VOGONS


Reply 1220 of 1316, by Shreddoc

User metadata
Rank Oldbie
Rank
Oldbie

Over the years, the majority of issues observed with PicoGUS self-builds has turned out to involve careful attention to U2/U3/U4. Even people with the skill + experience still have a tendency to rush it, and miss a small detail at first. I know I have, on more than one occasion.

Supporter of PicoGUS, PicoMEM, mt32-pi, WavetablePi, Throttle Blaster, Voltage Blaster, GBS-Control, GP2040-CE, RetroNAS.

Reply 1221 of 1316, by snipe3687

User metadata
Rank Member
Rank
Member
Shreddoc wrote on 2024-11-14, 21:54:

Over the years, the majority of issues observed with PicoGUS self-builds has turned out to involve careful attention to U2/U3/U4. Even people with the skill + experience still have a tendency to rush it, and miss a small detail at first. I know I have, on more than one occasion.

Interesting, could you elaborate on that? I originally built 5 of them and only 1 actually worked. the others had various issues. some I could flash them but couldn't get them to detect, some would detect but then would play garbled audio or crash the system and 1 worked fine but would crash in GUS mode. these were the chip down version. I built 5 of the module versions too and they were totally fine. you said U2-U4 specifically. what about those chips do you feel are problematic for self-installers? I'm always looking at ways to better my skills and I'd love to sort the other cards out at some point!

Reply 1222 of 1316, by Shreddoc

User metadata
Rank Oldbie
Rank
Oldbie
snipe3687 wrote on 2024-11-23, 21:27:
Shreddoc wrote on 2024-11-14, 21:54:

Over the years, the majority of issues observed with PicoGUS self-builds has turned out to involve careful attention to U2/U3/U4. Even people with the skill + experience still have a tendency to rush it, and miss a small detail at first. I know I have, on more than one occasion.

Interesting, could you elaborate on that? I originally built 5 of them and only 1 actually worked. the others had various issues. some I could flash them but couldn't get them to detect, some would detect but then would play garbled audio or crash the system and 1 worked fine but would crash in GUS mode. these were the chip down version. I built 5 of the module versions too and they were totally fine. you said U2-U4 specifically. what about those chips do you feel are problematic for self-installers? I'm always looking at ways to better my skills and I'd love to sort the other cards out at some point!

Take my comments as the opinions of a fellow amateur solderer, I'm no senior authority on the science. I've handmade about 25 x PicoGUS 1.x's over the past 2-3 years (time flies!). It sounds a lot but probably averages out to mere minutes per day. If I was spending even 1 consistent hour per day doing this, then it would different.*

But this is not my day job, so as a hobbyist, imperfections happen. At least 20 of those 25 builds required little fixes here and there, upon inspection after first attempt. Those fixes almost always included some aspect of U2/U3/U4/DAC, because those IC's are the smallest in pitch and have many pins clustered together.

The build experiences that others have posted in this thread over time seem to have mirrored that pattern (or so my vague meat-memory record says). And a proportion of those experiences involve surprise that it happened to them. That's no aspersion, it's just how it goes. For me too.

On a technical basis, there's the sheer size/pitch. But there are also other factors and perils. The providence of the parts (i.e. consistent quality providers). Technique. The flux and solder quantity and quality and temperature. The tip used. How much of a hurry the operator is in. The IC legs. The pads themselves. Damage inadvertently done during the course of multiple rework attempts is its own minefield. Read the last sentence again, because it's bitten me a few times. And of course, the classic bridges, which can come in all shapes, sizes and positions, from big obvious ones, to the occasional one so incredibly cunningly hidden that you can only award it a wry grin of acknowledgement when you finally spot it after desperately renting an electron microscope at $5000/hour (eliminate slight exaggeration at your discretion).

*There's a great variance in the equipment and consumables people use, too. It's no coincidence that quite a few people approach the PicoGUS and similar great open source hardware builds as learning experiences. That often means entry level equipment.

