ARTY A7 - Reasoning for 220 Ohm resistors on all ports?

[DmDerev]

I built the "shield" board for relatively high speed on the mezzanine connectors (~50-60MHz signals) and discovered that the main board has actually 220 Ohm resistors in series with all pins. When I looked at schematics, the values were shown as 220 (which on SMD resistors mean 22 Ohm - pretty usual and benign value for series termination on traces). But measurement shows 220 Ohm and as the result, the design does not work. It is inconvenient to rework all boards (I have few) and replace these 0402 resistors with jumpers. 

I am wondering what was the designers' train of thought when volume producing the boards which don't work with many practical signals? 

Even if the idea was to protect the I/O pins, it was not a great idea as using this relatively decent high speed chip is not possible without rework of the board!

[zygot]

I don't work for Digilent so I can't speak for them. This is a company whose products always seem to have me wondering "What were they thinking?".

What I can say is that they don't design or sell general purpose FPGA development boards. What they do design and sell are boards that use PMOD products and let the user build MicroBlaze-centric designs using the limited set of Vivado supplied IP and the IPI flow.. propagating the myth that you can do useful FPGA design and development using a GUI drag and drop design methodology. Most of their boards are fine as platforms for introductory University level digital design classes, not as a platform for prototyping an FPGA based real world concept. For general purpose FPGA development when you don't want to spend the time and money to build a custom board there are better platforms.

Now that I've said all of that I should mention that I have quite a few Digilent branded FPGA boards even though I've never used any of their PMODs in a design. In general the price is pretty good, and most board have usable external DDR memory and an Ethernet PHY. The pre-Series7, pre-FTDI boards used the Cypress CY3014 with both JTAG, UART and 8-bit parallel USB 2.0 connectivity. A few boards have FMC and usable transceivers via mDP connectors. The CMODs have decent IO pin availability but suffer form poor design choices limiting their usefulness. I have implemented many dozens of designs for the Digilent FPGA boards that I own in cases where the project requirements didn't include high performance custom interfaces to other hardware. A large ( > 5"x4" ) FPGA board with 4 or 5 12-pin PMODs for IO is just not that usable if you want to do prototyping that includes external interfaces or advanced IO capabilities of the FPGA devices. That just isn't part of the Digilent business model. I've tried prodding, irritating and pleading for years for some kind of migration out of the 1980's but to no avail. Geez, even a PMOD header with 12 usable pins including consistency with providing at least 1 clock-capable input pin would be something encouraging.

Obviously, the 220 ohm series resistors have no use except to limit current flow and provide modest over-voltage protection. For someone needing to maintain 50 or so of these over the course of a few years where hundreds of students are handling, using and misusing the equipment, I'm sure that this is a good feature. Then someone decided that using 16 unused IO pins on a 3.3V Vccio Bank and putting them on "high-speed differential PMOD" headers was a good idea. Kind of defeats the whole FPGA IO protection concept... and angers users who thought that they might try out LVDS or some other differential interface on their boards.

Sometimes companies get stuck in the past and can't see the future. Edited October 18, 2022 by zygot

[mwagner]

Well, I am not an interface expert, but there are series termination issues with resistor values in many ways. I think mostly people will drive wires like SPI, I2C and those require pull ups of around 120 Ohms and more. So implementing a 50 Ohms can not be usefull there. In the opposite it is possible to bridge a resistor with higher value easily by a parallel resistor. Maybe this is a solution.

P.S. I like those PMODs, because you can quickly assemble a system for test. Recently I needed a permanent I2C-Trigger and was able to do that this way in an hour.