Best of luck to you.

 

You have a lot of work to do.
First things:
1) What does a tester test - look at other manufacturers
2) What parts are used in the current tester?
3) I doubt you can re-engineer the USB protocol. it's probably just telling the box what to do.
4) It's a high speed design. That raises all sorts of issues)
5) If you had the software running on some platform, THEN it might be possible to reverse engineer to run on another.
6) Find an SFP receiver, the slot it plugs into and get a datasheet.

I am going to mention a thing that happened to me. I told my boss that the technology we needed wan;t available. He said the money is. One part of this involved controlling a monochometer. We were controlling it with a PDP-11 essentially as a real time program. You told the pdp-11 where the monochometer was and it kept track of the position with interrupts. It could also press the start/stop buttons.

It was a few years before a IEEE-488 interfacable monochometer was on the market.

I based my "conception" of writing a monochometer stepper motor driver for a stepper motor controller. Iwas going to control the stepper directly and not the "buttons" and determining position via interrupts.
This also required me to build a 4-slot filter wheel and shutters.

The stepper controller had the ability to do s-curves, and heal with limit and org sensors. using RS232/RS485. I would have had to come up with an ORG sensor.

This was part of a "large" $100,000.00 upgrade. there was no money for SMU (Source Measure Units) and by bass said they use feedback and he said no. For the SMU project, the system was a LOT slower than what we had. There was also no money for a $15,000.00 Monochometer and there wasn't one available. Eventually there was. The ORG sensor didn't work quite right either.

Performance was a little worse.

I nneded to say, I can do the project, but I need help. I could have used the PDP-11 as an interim controller using RS232
Creating a LabView driver was hard. I also had a hard-drive crash.

So, realize the bean counter's expectations and realize yours. If you need help - ask for it.
Managers count beans!

 

And this is exactly why I avoid places like here.

Thank you, KeepItSimpleStupid for your help and suggestions. I'll find other sources than this place.

It's good to know that you prefer sunshine and optimism in the face of hard reality. I still want to know how this works out for you.