--

P.S. Regarding the Integrated-DAC version of the 1.2 - the DAC IC in particular is a step-up in technical challenge, from the DAC Module version. Simply because it's not only at TSSOP scale, but it's also closely surrounded by a cavalcade of 0805's. Accordingly, I've found it a very good idea to solder the DAC chip relatively early in the build, and give it it's full microscopic inspection, continuity test, and rework prior to surrounding it.

Supporter of PicoGUS, PicoMEM, mt32-pi, WavetablePi, Throttle Blaster, Voltage Blaster, GBS-Control, GP2040-CE, RetroNAS.

Reply 1223 of 1316, by SaxxonPike

User metadata
Rank Member
Rank
Member

I had a bit of a scare recently with this card. I love it so much, and was really sad to see that something arced through it and sparked a small flame.

in-system.jpg
Filename
in-system.jpg
File size
513.33 KiB
Views
1177 views
File license
Public domain

I'm still trying to piece together how this could happen, but my partner looked at it under his scope and saw one of the traces was destroyed.

scope-1.jpg
Filename
scope-1.jpg
File size
69.54 KiB
Views
1177 views
File license
Public domain
scope-2.jpg
Filename
scope-2.jpg
File size
117.84 KiB
Views
1177 views
File license
Public domain

I was using this with the McCake Waveblaster attachment when it happened. Amusingly, the card's main functions continued to work (all hail the blinkenlight) but I'm probably not going to keep this connected in a system until we can investigate further.

I don't have skills adequate to build one of these myself, nor repair it. Not sure how to proceed.

Sound device guides:
Sound Blaster
Aztech
OPL3-SA

Reply 1224 of 1316, by Pickle

User metadata
Rank Member
Rank
Member

did the mcake survive?, seems the logical thing is that is arced from it to the picogus.
maybe its too much for the pico to power a rpi compute module? I run a rpi zero 2 off the waveblaster and had no issues.

Reply 1225 of 1316, by SaxxonPike

User metadata
Rank Member
Rank
Member
Pickle wrote on 2024-11-24, 00:56:

did the mcake survive?, seems the logical thing is that is arced from it to the picogus.
maybe its too much for the pico to power a rpi compute module? I run a rpi zero 2 off the waveblaster and had no issues.

The McCake is, I believe, fully powered by a 4-pin floppy drive power connection (there isn't anything I could find that suggests that it is powered from the host otherwise anyway.) It does make me wonder if this fed back into the PicoGUS somehow.

Sound device guides:
Sound Blaster
Aztech
OPL3-SA

Reply 1226 of 1316, by polpo

User metadata
Rank Member
Rank
Member
SaxxonPike wrote on 2024-11-23, 23:58:

I had a bit of a scare recently with this card. I love it so much, and was really sad to see that something arced through it and sparked a small flame.

Yikes! It looks like the digital ground going to the WT header took a lot of current and burned out. This is the second time I've seen an issue with a really bad short with the PicoGUS and a McCake on the wavetable header. The first I saw happened when someone plugged their McCake in "upside down" which caused +12V to short to ground through the wavetable header. They had a modern 900W ATX power supply, and unlike oldschool PSUs, modern ones can provide an absolutely enormous amount of current on the +12V rail, so a short to ground can do just what you saw. So a couple questions: are you using a modern PSU? And how did you have the McCake mounted? It should be like this as shown in root42's video on the PicoGUS:

picogus-mccake.jpg
Filename
picogus-mccake.jpg
File size
454.71 KiB
Views
1130 views
File license
Fair use/fair dealing exception

I've reached out to Serge from Serdaco to see if he has any ideas for how this could happen and how things are connected on the McCake side of the wavetable header. This is how things are on the PicoGUS side:

picogus-wtheader.png
Filename
picogus-wtheader.png
File size
203.76 KiB
Views
1130 views
File license
GPL-2.0-or-later

Reply 1227 of 1316, by SaxxonPike

User metadata
Rank Member
Rank
Member
polpo wrote on 2024-11-24, 03:18:

So a couple questions: are you using a modern PSU? And how did you have the McCake mounted? It should be like this as shown in root42's video on the PicoGUS:

I am using a Sparkle branded 300W PSU, it's old enough to have an AT power connector on it and has honestly done a better job at keeping systems older than Pentium 4 stable than modern PSUs.

