Post by thread:[PIC] PIC24 CAN

Robert Csaba Molnar wrote: > I have a 2 wire BUS connected to CAN_H and CAN_L, on both end of the > BUS I've connected two 120 ohm resistors. Then your impedence is too low. Each end needs to be terminated by a *single* 120 Ohm resistor, making the total bus impedence 60 Ohms. In your case it's 30 Ohms, which is lower than the drivers are intended to drive. > If I set the PIC24 CAN > controller in loop-back mode, I get the TX and RX interrupts working. > When configured in normal operation mode it doesn't work. You really think "doesn't work" describes the problem well enough for anyone to help!? Duh! Obviously more information is needed for anyone to help or for you to diagnose it. Do you see the message on the CANH and CANL lines? Are both floating about 2.5V with about 900mV excursions but in opposite directions for each bit? Is there another node on the bus so that a ACK is generated? Does the sending PIC give any error indication? Does it keep resending the message? What does the receiving node see? Have you put a CAN sniffer on the bus to see what frames are going by? ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000

Sorry for the ambiguous message. Each end is in fact terminated by a single 120 Ohm and the impedence (measured) is 60 Ohm. There are 2 nodes on the BUS, I didn't get any errors on the sending PIC and it does try to resend the message (there is an infinite loop on the sending PIC sending the messages). I do not have any means to see if there is anything on the CAN lines, nor a sniffer. What's odd is that I'm using DMA and the TX buffer interrupt flag is never set. --- On Tue, 12/14/10, Olin Lathrop <spam_OUTolin_piclistTakeThisOuTembedinc.com> wrote: From: Olin Lathrop <.....olin_piclistKILLspam@spam@embedinc.com> Subject: Re: [PIC] PIC24 CAN To: "Microcontroller discussion list - Public." <piclistKILLspammit.edu> Date: Tuesday, December 14, 2010, 2:53 PM Robert Csaba Molnar wrote: > I have a 2 wire BUS connected to CAN_H and CAN_L, on both end of the > BUS I've connected two 120 ohm resistors. Then your impedence is too low. Each end needs to be terminated by a *single* 120 Ohm resistor, making the total bus impedence 60 Ohms. In your case it's 30 Ohms, which is lower than the drivers are intended to drive. > If I set the PIC24 CAN > controller in loop-back mode, I get the TX and RX interrupts working. > When configured in normal operation mode it doesn't work. You really think "doesn't work" describes the problem well enough for anyone to help!? Duh! Obviously more information is needed for anyone to help or for you to diagnose it. Do you see the message on the CANH and CANL lines? Are both floating about 2.5V with about 900mV excursions but in opposite directions for each bit? Is there another node on the bus so that a ACK is generated? Does the sending PIC give any error indication? Does it keep resending the message? What does the receiving node see? Have you put a CAN sniffer on the bus to see what frames are going by? ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000

Robert Csaba Molnar wrote: > There are 2 nodes on the BUS, I didn't get any errors on the sending > PIC and it does try to resend the message (there is an infinite loop > on the sending PIC sending the messages). So is your firmware doing the resending or is the hardware? Your firmware should only send it once. If the hardware resends it, then that tells you it didn't get the ACK. If the firmware keeps resending it, then you don't know what's going on. > I do not have any means to see if there is anything on the CAN lines, Not even a oscilloscope? A oscilloscope is a very useful tool for debugging things like this, to the point where trying to do it without one is mostly a waste of time. ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000

A digital volt meter can at least tell you if something is being sent, although it can't tell you WHAT is being sent. Kerry Olin Lathrop wrote: {Quote hidden}> Robert Csaba Molnar wrote: > >> There are 2 nodes on the BUS, I didn't get any errors on the sending >> PIC and it does try to resend the message (there is an infinite loop >> on the sending PIC sending the messages). >> > > So is your firmware doing the resending or is the hardware? Your firmware > should only send it once. If the hardware resends it, then that tells you > it didn't get the ACK. If the firmware keeps resending it, then you don't > know what's going on. > > >> I do not have any means to see if there is anything on the CAN lines, >> > > Not even a oscilloscope? A oscilloscope is a very useful tool for debugging > things like this, to the point where trying to do it without one is mostly a > waste of time. > > > ******************************************************************** > Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products > (978) 742-9014. Gold level PIC consultants since 2000. > -- Internal Virus Database is out-of-date. Checked by AVG Anti-Virus. Version: 7.0.289 / Virus Database: 267.11.13 - Release Date: 10/6/05

