I didn't post on this forum lately, because I wanted to fix some issues with my mod first. Some of them I already fixed, which are published in the
prerelease v0.4.2-PR1 (something like a beta release but not in 100%).
Currently most of them are caused either by the user confusion with the modified firmware (too many available options for the user...) or because of the problems on Patreon platform.
I need to find new platform ASAP for the new users and some of the existing ones, which will want to go there.
Issues with the Patreon are huge, like displaying couple times bigger price (at least collected amounts are correct) and replies to my bug reports was like: "update Your browser, if this didn't help, update it again. If still it doesn't work it's Your fault for sure".
Also in one place, when something is free for the subscribers, there is not working button "buy". Later I had to answer questions like "why I need to pay for this second time". Just because of some error done in the code (Patreon, not my mod). Code which I never seen and I don't want to. I can write a book about software bugs on Patreon web pages.
In the past I was working for couple big corporations (one of them everybody knows for sure) as a software developer with web pages. Much bigger and more complicated than Patreon.
In those web pages, either for the end user or used internally (for example reporters uses backend web page to write article about something, weather report and things like that), there was a lot of bugs waiting in the line to fix. But when I was working in the team, we did not only new features or subpages but we also fixed a lot of bugs.
Also we tested everything in multiple web browsers, including older versions. Not like Patreon testing only with one, which I guess is the case.
I have strong feeling that they not fixing any bugs, but they hired the cheapest people with a web page engineering in their resumes, which turned out with ~10 bugs for each added feature.
It's very nice for me to see that many satisfied users of my mod. Some of them are not only supporting with the money (everything has some costs) by also actively, like a making fixes in my script for the issues they experienced, which I implemented with some changes (there is no such thing like a perfect code - no matter who will write it). On the side note, simplified installation via the .GEL file is still available.
Still there are some problems on the human side, like a not reading installation instruction. I tried to do my best to make it short and concisely as possible, with pointing out most important thing, which actually is the least read part - which is the "After installation" part. This contains only six very short sentences - less than one minute to read. But in some cases omitting this caused to lose precious hours for both sides.
I made an ignorant post on here yesterday.
Everybody in this world is making some mistakes from time to time. But not everybody is brave enough to admit it, try to fix it (if it's possible) or to prevent it in the future.
Maybe this is not the perfect place for such conversation, but in the aviation, if You do some unintentional mistake, You have to report it in 48 h, otherwise You can lose Your license. Personally I have only entry cards to restricted areas in the two biggest international airports in Europe
Anybody heard about famous actor Harrison Ford which almost crashed his private plane on the airliner with people inside and landed on taxiway instead of runway? He didn't lost his licence, not even for 5 minutes.
What are the prospects for a newer Android version?
I had this idea in mind. Little newer Android, like a 9.0 or 10.0 should take out some bugs and make possible to install newer versions of apps. And this also should make it easier to use kernel 5.1.x or newer, which should increase performance a lot (small/big cores in CPU handled more properly by the CPU scheduler in the kernel).
But there are some reasons to not do this. Android by itself is a huge cr*p. It uses (modified) Linux kernel, but this is the end of good things in this system.
Everybody want's to have OS, which is fast, fast to boot, not slowing the apps too much and to have full control over it. Android is not like that - You don't own Your oscilloscope, because Google (creator of Android) owns it. And it uses slow Java for ~95% of things.
Another issue with Android: about one month ago (or something like that) Google announced that they are closing Android source code. This probably is not the biggest issue, but it's not anything good.
In the same time, I already put in a lot of work to develop a working Debian (GNU/Linux) based system to work in this scopes ("Orange Rigol" in my footer), with the
5.10.209 kernel with some bug fixes, mipi-dsi driver which I ported (two weeks without any warranty to work it at all).
I don't mean to run any Android app on the GNU/Linux, which is not the greatest idea for some reasons.
Rigol oscillopscope app is almost like a regular Android App. Some part was written in Java (mostly UI) and some part was written in C++.
My idea is to took this compiled C++ part, which requires only one Android lib, make this simple lib from scratch to pass log data into some file or somewhere and wrap it around in another pure C code which will do same things as current Java code. This will make it faster to use it, to maintain, make (more or less) insane and useful features, easier to fix bugs and harder to make a new ones (bug = software engineer mistake).
