VOGONS


Tiny Vortex86-based DOS gaming PC - weeCee

Topic actions

Reply 20 of 952, by Mu0n

User metadata
Rank Member
Rank
Member
Mu0n wrote on 2021-05-19, 17:34:

Questions still orbiting in my mind:

1) what do we do with the EEPROM, must it be programmed ourselves, or the SoM just needs one and does its own thing with it? I have the very popular XGEcu TL-866II Plus everyone has, but I don't have adapters for SMT eeproms if I needed to link pins in specific places, and I haven't been warned from ICOP that I need to program one. My fear is that it's where you put the bios and you must find it, somehow.

Ah, you covered this in a previous video at this timestamp:

https://youtu.be/6cXdWMOl8QE?t=454

I'm trying to find that BIN file. As I understand it, the 24LC04B eeprom can store 500 bytes (4kbit). I found this vogons page with a bunch of driver files for DOS and various Windows:
http://vogonsdrivers.com/getfile.php?fileid=925
so far, every file I'm finding has more bytes. As per rasteri's video above, the end address is 0x01FF (so 511 bytes?), adding to the mystery.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 21 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

Is R9 6.8kOhm as specified in the details of that components in circuit maker, or 4.7kOhm as written on the label on the schematics? This is tied to U2 (the eeprom) so it must not have a super high impact if it's either...right?

R10, R19 are not populated (NP), was there a strong motivation to not use them?

Is my understanding correct in that the SoM will produce its own PC speaker square wave signal and send it eventually through a small filtering MIN (mono input) subcircuit plugged to pin 88 of the sound chip? There's no external hardware timer chip responsible for square waves? Gots to have the ol' pc speaker goodness.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 22 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

I found the answer for R19. It's a 4.7 kOhm resistor that either pulls down to ground (which is what was used and is labeled R20) or pulls up to 5V (which would be labeled R19). The relevant text in the datasheet is:

Mono Input
The mono input, MIN, is useful for mixing the
output of the "beeper" (timer chip), provided in
all PCs, with the rest of the audio signals. The
attenuation control allows 16 levels in -3dB
steps. In addition, a mute control is provided.
The attenuator is a single channel block with the
resulting signal sent to the output mixer where it
is mixed with the left and right outputs. Fig-
ure 29 illustrates a typical input circuit for the
Mono In. If MIN is driven from a CMOS gate,
the 4.7kΩ should be tied to AGND instead of
VA+. Although this input is described for a low-
quality beeper, the input is of the same
high-quality as all other analog inputs and may
be used for other purposes. At power-up, the
MIN line is connected directly to the MOUT pin
(with 9 dB of attenuation) allowing the initial
beeps, heard when the computer is initializing, to
pass through.

LWR_Recording.png

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 23 of 952, by rasteri

User metadata
Rank Oldbie
Rank
Oldbie

Haha you've been busy!

BIN file is included in the zip file of the PC104 soundcard. When I do a final release of the weeCee gerber files I'll include it with those.

Re: Open Source PC104 Soundcard

You should be able to find SMD adapters for the TL866 on ebay or aliexpress etc.

Reply 24 of 952, by weldum

User metadata
Rank Member
Rank
Member
weedeewee wrote on 2021-04-13, 12:06:
They don't say anything about USB drivers for any OS and dos only mentions network and possible serial interface. I bring it up […]
Show full quote
rasteri wrote on 2021-04-13, 11:32:
weedeewee wrote on 2021-04-13, 11:18:

Driver support for the different OSs will be annoying.

It runs pretty well on Win98- drivers are available for the graphics/sound/network/usb. 2000/XP supposedly work too but I don't have much interest in that really

Ndis2 drivers work on MS network client for DOS, I haven't tried the ODI or packet drivers yet though - http://www.dmp.com.tw/tech/vortex86dx/

They don't say anything about USB drivers for any OS and dos only mentions network and possible serial interface.
I bring it up because of this recent post getting bios mouse to work in 98se where no USB drivers were found and graphics drivers were also problematic.

Are you saying you are running Win98 with USB support? If so, you might want to let the other guy, weldum, know.

edit: watched build video, yes win98 has usb drivers apparently... 😀

thanks weedeewee, but i've concluded that the problem is in my unit, the bios crippled or something else, i've moved on to a different project currently.
also, this vortex86 is the DX version, mine is the first one without suffix. it's known that the dx version is capable of running windows embedded standard 2009 (a variant of XP) so far more compatible.