I have indeed connected the McCake as indicated also - with the 4-pin power toward the front of the case and the front-panel connector toward the rear bracket.

Sound device guides:
Sound Blaster
Aztech
OPL3-SA

Reply 1228 of 1316, by polpo

User metadata
Rank Member
Rank
Member
SaxxonPike wrote on 2024-11-24, 04:33:
polpo wrote on 2024-11-24, 03:18:

So a couple questions: are you using a modern PSU? And how did you have the McCake mounted? It should be like this as shown in root42's video on the PicoGUS:

I am using a Sparkle branded 300W PSU, it's old enough to have an AT power connector on it and has honestly done a better job at keeping systems older than Pentium 4 stable than modern PSUs.

I have indeed connected the McCake as indicated also - with the 4-pin power toward the front of the case and the front-panel connector toward the rear bracket.

Great, that's all good. Hopefully Serdaco can help provide some insight, becuase I'm stumped so far as to how that could have happened.

Reply 1229 of 1316, by myne

User metadata
Rank Oldbie
Rank
Oldbie

Sparkle had a terrible reputation when they were brand new.

25 years later...wow!

Any dead caps? They tend to fail short.

I built:
Convert old ASUS ASC boardviews to KICAD PCB!
Re: A comprehensive guide to install and play MechWarrior 2 on new versions on Windows.
Dos+Windows 3.11 auto-install iso template (for vmware)
Script to backup Win9x\ME drivers from a working install
Re: The thing no one asked for: KICAD 440bx reference schematic

Reply 1230 of 1316, by polpo

User metadata
Rank Member
Rank
Member

Thinking a bit more, it might not be a large amount of current on that digital ground (GND) trace itself but arcing between it and analog ground (AGND), which is right along side it on that layer. So something to check: in any normal circumstance, GND and AGND will be at the same potential. GND and AGND are tied together by a 0-ohm resistor or ferrite bead marked FB3 on the board next to the DAC chip (in old builds it's a ferrite bead, in new ones a 0-ohm resistor but in either case it'll have continuity). Does that component have continuity on a multimeter? If there was a large potential between GND and AGND, that resistor almost certainly would have popped before the arcing happened, so if there's no continuity that would point to that situation.

Now how could there be a large potential between GND and AGND? Just thinking out loud for possibilities: the ground of your audio equipment is somehow offset from normal ground, or something weird happened on the McCake, or something else I can't think of happened.

If FB3 has continuity, that points more towards there being a short between one of the voltage rails and digital ground on the McCake somehow.

Reply 1231 of 1316, by SaxxonPike

User metadata
Rank Member
Rank
Member
polpo wrote on 2024-11-24, 05:55:

Thinking a bit more, it might not be a large amount of current on that digital ground (GND) trace itself but arcing between it and analog ground (AGND), which is right along side it on that layer. So something to check: in any normal circumstance, GND and AGND will be at the same potential. GND and AGND are tied together by a 0-ohm resistor or ferrite bead marked FB3 on the board next to the DAC chip (in old builds it's a ferrite bead, in new ones a 0-ohm resistor but in either case it'll have continuity). Does that component have continuity on a multimeter? If there was a large potential between GND and AGND, that resistor almost certainly would have popped before the arcing happened, so if there's no continuity that would point to that situation.

Now how could there be a large potential between GND and AGND? Just thinking out loud for possibilities: the ground of your audio equipment is somehow offset from normal ground, or something weird happened on the McCake, or something else I can't think of happened.

If FB3 has continuity, that points more towards there being a short between one of the voltage rails and digital ground on the McCake somehow.

I'll check continuity now. Edit: FB3 has continuity.

When this happened, I only had keyboard+VGA connected. There wouldn't be a ground connection to external audio equipment in this case.

Sound device guides:
Sound Blaster
Aztech
OPL3-SA

Reply 1232 of 1316, by myne

User metadata
Rank Oldbie
Rank
Oldbie

Fbs do have continuity though...?
They're supposed to filter high frequency noise while allowing LF through

I only know this because one died on a voodoo 2 and a bodge made it work pending digikey

Is it possible that that ground trace is simply routed too close to the cut edge and a 5/12v line on another layer and the cutting process caused enough disruption to both traces to create a time bomb?

