POWER SUPPLY BYPASS REVISITED. A short story.


This is by no means expert advice on power supply and bypass coupling for hi-speed FPGA's, just observations I've made regarding my multi-board video project. I do, however, consider myself above average regarding power supply PCB layout, currently 4-layers. So this is why I have waited so long on installing capacitors on board. When I got up to 4 boards, things started to fail @148.5MHz...

Currently there are 4 boards. Each board consists of a 100-pin QFP SyncRAM, a 144-pin QFP FPGA, 2x 20-pin SOIC FPGA PROMs. The final output board adds a 48-pin QFP videoDAC.

I should have an ammeter to see how much current each board is drawing, but I'm not going to be that specific. I would guess the SyncRAM is pulling close to 300mA, and the Spartan 6 FPGA maybe 150? So guess maybe close to 1/2amp per board. The power supply is a very old PC/AT style which I mainly utilize the 3.3V/28A lines. Also 5V for the final output board, which is filtered down to 3.3V for the final output videoDAC IC.

So I tried to get away with as much as I could, in regards to not mounting very minute 0603 capacitor packages, in this project which is now operating at 148.5MHz at a resolution of 1920x1080. I too, do not particularly enjoy fiddling with ultra small parts. However there is enjoyment in the succeeding and thus a learning curve begins.

Each board has a layout to accomodate 2x 22uF bypass caps for the 3.3V and 5V power supply main input.
Under the FPGA, there's room for 2x 4.7uF for 1.2V and 2.5V VCCINT and VCCAUX mains after the Voltage Regulators.
Also on every power pin of the SyncRAM and FPGA (above and underneath), there are provisions for .1uF (100nF) caps.

So what I had that barely worked:
3x PVB:
Final output board: 22uF and 4.7uF installed. .1uF installed on top of board for FPGA, FPGA PROMs & videoDAC. None for SyncRAM. None for underneath FPGA.
2nd board: 22uF and 4.7uF installed. .1uF installed on top of board for FPGA, FPGA PROMs. videoDAC not installed. None for SyncRAM. None for underneath FPGA.
3rd board: no caps at all.

With 3 boards I was running into intermittent issues with the video tearing. The 3rd board was outputting bad video every few seconds. The system was running at low voltages. I think I read 2.7V where 3.3V was necessary.
With 4 boards, the 4th board was outputting trash. It could not sync up to the 3rd board's H&V sync outputs. I had unplugged board 4 for a couple weeks. I thought it was a malfunction. Confident in my power supply routing I was, until...

Today, I added in the 22uF and the 4.7uF bypass caps for the main power and FPGA power regulators on boards 3 and 4.
All 4 boards are functioning properly now. Albeit with some noise from the 3rd and 4th board, but no major problems. I suspect the noise is from the lack of .1uF cap's on board 3 & 4, or some bad soldering on the large QFP's.

Interesting to note I had 4 boards working at a lower 1024x768 resolution with no problems at all @ 70MHz. Also no noticeable noise in the video.

_________________
65Org16:https://github.com/ElEctric-EyE/verilog-6502