CAN Part 8 - ISO 15765, KWP2000, ODB, and UDS (Automotive Diagnostic Command Set)

In the earlier parts of this blog we CAN hardware, software, APIs, databases, CAN Signals and CAN Frames.  In this part we are going to talk more about one of the software techniques that has been developed to get around some of the technical limitations to CAN to help allow for other types of data, with larger payloads, to be sent and received.

Standard CAN Example

As defined by the CAN 2.0 spec, a standard or extended CAN Frame consists of up to 8 bytes of payload, in 1 byte increments.  This is a hard limitation and cannot be changed, without changing to a new specification like CAN FD which also requires new hardware from multiple devices.  But often times when we start getting into automotive ECU development we find ourselves wanting to use CAN in ways that maybe it wasn't intended to be used for.  For example the CAN data we've been talking about so far has primarily been periodic information.  A device may send out the status of an analog input every 100ms by sending out a single frame that contains this signal information.  This information is great for looking at values that may change rapidly over time, and having a waveform of all data at small time increments like that is very useful.

Engine Trouble Codes Example (DTC)

But lets say for a moment that I want to use CAN for something else.  Let say for instance my check engine light has turned on in my car and I want to know why.  An automotive technician will plug in a device on the CAN bus and it will ask the ECU what is wrong.  The ECU will then respond and tell the scanner what the list of things are that are currently wrong, and a list of things that were wrong but aren't any more, to give some history of issues to the technician.  These trouble codes are known as Diagnostic Trouble Code (DTC), and aren't periodically sent, and instead only are sent when another device asks it to.  DTCs are the kind of data that should not be sent every 100ms because over the life of the ECU there is a small chance that a device that wants to know what the DTCs are, is even listening.  So this type of data is sent only on the event that it was requested.  Standard and Extended CAN data can be sent this way but there is another issue with DTCs.

The list of potential DTCs any ECU can have set could be in the hundreds.  And we cannot know how many are set, or which ones are set, until the ECU tells us.  This causes the response from the ECU to vary the amount of data it is going to send back depending on other conditions, and it will likely need to be able to send back more than just 8 bytes of data which a CAN frame is limited to.

ISO 15765 Multi Frame Messaging

This is where the ISO 15765 protocol comes in.  The protocol is used in places where more than 8 bytes of data need to be received, or more than 8 bytes of data needs to be sent as a response.  It is also used in places were the data that needs to be sent or received is not the kind of data that should be blindly broadcasted on a network autonomously.

How does it work?

The basics of the 15765 are pretty simple.  A device will send out a single frame telling another device that it has something it wants to say.  Now if the thing it wants to say is small in size, it can be all contained in that one frame.  But often times we need to send more data so the sender will wait for the receiver to get the data, and then respond.  In the response the receiver will state how many frames the sender can send at once, and how much time can be between each frame.  Older CAN devices had a small buffer size, and wouldn't be able to receive data too fast.  So the sender is supposed to receive this one frame, known as a flow control, and then send back more data in with the timing specified.  This helps prevent the device receiving the data from being overloaded, since the sender will only send the specified number of frames requested.  Once all of the data gets received, the receiving device will act on it, and then send a reply back.  Just like the request, the response can also be a single frame, or multiple frames.  If it is multiple frames then it will send the first frame, and then the original sender device will send back a frame stating how many frames it can receive, and what timing should be between them.  These steps are a bit confusing so here is a flow diagram explaining it, generated using the NI State Diagram Toolkit.  The state machine starts on the green node, and ends on the red one.

Looking at this flow diagram things might be a bit confusing, so lets break it down a bit more and talk about the various components and checks that go into using raw frames for performing 15765 queries.

CAN IDs

So in Part 2 we first talked about CAN Frames and mentioned the payload limitation, as well as the fact that each frame has an ID.  Well when using ISO 15765 there are only two IDs ever used.  One ID is for the request and one ID is for the reply.  The ID performing the request is known as the Command Receive Object (CRO), and the ID for the response is known as the Data Transmission Object (DTO).

This means that all requests from other devices on the bus will have the same IDs, and the same priority.  This also means that the reply sent back will go to all devices on the bus, and it is the the responsibility of the device to look at the payload sent back to determine if the reply it found was intended for that device, or if the this response was to another devices request that was issued.  On a normal automotive CAN bus these 15765 messages aren't going out often, and as mentioned with our DTC example is something that is useful for debugging but isn't intended to be part of normal operation.  As a result this issue of multiple requester at once is probably a rare case and in most situations every reply you see to an ISO 15765 message, is going to be from a request your device issued.

