Open airwoodix opened 4 years ago
The only difference between standard PoE and passive PoE is the lack of a power negotiation mechanism. Voltage is a secondary thing, standard PoE modules can sometimes work with voltages down to 12V, but it is usually 48V.
Ok. Thanks for that clarification.
By proper PoE converter on that adapter PCB you mean Active PoE compatible with 802.3af/at or just DC/DC converter with max Vin >=48V?
And as I understand, power would run from the adapter PCB to DiPho via VCC/GND pins on USB-C connectors, because PoE conversion will happen on the adapter PCB. But then eth transformer should also be located on that adapter PCB, am I right?
So in the end adapter PCB, will have similar functionality (+external power input) to this PoE splitter, but RJ45 and USB cables will be merged into one cable?
Yes, I mean using for exampel Silvertel converter + RJ45 connector. Look how it is done in Thermostat design.
@gkasprow I've just pushed the first revision/draft schematics of afe and mcu boards.
@pm-daniel please use the master branch. My files were just templates. Did you do some simulations of the AFE? Why this particular opamp was used?
It's a bad idea to use both SSTX pairs for Ethernet signals - you will get stubs. That's why we use Auto MIDIX to enable automatic pair detection. Just use SSTX 1 and SSTX2.
try to use a standard SWD connector, of course, if that fits
please use ports only for multi-page projects. for local signals use nets. See how other Sinara projects are organized
@gkasprow
pm-daniel please use the master branch. My files were just templates. Did you do some simulations of the AFE? Why this particular opamp was used?
try to use a standard SWD connector, of course, if that fits
I'll try to fit it
@gkasprow it looks like the default branch for this repo is main
- could we do something about it? Or maybe I should move all files to the main
branch
Just make sure the file you push is listed on Github.
I understand that I should use the current source as a stimulus?
yes, but please add photodiode mode.
I choose OPA141 because it has significantly smaller input noise and input bias current, is available in a smaller package than TL071 f.e. and its price is not horrible IMO.
right. But in our case, you have to prove you need low voltage input noise. We are working with high-value feedback resistor so the current noise may be dominating. Low input bias current is not that critical since you can easily compensate for it. But be careful because you can add some noise! Here is an online tool to play with TIA. Essentially, you have to show that you know what you are doing :)
@gkasprow sure. It's a busy time right now but I'll try to do some simulations (and documents) in the following week or two
@gkasprow After doing a few simulations it became obvious that OPA141 may not be the best pick (TL071 either). Because of the pretty bad output voltage swing characteristic - I totally forget about it (for TL071 it's +-1.5V from the supply voltage and +-0.3V from the supply voltage for OPA14). I've tried a few different opamps (AD8541, AD8539, ADA4891) and ADA4891 looks most promising and it has a very low input bias current, very good swing characteristic, and a similar price to OPA141. All those conclusions are based on transient simulations that I conducted in LTspice. I used the current source with parallel capacitance as stimulus and now I'm struggling with noise simulation as it looks like LTspice expects voltage source to be selected as a noise source. Is there a better way of simulating photodiode input in the LTspice? I've spent some time on research and it looks like everyone uses just the current source with capacitance (just like I did). Once I complete the noise and bandwidth simulation I will publish results on GitHub. Maybe you have any other recommendations regarding opamp choice?
Sorry for such a delay but I'm really busy finishing the current semester.
A command like .noise V(OUT) I1 dec 1000 10 10Meg
(where OUT is the output node, and I1 is the current source) should work fine in LTspice.
@dnadlinger thank you! After seeing your reply I've checked my simulation command again and it turns out that I had IN and OUT nodes defined in the wrong order. It works just fine now.
@pm-daniel can you open a new issue and post your noise simulations?
@gkasprow done #3
@pm-daniel Do you have an update on this project? How did the testing of the first boards go? Thanks!
@airwoodix Tests are not finished yet - they will be ready by the end of the month (hopefully next week). I'll publish the results on GitHub. Sorry for the delays but in the meantime, I'm writing my BEng thesis. Working a full-time job at Creotech (on non-related projects) also doesn't help. I hope you understand!
Of course! :slightly_smiling_face: Thanks a lot for the update. I'm looking forward to the testing results.
@pm-daniel can you post some photos of the PCBs?
It's mounted in measurement fixture for now so those are the best photos I can take at the moment
I'll upload better ones if needed after finishing the measurements
Continuing the discussion in sinara-hw/meta#52 (comment) on cheap monitoring photodiodes for Sinara.
Does this cover enough use cases?
I only see opportunities if this can be around 10 times cheaper than the Thorlabs product, so max 50 €/$ per channel.