DT: R7-5800X3D/R5-3600/R3-1200/P-G5400/FX-6100/i3-3225/P-8400/D-900/K6-2_550
LT: C-N2840/A64-TK57/N2600/N455/N270/C-ULV353/PM-1.7/P4-2.6/P133
TC: Esther-1000/Esther-400/Vortex86-366
Others: Drean C64c/Czerweny Spectrum 48k/Talent MSX DPC200/M512K/MP475

Reply 25 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

I tried making my own based upon the datasheet for the sound chip CS4237B, I got to this, occupying about half of the 512 available bytes
EDIT - it's wildly different, so don't use this pic as a reference, people!

NqREdkW.png

When I freeze frame your video about your PC-104 sound card, we can see the TL866ii plus programmer app and it seems you have more occupied bytes and more strings, but they look like they're the assembly comments. I wonder if my file above is functionally the same. I'm also not sure how strings are dealt with in this system, are they simply wrapped with a 00 byte at the start and one more at the end? I'll guess I'll find out by scrounging in your PC-104 files. Thanks!

I'm having a blast pushing my electronics knowledge further. I just got a usb 'macroscope' today and I will definitely help me with the soldering which could start this week when I get my components tomorrow, my PCBs Thursday and my SOM Friday! Having gone through Ben Eater's 6502 breadboard kit (utterly fantastic series, very well done in a hands on, problem solving approach) definitely helped me feel secure every step of the way in this project.

Last edited by Mu0n on 2021-05-24, 19:43. Edited 1 time in total.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 26 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

Oh yeah, about aliexpress eeprom adapters, NAH, I'm definitely not waiting SEVERAL weeks for that stupid part. I'll try to insert wrapped wires on the legs of the eeprom and cross my fingers there's no too much interference and enough current.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 27 of 952, by weedeewee

User metadata
Rank l33t
Rank
l33t

It's one of those I2C eeproms, there's so many ways to get those written. I remember using the parallel port for it.
though you could also use a vga port, memory dimm, smbus port, some networkadapters, soundcards, FTDI (usb serial) or CH340 series, and ardiuno... indeed, no real need for a dedicated eeprom programmer.

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 28 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

I received my Mouser parts AND the pcbs yesterday.

With some misplaced bravado, I set out to extract (went well) and ROHS solder (went well) my first 0603 component, C35, around the voltage regulator. I've watched a healthy amount of tutorials about flux, was equipped with it, have some experience with it. Using 2 needlenose pliers (which I have) is slightly better than one and my Plugable USB microscope really helps at that refresh rate and resolution. The limiting factor by far is my soldering iron, which is this handheld thing with no station, with a tiny "temperature wheel" which by now, I understand only controls power approximately, not temperature and this thing, in retrospect, gets TOO HOT.

It was fine with my first 3 components, C35, C29 and U4 (completing the voltage regulation 'zone', a good, short starting point). But then, I wanted to go right in with the sound chip.

Methodology:
-injected no-clean flux first (mistake?)
-placed the chip, took a bit of time but I was VERY happy with my placement
-tried to get in a tiny amount of solder on the iron, problem strikes

That's when I realized that my ROHS 2.2% flux solder burns real fast on my iron. The tip becomes oxidixed within a minute, so alternating between microscope view placement and making sure the tip is tinned is stressful. The heat never drops and I suspect it's a good 50-100 C above what the small wheel control says. I'm constantly having to use the brass brush thing, re-tin, etc. After a while, I had a bad spiral of adding more flux, trying to drag solder across the chip's legs. I could not replicate the light touch of the iron and see solder sort itself out at the legs. The temperature situation is just catastrophically out of control.

The solder created a full side bridge at the top of the chip's legs rather than at their end on the pad. Using a solder wick to remove the excess (a step I've done countless times to great success with bigger components, managing its internal flux, adding some when needed + adding solder to help suck old solder, etc) did not work. It's hard getting a proper view of the legs' 3D structure and aiming exactly at the right place while you try to both position the wick and iron tip. To top it all off, one of the adjacent pads got lost (one of C1's pads).

My iron (well, the 2 of them) come from kits such as these: https://www.amazon.ca/gp/product/B07H9MJRLZ/
They've been useful to repair vintage macintosh analog and logic boards. They were fine when I replaced electrolytic SMT caps with tantalum ones, but they were much bigger packages than what is attempted here.

So...it's time to order lead solder (ROHS is for a higher level) and a proper station then. Thankfully this will serve in many more projects over the years.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 29 of 952, by SScorpio

User metadata
Rank Member
Rank
Member
Mu0n wrote on 2021-05-26, 14:11:

That's when I realized that my ROHS 2.2% flux solder burns real fast on my iron. The tip becomes oxidixed within a minute, so alternating between microscope view placement and making sure the tip is tinned is stressful. The heat never drops and I suspect it's a good 50-100 C above what the small wheel control says. I'm constantly having to use the brass brush thing, re-tin, etc. After a while, I had a bad spiral of adding more flux, trying to drag solder across the chip's legs. I could not replicate the light touch of the iron and see solder sort itself out at the legs. The temperature situation is just catastrophically out of control.

