SoftI2CMaster with megaTinyCore #664

TheFlav · 2022-03-09T15:15:23Z

TheFlav
Mar 9, 2022

I have a project where I want my megaTinyCore attiny to be an i2c slave, but I want it to aggregate data from its own GPIO and some other i2c slaves (and then serve that data over i2c to the main host system). I will likely be using 2-series parts for this (attiny427 and attiny1627).

I thought that using the Wire.h library for serving the data to the host would be the best approach.

To read data from other i2c slaves, I thought that the SoftI2CMaster library would be the best option. Please tell me if there are other good options that I should consider.

I had problems using SoftI2CMaster, as the assembly code didn't like working with megaTinyCore. In my early searching on the problem, I noticed that ATTinyCore has a SoftI2CMaster library included (https://github.com/SpenceKonde/ATTinyCore/blob/master/avr/libraries/Wire/src/SoftI2CMaster.h), but megaTinyCore does not. That makes me wonder why.

Note that I am using the latest code from https://github.com/felias-fogg/SoftI2CMaster which has been updated recently and fairly frequently, and I only mentioned the ATTinyCore link, because it lead me to believe that SpenceKonde might have some knowledge of the library and how it relates to attiny parts. It works! I just need to make the mods below.

Moving on, I'm no AVR assembly expert, but I dug into it a bit. It seems that the inline ASM constraints were the main issue here. I dug a bit, and I figured that the memory mapping of the GPIO registers was the problem here.

To use SoftI2CMaster, you have to define the pins like this.

#define SDA_PIN 3
#define SDA_PORT PORTA
#define SCL_PIN 5
#define SCL_PORT PORTB

#include <SoftI2CMaster.h>

I tried switching that to VPORTA or VPORTB, becuase the PORTx mappings are larger than 1 byte (which the inline ASM constraints didn't like). That didn't work.

Anyway, fast forward a bit, and here's what I came up with that seems to work (but I think is a hack that should be implemented in a better way for mass consumption).

I changed the library's SoftI2CMaster.h to add an #if around the definitions like this.

#if !(defined(SDA_DDR) && defined(SDA_OUT) && defined(SDA_IN) && defined(SCL_DDR) && defined(SCL_OUT) && defined(SCL_IN))
#define SDA_DDR         (_SFR_IO_ADDR(SDA_PORT) - 1)
#define SCL_DDR         (_SFR_IO_ADDR(SCL_PORT) - 1)
#define SDA_OUT         _SFR_IO_ADDR(SDA_PORT)
#define SCL_OUT         _SFR_IO_ADDR(SCL_PORT)
#define SDA_IN      (_SFR_IO_ADDR(SDA_PORT) - 2)
#define SCL_IN      (_SFR_IO_ADDR(SCL_PORT) - 2)
#endif

Then, in my .INO, I would explicity set these (instead of setting SDA_PORT and SCL_PORT). That looks something like this.

#define SDA_DDR         0x00 //VPORTA_DIR = 0x00, VPORTB_DIR = 0x04, VPORTC_DIR = 0x08
#define SCL_DDR         0x00 //VPORTA_DIR = 0x00, VPORTB_DIR = 0x04, VPORTC_DIR = 0x08
#define SDA_OUT         0x01 //VPORTA_OUT = 0x01, VPORTB_OUT = 0x05, VPORTC_OUT = 0x09
#define SCL_OUT         0x01 //VPORTA_OUT = 0x01, VPORTB_OUT = 0x05, VPORTC_OUT = 0x09
#define SDA_IN          0x02 //VPORTA_IN  = 0x02, VPORTB_IN  = 0x06, VPORTC_IN  = 0x0A
#define SCL_IN          0x02 //VPORTA_IN  = 0x02, VPORTB_IN  = 0x06, VPORTC_IN  = 0x0A

#define SDA_PIN 5
#define SCL_PIN 4
#define I2C_FASTMODE 1

#include <SoftI2CMaster.h>

So, I hope that's helpful. From my quick research, I think this would work for all the megaTinyCore parts, but I also think there's likely a much better way to do this. I would guess that SoftI2CMaster.h could be made to use the "#define SDA_PORT PORTA" method and map that to VPORTA with the proper offsets for the _DIR, _OUT, and _IN registers.

My point here is to start some discussion in case this helps anyone else or helps to add SoftI2CMaster into megaTinyCore.

Also, I would still like to know if SoftI2CMaster is even the right thing I should be using.

Thanks!

