TCAN2451-Q1: Schematic Review

Hi Kalpak

1. Review Below

a) If WAKE pins are unused they should be directly tied to ground to prevent parasitic wake-up. 

b) HSS1 - I see you have a 100nF before the LED - just ensure that this cap is as close to HSS1 pin as possible in layout. 

c) HSS1 - typically we would see the current limiting resistor before the LED - not after. 

d) VHSS/VSUP/VSUPB - so on this connection we generally advise a slightly different setup - while your setup isn't the worst as it does include a good amount of filtering there are some improvements that can be made - please see reference image below:

C1 = Minimum of 22uF - but could be more depending on application needs; this is to help protect against battery / ISO pulses 

C2 = 100nF - high frequency decoupling right after the reverse protection diode.

CF = 2.2uF - this value can generally be adjusted for application - however in general I'd suggest the 2.2uF

CD = 47uF needs to be at least 4 times C5

C5 = 10uF - may need to adjust as needed for application - the 10uF was used for EMC cert. with 0.8A load on VCC1

C3,C4,C6 = 100nF - as close to their respective pins as possible

LFLT = 1.5uH - but can adjust as needed. 

The biggest thing though is where the filtering is placed - the PI-Filter should be placed between VSUP/VHSS and VSUPB - VSUP and VHSS are not going to have as much EMI generation as VSUPB will - so we suggest separating the inputs with the PI filter. 

When looking at your schematic directly - it looks like for the most part you followed the suggested values  - you are using a ferrite bead instead of an inductor - so they may cause some differences but overall the bead should act as an inductor - the only thing is that the true inductor may have a bit better lower frequency response. 

However - the pi filter in your circuit comes before all three supply pins - the pi filter should be between them. 

e) Are you not using VCC2 to power VCAN? It is okay if you are using an external 5V to power VCAN and using VCC2 for something else - but that is the most common use of VCC2 but I see that VCC2 output and VCAN input are different labels. I just wanted to ensure that was the intention - because there is no direct issue here. ESR for the VCC2 output should be between 0.001 ohm and 1 ohm. 

f) The 100pF filtering caps on the CANH/CANL lines are a little large - generally we suggest to cap them to ~40pF max - however - there are a lot of applications that use 100pF - so you can use them but it could reduce the effective max bus length. I wouldn't suggest removing them - but you may want to be able to support multiple values here if issues arise. 

Other than that your schematic looks pretty good. 

2. SPI is referred to VCC1 voltage. So if VCC1 is 3.3V SPI is referenced to a 3.3V domain. 

3. The nINT pin is push-pull output - so the pull-up isn't needed - but many people still add it regardless so they have a default state always set on the output - it isn't required to do that but it doesn't hurt the device/application to add that either. 

4. Its main purpose is to disable action taken by SBC due to a missed watchdog trigger. I.e. If you hold the SW pin in its active state (which is default high) when the SBC misses a watchdog trigger it will just ignore it - this is helpful during initial development since missed watchdogs typically cause the device to reset or if the device has reset too many times in a row the device transitions to sleep/fail-safe depending on configuration. Since Watchdog is one of the last features that should be implemented for ease of design the SW pin can be helpful during development. When I have to work in lab on this device I pull the SW pin high so I don't have to worry about watchdog triggers. It can also be used as a digital wake-up pin - but I haven't seen anyone use it that way yet - it is a feature included in the device though. 

Please let me know if you have any other questions!

Best,

Parker Dodson

Hi Kalpak,

1. Yes you can externally pull-up SW with VCC1 - however - if this is done missed watchdogs will not cause reaction from SBC - in system typically the watchdog is wanted. If you don't want the watchdog in the system at all - the better way to approach the issue is to disable the watchdog through registers 13h and 14h (disable in normal and standby) so you don't have leakage current from pulling up SW all the time - however there it is not harmful at the device level to pull SW to VCC1 - it would be more system impact I'd be worried about - but only if you eventually want to have the watchdog timer enabled and able to cause the SBC to react. 

