Firmware V3.0

[flel]

Thank you for the great update!!

[sarukas]

Thanks Thomas! The finnicky video expander now is the star of the show! Amazing concept, but I am just as grateful that you have made these small improvements elsewhere, they are very important!

Regarding the manual: it is now the weakest part of the ecosystem, because many features are there but described thinly and without all exceptions or examples (recent example I came across: ECLD/ECLP mapping destination - works only in modular track.). Also mapping code snippet chatper would be amazing. 

Would it be possible for you to leverage any AI LLM in this day and age, and gradually rewrite it chapter by chapter, incorportaing all typical user questions and all your relevant answers from these forums? It can be easily done if you dedicate a bit of time to learn how. I would recommend NotebookLM which strictly follows source material in generating answers.

Heck, if you would not do it, then I could help on this if you could give me bulk export of forum material (db dump) and an editable manual. A day's work to produce possibly a 400 page manual, and then we can finish editing and reviewing it together over the nect year :-)

[vidret]

Great update, getting on this.

Don't have a video expander, but I really appreciate all the small fixes Thomas!

[XORadmin]

Thanks Thomas! The finnicky video expander now is the star of the show! Amazing concept, but I am just as grateful that you have made these small improvements elsewhere, they are very important!

Regarding the manual: it is now the weakest part of the ecosystem, because many features are there but described thinly and without all exceptions or examples (recent example I came across: ECLD/ECLP mapping destination - works only in modular track.). Also mapping code snippet chatper would be amazing. 

Would it be possible for you to leverage any AI LLM in this day and age, and gradually rewrite it chapter by chapter, incorportaing all typical user questions and all your relevant answers from these forums? It can be easily done if you dedicate a bit of time to learn how. I would recommend NotebookLM which strictly follows source material in generating answers.

Heck, if you would not do it, then I could help on this if you could give me bulk export of forum material (db dump) and an editable manual. A day's work to produce possibly a 400 page manual, and then we can finish editing and reviewing it together over the nect year :-)

I think we had these discussions many times about the manual already here. In fact I think you noticed that I did quite some work on the latest manual streamlining it a bit more. I did try to involve AI tools to improve stuff but the results were very average like someone explaining stuff they don't know plus all the wrong assumptions. Going through all the results is a daytime job and very capable of adding many unseen errors. In the end I stopped with that and improved what I thought must be improved while keeping in my mind that I got to release at some point.  

Coming to your ECLD example. The Mapping destinations use just available functionality. If the track doesn't have euclidean or whatever functionality then the destinations don't do that either. I could add this to every single position in the manual. I could add examples to every single functionality in the manual and it would end up in a 500 page manual while it is already much too big because they are so many functions. 
This is where a forum like this is for (or videos of peoples) . Sharing tricks and informations etc.  The sharing information part is actually very quiet except for my input. 

Also I actually did create podcasts with NotebookLM etc which is fun but ended up totally useless because it was filled with so many wrong information. The only consistent thing was ' Oh Jeff I am so glad you mention this', ' Yes Jane, this function totally changed my life' ... and so on....

We did also feed the manual and the forum DB to a LLM which gave quite some good results but only to a certain level. Still much misinformation has been introduced which is probably more frustrating than doing a document search for X or Y if you want to know something.  We'll do that for the V3.0 release as well and open it so everyone can use it. 

The editable manual is always available on Github. 

That being said, I will not put too much extra work into the manual but do only what is needed to be complete at all times with all updates. I will put my energy into video tutorials in the next weeks and months though, something that is surely much more worth than even more theoretical text.
And more important I got to figure out how to keep the company running since a whole market disappeared over night (a dark night in january for many) and my savings are shrinking. The latest update didn't help with that either, even though it was quite big.

[sarukas]

Great to hear Thomas that you have already tried my suggestions on improving the manual! At work I use LLMs extensively and the problems you are mentioning, while real, can be overcome with the proper context engineering technique.Claude code is especially recommended for technical documentation writing, which this manual is a case of. The key concept to apply is that you need to provide full context to LLM otherwise it just makes assumptions. A crucial piece of context in your case are the existing manual, this forum, and the source code itself.You can generate detailed guiders, diagrams or dataflows from the code with LLM to get highly accurate representations of how the code works. The main obstacle is the learning curve - it is about 3-6 months of constant work till you get good at this. Afterwards, you are 10x more productive. During the process - 0.5x productive.

[TheSandbag]

I don't know if I'm being dumb, but I can't find what the video modes map to anywhere

[TheWogglebug]

As usual this is beyond rad. Can't thank you enough for your work on this Thomas.

[septcercles]

Thanks so much for all this majors improvements and new features <3

[Manbearpig]

I'm having trouble updating the Nerdseq firmware to 3.0 and the CV16 firmware to 1.2.

The Nerdseq User Manual says to drop the firmware file into the "/firmware" folder but on my SD card there is no folder called this. The firmware instructions from the CV16 thread on this forum also mentions a folder called " nsa_cv16_1_1.nxp" as an example of a firmware folder (?) but there's no folder called that on my SD card either. My computer does not allow me to use the character "\" when trying to name a new folder on the SD card. And when I do create a folder on the SD card, it doesn't seem to stay on the card (I can create the folder, but when I re-connect the SD card to my computer, it doesn't see the folder). 

The only thing on the SD card I use with my Nerdseq is a file called uupd.bin

Any advice? Thanks!

[XORadmin]

I'm having trouble updating the Nerdseq firmware to 3.0 and the CV16 firmware to 1.2.

The Nerdseq User Manual says to drop the firmware file into the "/firmware" folder but on my SD card there is no folder called this. The firmware instructions from the CV16 thread on this forum also mentions a folder called " nsa_cv16_1_1.nxp" as an example of a firmware folder (?) but there's no folder called that on my SD card either. My computer does not allow me to use the character "\" when trying to name a new folder on the SD card. And when I do create a folder on the SD card, it doesn't seem to stay on the card (I can create the folder, but when I re-connect the SD card to my computer, it doesn't see the folder). 

The only thing on the SD card I use with my Nerdseq is a file called uupd.bin

Any advice? Thanks!

Oh if you got the file uupd.bin on the sdcard then it is broken and you need to replace the sdcard. 
It rarely happens but it can happen with sdcards. 
It’s not a NerdSEQ issue but a thing with sdcards in general (just google for uupd.bin)

I can send you a new card if you need one.