Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Torfinn Ingolfsen
    @tingox_gitlab

    I'm not sure that will work; I tried loading the precompiled bitstream (from https://github.com/litex-hub/linux-on-litex-vexriscv-prebuilt) but then fujprog complained

    tingo@kg-core2$ fujprog build/ulx3s/gateware/top.svf
    ULX2S / ULX3S JTAG programmer v4.6 (git 0a4cc36 built Jul 22 2020 22:13:24)
    Copyright (C) Marko Zec, EMARD, gojimmypi, kost and contributors
    Using USB cable: ULX3S FPGA 12K v3.0.8
    
    Found LFE5U-12F device, but the bitstream is for LFE5U-45F.
    
    Failed.

    unless there is a way to override fujprog checking?

    Lawrie Griffiths
    @lawrie
    @emard I have managed to coax my Mac Plus implementation (with a few temporary hacks) to get to the point where it prompts for a floppy disk. The floppy disk implementation looks complex, so I will probably use some of the code from the Mist/Mister implementation. As far as I can see, the Mister implementation downloads the whole 400kb floppy disk image into SDRAM.
    I think the Mac Plus version supports 800kb floppy disks as well as 400kb ones.
    The interface that the rom uses is the Integrated Woz Machine - http://www.brutaldeluxe.fr/documentation/iwm/apple2_IWM_Spec_Rev19_1982.pdf
    emard
    @emard
    @lawrie hm, this looks like some apple's solution to support new formats 3.5" 400/800KB with backwards compatile support of old 5.25" floppiys for apple ][ so my guess is that track and block access could be similar to apple ][ otherwise the chip would be too expensive. Still I don't get what would such compatibilty offer to users in practice as 5.25" won't fit to 3.5" drive.
    Goran Mahovlic
    @goran-mahovlic
    @tingox_gitlab you will need to use ecppack and ecpunpack to set right id, but it will work!
    not for 45F
    but 12F/25F
    but maybe some tweak is still needed to fit litex on 25F - I am not sure in that - recently I did saw some messages on twitter about it...
    Erik Piehl
    @Speccery
    I've continued to work on my icy99 core and posted a Youtube video of the current status, in case you're interested. https://youtu.be/zdST3wz00KU
    It's just a quick video, no production value here for sure. But I now support the 96x64 LCD and added disk drive support (via another TI-99/4A related project called TIPI, running a whole bunch of services on a Raspberry PI connected to the ULX3S). I have not been involved in the TIPI development, but it is a very interesting project.
    One question I do have: when I added audio support (yes, finally, using the 3.5mm output) I did a very quick and dirty implementation, just using the 4 bit resistor DAC "as is". I want to increase the resolution of the DAC. I could device my own delta-sigma conversion, but I think I saw somewhere this kind of an implementation already with 4 bit output. In my previous design I just used a 1 bit sigma-delta conversion at 100MHz.
    Erik Piehl
    @Speccery
    Oh and another thing I added was cursor key navigation with the PS/2 keyboard. Pressing F1 brings up the overlay screen, and cursor keys operate as directional keys. It's nice to have the standard way of pushing all 4 keys simultaneously, but since I have a keyboard anyway, using a single function key is much better.
    Erik Piehl
    @Speccery
    @lawrie I've been trying to test your ulx3s_ql project. I just run synthesis and loaded toplevel.bit from the ESP32. Do you have somewhere instructions how to bootstrap the QL core? Sorry I haven't read any of the source code yet. Just as I wrote this, I noticed the short document in the esp32/osd directory. Will try that out.
    Erik Piehl
    @Speccery
    And I did, with j1s.mx1 it does boot up as a QL. Very nice!
    Lawrie Griffiths
    @lawrie
    I think js.rom works now. I don't think it needs the patched js1.rom.
    Erik Piehl
    @Speccery
    Thank you. Does it support some form of software loading into the QL?
    emard
    @emard
    @Speccery wonderful TI99/4A setup! There exist similar small 240x240 color SPI LCD, similar size to SSD1331 and pin compatible, they may have pixel resolution almost enough for whole TI text screen.
    For QL yes rom should also support direct loading should be just point cursor and it should load
    I have normal ROM and Minerva and can switch between them using ESP32
    Lawrie Griffiths
    @lawrie
    @Speccery Yes, you can select an mdv file with the OSD, for example quill.mdv is in the roms directory. You can then do dir mdv1_ to look at the directory, and the microdrive has a boot file you can load and run.
    I have not implemented microdrive writing. I hope to do that sometime.
    I need better README files for my projects, as I forget how to use them.
    Your video is a good way of documenting how things work.
    Lawrie Griffiths
    @lawrie
    TiPi looks interesting.
    As @emard says, you would get more on a 240x240 st7789 display.
    My brother-in-law (Harry Moyes) has now got a Ulx3s, and he is a keen QL user, so I will probably have to get back to looking at the QL. I think he wanted floppy disk support and was thinking of implementing a QL expansion port.
    Erik Piehl
    @Speccery
    I have a small 240x240 display, 1.3" The pinout however on that one is different from the 96x64 system - of course I can still hook it up with cables. Is the one you're referring to something that has the same pinout?
    I also have a QL. Bought it on unknown condition, fixed it (it had problems with RAM) and I have now also a Gold card for it. The microdrives on it don't work, but I've now got a Gotek drive connected to the Goldcard and that works. The Goldcard has a 16MHz 68000 and 2Megs of RAM, so it makes for a nice QL. I'm still on the discovery phase for it, for example looking for ways to convert microdrive images to floppy images.
    Lawrie Griffiths
    @lawrie
    I had a QL when they came out, but have not used a real one since the mid 80s. My brother-in-law has some form of enhanced one.
    Compatible 7-pin 1.3" st7789s are cheap on ebay etc. The larger 8-pin st7789s are also usable. The 8th pin does not need to be connected.
    Erik Piehl
    @Speccery
    Ok thanks, need to order some then.
    emard
    @emard
    @Speccery if you have any 240x240 and it is very likely that inside is some ST7789, then yes you can connect with wires. Some have backlight on by default, some need some pin connected to 3.3V or GND to enable backlight, that's usually all the difference between them. I have some demo on ulx3s-misc so you check will yours work: https://github.com/emard/ulx3s-misc/tree/master/examples/hex/lcd_st7789_hex/proj
    emard
    @emard
    Have you seen Goran's tweeter, new v3.1.4 is assembled and passed self-test. It has some new features and needs further testing https://publish.twitter.com/?query=https%3A%2F%2Ftwitter.com%2FRadionaOrg%2Fstatus%2F1332696542499770370&widget=Tweet
    Paul Ruiz
    @pnru_gitlab