The solder created a full side bridge at the top of the chip's legs rather than at their end on the pad. Using a solder wick to remove the excess (a step I've done countless times to great success with bigger components, managing its internal flux, adding some when needed + adding solder to help suck old solder, etc) did not work. It's hard getting a proper view of the legs' 3D structure and aiming exactly at the right place while you try to both position the wick and iron tip. To top it all off, one of the adjacent pads got lost (one of C1's pads)

Do you have a reel of solder wire you are dragging across the legs of the chip? You should be applying flux to the pads and then adding solder to the pads themselves. You then place the chip and do the drag and let the heat melt the solder. You could also use solder paste which melts about as fast as the flux, the joint should be formed immediately not requiring minutes per side of the chip.

Reply 30 of 952, by Mu0n

User metadata
Rank Member
Rank
Member
SScorpio wrote on 2021-05-26, 14:40:

Do you have a reel of solder wire you are dragging across the legs of the chip? You should be applying flux to the pads and then adding solder to the pads themselves. You then place the chip and do the drag and let the heat melt the solder. You could also use solder paste which melts about as fast as the flux, the joint should be formed immediately not requiring minutes per side of the chip.

I do. ROHS solder reels is what I have in the highest quantity. I also have a non-descript feeding tube that might or might not be leaded solder, but it's pretty thick for this job. I will definitely get better solder to reduce the difficulty of the job.
I will attempt the solder-on-pad first, I've seen it done and I wonder why I didn't remember it. Thanks!

I thought about going the solder paste route (type 5) but I thought I could wing it with the flux I already have.
I have 7 remaining spare boards. It's not impossible I've destroyed the sound chip, but I have 2 spare. My possible routes are:

-Keep this board + cross fingers the sound chip is ok + create a new pad for C1's missing one with a small wire and scratched mask in the adjacent trace + remove the sound chip + clean the pads + resolder sound chip

-Use a new board + transfer C29,C35 and U2 from the old to the new + use a new sound chip

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 31 of 952, by rasteri

User metadata
Rank Oldbie
Rank
Oldbie

I've soldered QFPs with leadfree before but it's not ideal. It's all about adding loads of flux and encouraging the solder to go to separate pins by heating several pins at once.

Reply 32 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

I went on and calculated my total costs in the past months for the project. I set out to build 2 units (highlighted below in the chart) to save on costs. Keep in mind these purchases:

-SOM from ICOP, pcb mating pins for SOM, Hammond case, CS4237B-KQ audio chip, ps2 split cable, microsd 32 GB, PCBs, S2 WaveBlaster (for 1 unit, not both), electronic components including the above-10 bucks ethernet port that stings

I did not count:
-Soldering station with controlled temp (it was time!), usb microscope, non-ROHS solder, flux syringe (these are long due purchases that will serve me well in the future):
-PS2 mouse and PS2 keyboard (I already have one of each)
-USB power adapter (I already have a few)
-Db15 joysticks (I have 4!)
-DB15 midi/joystick split cable (I made one!)
-Roland MT-32 and Roland SC88-ST (silent flex on these...)

Caveats:
-I live in Canada so shipping to me can be much higher than other parts of the world (especially compared to the US) and we get hit by duties quite severely sometimes: example: each sound chip was 9.87 CAD, but the shipping was 35 CAD and the duties an additional 21 CAD.
-It could be possible to shop around and reduce costs at the expense of waiting more. I did not feel I had more time to invest in that aspect.
-Some shipping + duties might scale up at some point, since it's hard to predict, I assumed fixed costs.
-I assumed a 1.21 CAD to 1.00 USD ratio on May 27th 2021.
-Commas are used as a decimal part separator in French, that's the language used in my installation of Excel.

Cost analysis according to the number of units you want built:
IEsJvwG.png

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 34 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

I finished the assembly late last night and... It wouldn't display anything. The blue LED on the SOM lit up, I was finding 4.73 V everywhere I tested that should receive it, etc.

This morning, I started testing my 2nd SOM, same behavior. I then realized I should make sure the top of the mating SOM pins that go into it are 100% free of rogue solder blotches and there were 2 with tiny amounts of it. I don't have proper length screws to secure the SOM to the board so I ended up using the ones that came with the Hammond case, they are just barely long enough if you screw from the bottom. It then booted with video signal... in amber (not 100% secure VGA cable).

