News:

;) This forum is the property of Proton software developers

Main Menu

Higher Income

Started by dondwsmith, Feb 21, 2024, 10:16 AM

Previous topic - Next topic

dondwsmith

Good day Les

I have been reading past posts and I was thinking on your need for more income and I have 2 suggestions.

1/ Microchip tends not to obsolete products they just keep increasing the prices. This price increase creep often goes unnoticed until you get an order for a low volume old product and suddenly your MCU cost has gone up 2 to 3 times. Now of course you do not have time to rewrite the ASM code for a new processor.
There is a lot of code out-there for the Microchip MPASM and linker but the newer cost effective MCU's are only programmable in PIC-AS and converting and testing is onerous.

If you could offer a system to write MPASM code to the newer MCU's in a much easier conversion than that offered by Microchip I am sure you would get an increase in sales plus the advantage of of putting your compiler in front of a lot of new customers. I also think you would get a lot of free exposure on all the PIC user groups which again would put your compiler in front of a lot of potential new users.

2/ I am operating in South Africa and the distributor for Microchip used to register our designs and then give us the 5k pricing on the MCU's for orders of a few hundred up, but since Covid and the purchase of Atmel, Microchip does not allow this anymore. So our processor prices have increased a lot.

We have been looking at the RISC-V chips from WCH in China. They are offering an 8 and 16 pin SO package device for about US$ 0.20 for 50 units. These low end devices are a 32 bit 48Mhz Mcu with 2kbyte Ram and 16 kbyte Flash. The do have bigger and faster devices but all in narrow pin spacing surface mount packages

I see that many Universities are using RISC-V for teaching Computer Science. If you supported these processors you would be getting in before any other professional Basic Compilers for embedded applications exist. Especially if you offered site licenses at discounted rates.

Frizie

What you are asking for here is not something that can simply be made, it takes a lot of effort, time and energy.
In addition, Les suffered a brain injury a number of years ago, which still bothers him a lot  :-\
Ohm sweet Ohm | www.picbasic.nl

Stephen Moss

Quote from: dondwsmith on Feb 21, 2024, 10:16 AMI see that many Universities are using RISC-V for teaching Computer Science. If you supported these processors you would be getting in before any other professional Basic Compilers for embedded applications exist

Then they must already have a tool chain for programming/using them, so why would they switch?
Particularly, when many people and Universities tend to look down on BASIC, considering it slow and cumbersome, likely as a result of Microsoft's QBASIC which was interpolated and thus slow, and not taking into account that compiled BASIC languages having been around for years that should be able to produce code that execute as fast as that written in any other language.

Additionally, certainly the major Universities would generally be looking to use whatever language (i.e. C++ or Python)/toolchain that is likely be used in industry so that their students are better prepared for employment after graduation. I am not saying outright that your suggestion would not work but I think Universities would need a lot of persuasion to change.

top204

#3
As Frizie stated, creating an efficient compiler is not something that can be accomplished overnight! It takes months of research and work to get a Beta compiler working with efficiency, then more months ironing out the querks and benefits of the microcontroller's architecture. It is possible to knock together a dreadful compiler full of holes and extremely inefficient, in a few months, as is now available online with several languages. But that is not my way.... If I create something, it has to be as good as I can create it, not just "knock something up", and that takes time!

In the mean-time, I also have to make a living, and actually create some money to live on! R&D is all well and good if a person is getting paid to do it, or the person is so wealthy they have lots of time to spare. Both of which I am not, I'm afraid. :-(

Also, as Stephen pointed out, universities are extremely elitist, and simply follow the 'sheep' to tell the 'sheep' what to do whenever possible, so they use the most popular languages, even when some of those languages are truly dreadful. :-) So creating a language for universities is the same as creating one for a brick wall to use. i.e. It will not! :-)

trastikata

Quote from: 8100xt2 on Feb 26, 2024, 01:06 PMYes we all should stay in the past ! strange moderator here !

I am not sure what you mean by that ... you are free to choose whatever compiler suits you best.

If you think the best way for one to optimize own income is to do what you say ... why don't you do it yourself?

trastikata

Quote from: 8100xt2 on Feb 26, 2024, 02:17 PMDo you hear me complaining ?  ;D  Not me !!!!

Then I guess you could finance the R&D costs for the additional compilers so we ALL can benefit from it?

trastikata

Quote from: 8100xt2 on Feb 26, 2024, 02:27 PMNo we make it abandonware ! bye bye

Good luck to you too.

RGV250

Hi,
Not quite sure what you meant by the last post, it is getting updates and fixes more than other software I have used.
Only a month ago  https://protoncompiler.com/index.php/topic,1972.msg15379.html#msg15379

Bob

top204

#8
That user has now been deleted from the forum, so unfortunately, your good answers to questions he posted look a bit strange. :-)

As a note... I have the bits and pieces ready for the upgrade to the Q40 and Q41 devices, so I am in the process of putting them together in an upgrade. It is just a matter of finding the time to do it all, but I will try my damnest to get it done ASAP.

