Development board with FPGA GW1N-LV9LQ144C6/I5

[FlyingDutch]

Hello forum,

I designed a simple development board with FPGA from GowinSemi - IC model: GW1N-LV9LQ144C6/I5. Here is link to mouser.com with this product:
https://www.mouser.com/ProductDetail/GOWIN-Semiconductor/GW1N-LV9LQ144C6-I5?qs=wnTfsH77Xs6N%2FyGezJZrJQ%3D%3D

https://www.gowinsemi.com/en/product/detail/46/

Tis is basic FPGA with 8640 LUT and 120 I/Os in LQFP-144 case. The amount of resources allow conduct experiments with Soft-CPU (for example with RISC-V soft-cores). My main goal was to design FPGA development which is cheap, but is suitable for beginners. Apart the FPGA itself the board is equipped with:
1) 24 MHz clock circuit
2) LDO voltage controllers and switches for configuring FPGA I/O Banks (three voltages: 3.3V , 2.5V and 1.2 V
3) FPGA Mode switches (for boot mode)
4) Setup circuitry (Reconfig, Done etc.)
5) FPGA reset circuit
6) USB to JTAG based on FTDI chip FT 2232HL (allowing programing by USB socket)
7) JTAG Header for alternating programming by "Gowin Cable" external programmer
Additional 32Mb SPI flash IC
9) 8MB of PSRAM (Pseudo RAM) alowing big frame-buffers
10) DSub15 VGA connector
11) 8 user LEDs
12) 8 DIP-Switches
13) 5 Push Buttons
14) TF (uSD) card module
15) Simple audio output
16) 59 I/O pins led out to the three goldpin connectors

I wanted the FPGA kit had the opportunity to program by USB socket and did not require expensive external programer (Gowin "USB Cable"), hence the use of the FTDI chip . I also wanted the FPGA board bring out a large number of I/O pins. I don't plan to output differential pins (LVDS) with impedance and length control on PCB. Here is finished schematic (PNGx and PDF).

It remains to design the PCB.
This is how (third image)  the initial arrangement of components on the PCB looks like. I will update the thread as I design the PCB.

Please be so kind to give me a feedback about this design 

Regards

[Gribo]

You left the regulator's thermal pads unconnected. You should connect them to a ground plane.

[FlyingDutch]

You are right - thanks

Regards

[colorado.rob]

I would drop the VGA and add an HDMI port. I think HDMI is more straight-forward than VGA. And finding a VGA monitor is getting difficult. I don't know that I have a VGA monitor anywhere at home.

You can either do the proper TMDS termination on the board, or go with a direct LVDS connection. The users might need to use a HDMI port extender to use an LVDS port with some devices. But it is more flexible.

Either that or make it a 40-pin LCD connector.

[nctnico]

A better option is to add a board-to-board connector with good signal integrity (IOW: not headers) so you can add whatever high-speed interface you like to the board.

[FlyingDutch]

I would drop the VGA and add an HDMI port. I think HDMI is more straight-forward than VGA. And finding a VGA monitor is getting difficult. I don't know that I have a VGA monitor anywhere at home.

You can either do the proper TMDS termination on the board, or go with a direct LVDS connection. The users might need to use a HDMI port extender to use an LVDS port with some devices. But it is more flexible.

Either that or make it a 40-pin LCD connector.

Hello @colorado.rob,
I thought about adding HDMI output. This board is designed for learning FPGAs and VGA projects in HDL are much simpler than HDMI, but maybe you are right. I will add HDMI output to this design, but I do not drop VGA connector.

Thanks for comment and regards

[FlyingDutch]

A better option is to add a board-to-board connector with good signal integrity (IOW: not headers) so you can add whatever high-speed interface you like to the board.

Hello @nctnico,

I designed this board for beginners for learning purposes and the ease of connecting to signals is essential. I am not willing to change this. I would like to design a second more advanced board with Gowin FPGA GW1N LV9 with BGA256 case. On this second board, I will lead to the connector high-speed signals (LVDS) with tracks designed with impedance and length control and more convenient connectors.

Thanks for your comment and regards

[up8051]

Beginners will buy "Tang Nano 9k" for 18$,  cheaper than chip itself from  mouser.

I would love to buy a development chip with LVDS outputs, LCD 40/50 pin connector and voltage generators for LCD, switched voltages of I/O banks, etc

Best regards
JarekC

[nctnico]

A better option is to add a board-to-board connector with good signal integrity (IOW: not headers) so you can add whatever high-speed interface you like to the board.

Hello @nctnico,

I designed this board for beginners for learning purposes and the ease of connecting to signals is essential. I am not willing to change this. I would like to design a second more advanced board with Gowin FPGA GW1N LV9 with BGA256 case. On this second board, I will lead to the connector high-speed signals (LVDS) with tracks designed with impedance and length control and more convenient connectors.