It does look like it's right on the edge.

I built:
Convert old ASUS ASC boardviews to KICAD PCB!
Re: A comprehensive guide to install and play MechWarrior 2 on new versions on Windows.
Dos+Windows 3.11 auto-install iso template (for vmware)
Script to backup Win9x\ME drivers from a working install
Re: The thing no one asked for: KICAD 440bx reference schematic

Reply 1233 of 1316, by SaxxonPike

User metadata
Rank Member
Rank
Member

I wonder if perhaps there was something happening with the 4-pin power cable that fed the McCake. Is it possible it would ground out through the PicoGUS if the ground path back through the 4-pin was lost somehow?

Edit: amusingly, the McCake seems unharmed after powering it on by itself.

Sound device guides:
Sound Blaster
Aztech
OPL3-SA

Reply 1234 of 1316, by rasz_pl

User metadata
Rank l33t
Rank
l33t
polpo wrote on 2024-11-24, 03:18:

This is how things are on the PicoGUS side:
picogus-wtheader.png

I never looked at picogus PCB. That layout needs some love. Route analog/digital signals over solid ground instead of kludges like physically separate analog/digital grounds - that stuff doesnt work in real life. Afaik this practice hails from the times of 2 layer boards and relying on hunches and bad intuition. Here from companies that knows what they are talking about
https://resources.pcb.cadence.com/blog/2021-s … l-ground-planes
https://resources.altium.com/p/how-to-use-a-s … ound-connection
Signals will always use shortest path for ground return EDIT because I wasnt clear: shortest from the track sending the signal, and paradoxically for UART_TX its going right to analog ground thru the fuse and then to rp2040. Left audio channel going over islands of no ground between pins 15-22 🙁 Same for Uart TX going over broken ground.

polpo wrote on 2024-11-24, 05:55:

digital ground (GND) trace itself but arcing between it and analog ground (AGND)

for that to happen someone would have to be running a welder or something 😀

polpo wrote on 2024-11-24, 03:18:

and digital ground on the McCake somehow.

I dont have McCake and its not an open design so cant check, but I bet all grounds are linked together (as they should be).

https://github.com/raszpl/FIC-486-GAC-2-Cache-Module for AT&T Globalyst
https://github.com/raszpl/386RC-16 memory board
https://github.com/raszpl/440BX Reference Design adapted to Kicad
https://github.com/raszpl/Zenith_ZBIOS MFM-300 Monitor

Reply 1235 of 1316, by esher

User metadata
Rank Newbie
Rank
Newbie
Shreddoc wrote on 2024-11-23, 23:08:
Take my comments as the opinions of a fellow amateur solderer, I'm no senior authority on the science. I've handmade about 25 x […]
Show full quote
snipe3687 wrote on 2024-11-23, 21:27:
Shreddoc wrote on 2024-11-14, 21:54:

Over the years, the majority of issues observed with PicoGUS self-builds has turned out to involve careful attention to U2/U3/U4. Even people with the skill + experience still have a tendency to rush it, and miss a small detail at first. I know I have, on more than one occasion.

Interesting, could you elaborate on that? I originally built 5 of them and only 1 actually worked. the others had various issues. some I could flash them but couldn't get them to detect, some would detect but then would play garbled audio or crash the system and 1 worked fine but would crash in GUS mode. these were the chip down version. I built 5 of the module versions too and they were totally fine. you said U2-U4 specifically. what about those chips do you feel are problematic for self-installers? I'm always looking at ways to better my skills and I'd love to sort the other cards out at some point!

Take my comments as the opinions of a fellow amateur solderer, I'm no senior authority on the science. I've handmade about 25 x PicoGUS 1.x's over the past 2-3 years (time flies!). It sounds a lot but probably averages out to mere minutes per day. If I was spending even 1 consistent hour per day doing this, then it would different.*

But this is not my day job, so as a hobbyist, imperfections happen. At least 20 of those 25 builds required little fixes here and there, upon inspection after first attempt. Those fixes almost always included some aspect of U2/U3/U4/DAC, because those IC's are the smallest in pitch and have many pins clustered together.