Note: I should have mentioned this up front, so I'll add it up there for anyone that comes along. I am using the latest code from https://github.com/felias-fogg/SoftI2CMaster which has been updated recently and fairly frequently. I think that my mention of the link to SoftI2CMaster.h in ATTinyCore may have have been misleading. I only mentioned that in reference to searching for SoftI2CMaster for attiny parts and figuring that SpenceKonde must have had some knowledge of it.

SpenceKonde · 2022-03-09T17:28:35Z

SpenceKonde
Mar 9, 2022
Maintainer

megaTinyCore does not have softI2CMaster built in because these parts are all capable of acting as master and slave, and there is no reason for a flaky software implementation. It was added to ATTinyCore because there are 3 parts that not only don't have hardware I2C master, (only the tinyx8 there does), but also doesn't have a USI. Therefore, before that was added, trying to #include Wire.h would error out, and people had no prayer of using a library that made use of Wire .h for I2C on those parts. Since two of the three parts (the 841 and 441 - the 828 being the third, and the clear runt of the litter) were the parts that initially motivated me to make ATTinyCore, and are otherwise among the best classic tinies, a solution was needed. The people on the USI parts also complained about the fact that Wire didn't work, and how they had to use USIWire or some other similar library, **and then modify the library for the I2C device to use that library**. That situation was just awful, and everyone involved including myself hated it. And so we lashed together the three different versions of Wire, made the API match, and used #ifdefs to select the one that works on the available hardware. The same thing was done for SPI, since some parts have harware SPI (x8, x7, x41 and 1634), but others don't. I would not expect softI2CMaster to work on a "modern AVR" without serious work by someone who understood the code deeply (ie,not me). Also, softI2CMaster ah... kind of sucked. The error reporting is bad, it hogs space, you can't configure the clock speed (the function call is there, but it's just a do-nothing to let sketches compile, because for most users then, the alternative was "give up on using the tinyAVR parts completely" (a lot of people re not comfortable touching the libraries. **I don't really understand why you want softI2CMaster on a modern AVR, period** - since I think 2.4. or 2.5, we have supported simultaneously acting as a master to and slave on the same I2C bus to handle the use case you are talking about. See the documentation for the Wire library included with the core. It has a lot more functionality (and takes like half the flash) than the "stock" version - it has been almost totally reimplemented. it's all done with real hardware TWI, and it works considerably better than SoftI2CMaser ever did.

…

On Wed, Mar 9, 2022 at 10:15 AM Ed Mandy ***@***.***> wrote: I have a project where I want my megaTinyCore attiny to be an i2c slave, but I want it to aggregate data from its own GPIO and some other i2c slaves (and then serve that data over i2c to the main host system). I will likely be using 2-series parts for this (attiny427 and attiny1627). I thought that using the Wire.h library for serving the data to the host would be the best approach. To read data from other i2c slaves, I thought that the SoftI2CMaster library would be the best option. Please tell me if there are other good options that I should consider. I had problems using SoftI2CMaster, as the assembly code didn't like working with megaTinyCore. In my early searching on the problem, I noticed that ATTinyCore has a SoftI2CMaster library included ( https://github.com/SpenceKonde/ATTinyCore/blob/master/avr/libraries/Wire/src/SoftI2CMaster.h), but megaTinyCore does not. That makes me wonder why. Moving on, I'm no AVR assembly expert, but I dug into it a bit. It seems that the inline ASM constraints were the main issue here. I dug a bit, and I figured that the memory mapping of the GPIO registers was the problem here. To use SoftI2CMaster, you have to define the pins like this. #define SDA_PIN 3 #define SDA_PORT PORTA #define SCL_PIN 5 #define SCL_PORT PORTB #include <SoftI2CMaster.h> I tried switching that to VPORTA or VPORTB, becuase the PORTx mappings are larger than 1 byte (which the inline ASM constraints didn't like). That didn't work. Anyway, fast forward a bit, and here's what I came up with that seems to work (but I think is a hack that should be implemented in a better way for mass consumption). I changed the library's SoftI2CMaster.h to add an #if around the definitions like this. #if !(defined(SDA_DDR) && defined(SDA_OUT) && defined(SDA_IN) && defined(SCL_DDR) && defined(SCL_OUT) && defined(SCL_IN)) #define SDA_DDR (_SFR_IO_ADDR(SDA_PORT) - 1) #define SCL_DDR (_SFR_IO_ADDR(SCL_PORT) - 1) #define SDA_OUT _SFR_IO_ADDR(SDA_PORT) #define SCL_OUT _SFR_IO_ADDR(SCL_PORT) #define SDA_IN (_SFR_IO_ADDR(SDA_PORT) - 2) #define SCL_IN (_SFR_IO_ADDR(SCL_PORT) - 2) #endif Then, in my .INO, I would explicity set these (instead of setting SDA_PORT and SCL_PORT). That looks something like this. #define SDA_DDR 0x00 //VPORTA_DIR = 0x00, VPORTB_DIR = 0x04, VPORTC_DIR = 0x08 #define SCL_DDR 0x00 //VPORTA_DIR = 0x00, VPORTB_DIR = 0x04, VPORTC_DIR = 0x08 #define SDA_OUT 0x01 //VPORTA_OUT = 0x01, VPORTB_OUT = 0x05, VPORTC_OUT = 0x09 #define SCL_OUT 0x01 //VPORTA_OUT = 0x01, VPORTB_OUT = 0x05, VPORTC_OUT = 0x09 #define SDA_IN 0x02 //VPORTA_IN = 0x02, VPORTB_IN = 0x06, VPORTC_IN = 0x0A #define SCL_IN 0x02 //VPORTA_IN = 0x02, VPORTB_IN = 0x06, VPORTC_IN = 0x0A #define SDA_PIN 5 #define SCL_PIN 4 #define I2C_FASTMODE 1 #include <SoftI2CMaster.h> So, I hope that's helpful. From my quick research, I think this would work for all the megaTinyCore parts, but I also think there's likely a much better way to do this. I would guess that SoftI2CMaster.h could be made to use the "#define SDA_PORT PORTA" method and map that to VPORTA with the proper offsets for the _DIR, _OUT, and _IN registers. My point here is to start some discussion in case this helps anyone else or helps to add SoftI2CMaster into megaTinyCore. Also, *I would still like to know if SoftI2CMaster is even the right thing I should be using.* Thanks! — Reply to this email directly, view it on GitHub <#664>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTXEW6AMZXYDGPNOUZBNADU7C6CTANCNFSM5QJ24CFA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