    @Speccery Excellent stuff on the Icy99 project. I'm seriously impressed.

    On the QL: If the original sound on your QL still works, it might be worth to compare its output with that of the FPGA version. What we have today is based on reverse engineering the 8049 code and creating an equivalent direct circuit -- we are not sure the output is (roughly) correct.

    On the microdrives there is an outstanding work item: implementing writes to the disk. The current FPGA/ESP32 design allows for it, but it is currently unimplemented.

    Lawrie Griffiths
    @lawrie
    @pnru_gitlab @emard I am currently having a problem with the SDRAM driver for my Mac 128k/Mac Plus implementation. It is the same driver as I used for the QL and a lot of the code is similar to the QL. I noticed the problem when I switched on interrupts. The interrupt vector are copied to the SDRAM for the Mac. It writes $401A42 to address $64 for the level 1 interrupt. hen it is read back to execute the interrupt routines it gets the second 16-bit word at address $66, but the word at $64, which should be $40 is garbage. $66 might be correct because it is the last thing written. Any ideas what might be wrong.
    One thing I changed is the HDMI and SDRAM clock speeds to 125 rather than 135. I am trying putting that back. I am also trying different phase values. The code is here - https://github.com/lawrie/ulx3s_mac128/blob/main/src/mac128.v, but without the change that enables interrupts.
    emard
    @emard
    @lawrie first quick thing you can try is compare 12F and 85F. If it is fmax issue, 12F would perform better
    Lawrie Griffiths
    @lawrie
    I may be using too much BRAM for a 12F, but I can try a 45F.
    Lawrie Griffiths
    @lawrie
    It doesn't fit on the 12F, and seems to fail on the 45F.
    Lawrie Griffiths
    @lawrie
    Actually, the phase is irrelevant as @pnru_gitlab's SDRAM controller does not use the clock with the phase shift.
    Paul Ruiz
    @pnru_gitlab
    Can't investigate right now but it might be that the bus signals during an interrupt are different from during a normal memory read and that the sdram controller is not reacting correctly
    Lawrie Griffiths
    @lawrie
    I did think that the SDRAM had passed a memory test, so it could be something like that.
    Torfinn Ingolfsen
    @tingox_gitlab
    if I understand this correctly, ecpunpack can unpack a .bit bitstream to a .config file, which I then can repack with ecppack, for example to change idcode in the bitstream. Unfortunateky, it doesn't look like ecpunpack can do the same with a .svf format bitstream, or?
    emard
    @emard
    @tingox_gitlab No but bitstream.svf contains same bitstream reverse hex printed
    there is bit2svf.py also in trellis so you need own svf2bit, should be rather simple to make
    Simon Thornington
    @sthornington
    not that anyone here is interesting I am sure, but I got the (LiteX) wishbone debug bridge over US2 working with the main/master branch of the valenty USB core (DummyUsb endpoint). Seems reliable with cdc support enabled. But, SoC reset causes both US2 debugging as well as US1 JTAG programming to fail until unplug/replug
    Simon Thornington
    @sthornington
    random question - for ulx3s ECP5 85F, what FMAX should I be "happy with" ? Trying to recalibrate coming from Xilinx, it seems like 100MHz is quite hard to achieve ?
    emard
    @emard
    For well designed pipelined small cpu/soc 100MHz (f32c) should be possible but on the edge. For SDRAM driver 125 MHz should also easily always possible (overclocked ram test go to 180-200MHz). Small parts like video DVI shifter can do 350 MHz (overclocked work up to 600MHz, no guarantee all boards). Big soc like linux about 50MHz for CPU core, RAM 100MHz.
    Simon Thornington
    @sthornington
    Thanks, that helps
    Simon Thornington
    @sthornington
    I'm placing a small design on a big FPGA, are there any guides to encouraging nextpnr to place things closer together? for instance, I get much higher fmax targeting a 12F than an 85F