Thanks for your comment and regards

I get that but boards like yours are a dime-a-dozen on Aliexpress. Recently I bought two Tang Nano boards for 11 euro each and my biggest complaint is that they don't have good connectors that allow use as a module on a serious prototype board. Besides that, having board-to-board connectors doesn't rule out to have pads for headers. They can both exist at the same time.

[asmi]

Hello @colorado.rob,
I thought about adding HDMI output. This board is designed for learning FPGAs and VGA projects in HDL are much simpler than HDMI, but maybe you are right. I will add HDMI output to this design, but I do not drop VGA connector.

VGA and HDMI are fundamentally the same (and not, that is not a coincidence!), but VGA requires digital-to-analog conversion while HDMI instead sends color values directly via serial links, so I would say HDMI is a bit simpler logically. One problem with HDMI is that it requires fast IO and preferably hardware SERDES if you want to achieve 1080p@60Hz (this requires 1450 Mpbs per lane!), and half of that for 720p@60Hz, but of course if 480p is all you need (which is what you are probably going to aim for with VGA), you can get away with relatively slow 270 Mpbs per lane.

With that said, I do agree that VGA is long obsolete and past it's usefulness except in a very few cases of recreating ancient hardware.

[asmi]

Besides that, having board-to-board connectors doesn't rule out to have pads for headers. They can both exist at the same time.

Actually it kind of does because those extra pads for headers will create stubs which negatively affect signal integrity due to reflections.

[nctnico]

Besides that, having board-to-board connectors doesn't rule out to have pads for headers. They can both exist at the same time.

Actually it kind of does because those extra pads for headers will create stubs which negatively affect signal integrity due to reflections.

That is not a huge problem for the frequencies these FPGAs are useful at. For a prototype spread over several boards, the signal integrity won't be excellent anyway.

[asmi]

I get that but boards like yours are a dime-a-dozen on Aliexpress. Recently I bought two Tang Nano boards for 11 euro each and my biggest complaint is that they don't have good connectors that allow use as a module on a serious prototype board.

Why don't you design your own board which would have whatever you want? These are very simple FPGAs, so designing a PCB for them is a rather trivial matter (no DDRx, rather easy power requirements, etc.).

[nctnico]

I get that but boards like yours are a dime-a-dozen on Aliexpress. Recently I bought two Tang Nano boards for 11 euro each and my biggest complaint is that they don't have good connectors that allow use as a module on a serious prototype board.

Why don't you design your own board which would have whatever you want? These are very simple FPGAs, so designing a PCB for them is a rather trivial matter (no DDRx, rather easy power requirements, etc.).

Why would I for a prototype? If I'm going to design a board, I'm not going to make something universal I could sell as a product. However, if somebody is asking for input to design such a product aimed at prototyping, I'm more than happy to help with suggestions. It is up to the person to deem my suggestions usefull or not. Looking at what market segment is interesting is part of product development.

[asmi]

Why would I for a prototype?

I always design a devboard whenever I evaluate a new FPGA for my future projects, or if my existing devboards can't accomodate peripheral I want to use. Don't really see a problem with that, these boards always get a lot of mileage

If I'm going to design a board, I'm not going to make something universal I could sell as a product. However, if somebody is asking for input to design such a product aimed at prototyping, I'm more than happy to help with suggestions. It is up to the person to deem my suggestions usefull or not. Looking at what market segment is interesting is part of product development.

His board is clearly intended for beginners, who probably don't even know what LVDS is, much less how to use it. Also high-speed connectors tend to be relatively expensive, and beginners tend to be very price-sensitive as they are not yet committed to FPGAs and might not be willing to invest into the devboard.

[FlyingDutch]

Beginners will buy "Tang Nano 9k" for 18$,  cheaper than chip itself from  mouser.

I would love to buy a development chip with LVDS outputs, LCD 40/50 pin connector and voltage generators for LCD, switched voltages of I/O banks, etc

Best regards
JarekC

Hello @up805,
I am going to design a second more advanced board with differential signals (LVDS) led out to better connectors and with tracks impedance and length matching. Also with 40 pin connector for LCD. A am also planning to place ADC with 20 MSPS speed and a similar speed DAC on board. This simpler board is for getting more experience and checking if everything is working properly.

Thanks for your comment and Regards

[FlyingDutch]

VGA and HDMI are fundamentally the same (and not, that is not a coincidence!), but VGA requires digital-to-analog conversion while HDMI instead sends color values directly via serial links, so I would say HDMI is a bit simpler logically. One problem with HDMI is that it requires fast IO and preferably hardware SERDES if you want to achieve 1080p@60Hz (this requires 1450 Mpbs per lane!), and half of that for 720p@60Hz, but of course if 480p is all you need (which is what you are probably going to aim for with VGA), you can get away with relatively slow 270 Mpbs per lane.

