site  contact  history  index

NVIDIA 470 and 525 SFSs created

June 25, 2024 — BarryK

I have created these two, installatable with SFSget, via the desktop "pkg" icon:

nvidia_6.6.32-470.256.02_amd64.sfs
nvidia_6.6.32-525.147.05_amd64.sfs

Was unable to compile the 390.* source, though it did compile with the 5.15.x kernel.

Also unable to compile the 555.* source.

Information about which SFS to use can be found here:

https://www.nvidia.com/en-us/drivers/unix/

These NVIDIA SFSs should just work. Install to the main filesystem, reboot, that's it.   

Tags: easy

Trike simplified steering linkage assembled

June 24, 2024 — BarryK

A couple of days ago, I posted about a simplified steering linkage that I can put together quickly:

Today manufactured the wheel-knuckle lever extensions. Used 50x3mm cross-section mild steel. Here is a sketch:

img1

Then assembled on the trike:

img2

As stated before, there are lots of compromises with this simple linkage.

Fleshing out a bit more how I calculated the amount that the inner wheel was turning a bit too much...

The SolveSpace diagram shows two concentric circles, going through each wheel. The diagram also shows the angle of each wheel, relative to the trike frame. 90 degrees would be the wheels pointing straight ahead. If the diagram is adjusted so that the outer wheel is following the outer radius, we can subtract the two wheel angles to see how much the inner wheel has turned in or out relative to the outer wheel.

There is a formula for calculating the ideal angle for each wheel: A=atan(L/R)

Where L is the length from the front wheels to the hub of the rear wheel. In my case 1200mm. R is the radius of the circle while cornering. A is the angle.

For the example in SolveSpace of inner circle radius 3138mm and outer radius 3583mm, calculating the two required wheel angles:

Aouter=atan(1200/3583)
Ainner=atan(1200/3138)
Aouter=18.52°, Ainner=20.93°
Difference=2.41°

In the SolveSpace diagram, the difference is 3.31°, so the inner wheel has turned 0.9° more than the Ackermann ideal.

EDIT 2024-06-25:
I assembled the arms, so as to evaluate basic steering functionality:

img2

Have made a list of required modifications, so tonight will completely dis-assemble the trike. Rebuild will probably be about a month from now. Intend for it then to be rideable and will use lock-nuts and thread-lock.      

Tags: light

EasyOS Scarthgap-series version 6.0.2 released

June 24, 2024 — BarryK

Lots of bug fixes. Please read the 6.0 announcement for more details about this new series:

Release notes for 6.0.2:

https://distro.ibiblio.org/easyos/amd64/releases/scarthgap/2024/6.0.2/release-notes.htm

Download:

https://distro.ibiblio.org/easyos/amd64/releases/scarthgap/2024/6.0.2/

NLUUG mirror:

https://ftp.nluug.nl/os/Linux/distr/easyos/amd64/releases/scarthgap/2024/

Feedback welcome at the forum:

https://forum.puppylinux.com/viewforum.php?f=63   

Tags: easy

Some more fixes

June 24, 2024 — BarryK

Continuing finding and fixing bugs.

8. pup_event AppInfo.xml syntax error
Fixed. See: https://forum.puppylinux.com/viewtopic.php?p=123975#p123975

9. Xarchive
/files and /files/downloads group getting changed. See: https://forum.puppylinux.com/viewtopic.php?p=124006#p124006

10. ffmpeg
pulseaudio dependency missing. See: https://forum.puppylinux.com/viewtopic.php?p=124025#p124025

Getting ready to release 6.0.2   

Tags: easy

Yet more buglets fixed in EasyOS 6.0

June 23, 2024 — BarryK

Gee, they are really coming out of the woodwork. Blog post yesterday:

Continuing...

5. German locale
Thanks to Caramel; superb detective work, chased this one down:

https://forum.puppylinux.com/viewtopic.php?p=123603#p123603

I have fixed that script.

6. Grisbi
Forum member nnriyer reported that a library is missing, needed by Grisbi email client. Yes, 'libetpan' package is required. Fixed.

7. losetup
FeodorF and Caramel discovered this problem, and solution:

https://forum.puppylinux.com/viewtopic.php?p=123852#p123852

Yes, there used to be 'losetup' and 'losetup-FULL'. This harks from the Puppy days. The former was a busybox applet, the latter from util-linux. Now we just have /usr/bin/losetup from util-linux. I have fixed the script; however, just in case we encounter old Puppy scripts that require losetup-FULL, I have created it as a symlink.

FeodorF has discovered another problem with the de locale. I will need to bootup with de locale. Intend to look at that tonight.   

Tags: easy

Trike simplified steering linkage

June 22, 2024 — BarryK

Continuing the tadpole trike front suspension project, here are recent posts:

...that last link describes a fairly sophisticated steering linkage mechanism, that provides Ackermann compensation and handles the shock-absorber suspension.

However, that linkage is complicated, and for now I just want to get the trike to a rideable state as soon as possible. So, have figured out a linkage design that is very simple and makes use of what is already available.

There are four tie-rods from the trike; the two shorter ones I will use. Also will attach them to the existing mount points on the arms-frame. This diagram shows how it goes together:

img1

As using the existing tie-rods and arm mount-points, it requires the knuckle-levers to be extended, as shown in red. The red lines are parallel to wheel direction and there is a 15mm offset where the tie-rods are attached.