What is it with microchip users who always want "the latest device", even though they are, essentially, the same as the previous devices? The Q4x devices do have a bit more RAM, but not much else is different in them that another 18F could not do just as well?? They do not operate any faster, and do not have any magic peripherals in them or a special core with hardware division or signed multiple etc, so why always the "latest model"? OK, they are a little less expensive (a few 10s of pence), and some of the wait times are a bit shorter, but for the vast majority of microcontroller users, these are not a problem.

The one problem I now have is that microchip have stopped with their free samples, so I have to perform some best guesses with some of the library routines that user peripherals, and as I have had to do with previous devices that change a lot, and add some 'other' library routines that are selected within the device's PPI file.

trastikata

Quote from: top204 on Feb 26, 2024, 03:24 PMThe one problem I now have is that microchip have stopped with their free samples, so I have to perform some best guesses with some of the library routines that user peripherals, and as I have had to do with previous devices that change a lot, and add some 'other' library routines that are selected within the device's PPI file.

This is something we could help with, post a list with the types and preferred package and we can provide them for you.

Regards

Frizie

Good idea Trastikata, I'm in favor of it!  :P

It may be just me, but I don't find the datasheets of the newer PICs nearly as clear as the datasheets of the older PICs.
Less information, less clear.
I understand that as a PIC gets more and more peripherals, it becomes increasingly difficult to keep the data sheet clear and, above all, informative.
But this is partly due to Microchip itself, which has to mess everything up every time (which means, among other things, that compiler writers have to rewrite parts again and again).

Maybe I still have to get used to the newer datasheets. ;D
Ohm sweet Ohm | www.picbasic.nl

dondwsmith

The reason people want to use the newest processor is that after upgrading 2 or 3 times over the lifespan of a project you want to use the newest part as this will give you the longest time before you have to upgrade again. It also means that availability is better as the new parts don't have high usage yet. I recent asked the rep at Avnet what parts they recommended for a new project. The reply was the PIC16F18174, PIC16F17174, PIC16F19176 & PIC18F45Q43. As far as I can see only the PIC18 MCU is supported at present.

okmn

#12
Quote from: top204 on Feb 26, 2024, 03:24 PMThat user has now been deleted from the forum, so unfortunately, your good answers to questions he posted look a bit strange. :-)

As a note... I have the bits and pieces ready for the upgrade to the Q40 and Q41 devices, so I am in the process of putting them together in an upgrade. It is just a matter of finding the time to do it all, but I will try my damnest to get it done ASAP.

What is it with microchip users who always want "the latest device", even though they are, essentially, the same as the previous devices? The Q4x devices do have a bit more RAM, but not much else is different in them that another 18F could not do just as well?? They do not operate any faster, and do not have any magic peripherals in them or a special core with hardware division or signed multiple etc, so why always the "latest model"? OK, they are a little less expensive (a few 10s of pence), and some of the wait times are a bit shorter, but for the vast majority of microcontroller users, these are not a problem.

The one problem I now have is that microchip have stopped with their free samples, so I have to perform some best guesses with some of the library routines that user peripherals, and as I have had to do with previous devices that change a lot, and add some 'other' library routines that are selected within the device's PPI file.




The reason why I tend to use every new PIC MCU is that they are 50% cheaper, especially 16F152xx,and 3xx series and other 16F1804x or 14x...even their adc are 12b (some models)also new 18F seies has 2 or 4 Opamp like Q20,Q40,Q41,Q71 series.


and these days,

They are trying to put Nuvoton on the market!!! 32 bit MCU for 0.39 USD,!!! This is a market, I know the efforts to own a market, yes.

  However, this being the case, to employers who expect the most cost-effective designs from us,

We have a compulsory expectation from Les, who is the only solution point for 30-year-old Electronicians like me who have to do something to compete with very cheap MCU prices in the market (for example, Nuvoton).

because I/we only use PIC MCU just like Siemens!!! (Siemens also uses pic18F45k22-E/PT in its industrial PS-8642 POWER SUPPLY 110-220VDC or 230VAC 45W products.)

Dear Les,

  You are a very good programmer in your field, that's for sure. As the owner of this topic said, what would a new compiler look like?

  Which, I'm sure you already have an idea - you even have a project about writing a completely new compiler that will quickly adapt to microchip's new mechanism (compatible with assemblies and/or c, c++ libraries)

If it is suitable for the majority, should everyone buy the current version from Ebay every month, over and over again, so that you can have time for a new project thanks to the income from those sales?