The CAN IDs used in 15765 can vary from device to device, and will be specific to your device.  It is common to have different request and response IDs for each device on the bus so this information needs to be looked up in the documentation for the device you are trying to talk to.  In the following examples we will use CAN ID 0x70 for the request and 0x7F for the response.

Simple Example

The following is a simple example where we send a small amount of request data, and we receive a small reply following the state machine from above.

Write Single Frame

At the start of our state machine we have a state where we either write a Single Frame if our payload is less than or equal to 7 bytes, or we will write the first frame in many if it is more.  Lets start with writing the single frame and how it looks.  Lets say we intend on writing the following 3 bytes of data 0x22 00 02, when we do this using 15765 our Single Frame we send will look like this:

--> ID 0x70 Payload 0x03 22 00 02 00 00 00 00

The padding of the frame to 8 bytes isn't always required and from now on will not be included but some hardware may expect it.  The rest of the payload is pretty simple.  The first byte represents the number of bytes in our transmission.  Because this first byte always says how many bytes to expect, it leaves us with only 7 bytes of usable payload.

Some messages don't require a response, and if we are sending a Single Frame, and we don't care about the response we are done.  If we do care about the response we then go and start reading frames looking for the response in the Read Reply state.

Read Reply

So in our example we have successfully sent one frame which contained all the data we wanted to send.  The next step is to look for a reply.  The only reply we care about, is one on the CAN ID associated with a response and in our case is 0x7F.  All other frames can be ignored.  If too much times goes on and we haven't seen a frame with ID 0x7F we go to Done where we can throw an error.  This timeout is usually on the order of 1 second but can vary depending on hardware.  Assuming we do see a frame with ID 0x7F the next step is to identify what kind of frame it is.

Process Read (Single Request)

So we sent a frame, and we just got one back that is in response to what we sent.  Now lets identify what kind of frame it is in Process Read state.  Just like the first step in our state machine, the reply could be a single frame taking 7 or less bytes, or it may need to be multiple frames.  For simplicity lets say the reply is also 7 bytes or less and consists of the following data, 0x01 02 03 04.  As a result the reply may look something like this:

<-- ID 0x7F Payload 0x62 04 01 02 03 04

The first byte of the response is 0x62 because the first byte of the request was 0x22, and the response needs to be the same as the request, but with 0x40 added.  This helps us know the response we found was to a request we issued.  If the first byte was anything else we know the message was not for us.  The only other valid first byte that is for us is 0x7F which is a negative response which we'll get into later.

So 0x62 means the reply is for us, and 0x04 means we have 4 bytes of data to receive, and then after that we have the 4 bytes of data.  So in this example we sent the three bytes of data 0x22 00 02, and we received four bytes of data 0x01 02 03 04.  Using another protocol on top of 15765 known as UDS (ISO 14229), this example data corresponds to asking the device to read an identifier (memory address) 0x00 02, and the device returning the memory at that address which was 0x01 02 03 04

--> ID 0x70 Payload 0x03 22 00 02 00 00 00 00    #Request to read data at address 0x00 02
<-- ID 0x7F Payload 0x62 04 01 02 03 04              #Response data at that address 0x01 02 03 04

 Negative Response

If we got a negative response from the device it would look something like this:

<-- ID 0x7F Payload 0x7F 22 11

The 0x7F means this is a negative response, to the request 0x22, and the status of the response is 0x11.  There can be many custom response codes but there are a set of them that are mostly standard.

Negative Response Table
Response Code Response
0x10 General reject
0x11 Service not supported
0x12 Sub-Function not supported
0x13 Incorrect message length or invalid format
0x14 Response too long
0x21 Busy repeat request
0x22 Conditions not correct
0x24 Request sequence error
0x25 No response from sub-net component
0x26 Failure prevents execution of requested action
0x31 Request out of range
0x33 Security access denied
0x35 Invalid key
0x36 Exceeded number of attempts
0x37 Required time delay not expired
0x38 - 0x4F Reserved by Extended Data Link Security
0x70 Upload/Download not accepted
0x71 Transfer data suspended
0x72 General programming failure
0x73 Wrong Block Sequence Counter
0x78 Request correctly receive, but response is pending
0x7E Sub-Function not supported in active session
0x7F Service not supported in active session

Looking at the state machine the only other option is needing to reissue the request.  If this happens the sender should send the request again.  One other negative response that doesn't end the state machine is if the device asked for more time 0x78.  If this negative response is seen  the receiver should reset the timeout value and continue looking for the next positive, or negative response.

