Question Possible direct drive alternative?

[DekuNut]

So the Wii Direct Drive is pretty darn expensive and I dont really see any quality benefits in comparison to the FujiFlex and HDMI driver board combo.
The problem with that however is that the driver boards are big, bulky and unneccessary and mostly use rtd2660 chips or derivatives which are hard to program if you are sourcing them yourselves for custom pcbs, meaning you have to buy a driver board to harvest the chip which imo is just wasteful.

So, while trying to understand the RGB TTL interface that most of the screens used in portables use, I stumbled upon a post in a german microcontroller forum where someone suggested that since dvi is basically a serialized RGB TTL signal, you could use a dvi deserializer from TI do drive the display that way.

One such chip is the TFP401 which can be had on Mouser for just 10€ and even cheaper on chinese sites.

Together with the fujiflex then, this could make for a cheaper direct drive alternative as in total it might cost 40€ at the most including pcbs etc.
OSHpark FujiFlex PCB 13$
Fujiflex bom 10$
Deserializer + PCB 15$
= 38$, which you can make even cheaper by ordering from jlc or something like that

Now, is this possible and reliable? I dont know, so before spending some sleepless nights trying to come up with something, I wanted to know the opinion of the experts.
What are the cons of this and do you think this would be a good idea?

TFP401 Datasheet: here

 

[YveltalGriffin]

You bring up some good points! Hopefully I can clarify some things and answer your questions.

So the Wii Direct Drive is pretty darn expensive and I dont really see any quality benefits in comparison to the FujiFlex and HDMI driver board combo.

Direct Drive does have quite a few benefits over GCVideo DVI (FujiFlex) plus an HDMI driver board. The big one is power consumption:

• DD draws ~350mA @ 3.3V (1.15W)
• A typical RTD2660 driver board draws 1-2W depending on how you power it, and GCVideo DVI draws ~450mW. That's 1.5W minimum, 2.5W maximum!

In a typical Wii portable (8-12W total), this is a sizable difference!

There are also video quality and user experience benefits:

• GCVideo DVI exhibits framedrops / video dropout in Resident Evil and Rogue Leader games. As far as I know, DD doesn't suffer from this.
• DD eliminates the black borders around games found with HDMI
• DD eliminates the blue source screen HDMI driver boards have
• DD has zero video delay when the Wii boots, unlike HDMI driver boards which take multiple seconds to lock onto a signal

One such chip is the TFP401 which can be had on Mouser for just 10€ and even cheaper on chinese sites.

The TFP401 does the same thing the RTD2660 does: it converts HDMI to the parallel 24-bit RGB the LCD needs. BUT, the TFP401 does not have a video scaler like the RTD266x. This means you need to feed it the exact LCD resolution for it to work. This really limits the TFP401's usefulness in a system where you can't easily control the HDMI output modeline (e.g. unmodified GCVideo DVI). (For something like a Linux SoC or computer, it's perfect! The Retrolite CM4 uses a TFP401 in case you are interested in a well-documented reference design.)

Re: cost, RTD266x chips can be had for like $2 from China, so they also beat the TFP401 in that arena.

Together with the fujiflex then, this could make for a cheaper direct drive alternative

@thedrew has done exactly this and calls it "Digidrive". He hasn't shared it on the forums, only in the Discord, but you can see an early prototype in his worklog here. The complexity of such a design is waaay higher than the single FPGA solution of DD.

38$, which you can make even cheaper by ordering from jlc or something like that

I'm obligated to mention that the BOM cost for RVL-DD is lower than what you've quoted for a Digidrive solution. The costliest component in RVL-DD is the FPGA, which costs ~$15. Think of RVL-DD's retail cost as supporting the ongoing firmware and ecosystem development 4LT provides to the community.

With that being said, an OSHW direct drive is in the works to provide a low-cost DIY solution.

 

[DekuNut]

Thanks so much for your detailed reply, I was also thinking about maybe doing a whole ass reimplementation of DD but I know nothing about fpga programming so im glad that a low cost solution is in development. It has been some time since I was active on the forums and I dont really use discord much so I thank you for pointing out these 2 reference projects.
Also I get that Aurelio, Gman, Shanks and everyone else involved in 4layer obviously deserve something for developing these products and providing them. Still, I dont like spending so much money so I always try and find a way to do something my own way thats hopefully cheaper. So, im sorry to bother you, but I still have some questions.