--

____________ Spence Konde Azzy’S Electronics New products! Check them out at tindie.com/stores/DrAzzy GitHub: github.com/SpenceKonde ATTinyCore <https://github.com/SpenceKonde/ATTinyCore>: Arduino support for all pre-2016 tinyAVR with >2k flash! megaTinyCore <https://github.com/SpenceKonde/megaTinyCore>: Arduino support for all post-2016 tinyAVR parts! DxCore <https://github.com/SpenceKonde/DxCore>: Arduino support for the AVR Dx-series parts, the latest and greatest from Microchip! Contact: ***@***.***

0 replies

SpenceKonde · 2022-03-09T17:52:02Z

SpenceKonde
Mar 9, 2022
Maintainer

Oh - and don't use anything from the masteer branch of attinycore, it in not where any of the updates are. 2.0.0-dev will replace master within a few weeks, encompassing 18 months of changes, some of them being the largest in the history of that core. Wire has had major refactoring done twice in that time iirc

…

____________ Spence Konde Azzy’S Electronics New products! Check them out at tindie.com/stores/DrAzzy GitHub: github.com/SpenceKonde ATTinyCore: Arduino support for almost every ATTiny microcontroller Contact: ***@***.***

On Wed, Mar 9, 2022, 10:15 Ed Mandy ***@***.***> wrote: I have a project where I want my megaTinyCore attiny to be an i2c slave, but I want it to aggregate data from its own GPIO and some other i2c slaves (and then serve that data over i2c to the main host system). I will likely be using 2-series parts for this (attiny427 and attiny1627). I thought that using the Wire.h library for serving the data to the host would be the best approach. To read data from other i2c slaves, I thought that the SoftI2CMaster library would be the best option. Please tell me if there are other good options that I should consider. I had problems using SoftI2CMaster, as the assembly code didn't like working with megaTinyCore. In my early searching on the problem, I noticed that ATTinyCore has a SoftI2CMaster library included ( https://github.com/SpenceKonde/ATTinyCore/blob/master/avr/libraries/Wire/src/SoftI2CMaster.h), but megaTinyCore does not. That makes me wonder why. Moving on, I'm no AVR assembly expert, but I dug into it a bit. It seems that the inline ASM constraints were the main issue here. I dug a bit, and I figured that the memory mapping of the GPIO registers was the problem here. To use SoftI2CMaster, you have to define the pins like this. #define SDA_PIN 3 #define SDA_PORT PORTA #define SCL_PIN 5 #define SCL_PORT PORTB #include <SoftI2CMaster.h> I tried switching that to VPORTA or VPORTB, becuase the PORTx mappings are larger than 1 byte (which the inline ASM constraints didn't like). That didn't work. Anyway, fast forward a bit, and here's what I came up with that seems to work (but I think is a hack that should be implemented in a better way for mass consumption). I changed the library's SoftI2CMaster.h to add an #if around the definitions like this. #if !