Page 1 of 1
Known current problems with MiSTer clones
Posted: Sat Sep 28, 2024 9:28 am
by akeley
There are many people interested in the DE10 Nano board & add-on clones from QMtech & Taki Udon, so I thought it would be a good idea to gather information about their known and alleged problems (plus solutions) in one place, so users can make informed purchase decisions.
I'd appreciate it if we could keep things factual in here and reserve opinions about these devices for other threads.
Please note that other than the following problems, the clones are generally reported to work perfectly fine. The problems might (or might not) be fixed in future revisions.
This thread is of course a WIP, please contribute any more information you have.
As of Sepetember 2024:
QMTech
1) problem: Direct Video dongles not working
solution: use powered dongles
MiSTer Pi
1) problem: low quality audio from I/O board
solution: RMA the board
solution2: use audio DAC+Toslink cable (to be confirmed if it works ok, problems reported)
solution3: buy I/O board from official seller (confirmed working)
2) problem: Nintendo 64 core issues?
solution: core fix being worked on
3) problem: RAM issues (same as #2 perhaps)?
Also, a suggestion for @Xbytez: maybe it'd be a good idea to make a separate Clones subforum in the Hardware section?
Re: Known current problems with MiSTer clones
Posted: Sat Sep 28, 2024 9:35 am
by rickdangerous
1) #2 works
3) #2 may not work with current v3 but we are testing new BGA modules.
Re: Known current problems with MiSTer clones
Posted: Sat Sep 28, 2024 3:38 pm
by rickdangerous
Apparently Robert, the N64 core developer, is making the cores compatible with all clone boards.
Re: Known current problems with MiSTer clones
Posted: Sun Sep 29, 2024 7:23 am
by FPGA64
On Discord Robert wrote this
let me try it...
When reading data from sdram, you need to decide when to look at the data. As it is changing with the sdram clock speed, the data will not be stable all the time but only for some time. There is a time when it begins to get stable(170°), a stable time and a time where it gets unstable again(300°).
Previously we sampled right at the border(180°) and on some new boards it will even shift in the unstable area with that.
The testbuild shifts the sample point. Not yet clear if really in the middle of the "eye", but probably enough to fix it already
Re: Known current problems with MiSTer clones
Posted: Sun Sep 29, 2024 8:00 am
by akeley
Thank you both for the RAM info. So, is the "RAM problem" the same as "N64 problem" (#2 & 3 on my list)? I don't know, I just saw them mentioned on two separate occasions.
Re: Known current problems with MiSTer clones
Posted: Sun Sep 29, 2024 8:08 am
by FPGA64
That was Roberts experimental fix for the N64 core
Re: Known current problems with MiSTer clones
Posted: Sun Sep 29, 2024 9:10 am
by Armakuni
The QMtech board Direct Video needing a HDMI to VGA with a +5v power is not an issue there are plenty available and cheap from the likes of Ali and some have better DACs than the AG2600
The N64 issue has been now solved and this affected both Taki and QMtech boards, even some DE10 setups had issues with Conker
The Taki IO board analogue audio noise only affects a certain batch and he says to RMA them
The SDram used by QMtech on their soldered on boards is a slightly different spec Alliance, this is why it can be stable at 160mhz. He mentioned this on Twitter
They are all minor issues
Re: Known current problems with MiSTer clones
Posted: Sun Sep 29, 2024 9:37 am
by akeley
A low quality audio from a board advertised as 9.2 design is certainly not a "minor issue". Similarly, if you don't know about the DV dongle needing voltage (standard boards don't) then it is a problem.
The purpose of this thread is to present clear information about these things (which these days can be scattered across different media).
Thanks for the info anyhow, so is the N64 problem now officially fixed or "experimental" as it's been said above?
Re: Known current problems with MiSTer clones
Posted: Sun Sep 29, 2024 9:40 am
by FPGA64
Its Experimental so far. But it seems to be working but as with all, testing is needed