> What's odd is that I'm using DMA and the TX buffer interrupt flag is never set. On the Transmit end? I haven't played with CAN, so assume this interrupt should be set when the buffer has been transmitted - presumably with an appropriate acknowledgement. If this is the case then not getting an acknowledgement is a likely cause. If there is no acknowledgement, are both nodes working at the same baud rate ?? But as Olin says, some basic test equipment to see what happens on the data lines is required. -- Scanned by iCritical.

Kerry Wentworth wrote: > A digital volt meter can at least tell you if something is being sent, Not neccessarily. The bits go by so fast that the average voltage may not change enough so show up on a voltmeter. If you just want to know something was sent, connect a CAN line to a audio preamp input. Each message should sound something like a click. If you get nothing, then you know the bus lines aren't wiggling at all. Do this test on both of the CAN lines independently to make sure they are both wiggling. Again however, without a 'scope, this is going to be quite a time sink. ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000

I don't have an oscilloscope, all I have is a digital volt meter. Using the volt meter all I see is 2.5 Volts. I will try to get an oscilloscope and see if there is something sent on the wire. Both nodes run at same baud, same PIC24 and almost the same software (a min or change in one of the nodes to keep sending the messages). --- On Tue, 12/14/10, .....alan.b.pearceKILLspam.....stfc.ac.uk <EraseMEalan.b.pearcespam_OUTTakeThisOuTstfc.ac.uk> wrote: From: alan.b.pearcespam_OUTstfc.ac.uk <@spam@alan.b.pearceKILLspamstfc.ac.uk> Subject: RE: [PIC] PIC24 CAN To: KILLspampiclistKILLspamMIT.EDU Date: Tuesday, December 14, 2010, 4:07 PM > What's odd is that I'm using DMA and the TX buffer interrupt flag is never set. On the Transmit end? I haven't played with CAN, so assume this interrupt should be set when the buffer has been transmitted - presumably with an appropriate acknowledgement. If this is the case then not getting an acknowledgement is a likely cause. If there is no acknowledgement, are both nodes working at the same baud rate ?? But as Olin says, some basic test equipment to see what happens on the data lines is required. -- Scanned by iCritical.

It's certainly possible that a single byte might go by too fast for the voltmeter to respond, but the OP indicated that the bytes are being resent continuously, and if they are not, he could modify the software so that they are. Obviously, a scope is the tool of choice, but if the only tool available is a volt meter, then it is what will have to do. Kerry Olin Lathrop wrote: {Quote hidden}> Kerry Wentworth wrote: > >> A digital volt meter can at least tell you if something is being sent, >> > > Not neccessarily. The bits go by so fast that the average voltage may not > change enough so show up on a voltmeter. > > If you just want to know something was sent, connect a CAN line to a audio > preamp input. Each message should sound something like a click. If you get > nothing, then you know the bus lines aren't wiggling at all. Do this test > on both of the CAN lines independently to make sure they are both wiggling. > > Again however, without a 'scope, this is going to be quite a time sink. > > > ******************************************************************** > Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products > (978) 742-9014. Gold level PIC consultants since 2000. > -- Internal Virus Database is out-of-date. Checked by AVG Anti-Virus. Version: 7.0.289 / Virus Database: 267.11.13 - Release Date: 10/6/05

--- On Tue, 14/12/10, Olin Lathrop <RemoveMEolin_piclistTakeThisOuTembedinc.com> wrote: > Again however, without a 'scope, this is going to be quite > a time sink. Having gone through this process myself a few years ago I can only agree with what Olin has said. I don't know if the Newer CAN modules are easier to use, but the one in the 18F4585 is quite complex to set up and debug. I would suggest that to start with you set up the Rx side to accept ALL messages it see on the wire. Don't try to set up any masks or filters until you can get it working like that. Who's code are you using to setup the CAN module ? PeterO