(defined(SDA_DDR) && defined(SDA_OUT) && defined(SDA_IN) && defined(SCL_DDR) && defined(SCL_OUT) && defined(SCL_IN)) #define SDA_DDR (_SFR_IO_ADDR(SDA_PORT) - 1) #define SCL_DDR (_SFR_IO_ADDR(SCL_PORT) - 1) #define SDA_OUT _SFR_IO_ADDR(SDA_PORT) #define SCL_OUT _SFR_IO_ADDR(SCL_PORT) #define SDA_IN (_SFR_IO_ADDR(SDA_PORT) - 2) #define SCL_IN (_SFR_IO_ADDR(SCL_PORT) - 2) #endif Then, in my .INO, I would explicity set these (instead of setting SDA_PORT and SCL_PORT). That looks something like this. #define SDA_DDR 0x00 //VPORTA_DIR = 0x00, VPORTB_DIR = 0x04, VPORTC_DIR = 0x08 #define SCL_DDR 0x00 //VPORTA_DIR = 0x00, VPORTB_DIR = 0x04, VPORTC_DIR = 0x08 #define SDA_OUT 0x01 //VPORTA_OUT = 0x01, VPORTB_OUT = 0x05, VPORTC_OUT = 0x09 #define SCL_OUT 0x01 //VPORTA_OUT = 0x01, VPORTB_OUT = 0x05, VPORTC_OUT = 0x09 #define SDA_IN 0x02 //VPORTA_IN = 0x02, VPORTB_IN = 0x06, VPORTC_IN = 0x0A #define SCL_IN 0x02 //VPORTA_IN = 0x02, VPORTB_IN = 0x06, VPORTC_IN = 0x0A #define SDA_PIN 5 #define SCL_PIN 4 #define I2C_FASTMODE 1 #include <SoftI2CMaster.h> So, I hope that's helpful. From my quick research, I think this would work for all the megaTinyCore parts, but I also think there's likely a much better way to do this. I would guess that SoftI2CMaster.h could be made to use the "#define SDA_PORT PORTA" method and map that to VPORTA with the proper offsets for the _DIR, _OUT, and _IN registers. My point here is to start some discussion in case this helps anyone else or helps to add SoftI2CMaster into megaTinyCore. Also, *I would still like to know if SoftI2CMaster is even the right thing I should be using.* Thanks! — Reply to this email directly, view it on GitHub <#664>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTXEW6AMZXYDGPNOUZBNADU7C6CTANCNFSM5QJ24CFA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

0 replies

TheFlav · 2022-03-09T17:55:03Z

TheFlav
Mar 9, 2022
Author

I'm no I2C/TWI expert, so this is all good info. I can look more into using the attiny as a slave and a master on the same bus. However, my host system is the Raspberry Pi, and the datasheet for the BCM that we are testing with shows that it can do "I2C single master only operation". I assume that means that the Pi will not be happy if there are other devices acting as a master on the I2C bus.

I had hoped to insulate this by having a "main" i2c bus and then a secondary i2c bus where the attiny talked to the other peripherals (that the Pi didn't need to talk to).

I'm definitely open to suggestions.

Thank you very much!

P.S. I'm not using attinycore at all. I'm only using megaTinyCore. I just came across the SoftI2CMaster stuff in attinycore when I was trying to figure out what was going wrong.

0 replies

SpenceKonde · 2022-03-09T18:08:48Z

SpenceKonde
Mar 9, 2022
Maintainer

The tricky thing about that term is that it sometimes means what you interpreted it as. Other times, it means the device in question acting as either a master or slave. The AVR Dx parts address this with "dual mode" which moves the slave function tona different pair of pins. But it's not offered on the tinies. If they eve5 get around to releasing the DD- series, that will be available in 14 and 20 packages (and support dual mode) and will come in 16, 32, and 64k flash sizes. That obviously raises some questions about the future of the tinyavr product line, particularly seeing as there would be a part number conflict between a furure 3-series part with 16k flash and 14 pins ans an existing part...