Stephen Moss

Quote from: dondwsmith on Feb 27, 2024, 09:43 AMThe reason people want to use the newest processor is that after upgrading 2 or 3 times over the lifespan of a project you want to use the newest part as this will give you the longest time before you have to upgrade again.
That makes no sense, if you have a device that is suitable for your needs then the only reason it would be an issue is if you are taking so long to develop a product that your original device becomes discontinued, in which case...
  • That is somewhat on you for taking many years to develop the product and thus still likely to happen even if you start with the latest device
  • If 1 happens there is probably a similar newer equivalent available that would only require minimal code adjustments to be suitable. Provided that key pins like power, clock and MCLR are in the same location as many newer device have PPS so you can probably make the peripheral pins match even if they initially default to a different location.
  • Unless there is no other option it is probably better not to use the latest devices as soon as they come out, but to wait a while in case there are issues with it and an errata has to be issued. Provided you select something less than 5 years old it is likely to be available for at least another 10 years provided it is selling in reasonable quantities.

If you are instead talking about product support over a long lifespan then surely you would keep track of whether or not that particular microcontroller is being discontinued and if so purchase a quantity of them as replacements while they are available so that you can continue ongoing support of the end product. 

Quote from: dondwsmith on Feb 27, 2024, 09:43 AMI recent asked the rep at Avnet what parts they recommended for a new project. The reply was the PIC16F18174, PIC16F17174, PIC16F19176 & PIC18F45Q43. As far as I can see only the PIC18 MCU is supported at present.
If you literally just asked them what they thought was good for a new project without giving them a specification to work to, i.e. 5V or 3.3V, memory size, what peripherals you need then that recommendation is virtually useless as they probably recommended what they wanted to sell rather then something that is suited to the needs of a particular project.

Putting those device numbers into the IDE both the 16F19176 & 18F45Q43 are highlighted so should be supported.

SeanG_65

Quote from: Frizie on Feb 24, 2024, 09:26 PMWhat you are asking for here is not something that can simply be made, it takes a lot of effort, time and energy.
In addition, Les suffered a brain injury a number of years ago, which still bothers him a lot  :-\

He's still WAY smarter than me.

JonW

#15
All of these topics on increasing income and promoting the compiler are far from trivial to implement.  Expansion/success is a real issue for almost all small businesses where sales are directly proportional to marketing effort and spending.  YouTube is a great platform, but creating exceptional content. takes a lot of effort, skill and cost. So it's chicken and egg; without investment, you can't afford the marketing; without sales, you don't get the revenue to invest...

For all its faults, Arduino is a good platform and for the millions of makers (hobbyists) out there, it's perfect.  Ide is free, hardware is diverse, and it is low-cost, and despite how bloated it is, it works well for its purpose.  As Steven has clearly stated, extremely complex functions can be coded in literally minutes without hardly any knowledge of the hardware or firmware.  You have to give it credit for that. Still, the main difference is it Les vs a massive team of engineers, plus professional marketing companies with the addition of endless investment from the silicon manufacturers, not to mention millions of followers generating content and code libraries.

I may be wrong here, but I get the feeling Les is already struggling to keep the current compiler up to date, fix bugs, and support what MCU he deems will benefit the compiler and its users. He has a lot on his plate due to medical issues and needs to keep his household running financially on top of everything. Adding more or newer devices is obviously a significant undertaking and a huge risk as it may not yield a return on his investment.

When you consider everything in the posts over the last week concerning adding newer devices, newer processors, investment in YouTube, etc., if I were Les, I probably wouldn't do it as the return on investment is too risky, and rightly so his priority is his health and his family.

If you are a commercial entity or want to contribute, then there may be an opportunity for you to pay Les for a custom library you need to develop, etc.  Time is money... He may not want to do this or have the time, but he would likely make 10x than he does selling compilers.  I know this has been said many times, but a support contract/payment per annum is one way to get a steady, recurring revenue.


Be careful with new devices. We fell into this issue with Q devices and even paid to get the device added to the compiler. After wasting a lot of time, we finally managed to get the part running; the market was flooded with older Errata issue devices.  If cost is paramount, then Microchip will never get to the price point of the big chip guys; for extremely low cost/high volume, there are lots of quality feature-packed MCUs sub $0.10 out there with good toolchains.

Frizie

Quote from: JonW on Feb 29, 2024, 08:54 AM...investment in YouTube, etc., if I were Les, I probably wouldn't do it as the return on investment is too risky, and rightly so his priority is his health and his family.

It doesn't have to be Les who makes a course on YouTube.
Someone who is handy with YouTube, knows Positron and has the time and like to make this can do this just fine.

And indeed, what you write next is completely true.

(Les should have co-writers ;) )
Ohm sweet Ohm | www.picbasic.nl