Page 48 of 60

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 28 Nov 2020 15:51
by terriblefire
Will adjust that a bit.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 28 Nov 2020 16:34
by arkadiusz.makarenko
terriblefire wrote: 28 Nov 2020 15:51 Will adjust that a bit.
Thank you. :)

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 28 Nov 2020 18:39
by arkadiusz.makarenko
I have cleared all C compilator warnings, which solved one small memory leak which I wasn't able to pin down before.
This allowed me to tidy up main a little as well where I had to find work around issues.

EDIT.
After tidy up and solving some issues in code, I had to tidy up variable definitions etc. and now seems to be ok.
I think I have somewhere wrongly defined or init variable or mess with poiter.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 29 Nov 2020 11:44
by arkadiusz.makarenko
Looking at current project I can downgrade project from Stm32f722 to stm32f730. It is half the price, and there would still be like 15-20k left in flash for any potential upgrade.
I will have a look at cpld as well it might be possible to downgrade it from 72 to 36 cells, but it wouldn't have any potential to grow at all. So need to think about that one.

EDIT
CPLD - I take it back, I use 18 cells at this moment, so there would still be another half of the chip to use. Plenty space as at this point it would be used only really to decode few other addresses.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 29 Nov 2020 12:03
by terriblefire
arkadiusz.makarenko wrote: 29 Nov 2020 11:44 Looking at current project I can downgrade project from Stm32f722 to stm32f730. It is half the price, and there would still be like 15-20k left in flash for any potential upgrade.
I will have a look at cpld as well it might be possible to downgrade it from 72 to 36 cells, but it wouldn't have any potential to grow at all. So need to think about that one.

EDIT
CPLD - I take it back, I use 18 cells at this moment, so there would still be another half of the chip to use. Plenty space as at this point it would be used only really to decode few other addresses.
Yeah the CPLD can go down to a 36 no problem. I had envisioned it might do more stuff but really its just an address decoder. Keep the ARM big though. There is soundcard potential there too.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 29 Nov 2020 13:54
by arkadiusz.makarenko
terriblefire wrote: 29 Nov 2020 12:03
arkadiusz.makarenko wrote: 29 Nov 2020 11:44 Looking at current project I can downgrade project from Stm32f722 to stm32f730. It is half the price, and there would still be like 15-20k left in flash for any potential upgrade.
I will have a look at cpld as well it might be possible to downgrade it from 72 to 36 cells, but it wouldn't have any potential to grow at all. So need to think about that one.

EDIT
CPLD - I take it back, I use 18 cells at this moment, so there would still be another half of the chip to use. Plenty space as at this point it would be used only really to decode few other addresses.
Yeah the CPLD can go down to a 36 no problem. I had envisioned it might do more stuff but really its just an address decoder. Keep the ARM big though. There is soundcard potential there too.
I will take my time to review it, but at this moment I think the only difference is in size of flash memory. RAM on both chips is the same. Initially firmware growth was substantial due to usb stack and usb libraries,now it is much slower.
Currently there is about 30-35k taken (25k ? with size optimalization) F730 wouls still have like 25-30k flash left for growth.

Ps. Soundcard any more details, please?

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 29 Nov 2020 15:00
by terriblefire
arkadiusz.makarenko wrote: 29 Nov 2020 13:54
terriblefire wrote: 29 Nov 2020 12:03

Yeah the CPLD can go down to a 36 no problem. I had envisioned it might do more stuff but really its just an address decoder. Keep the ARM big though. There is soundcard potential there too.
I will take my time to review it, but at this moment I think the only difference is in size of flash memory. RAM on both chips is the same. Initially firmware growth was substantial due to usb stack and usb libraries,now it is much slower.
Currently there is about 30-35k taken (25k ? with size optimalization) F730 wouls still have like 25-30k flash left for growth.

Ps. Soundcard any more details, please?
Since you are already implementing most of the clockport ... maybe emulate a clockport sound card..

There are 2 DACs on the arm that come out to that 3 pin header....

I'd need to dig into the details of how those soundcards work for more ideas.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 29 Nov 2020 15:32
by arkadiusz.makarenko
terriblefire wrote: 29 Nov 2020 15:00
arkadiusz.makarenko wrote: 29 Nov 2020 13:54
I will take my time to review it, but at this moment I think the only difference is in size of flash memory. RAM on both chips is the same. Initially firmware growth was substantial due to usb stack and usb libraries,now it is much slower.
Currently there is about 30-35k taken (25k ? with size optimalization) F730 wouls still have like 25-30k flash left for growth.

Ps. Soundcard any more details, please?
Since you are already implementing most of the clockport ... maybe emulate a clockport sound card..

There are 2 DACs on the arm that come out to that 3 pin header....

I'd need to dig into the details of how those soundcards work for more ideas.
I wasn't aware that there are clockport soundcards...

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 29 Nov 2020 17:38
by 8 Bit Dreams
Oh yes.. There are few :idea:
IMG-20201117-WA0015.jpg
Delfina, for example.. 🤔

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 29 Nov 2020 21:45
by arkadiusz.makarenko
terriblefire wrote: 29 Nov 2020 12:03 Keep the ARM big though. There is soundcard potential there too.
Now I get it. MP3 De-Codecs only are like 50k.
This is quite impressive on paper : https://www.st.com/en/embedded-software ... l#overview
So features like mp3 player should be doable as Amiga has just to supply data to stm32 buffer quickly enough.