1. Assuming there is no minimum hdmi resolution, which should be the case I hope, how hard would it be to modify gc video to output 480p over hdmi?

2. In the spirit of power conservation, I noticed that fujiflex requires the 1.8v rail, which can be safely undervolted to about 1.45v or so I believe, would this hinder the operation of the fpga?

3. Kind of off topic, but your Kawaii project utilises the case as a passive heatsink. What software or calculations did you use to make sure that the Wii wont overheat with this case design?

 

[YveltalGriffin]

1. Assuming there is no minimum hdmi resolution, which should be the case I hope, how hard would it be to modify gc video to output 480p over hdmi?

GCVideo DVI already outputs 480p HDMI, but you'd need to adjust the reblanker to output the 800x480 modeline the LCD requires, and probably tweak some other stuff, too. The Xilinx toolchain is pretty nasty and I wouldn't recommend going that route.

2. In the spirit of power conservation, I noticed that fujiflex requires the 1.8v rail, which can be safely undervolted to about 1.45v or so I believe, would this hinder the operation of the fpga?

Fujiflex uses 1.8v to power the FPGA 1.2v core rail:

If you undervolt the 1.8v rail too much, the LDO won't have enough voltage headroom and will operate in dropout, potentially causing the LDO output voltage to fall below 1.2v. You might get away with undervolting the FPGA, but I would recommend powering Fujiflex's 1.8V input with 3.3v instead to avoid issues.

If you are making a custom GCVideo DVI board, you can power the core rail with a lower dropout LDO, or a buck converter (with proper filtering) for optimal power efficiency.

3. Kind of off topic, but your Kawaii project utilises the case as a passive heatsink. What software or calculations did you use to make sure that the Wii wont overheat with this case design?

No math or sims (although they're simple to do in Fusion and I should try one while I wait for the shells!) Wesk conducted some very promising passive cooling tests with FD Omega trims, and I had some vague intuition about thermals from Shinobi Scaler and work stuff that made me feel like ~3W was workable. We'll see if it pans out.

 

[Chickacruh]

@DekuNut, If you are comparing the ElectronAVE kit to the RVL-DD the price isn't even drastically that different. The RVL-DD kit is just 20 dollars more, and that makes sense considering it is a more specialized kit. Sure FugiFlex is cheaper, but you need to assemble it yourself and that introduces another variable of hassle. So considering the fact that ElectronAVE kit and RVL-DD are fully assembled and tested seems like a fair price to me.
Also what are you trying to make? If it's a portable, I would think it would just be easier to use the RVL-DD because it interfaces directly with the screens.

 

[DekuNut]

Yeah I know that its more of a hassle to assemble it on your own but I dont mind it if I can save some money. Im currently in the concept stages for my second portable which im trying to make as small as possible because my first one was quite big and I can already emulate Wii games on my phone, just with no controller so I wanted something small with controllers built in. Im currently searching for small lipo batteries, a way to thinly cool the Wii, optimized power delivery and other optimizations to make it as small as possible. Im not quite ready to start building yet as I currently lack the funds and the time to embark on another portable, but it will hopefully utilize an omega trim that will also be undervolted. Wiis are quite cheap here so Im not really concerned with that, its just that certain pcbs like the pms or rvl dd are too expensive for me, which is why I often try to reimplement them. Thats also the whole reason why im studying electrical engineering soon, as I wanted to clone the rvl pms in 2020, so I started learning pcb design and kicad because of that.

I know that they are priced quite fairly, but even when I personally sell stuff I only really include the hardware price with some profits for me without including the price for the firmware, so this pricing model is kind of always in the back of my mind when buying stuff. Still, while I personally am not ready to pay that much right now, its an appropriate price.

 

[Aurelio]

We aren't really including development costs in our price, as those get compensated with time. When pricing something like this you have to take into account multiple factors, including the BOM, the assembly costs, margin to cover defective units and more others. On top of that there's of course some profit, but that's the key point of running a business.