THVD8000: Vehicle use

Hi Matt,

It is possible - the THVD80X0 is very flexible - but there are some tradeoffs with using it compared to normal RS-485. I will note however that the biggest advantage of using this device is to save on cabling costs as you only have 1 shared bus for power and data - while I definitely understand why that would be wanted in a vehicle there will be some tradeoffs that may not perfectly align with what your needs are. 

1. The modulation frequency is suggested to be 10x higher than the data rate - if the ratio is lower we no longer guarantee that duty cycle distortion added by device is capped at +/-2%. This does mean that there will be more energy in higher frequency bands using this device compared to classical RS-485 - but the coupling network that we suggest to use with these devices is designed to keep as much of the data signal within the bus as possible. Higher modulation frequencies will also reduce effective bus length - essentially if you take the below graph, divide the X-axis by 2 and change it to modulation frequency instead of BPS you can get an approximate modulation frequency versus bus length chart. The THVD8000 highest frequency output would be similar to 10Mbps devices - so while the frequency added is higher than if you were to use standard RS-485

2. For noise -

A) In THVD80x0 designs its not uncommon to see the use of split terminations and common mode chokes to help combat noise in the system if the system is more sensitive . But RS-485 architecture is already pretty noise resistant + this device family does have spread spectrum clocking on its outputs to help smooth the frequency response on the output

B) Also it is very important to note that we assume the power supply and power loads are "AC Ground" - meaning there is a very low impedance pathway for the modulated data signal - on DC Power systems this is usually handled by the bulk capacitance of the supplies/loads - but AC systems this capacitance  may need to be added at the cost of increased power usage - if you don't the THVD80x0's signal can and most likely will show up at the power nodes - so that capacitance on the power nodes is very important. 

C) If your data rate is 30kbps or lower I'd use the THVD8010 as it has a pretty wide hysteresis curve - so it is much more noise immune than the THVD8000 - where the THVD8000 is better if you  need data rates from 30kbps to 500kbps (can push it to 1Mbps at the cost of more duty cycle distortion - probably around +/-4% there) 

D) These devices work best daisy chained - as neither the THVD8000 or THVD8010 can transmit along long unterminated stubs without reflecting energy. If you have more than 2 nodes in the system than the non-terminal nodes are unterminated so using a daisy chain will minimize stub length - you generally should get about 0.5 meters before most issues - but it will vary based on cabling. If you have a non-ideal setup you can still use the THVD80x0 but we would also suggest you use a THS6222 line driver in addition to that and terminate every single node - so that would be power intensive to do - but would help clamp down on noise. 

3. We haven't seen a ton of trouble with cable mismatches - generally speaking RS-485 expects up a +/-20% mismatch in cabling impedance and usually that will not be enough to cause major issues. Ideally the input of one THVD80x0 node should either be an end node that is terminated with 120 ohms or an unterminated stub that we can largely ignore. Cable mismatch - just like in any RS-485 system - can be problematic and should be as close to 120 ohms as possible. 

4. One potential issue - I don't think it should be too concerning on a vehicle - but for standard implementations (what is discussed directly in the datasheet) it is suggested to go up to 36V supplies (realistically pushing it to maybe 50V with some additional protection diodes should be okay); above that and we generally start suggesting additional line drivers and a different coupling network (isolated transformer coupling data onto power node directly) as that architecture could support up to ~1kV to 1.2kV depending on coupling network. 

5. The biggest issue honestly is solution size; The standard implemenation requires inductors (2 per power source/power load connected to shared bus) and in low data rate applications generally slower modulation frequency will get the best nominal performance - but the parallel combination of the inductors impedance + the transceivers input impedance must stay above 375 ohms at modulation frequency - so that leads to some big inductors on slower speeds. The more nodes you have the more inductors will be needed and the larger they will get. Also the amount of current running through them is also going to be a factor in how large they are. Depending on the nodes and additional space - the solution size may be quite large - but you will save in cabling. 

Please let me know if you have any questions - but ultimately I think this is possible with some tradeoffs and these designs do typically take more effort/time than standard RS-485 because there are more elements to consider  - however if the cost savings from reducing amount of cabling in your system is worthwhile than it is probably worth exploring as a potential option. Please see the product page for these parts as we do have  decent amount of information around varied use cases - this is probably one of our most well documented RS-485 type part. 

Best,

Parker Dodson