Did you try the meter on AC volts? Can you slow down the baud rate? Can you start/stop transmissions? Kerry Robert Csaba Molnar wrote: {Quote hidden}> I don't have an oscilloscope, all I have is a digital volt meter. > Using the volt meter all I see is 2.5 Volts. I will try to get an oscilloscope and see if there is something sent on the wire. > Both nodes run at same baud, same PIC24 and almost the same software (a min or change in one of the nodes to keep sending the messages). > > > > --- On Tue, 12/14/10, spamBeGonealan.b.pearcespamBeGonestfc.ac.uk <TakeThisOuTalan.b.pearceEraseMEspam_OUTstfc.ac.uk> wrote: > > From: RemoveMEalan.b.pearceTakeThisOuTstfc.ac.uk <alan.b.pearceEraseME.....stfc.ac.uk> > Subject: RE: [PIC] PIC24 CAN > To: EraseMEpiclistMIT.EDU > Date: Tuesday, December 14, 2010, 4:07 PM > > >> What's odd is that I'm using DMA and the TX buffer interrupt flag is >> > never set. > > On the Transmit end? I haven't played with CAN, so assume this interrupt > should be set when the buffer has been transmitted - presumably with an > appropriate acknowledgement. If this is the case then not getting an > acknowledgement is a likely cause. If there is no acknowledgement, are > both nodes working at the same baud rate ?? > > But as Olin says, some basic test equipment to see what happens on the > data lines is required. > -- Internal Virus Database is out-of-date. Checked by AVG Anti-Virus. Version: 7.0.289 / Virus Database: 267.11.13 - Release Date: 10/6/05

I've tried to slow down the baud rate an mostly everything that can be done via software. I will get an oscilloscope and see if anything goes out on the wire. --- On Tue, 12/14/10, Kerry Wentworth <RemoveMEkwentworthEraseMEEraseMEskunkworksnh.com> wrote: From: Kerry Wentworth <RemoveMEkwentworthspam_OUTKILLspamskunkworksnh.com> Subject: Re: [PIC] PIC24 CAN To: "Microcontroller discussion list - Public." <RemoveMEpiclistTakeThisOuTspammit.edu> Date: Tuesday, December 14, 2010, 4:42 PM Did you try the meter on AC volts? Can you slow down the baud rate? Can you start/stop transmissions? Kerry Robert Csaba Molnar wrote: {Quote hidden}> I don't have an oscilloscope, all I have is a digital volt meter. > Using the volt meter all I see is 2.5 Volts. I will try to get an oscilloscope and see if there is something sent on the wire. > Both nodes run at same baud, same PIC24 and almost the same software (a min or change in one of the nodes to keep sending the messages). > > > > --- On Tue, 12/14/10, EraseMEalan.b.pearcespamspamBeGonestfc.ac.uk <RemoveMEalan.b.pearceKILLspamstfc.ac.uk> wrote: > > From: alan.b.pearceSTOPspamspam_OUTstfc.ac.uk <spamBeGonealan.b.pearceSTOPspamEraseMEstfc.ac.uk> > Subject: RE: [PIC] PIC24 CAN > To: KILLspampiclistspamBeGoneMIT.EDU > Date: Tuesday, December 14, 2010, 4:07 PM > > >> What's odd is that I'm using DMA and the TX buffer interrupt flag is >> > never set. > > On the Transmit end? I haven't played with CAN, so assume this interrupt > should be set when the buffer has been transmitted - presumably with an > appropriate acknowledgement. If this is the case then not getting an > acknowledgement is a likely cause. If there is no acknowledgement, are > both nodes working at the same baud rate ?? > > But as Olin says, some basic test equipment to see what happens on the > data lines is required. > -- Internal Virus Database is out-of-date. Checked by AVG Anti-Virus. Version: 7.0.289 / Virus Database: 267.11.13 - Release Date: 10/6/05

On 14/12/2010 14:57, Robert Csaba Molnar wrote: > I've tried to slow down the baud rate an mostly everything that can be done via software. I will get an oscilloscope and see if anything goes out on the wire. An oscilloscope is essential for anything more than basic electronics, so that is definitely a good idea. In the meantime though a couple of ideas: The audio input Olin mentioned is a good idea, nice and simple. Or if you have a PicKit2 it has a logic analyser you could use (just in case you do and you didn't know..) Or a PC sound card with a freeware scope program (only something like max 128Ksps depending on hardware, but suitable for low baud rates - make sure voltages are within card input range, divide down/buffer if necessary) Or use a free input on the PIC (or a separate PIC) to sense basic activity. If you want to do something a little more involved, someone posted a PIC based logic analyser project a few days ago which might be useful for you to check out. Quite a few PIC based scopes out there too.

