GTH transceiver voltages

[archangel-lightworks]

Probably a basic question, but one I couldn't make clear from the documents. 

From UG 576 table 3-35 it looks like I'm able to configure the voltage swing between 134 and 1000 mV.

From DS925 Table 96:

Differential peak-to-peak input voltage                      Min                Max
(external AC coupled)                                                150               1250 mV

If I'm not wrong, Diff Vpp = 2 * Vswing, and in theory, I should be able to connect a GTH TX to a GTH RX, but it looks like if I select 1000 mV swing and I have 2 Vpp I would be beyond the input specs (?).

Also I couldn't find the GTH transceiver output impedance written anywhere in UG 576 or DS 925.

[eschidl]

Hi @archangel-lightworks ,

the values in table 3-35 are differential values, not single ended.

For the impedance, have a look at figure 3-32.

[archangel-lightworks]

"voltage swing" tends to be a single-ended one-wire measure. It could be differential but I think this is something worth mentioning somewhere in a 517-page document to prevent users to scratch their heads, get the answer one by one in a forum, and in the worst case, make a design mistake.

What if I am challenged to explain that? I won't look much professional saying "oh, I asked in the Xilinx forum and I got this answer, here is a screenshot...". Xilinx is a leading company and you guys need to be up to that. Please take it as constructive criticism.

Similar thing for the impedance, I would expect that to be explicitly written, again somewhere in a 517-page document, instead of a diagram just showing a resistor symbol, something like a pull-up, that seems to insinuate to be the value of the output impedance. That's simply not serious. On top of that it shows a variable resistor, so what are the limits? Even more, it's a differential pair and differential impedance is not exactly twice as much as the single-ended impedance. 

[archangel-lightworks]

UG 576 v1.6 I assume is a document that has had at least 6 releases. The output impedance of a high-speed transmitter is a primary parameter and should be one of the first values presented. Something that will always has to be done is routing those pins on a PCB (buying dev boards is not always an option)

[eschidl]

Hi @archangel-lightworks ,

The parameter you are changing is TXDIFFCTRL. It has the differential already in it - TX differential control. I hope that is explanation enough.

Also, please have a look at the datasheet DS925, table 96. There is Rin and Rout mentioned. Both to be 100Ohm.
The variable resistor in the image shows that the resistor is calibrated. You will find this similarly in the picture for the RX input. This is done through the RCAL circuit which is described in the user guide as well.

[archangel-lightworks]

"The parameter you are changing is TXDIFFCTRL. It has the differential already in it - TX differential control"

The transceiver is physically differential but not all measurements need to be. Most of us have two legs, so I suppose when you say how long your legs are you add the left and right measurements. And as a justification, you say "humans are bipeds".

Female models are expected to be some 180 cm tall and have a leg length of 80 cm. Take the latter measurement as "differential" and that's the type of mistake that crashed the Mars Lander in 2016 and God knows how much money with it. We need to be specific, even if looking pedantic.

Rin and Rout... Impedance is more than resistance. Also, many times series resistors are added for impedance matching. "Rin" and "Rout" might be understood as that and not necessarily matching the impedance.