…

____________ Spence Konde Azzy’S Electronics New products! Check them out at tindie.com/stores/DrAzzy GitHub: github.com/SpenceKonde ATTinyCore: Arduino support for almost every ATTiny microcontroller Contact: ***@***.***

On Wed, Mar 9, 2022, 12:55 Ed Mandy ***@***.***> wrote: I'm no I2C/TWI expert, so this is all good info. I can look more into using the attiny as a slave and a master on the same bus. However, my host system is the Raspberry Pi, and the datasheet for the BCM that we are testing with shows that it can do "I2C single master only operation". I assume that means that the Pi will not be happy if there are other devices acting as a master on the I2C bus. I had hoped to insulate this by having a "main" i2c bus and then a secondary i2c bus where the attiny talked to the other peripherals (that the Pi didn't need to talk to). I'm definitely open to suggestions. Thank you very much! P.S. I'm not using attinycore at all. I'm only using megaTinyCore. I just came across the SoftI2CMaster stuff in attinycore when I was trying to figure out what was going wrong. — Reply to this email directly, view it on GitHub <#664 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTXEW6I7V6F2TA2AEAVJFLU7DQYZANCNFSM5QJ24CFA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you commented.Message ID: ***@***.*** com>

0 replies

trimarco232 · 2022-03-09T18:51:21Z

trimarco232
Mar 9, 2022

"So, I hope that's helpful. "
Hi, thanks
I have used SoftI2CMaster with older atmegas, so yes, it might be useful in some special cases, like :

i2c pins are all used by other peripherals
i2c pins can't be achieved because of pcb congestion
some i2c devices have few i2c adresses, so a second i2c port can be a solution

4 replies

TheFlav Mar 9, 2022
Author

I will say that, in my VERY limited testing, the SoftI2CMaster seems to be working well enough for what I might want to ultimately use it for. At the moment, this is just an idea that I wanted to flesh out a bit more. I don't know what we'll do in the actual design down the road. For now, our attiny1627 is fine without another slave i2c device, but we're considering adding more features.

There are some other reasons that it might be nice to have everything on one i2c bus, and there are other reasons that it could be good to have physical separation.

SpenceKonde Mar 10, 2022
Maintainer

Well, if it works, that's good then. I am surprised to hear it!

That library is nothing unique to ATTinyCore, all we did was integrate it into Wire.h s that there was a single library to work on parts that need it. We do not maintain it independently, If you find that there are changes needed to make it work with the new parts, or to otherwise improve it, you want to get those back to the maintainer of that library.

TheFlav Mar 10, 2022
Author

I should have mentioned this up front, so I'll add it up there for anyone that comes along. I am using the latest code from https://github.com/felias-fogg/SoftI2CMaster which has been updated recently and fairly frequently. I think that my mention of the link to SoftI2CMaster.h in ATTinyCore may have have been misleading. I only mentioned that in reference to searching for SoftI2CMaster for attiny parts and figuring that SpenceKonde must have had some knowledge of it.

(I also added the same text as an update to my original post.)

SpenceKonde Mar 10, 2022
Maintainer

Probably should update the version used in the legacy core. Ugh. I don't want to have to do that.

TheFlav · 2022-03-10T14:54:27Z

TheFlav
Mar 10, 2022
Author

I created a discussion about this at SoftI2CMaster's Github page.
felias-fogg/SoftI2CMaster#74

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

SoftI2CMaster with megaTinyCore #664

{{title}}

{{editor}}'s edit

{{editor}}'s edit

Replies: 6 comments 4 replies

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

Select a reply

SoftI2CMaster with megaTinyCore #664

TheFlav Mar 9, 2022

Replies: 6 comments · 4 replies

SpenceKonde Mar 9, 2022 Maintainer

SpenceKonde Mar 9, 2022 Maintainer

TheFlav Mar 9, 2022 Author

SpenceKonde Mar 9, 2022 Maintainer

trimarco232 Mar 9, 2022

TheFlav Mar 9, 2022 Author

SpenceKonde Mar 10, 2022 Maintainer

TheFlav Mar 10, 2022 Author

SpenceKonde Mar 10, 2022 Maintainer

TheFlav Mar 10, 2022 Author

TheFlav
Mar 9, 2022

Replies: 6 comments 4 replies

SpenceKonde
Mar 9, 2022
Maintainer

SpenceKonde
Mar 9, 2022
Maintainer

TheFlav
Mar 9, 2022
Author

SpenceKonde
Mar 9, 2022
Maintainer

trimarco232
Mar 9, 2022

TheFlav Mar 9, 2022
Author

SpenceKonde Mar 10, 2022
Maintainer

TheFlav Mar 10, 2022
Author

SpenceKonde Mar 10, 2022
Maintainer

TheFlav
Mar 10, 2022
Author