The build experiences that others have posted in this thread over time seem to have mirrored that pattern (or so my vague meat-memory record says). And a proportion of those experiences involve surprise that it happened to them. That's no aspersion, it's just how it goes. For me too.

On a technical basis, there's the sheer size/pitch. But there are also other factors and perils. The providence of the parts (i.e. consistent quality providers). Technique. The flux and solder quantity and quality and temperature. The tip used. How much of a hurry the operator is in. The IC legs. The pads themselves. Damage inadvertently done during the course of multiple rework attempts is its own minefield. Read the last sentence again, because it's bitten me a few times. And of course, the classic bridges, which can come in all shapes, sizes and positions, from big obvious ones, to the occasional one so incredibly cunningly hidden that you can only award it a wry grin of acknowledgement when you finally spot it after desperately renting an electron microscope at $5000/hour (eliminate slight exaggeration at your discretion).

*There's a great variance in the equipment and consumables people use, too. It's no coincidence that quite a few people approach the PicoGUS and similar great open source hardware builds as learning experiences. That often means entry level equipment.

--

P.S. Regarding the Integrated-DAC version of the 1.2 - the DAC IC in particular is a step-up in technical challenge, from the DAC Module version. Simply because it's not only at TSSOP scale, but it's also closely surrounded by a cavalcade of 0805's. Accordingly, I've found it a very good idea to solder the DAC chip relatively early in the build, and give it it's full microscopic inspection, continuity test, and rework prior to surrounding it.

I am partially working as electronic engineer, i mostly work with Autotronik B385 PNP machine over 5 years, so i meet the solder problems every production cycle. Most of them are connected with solder paste, pcb design (thermal relief) and thermal profile of the heating machine. If the thermal relief is not in the rules, we will get a lot of tombstones. If the paste is not well mixed up, we will get a lot of shorts on small step pins and/or unsolder points. The air flow in heating machine can also be the reason of some parts changing their positions (tombstones, shorts and so on). This things after inpection i am fixing manually with solder rework gun, flux and solder iron. But with home projects, like PicoGUS, i am working at home, i have very little time on hobbies, so the work "in hurry mode " results in such typical mistakes.

Reply 1236 of 1316, by Shreddoc

User metadata
Rank Oldbie
Rank
Oldbie
esher wrote on 2024-11-28, 08:41:
Shreddoc wrote on 2024-11-23, 23:08:
Take my comments as the opinions of a fellow amateur solderer, I'm no senior authority on the science. I've handmade about 25 x […]
Show full quote
snipe3687 wrote on 2024-11-23, 21:27:

Interesting, could you elaborate on that? I originally built 5 of them and only 1 actually worked. the others had various issues. some I could flash them but couldn't get them to detect, some would detect but then would play garbled audio or crash the system and 1 worked fine but would crash in GUS mode. these were the chip down version. I built 5 of the module versions too and they were totally fine. you said U2-U4 specifically. what about those chips do you feel are problematic for self-installers? I'm always looking at ways to better my skills and I'd love to sort the other cards out at some point!

Take my comments as the opinions of a fellow amateur solderer, I'm no senior authority on the science. I've handmade about 25 x PicoGUS 1.x's over the past 2-3 years (time flies!). It sounds a lot but probably averages out to mere minutes per day. If I was spending even 1 consistent hour per day doing this, then it would different.*

But this is not my day job, so as a hobbyist, imperfections happen. At least 20 of those 25 builds required little fixes here and there, upon inspection after first attempt. Those fixes almost always included some aspect of U2/U3/U4/DAC, because those IC's are the smallest in pitch and have many pins clustered together.

The build experiences that others have posted in this thread over time seem to have mirrored that pattern (or so my vague meat-memory record says). And a proportion of those experiences involve surprise that it happened to them. That's no aspersion, it's just how it goes. For me too.