More Complicated Example

The following example is a bit more complicated and will involve sending a larger amount of request data, and we receive a larger amount of data in a reply, again using the state machine from earlier.

Write First Frame

The example listed above is a very simple one.  We send a single frame, and we get a single frame.  but 15765 is more useful when sending large amounts of data.  So lets walk through an example of sending more data.  If we want to send more than 7 bytes of data we still need to start by sending the First Frame and then we will send more when we are ready.  Lets say for instance we want to write the following 17 bytes of data 0x2E 00 02 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D, our First Frame will look like this:

--> ID 0x70 Payload 0x10 11 2E 00 02 00 01 02

Here the first two bytes are used telling the receiver what the transmission will contain.  The second nibble in the first byte of the payload being a 1 is telling the receiver that there is more data than just this one frame.  This leave the remaining lower nibble, and 2nd byte to be used to tell the receiver how many bytes to expect and in our case it is 0x00 11 or decimal 17 which is the full payload we need to send.  Since the first two bytes of the payload are taken by the number of bytes to be sent, this leaves us with only 6 bytes to send usable data through, and so we take the first 6 bytes of our 17 bytes and send it.

Reply is Flow Control

So after sending the first frame, we will wait reading new frames just like the previous example, but we expect to see a flow control frame.  This is a frame that tells us how many frames the device can read at once, and how much time needs to be between each one.  The reply will look something like this:

<-- ID 0x7F Payload 0x30 05 0A

Here the 0x30 tells us this is a flow control message, 0x05 means the device can accept the next 5 frames before needing to reissue the flow control.  This helps so that we don't overload the device and we know that we are sending data to a device that is still listening.  And 0x0A is the separation time needed between each frame.  The units of this are in milliseconds from 0 to 127, and in microseconds from 100 to 900.  This doesn't need to be precise and is a lower bounds, so the device should be fine if we send them a little slower than that, but we shouldn't send it faster than requested.  A zero for the number of frames means just send them all, and a zero for the separation time means no delay between frames is needed.

Write Chunk of Data

After we sent the first frame, and we received the flow control response, we are now able to start sending chunks of data.  We need to abide by the flow control rates, and number of frames.  In this case we are only going to need to send 2 more frames and the device said we can send up to 5, as long as we separate them by 10 milliseconds.  Because of that our next two frames we write will look like this:

--> ID 0x70 Payload 0x21 03 04 05 06 07 08 09
--> ID 0x70 Payload 0x22 0A 0B 0C 0D

The first byte in the payload here goes from 0x21 and increments to 0x2F then back around to 0x21 with each frame sent in this chunk.  In our case we only needed to send 2 frames.

Read First Frame Reply

So we sent some data, waited for the device to reply, then we were able to send the rest of our payload.  Now we need go into perform similar work as we did in the first example.  We continue to read data looking for a reply.  Just as in the earlier example the reply may tell us to re-transmit a request, wait longer for a reply, or get the reply.  The reply can be a single frame as in the previous example, or we can have a multi frame reply, just like our multi frame request.  We've already seen a single frame reply example so lets look at the multi frame reply.  In this example lets assume the device wants to reply with the following 21 bytes of data 0x6E 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14.  This will start with the first few bytes of the payload as before:

<-- ID 0x7F Payload 0x10 15 6E 01 02 03 04 05

Here we see the second nibble telling us this is the first frame of many because it is a 1, and the first nibble and second byte is 0x00 15 telling us there are 21 bytes total to receive.  Then the remaining 6 bytes is the first 6 of the reply.

Write Flow Control

So we've sent our request, and we got the first frame of the reply, but now we need to tell the device how many frames it can send us, and how much time can be between each frame.  In most cases our hardware can just read as many frames as possible since they will sit in our hardware buffer, and we don't really need any time between.  But just for an example we will tell the device we can only read 4 frames at a time, and we need 20 milliseconds between each:

--> ID 0x70 Payload 0x30 04 14

After we send this the device is free to send up to 4 frames with 20 milliseconds between them.  To tell the device to send all frames with no delay replace the last two bytes to 0.