After properly pushing it all the way in the case and screwing both plates, it boots fine now. I'm now onto the next problem: no sound. I do hear a slight pop when I boot it up, but no PC Speaker sound (I tested Commander Keen 1). So far, I've tried the CS4232_4237 drivers from here on Vogons, these 2 versions : the latest DOS release and the very next one after. I'm just starting to figure out what all the addresses (which I can attempt to set in a setup utility in the latest dos version) should be so pointers would be greatly appreciated here. I'm running DOS 6.22 and will later mess with Win98SE. The CWBinit.exe program states it can't find the Crystal chip.

Still, for a novice at soldering that had done around 15 SMT tantalum caps with packages bigger than 1026. Undertaking this project: generating Gerber files, back and forth with pcbway, dealing with 0603 and 1206 and QFP with a total of 400-500 solder points and that it freaking works as much as it is right now, this is a gigantic win for me.

Edit- I'll give unisound a shot UNISOUND - Universal ISA PnP Sound Card Driver for DOS v0.81b

Last edited by Mu0n on 2021-06-06, 14:31. Edited 1 time in total.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 35 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

Unisound gives: No PnP sound card detected

CWD-v180 (1997 dos latest): cs4232c.exe reports 'Cannot find Crystal chip' and cwdiag reports 'Fail' on SoundBlaster, MPU401, FM/OPL3 and Windows Sound System.

CWD-v250 (1997 w+wavestrm): cwbinit reports 'Cannot find Crystal chip'

I've even reflowed ALL my chip's legs with tons of flux and a tiny, tiny bit of solder on 2 sides, I think it looks good but an expert might disagree. I could post very close up pics of it.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 36 of 952, by rasteri

User metadata
Rank Oldbie
Rank
Oldbie

Heh excellent EEPROM work 😀

The fact that unisound can't detect anything means it can't see the crystal chip on the ISA bus at all, if the I2C EEPROM was just dead or anything it would still detect it (but be unable to assign it an IRQ etc).

Do you have a scope? You could check the crystal is at least giving out a 16mhz clock (set your probe to 10x).

Also which rev of the board is this? possibly it still needs the bodge?

A picture of the board might show us anything you might have missed.

Reply 37 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

Scope: at work only. I could go back this week.

Rev: the board I used and had made is an "early v0.3". You made several changes (but are still labeling it as v0.3 as of right now) after I cloned the project. Here's what circuitmaker detects as differences

YLcxBCh.png

And here's a screencap of the underside, no top overlay, no top 3d shown for visual clarity, MY VERSION

rhOtKh2.png

YOUR VERSION

AvCPuGU.png

Here are pics of my board (ignore the tape on the sdcard, it holds it together as the autoeject mechanism became insta borked on it, gotta remove the spring). When I freeze frame your video (linked in the first post), we see that you have R10 populated, but your schematics lists it as NP. I'm not 100% convinced which way it has to be by reading the datasheet on the crystal chip.

hhWbYxW.jpeg
7HCIYqW.jpeg

Here are close shots of the Cs4237B-KQ pins

Pins 1-25
S2LIlgF.jpeg
Pins 25-50
OZSh3R9.jpeg
Pins 51-75
PYBGezz.jpg
Pins 76-100
DyKnaqV.jpeg

Closest shot I can do, near pin 20:
nhXBR8n.jpeg

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 38 of 952, by Mu0n

User metadata
Rank Member
Rank
Member

Changed the crystal for my spare one, same behavior.

I'm wondering about some BIOS settings:

PCIPnP (probably doesn't matter): Plug and Play OS: No?

Southbridge/ISA clock speed: 8 MHz (default)?
Southbridge/DMA speed: 4 MHz (default)?

What bodge wires did you need to do in your earlier revs and why? One looked (in your video) like it was around pin 89 (left side of the sound chip) to make it link to ground, but it did not strike me as an obvious gain from the datasheet.

I'll check my isa lines from the sound chip, from the SOM once more...

Last edited by Mu0n on 2021-06-08, 15:33. Edited 1 time in total.

1Bit Fever Dreams: https://www.youtube.com/channel/UC9YYXWX1SxBhh1YB-feIPPw
AnyBit Fever Dreams: https://www.youtube.com/channel/UCIUn0Dp6PM8DBTF-5g0nvcw

Reply 39 of 952, by Tiido

User metadata
Rank l33t
Rank
l33t

There is a chance tha there must be some configuration done on the chipset to make the ISA bus see all addresses. On one PC104 module using AMD elan520 chipset I had to make a little utility to conf ISA to see more Io ranges or many things didn't work. I recall that by default it only forwarded 2xx and 3xx but not more, so I changed the default conf to get a reasonable coverage and then many more things began to work.

T-04YBSC, a new YMF71x based sound card & Official VOGONS thread about it
Newly made 4MB 60ns 30pin SIMMs ~
mida sa loed ? nagunii aru ei saa 😜