2. The Major difference is that the GFO pin is a general field output - so while it can be used to signal to processor for specific interrupts - those are the alt functions of the pin. The main and default function of the GFO pin is essentially a digital output that can output high/low signals for use in the system. The main application we foresaw this pin being used for is for channel expansion - essentially the GFO pin can be used to inhibit/enable an external CAN transceiver or LIN transceiver in case you need extra CAN or need a LIN expansion for the system. nINT is a generic interrupt pin and doesn't really discriminate against the types of interrupts it reacts to. Most systems I have seen typically have GFO floating - essentially if you don't see a need for it your system it can be left floating. 

3. So the LIMP/LSS pin has two main functions - either being LIMP or a Low Side Switch (LSS). 

LIMP is for LIMP home mode -  limp mode is a security feature in cars which activates when the engine or transmission control unit picks up a fault. Once it detects a problem, limp mode will cause the less important parts of the car, such as air conditioning, to switch off, and the speed of the car will be reduced. The LIMP pin will flag for a few reasons - but primarily missed watchdog triggers (if watchdog is enabled and SW isn't in active state) as missed watchdog triggers could indicate a communication failure between controller and SBC or when the SBC enters fail-safe mode due to a fault the LIMP pin will be pulled low. The LIMP signal goes elsewhere in the system to signal that there is a fault and that less important features of the vehicle are shutdown/reduced - essentially it is a way to help mitigate catastrophic disasters at the vehicle level. The LIMP pin on the device is just a way for the SBC to indicate to a system that there is a fault - and it doesn't require VCC1 or VCC2 to operate as the pull-up should be to battery level. It is application and specific system specific if you utilize this pin. The actual action the entire system takes depends on how the system is defined - the SBC just uses it to indicate potentially critical faults to main control unit.

LSS - is is a switch that connects to ground - it can be on, off, or set to a PWM frequency. Essentially a high side switch will go voltage source -> switch -> load -> ground and a low side switch goes voltage source -> load -> switch -> ground ; but it is limited to 25mA instead of 100mA like the HSS module. I haven't really seen this used in applications - but if LSS is needed in system LIMP must be disabled. Low side switches typically are a bit better with floating loads than HSS modules are - but they can introduce larger ground currents. 

Please let me know if you have any further questions!

Best,

Parker Dodson

Hi Kalpak,

It will be high impedance until VCC1 is available then it should be opposite of active state - so by default the pin should be low when left floating if the default conditions are kept (which has SW active high). 

Best,

Parker Dodson

Hi Kalpak,

I am really sorry I missed this reply. Apologies for that. 

1. Watchdog trigger is received via the SPI bus - there is a register that you write 0xFF to to trigger the watchdog for window and timeout - and there are specific registers if you use Q/A watchdog since the response will not be 0xFF. 

2. In general they don't need to have a pull-up or pull-down to work as the input is just looking for edge or pulses (if configured) - however most implementations pull up the WAKE pin to the battery voltage with about a 3k resistor. I.e. it is better to have some kind of default voltage on the line for best results - it can be a pull-down or pull-up depending on what the WAKE input is. 

3. So this is a bit more complicated - because there are multiple steps and the nINT pin being turned low is one of the last things that happen. So first the wake up signal - if you are doing edge detection which is the default - needs to be at a proper level for at least 140us - that will cause the device to transition to restart mode, then VCC1 will start ramping -which the time of this event is going to be variable with system load and output voltage, when VCC1 is at a proper level CRXD will go high. If VCC1 stays at proper level (which it should unless there is a problem) for t_RSTN_act (typically 2ms for configuration 1 and 15ms for the alternate configuration) then the device will switch to standby mode - at which time the CRXD pin will latch low or toggle (based on configuration) and the nINT pin will pull low when the CRXD pin starts latching - as the nINT pin is not available when VCC1 is off - we don't have a specific time for the nINT - but when Standby mode is entered the nINT pin will react to any interrupt (which it will have at least a LWU interrupt).

So basically 140us to detect wake up signal + VCC1 rise time (dependent on load) + t_RSTN_act (configuration option 1 typically 2ms with a range of [1.5ms,2.5ms] and configuration option 2 has typically 15ms with a range of [10ms, 20ms]) + some us level transition time for the nINT pin. The largest time is going to be the t_RSTN_act time and will take most of the delay - so it will be ms levels of delay under normal conditions. 

Best,

Parker Dodson