On Tue, Dec 14, 2010 at 1:40 PM, Oli Glaser <EraseMEoli.glaserEraseMEtalktalk.net> wrote: > An oscilloscope is essential for anything more than basic electronics, > so that is definitely a good idea. <snip> > If you want to do something a little more involved, someone posted a PIC > based logic analyser project a few days ago which might be useful for > you to check out. Quite a few PIC based scopes out there too. I second the idea of a logic analyzer. I haven't really checked out the one that was just posted (though I did bookmark the link for future review) as I have a working computer based analyzer for now. They really help out when debugging communications, especially if they have a protocol interpreter for the protocol you're working on. I just wish the manufacturer of my unit (Intronix) would either add a few new interpreters or add a method of allowing the user to define them. I'd love a few more than what comes standard with the software. Their current answer is to export the data and do the analysis in a separate program. Josh -- A common mistake that people make when trying to design something completely foolproof is to underestimate the ingenuity of complete fools. -Douglas Adams

Thanks for the replies. I have PickKit2, I'll look into the logic analyzer. I use MPLAB and the C30 compiler from Microchip. The application is a simple. Controls access to doors. There will be up 20-30 devices connected, each controlling 1 or more doors. I managed to get everything working except the ECAN controller. In the meantime I'll try to get my hands on an oscilloscope, the audio trick is a nice idea. I took out the transceiver and set up the pins connected to it as IO, but i always get 0 on those pins. Thanks again, I'll keep you posted with my results. Robert --- On Tue, 12/14/10, Josh Koffman <@spam@joshybear@spam@spam_OUTgmail.com> wrote: From: Josh Koffman <spamBeGonejoshybearKILLspamgmail.com> Subject: Re: [PIC] PIC24 CAN To: "Microcontroller discussion list - Public." <.....piclistspam_OUTmit.edu> Date: Tuesday, December 14, 2010, 9:07 PM On Tue, Dec 14, 2010 at 1:40 PM, Oli Glaser <TakeThisOuToli.glaser.....TakeThisOuTtalktalk.net> wrote: > An oscilloscope is essential for anything more than basic electronics, > so that is definitely a good idea. <snip> > If you want to do something a little more involved, someone posted a PIC > based logic analyser project a few days ago which might be useful for > you to check out. Quite a few PIC based scopes out there too. I second the idea of a logic analyzer. I haven't really checked out the one that was just posted (though I did bookmark the link for future review) as I have a working computer based analyzer for now. They really help out when debugging communications, especially if they have a protocol interpreter for the protocol you're working on. I just wish the manufacturer of my unit (Intronix) would either add a few new interpreters or add a method of allowing the user to define them. I'd love a few more than what comes standard with the software. Their current answer is to export the data and do the analysis in a separate program. Josh -- A common mistake that people make when trying to design something completely foolproof is to underestimate the ingenuity of complete fools. -Douglas Adams

As promised here is the result: the transceivers on both devices I used (2 out of total 3) were bricked... just my luck. Replaced them and everything works fine. Probably someone connected them on a bus without the 120 Ohm resistors. Thank you all for your suggestions :) Robert --- On Tue, 12/14/10, Jan-Erik Soderholm <TakeThisOuTjan-erik.soderholmKILLspamspamtelia.com> wrote: From: Jan-Erik Soderholm <.....jan-erik.soderholmRemoveMEtelia.com> Subject: Re: [PIC] PIC24 CAN To: "Microcontroller discussion list - Public." <RemoveMEpiclistspamBeGonemit.edu> Date: Tuesday, December 14, 2010, 10:48 PM On 2010-12-14 21:32, Robert Csaba Molnar wrote: > I took out the transceiver and set up the pins connected to it as IO, > but i always get 0 on those pins. > Either the pins are non-functional or mis-configured. As long as you can't get the pins to work as normal I/O pins, there is no reason to try them as CAN pins.