On a technical basis, there's the sheer size/pitch. But there are also other factors and perils. The providence of the parts (i.e. consistent quality providers). Technique. The flux and solder quantity and quality and temperature. The tip used. How much of a hurry the operator is in. The IC legs. The pads themselves. Damage inadvertently done during the course of multiple rework attempts is its own minefield. Read the last sentence again, because it's bitten me a few times. And of course, the classic bridges, which can come in all shapes, sizes and positions, from big obvious ones, to the occasional one so incredibly cunningly hidden that you can only award it a wry grin of acknowledgement when you finally spot it after desperately renting an electron microscope at $5000/hour (eliminate slight exaggeration at your discretion).

*There's a great variance in the equipment and consumables people use, too. It's no coincidence that quite a few people approach the PicoGUS and similar great open source hardware builds as learning experiences. That often means entry level equipment.

--

P.S. Regarding the Integrated-DAC version of the 1.2 - the DAC IC in particular is a step-up in technical challenge, from the DAC Module version. Simply because it's not only at TSSOP scale, but it's also closely surrounded by a cavalcade of 0805's. Accordingly, I've found it a very good idea to solder the DAC chip relatively early in the build, and give it it's full microscopic inspection, continuity test, and rework prior to surrounding it.

I am partially working as electronic engineer, i mostly work with Autotronik B385 PNP machine over 5 years, so i meet the solder problems every production cycle. Most of them are connected with solder paste, pcb design (thermal relief) and thermal profile of the heating machine. If the thermal relief is not in the rules, we will get a lot of tombstones. If the paste is not well mixed up, we will get a lot of shorts on small step pins and/or unsolder points. The air flow in heating machine can also be the reason of some parts changing their positions (tombstones, shorts and so on). This things after inpection i am fixing manually with solder rework gun, flux and solder iron. But with home projects, like PicoGUS, i am working at home, i have very little time on hobbies, so the work "in hurry mode " results in such typical mistakes.

Interesting, this shows how a professional solderer and a hobbyist both face some of the same challenges, when doing these manual home projects.

My experience level is the opposite. No professional experience of soldering. All hobby. So, over the years, when I want to experience a new part of the skill - like the paste, thermal profiles, heat plate, hot air, inspection, etc - then I must do it myself at home. And so after some time, home is now where my best (only) equipment and skills are, this serves me well for projects like this. But only after much learning and mistakes. Which still happen, I think this will be the one factor common to the entire timeline haha.

Supporter of PicoGUS, PicoMEM, mt32-pi, WavetablePi, Throttle Blaster, Voltage Blaster, GBS-Control, GP2040-CE, RetroNAS.

Reply 1238 of 1316, by FreddyV

User metadata
Rank Oldbie
Rank
Oldbie
zuldan wrote on 2024-12-02, 11:52:

Is there a way to have a USB mouse and MPU-401 mode with full intelligent mode at the same time?

Hi,

There is a list of games needing the Interrupt in the MPU4101 mode :
https://github.com/bjt42/softmpu/wiki/Compati … gent-Mode-Games

This can't be used with mouse, as the mouse also need the interrupt.

Reply 1239 of 1316, by digger

User metadata
Rank Oldbie
Rank
Oldbie
FreddyV wrote on 2024-12-02, 14:35:
Hi, […]
Show full quote
zuldan wrote on 2024-12-02, 11:52:

Is there a way to have a USB mouse and MPU-401 mode with full intelligent mode at the same time?

Hi,

There is a list of games needing the Interrupt in the MPU4101 mode :
https://github.com/bjt42/softmpu/wiki/Compati … gent-Mode-Games

This can't be used with mouse, as the mouse also need the interrupt.

If I'm not mistaken, multiple devices on the same ISA card could share an IRQ, but it would have to be handled properly both device-side and driver-side. Perhaps CuteMouse could be enhanced to support something like that on this card?

Alternatively, again if I'm not mistaken, one of the Microsoft bus mouse standards (perhaps the later InPort standard, but it could also have been the older standard before it) supported an optional IRQ-less operating mode for mice. My guess would be that it worked through some kind of polling mechanism, piggybacking on the timer interrupt. Of course that would probably come at the price of some performance, particularly on slower systems. This too would require explicit support in the mouse driver.