There are all kinds of compromises.

There is Ackermann compensation, but not perfect. The lower-left curve is a 2m radius, the outer curve is a 2.4m radius. Circles are concentric. If the steering is turned so that the outer wheel is following the 2.4m radius, the inner wheel has turned a bit too much; 1.55° too much.

Turning a larger circle, 3.2m inner radius and 3.6m outer radius, the inner wheel turns 0.9° too much.

Another compromise is that the steering arms swivelling is limited by the seat. This is going to limit turning to about a 2m radius. I guess can live with that.

If a wheel hits a bump, the tie-rod will cause the wheel to turn in or out slightly. I think that the tie-rod can be connected such that the wheel will turn slightly inward with a small bump, and slightly outward with a bigger bump. That situation would be improved with longer tie-rods; however, the other two that I have are too long, and I couldn't figure out how to fit them in, at least not without a huge amount of extra construction.

The only thing required to construct is the knuckle-lever extensions, that I will probably do on Monday.

Pretty much definitely will be wanting to modify this steering design later, but this will get me out onto the road quickly. I'm looking forward to experiencing how much the compromises affect the actual riding handling.

The SolveSpace design is here, with false ".gz".  

Tags: light

More buglets fixed in EasyOS 6.0

June 22, 2024 — BarryK

This is what happens when there is a major new release version; we get it out there with lots of users, and discover bugs. I posted a couple of days ago about a XKB Configuration Manager bug:

Here are some more...

1. clang
With the devx SFS loaded, clang seemed to be broken. Wasn't sure what was wrong; anyway, compiled llvm/clang 18.1.5 in a running EasyOS 6.0 and created a PET package. This will be in the next devx, instead of the package compiled in OE.

Thanks to the Linux From Scratch guys for guidance how to compile llvm/clang:

https://www.linuxfromscratch.org/blfs/view/svn/general/llvm.html

2. Grisbi
In the "Business" category of the menu, the icon for "Grisbi finance management" is missing. Fixed.

3. Chrome and Ungoogle Chromium
I have removed these from Appi and Flapi. Reason is, they are in the menu "Internet" category, and it is much better to install them this way. One advantage is they can be updated from the menu (see snapshot below) and the latest official package is downloaded so don't have to wait on AppImage and Flatpak updates.

4. Firefox
There is also "Download latest Firefox" in the menu:

img1

Rick sent me an email, reporting that after installing Firefox via "Download latest Firefox", there was no menu entry. I have fixed that and you can see above "Firefox web browser" entry.

You can update Firefox at any time, just by choosing "Download latest Firefox". Ditto for the other browsers. Because of that, I have disabled updating from within Firefox. There is now this file, at /usr/lib/firefox/distribution/policies.json:

{
	"policies":
	{
		"DisableAppUpdate":true,
		"DontCheckDefaultBrowser":true,
		"ExtensionUpdate":false,
		"HardwareAcceleration":false
	}
}

...this overrides user setup choices.

For example, the user cannot enable hardware acceleration; would have to edit that 'policies.json' file and delete that line.

The reason I have disabled hardware acceleration is that with it enabled Firefox behaves very badly on some hardware -- weird glitches when rendering. I want the best out-of-the-box experience for all users.

I don't recall why I also put in to disable extension updating. Anyway, if you want to discuss this, please do so at the forum, here.

5. German locale
If the German (de) locale is chosen, there is a black screen at the third bootup. Thanks to Federico and Caramel who chased this bug down, see here. I have not yet implemented the fix.

Notice in the above snapshot, there is no "Download latest Chromium". The reason for this is that Chromium is built-in to easy.sfs and was compiled by me. I compiled it with support for extra video codecs, whereas the official packages from Google do not support "proprietary codecs" and won't play many YouTube videos.

The downside is that you to update Chromium you have to wait until I compile it and include it in a future release of EasyOS. If you want more frequent updates, consider Ungoogled Chromium. 

Tags: easy

Trike 320 rear fork large gap fixed

June 21, 2024 — BarryK

I posted about this problem:

The rear fork is this part shown in black:

img1

Thanks to you guys who gave feedback about bending the fork closer. The wheel hub is 135mm, which is a common standard for bicycle rear wheels. However, the fork spacing is 140mm and is too stiff to pull in completely.

At the workshop I attend, received help from Terry, who is a retired school metalwork teacher. We set the fork into a clamp:

img2

Terry advised to do the bend cold, not to heat the fork, so that's what we did.

At the back, we placed an anvil, on which to rest the other end of the fork. The clamp is used by the woodworking guys. I suppose that an ordinary large g-clamp could be used ...hmmm, maybe not, as some considerable force is required. We lubricated the clamp thread to help.

The thing is to clamp to a certain point, where upon release the fork and it springs back just a couple of millimeters closer than the original 140mm. We were very careful about this, clamping the fork, then releasing, then clamp a bit more, and so on.

Very interesting; it kept springing back to 140mm. But a point is reached where the frame will just start to deform. For us, it was when we clamped the fork to 110mm spacing, then released and it bounced back to 138mm.

138mm, that is perfect. Now, can tighten the skewer knob a little, then flip the quick-release lever and the fork is now properly gripping the wheel hub. This is good news.    

Tags: light