Read Consecutive

So we sent a request, and we got the first 6 bytes of the reply, and we sent the flow control telling the device to send the rest which will look like this, which mirrors how the request message sent multiple frames earlier:

<-- ID 0x7F Payload 0x21 06 07 08 09 0A 0B 0C
<-- ID 0x7F Payload 0x22 0D 0E 0F 10 11 12 13
<-- ID 0x7F Payload 0x23 14

After this it is the responsibility of our software to concatenate all of the data into the request and response.  Here is the full transaction that takes place:

--> ID 0x70 Payload 0x10 11 2E 00 02 00 01 02     #First frame of multiple in request
<-- ID 0x7F Payload 0x30 05 0A                             #Flow control telling us how to send the request
--> ID 0x70 Payload 0x21 03 04 05 06 07 08 09     #More data in request
--> ID 0x70 Payload 0x22 0A 0B 0C 0D                 #More data in request
<-- ID 0x7F Payload 0x10 15 6E 01 02 03 04 05     #First frame of multiple in reply
--> ID 0x70 Payload 0x30 04 14                               #Flow control telling how we can receive data
<-- ID 0x7F Payload 0x21 06 07 08 09 0A 0B 0C   #More data in response
<-- ID 0x7F Payload 0x22 0D 0E 0F 10 11 12 13    #More data in response
<-- ID 0x7F Payload 0x23 14                                    #More data in response


(If you made it this far congratulations, that was a very large wall of text in those two basic examples.)


Other Protocols

The ISO 15765 just defines how to read and write multi-frame data, but doesn't define what that data is.  The Unified Diagnostic Services (UDS), Keyword Protocol 2000 (KWP2000), and On-Board Diagnostic (OBD) and three software layers that each sit on ISO 15765 to perform operations like reading and clearing DTCs.

Automotive Diagnostic Command Set (ADCS)

The NI toolkit ADCS is basically the software implementation of ISO 15765, KWP2000, OBD, and UDS for NI-CAN and NI-XNet devices.  This is fine if you have these hardware options, but the software to implement this in any other language is laid out before you.  So I decided to take the ADCS toolkit, and replace the core calls to the NI DLL driver, which is locked down, and replace it with my G implementation.  I posted this idea on the Idea Exchange.  I also ran into several bugs with the toolkit and getting support wasn't easy, so I eventually posted the basic G implementation of the toolkit for NI-CAN hardware here.  With the G implementation before us, all that is needed to this functionality in other hardware is to just replace the CAN Read, and CAN Write functions with the hardware API of your choice.  This means that ISO 15765, and really the ADCS functionality is now available for any CAN hardware in LabVIEW.  I've created a set of code and wrappers and inlcuded them in the Hooovahh CAN ISO 15765 package, which works with the CAN Drivers package for 3rd party devices from earlier.


So with this newly created G code and palette, what steps are needed to read the DTCs on an ECU, and then clear the DTCs?  Take a look at the following example.


Here we open the first Vector port, initialize the protocol based on the request and reply ID, read the DTCs, Clear DTCs, then close the references  There are some default values on some subVIs but the values used are the most common.  All of the ISO15765 protocol, and the UDS protocol that sits on top of it is all available in the G implementation inside those VIs.  These protocols and APIs work with any KVaser, Vector, or Intrepid CAN hardware device, since it uses the Hooovahh CAN Drivers package from earlier to perform the raw CAN frame reading and writing.  The hardware drivers aren't wrapped into a nice hardware abstraction class, but they probably should be, I just didn't get around to it.  Regardless the point of this code is not to give a working solution for your project, but instead to demonstrate how ISO 15765 operates, and demonstrate that it can be used on any CAN hardware as long as you can read and write raw CAN frames.  This means adding support for other hardware types like NI-CAN, NI-XNet, or any other device can be done without too much extra work.

Part 9...

In this blog series we've covered a lot of information.  From hardware, to APIs, to CAN Frames, CAN Signals, databases, advanced XNet features, and now a software layer on top of CAN to help facilitate multi frame data, and event based communication.  What else could we possibly talk about?  The subjects are now going to get a bit more specific.  So in Part 9 we will have a quick discussion on the availability of what I call the bucket filling problem in XNet, and how to handle functions like CRC, and incrementing counter signals.