THVD2450: High power consumption

The THVD2450 and THVD2450V are different devices.

For unidirectional signals, you do not need a termination resistor at the driver, so the load is lower.

But to reduce power consumption for distances less than 10 m, do not use RS-422 but LVDS. Use drivers like the DSLVDS1001/DS90LV011A (1 channel) or DS90LV027 (2 channels), and receivers like the DS90LV012A or DS90LV028A. With the 100 Ω termination, typical power consumption is 26 mW per driver and 18 mW per receiver.

David,

first - 1MHz = 2Mbps for RS-485 and RS-422 devices - they aren't equivalent. So that is how this is going to be used - when I use MHz I am referring to frequency not data-rate (data rate is 2 * frequency) 

1. 325mW is not a spec in the datasheet - so I don't know where the customer is getting this value.

2. The power dissipation table is only rated for VCC = 5.5V ; TA = 125C; RS-485 loading; no common mode loading (i.e. no other transceivers on line during test). No loading is shown, VCC =/= 5.5V, and TA isn't specified in the customers test setup. So however they are estimating power they are doing it incorrectly and they can't use the power dissipation table. 

3.  The typical power draw from a 3.3V source with a 54 Ohm load between A and B (2 120 resistor between A and B, one at each end of bus)  is shown below:

At 20Mbps (10MHz) we have a typical ICC of ~52.5mA (so 3.3V * 52.5mA = 173.25mW) and at 40Mbps (20MHz) we have ICC ~ 64mA (so 3.3V * 64mA = 211.2mW) 

So with 72 IC's you are looking at a typical power draw of: 12.474W at 20Mbps (10MHz) or 15.2064 W at 40Mbps (20MHz). However there are a couple caveats.

1. This assumes no common mode loading and no additional bus capacitance parasitic or otherwise. More capacitance on the output due to parasitics, bus parameters, or filtering elements will add to the total amount of power burned by the device itself - so any deviations from an ideal setup - which customer will not share so I can't give them a definitive answer because I don't know what they did on the RS-485 bus - if they won't share - no body besides themselves can predict the power because only they know what's connected.

2. This assumes an operating temperature of 25C - which isn't realistic in many applications - so the power burned could be higher than the expected power output.

Now onto the customer system specifically:

1. Since they won't share schematic I can't estimate power burned besides typical loading conditions. I imagine they are not using a standard RS-485 loading - as if they were I don't think they'd be so apprehensive to share as the RS-485 bus is standardized - so it should be essentially be the same across multiple use cases regardless of who designed it. If they are using the standard bus then you should be pretty close to the power shown in my calculations above  - there may be some higher power burned due to higher operating temps but its based on device data - which the customers estimations are not. 

2. For 72 units - which means 72 loads from the sounds of the application  RS-485 guarantees 1.5V across 54 Ohm resistor -  That is a power that must be delivered to the load of 41.67mW per load - multiply that by 72 and you get 3W just from the load current  + ~600mW from the idle current (assuming typical Iq current for every device). So at a constant logic value you could burning ~3.6W in the system typically and then when you increase the data rate the power consumed will increase + the voltage across the loading resistors is going to be closer to 2.5V or 3V for most use cases with a VCC = 3.3V - so instead of 41.67mW per load it most likely will be closer to 115mW - 167mW per load. So I think its extremely unlikely to get <5W for as many IC's as is being used. This is less a statement about this THVD2450 and more a statement on RS-485 in general - they will be hard pressed to get that low of power consumption out of RS-485 when using proper loading.

3. There could be one work-around to reduce power usage - and that is to not terminate the system as suggested by RS-485 standard. Essentially a considerable amount of power consumed is due to the termination resistor. While it may not hit <5W it would be less than if termination resistors are not used. However you may not be able to remove the terminations. Essentially if your bus is short enough you can treat the entire bus as a lumped circuit model - and termination isn't needed as reflections aren't considered in a lumped model. 

Essentially if the bus's electrical length is > 1/10th of the signal's wavelength you can't remove termination resistors or else there will be reflections. 

So ultimately:

1. The power consumption should be closer to 15W typical for all units assuming properly terminated RS-485 buses are used. This is worst case assuming all IC's are communicating at the same time - the IC's will use considerably less power when not actively driving data.

2. This typical power consumption is extremely sensitive to system loading - if there are deviations from standard RS-485 loading they could see higher power consumption. If they are unwilling to share more of a schematic (which is fine)  I can't give them a more precise power estimation formula and what I have given above is a good benchmark to start looking at. 

3. With the size of the system its extremely unlikely to get <5W when using proper RS-485 setup. This has more to do with RS-485 in general than it does the THVD2450.

4. If the bus length is < 1/10th of the signals wavelength (including relevant harmonics) then termination may not be needed and you will burning considerably less power as the output driver current is going to be much lower so most of the power is just quiescent current in these situations. But it can't be applied to every system as most require the termination to prevent EMI issues due to reflections. 

Best,

Parker Dodson

The driver supply current includes the bus current. M-LVDS uses two termination resistors, which is not necessary in your application. With a single 100 Ω termination at the receiver, the current will be lower (probably similar to that specified for the DS92LV040A).

There are no LVDS drivers for a lower supply.

Hello David,

Looking at the Supply Current table in the datasheet, the total device current with 4 drivers enabled is 75mA max. Total device current with 4 receivers enabled is 38mA max.  

With this information above the total current consumption is Driver (~57mA max) and Receiver (~8mA max) = 65mA.

Just a note, the DS91M040 function is M-LVDS and not LVDS. M-LVDS uses a higher current driver since it's mainly used for backplane applications. LVDS devices can be used for lower current consumption but for distances less than 10 meters.

Regards,

Josh