Some EE probably will not understand what I just said. I mean to rewrite it, but not completely at first. At first will go Java part, which is tied hardly to Android only and is a huge bottleneck both in scope performance and in developing any changes. Later this will allow to rewrite some compiled parts in C++ into pure C, which will help with the things that I just mentioned.
I only noticed some small problems with the references
In the v0.4.1? Anyway, I need to fix it.
I thought it would be boot faster with a faster card... 
Those scopes uses RK3399 as the main CPU (there are three physical processors). It can boot from one of the two MMC (SD card = MMC memory) ports. After initial reset (after powering it up) it executes small internal code (internal ROM), which decides which MMC memory chose to boot from.
This internal bootloader tries to read keys in the another bootloader (named U-Boot) located at the almost beginning of the SD card.
Little later, this U-Boot bootloader does some crazy staff (which is needed since this is not the x86 regular computer) and executes commands from the config file and eventually DT file. It makes initial setup of the SD card, like a frequency, voltages and other magic staff.
At the end of above things, it loads Linux kernel (again... Android uses modified Linux kernel) and couple more things into external DDR3 memory and executes one simple CPU instruction to jump with the code execution into Linux kernel.
Almost at the beginning of booting Linux, it reads DT in which there is a lot of hardware information. Including SD card max frequency and other things.
Later it tries to run first process (process is something like a app) called init. It doesn't care what is this in reality, it just execute this. In this "init", seats Android staff.
In my mod I already modified this DT for the better performance and to have more waveform updates per second. But I didn't even touched SD card related parts in it (beside of enabling unused MMC port for newer scopes - another long story). I think there is a chance to make it go faster.
But (!!!), as they say: "a chain is as weak as its weakest link".
Today we have one huge issue with almost all computers. Most corporate software is done with idea: make this just work and nothing else. You can use the slowest libs software and methods available on the internet. Just make it quick, because people will buy it anyway.
In case of the Rigol firmware (app mostly), if they figured out how to enable compiler optimizations, which is literally one minute of work, this app will be 5-20 (depends in which exact part) times faster. But guess what: they didn't. I think they never figured out this at all. Some people are like: lets learn first better programming language and make some money...
10-20 and more years ago, people was doing something with passion. And this was selling good. Now, You have to buy 10 times more expensive CPU for Your PC, because somebody was not aware of existing such thing as compiler optimizations. I think it's obvious it's going into very bad way.
Speaking of the CPU-s. I just found brand new fan and heatsink for my work laptop, because original one is making noise from the ball bearing and it's not effective as it was before.
12 year old business and very ergonomic laptop. It's highly ergonomic and fast enough. But Im not using Windows "rubbish" and in same time it's 10 times better and more reliable that today new and cheapest ones. BTW it's hard to buy this exact model or parts for it because of high demand.
Back to topic: Orange Rigol (modified Debian) with mentioned 5.10.209 kernel and with much more things executed at boot time, is ready to use after ~30 seconds.
If I will finish my work with rewriting Java part of the app, it should start in about 5 seconds (instead of ~12) and run much faster. After rewriting some parts from the existing .so (C++) lib into C, it will be also faster. So that should give ~35 seconds from pressing power button to using scope as a scope.
with just 2 channels
3 channels to be exact. Last one (used only as a EXT trigger in the stock app) has some missing resistors for AFE, but we have
very little data if this has to be changed as other channels or not, to work properly - I asked him to do self calibration and to try if this will fix the issue, but he didn't replied yet.
Norbert's firmware really makes this a spectacular scope.
Nah... this is just the beginning of this adventure
Speaking of the adventure, some people was screaming at me that my price is way too big for simple hack. For such people I wish good luck with 6 months of reverse engineering hardware and software and another 6 moths of making same thing. If somebody will do the same thing twice as fast, I will call such person as a genius.
And how much I earned for this "overpriced" "simple hack"? In the attachment there is the answer for this.
Sadly there is no possibility to change language in the Patreon (...) and they like to show amounts without some fees or sometimes 10 times more that actually is. I added numerical months to the screenshot.
First release was at 15.05.2025. Without time spent on rev. eng. and without all fees and taxes, it's 3028 $ per 5.5 months or 550 $ per month. Which is also the 41% of the minimal wage in Poland or about 21% of minimal wage in Germany.
This post is more or less messy - I just spent whole night fixing one scope remotely - something made one important file to be empty for some reason.
Home
Help
Search
Login
Register