With that said, I do agree that VGA is long obsolete and past it's usefulness except in a very few cases of recreating ancient hardware.

Hello @asmi,
I agree with you, but my experience as a beginner is such that at the start of learning it is easier to understand how VGA projects work (I mean in HDL code). This the reason I put the VGA output on this board.

Thanks for your comment and regards

[asmi]

Hello @asmi,
I agree with you, but my experience as a beginner is such that at the start of learning it is easier to understand how VGA projects work (I mean in HDL code). This the reason I put the VGA output on this board.

I don't understand how it's so - it's the same code, up to the last stage, which goes into DACs (for VGA) vs SERDES (HDMI). My concern with VGA is that it's extremely unlikely that a future user will actually have a VGA-capable display unless he bought one specifically for that reason (which is unlikely because beginners typically are not willing to invest serious money into a display that won't be useful for anything else, as all other boards typically have only digital outputs, like HDMI or even DisplayPort), and as time goes on, it becomes progressively harder to find.

[nctnico]

IMHO an LCD panel interface (either parallel or LVDS) makes more sense. These are cheap from various sources. The only problem is that the connector typically isn't standard but going for something that also works on the Raspberry pi is likely a good enough default.

[asmi]

IMHO an LCD panel interface (either parallel or LVDS) makes more sense.

It's exact same as HDMI and VGA (sans the output stage)! And DisplayPort is again the same - apparently people designing all these interfaces really hate change, so they tried their absolute hardest to shoehorn old analog approach (with blanking and stuff, which make no sense for modern displays) into digital protocols. I would agree though that even 720p would probably be too much for such small density device, while 800x480 LCD panel would be just about right. But this again raises a concern about increased price, because it's unreasonable to expect users to already have such a panel.

[langwadt]

IMHO an LCD panel interface (either parallel or LVDS) makes more sense.

It's exact same as HDMI and VGA (sans the output stage)! And DisplayPort is again the same - apparently people designing all these interfaces really hate change, so they tried their absolute hardest to shoehorn old analog approach (with blanking and stuff, which make no sense for modern displays) into digital protocols. I would agree though that even 720p would probably be too much for such small density device, while 800x480 LCD panel would be just about right. But this again raises a concern about increased price, because it's unreasonable to expect users to already have such a panel.

https://wiki.sipeed.com/hardware/en/tang/Tang-Nano-9K/Nano-9K.html

on Aliexpress it is half the price of the FPGA alone on mouser

[asmi]

https://wiki.sipeed.com/hardware/en/tang/Tang-Nano-9K/Nano-9K.html

on Aliexpress it is half the price of the FPGA alone on mouser

Aaand? What's your point?

[nctnico]

IMHO an LCD panel interface (either parallel or LVDS) makes more sense.

It's exact same as HDMI and VGA (sans the output stage)! And DisplayPort is again the same - apparently people designing all these interfaces really hate change, so they tried their absolute hardest to shoehorn old analog approach (with blanking and stuff, which make no sense for modern displays) into digital protocols. I would agree though that even 720p would probably be too much for such small density device, while 800x480 LCD panel would be just about right. But this again raises a concern about increased price, because it's unreasonable to expect users to already have such a panel.

Small TFT panels are a dime-a-dozen.

https://wiki.sipeed.com/hardware/en/tang/Tang-Nano-9K/Nano-9K.html

on Aliexpress it is half the price of the FPGA alone on mouser

Aaand? What's your point?

The point is that what the OP wants to make can be bought off-the-shelve for a far lower price than the OP can buy the parts for. So how to sell this board into a cost conscious market? This can be done by offering a very good ecosystem but how to tie that to the product of the OP? Maybe the beginners market isn't even interesting but a higher end board could sell better to a more professional market segment.

[asmi]

The point is that what the OP wants to make can be bought off-the-shelve for a far lower price than the OP can buy the parts for.

And? What about all other stuff?

[nctnico]

The point is that what the OP wants to make can be bought off-the-shelve for a far lower price than the OP can buy the parts for.

And? What about all other stuff?

Random Google find:

https://learn.lushaylabs.com/tang-nano-9k-graphics/

With the endless flood of 'shields' that are pluggable into a breadboard, the possibilities are limitless. For as long as you don't need high speed interconnects or a more permanent solution... Which circles back to the market segment that isn't addressed by all pin-header based boards. Think about why the Raspberry Pi compute modules that fit in an SO-DIMM slot exist. An M.2 or mSata slot could also be an interesting form factor for an FPGA board. And these connectors don't add any cost to the board at all.

Just thinking out loud here