So, I'm trying to troubleshoot some RS485 problems that I think are due to leakage paths on our PCB, and I looked at the datasheet to see if I could tell what the internal circuit was like, and what sort of resistance I should expect from "A" and "B" inputs to ground. What I found was resistors measured in kW. Not what I expected.

I tried googling this, and got nowhere. I did find an RS485 design guide (also from TI) which also uses this unit in some places. What is it? Should I just treat it as ohms and move on with my life?

Aside from the units of impedance question, I suppose I may as well ask for help with my real issue while I'm here. Based on measurements from working boards, it seems like the resistance from "A" or "B" to device ground (pin 5) is anywhere from 2-20M. Meanwhile, I've got a malfunctioning board which measures all over the place, but generally between 30-70k from either pin to ground. I'm assuming that whatever leakage path is causing this difference in resistance readings is also the reason for our data corruption issues, but it's weird that the diagram in the datasheet seems to imply there should be lower resistance paths to ground within the chip... assuming "W" actually equals ohms, then the chip has a 54k path to ground from each input. Is that ground an internal signal ground that's separate from the chip ground on pin 5?

It's not critical that I understand every bit of these impedance questions, but it would help a lot with diagnostics if I could tell people the resistance from A or B to ground should always be at least [insert value here] ohms. Right now my real world measurements and what I think I'm reading in the datasheet aren't lining up at all.

I don't think it should matter, but here's how the RS485 chip is incorporated into our circuit. There aren't any deliberate paths to ground for those data lines, so theoretically, measuring corresponding pins at J2 should be the same as measuring input pins on the RS485 chip.


Any insights on the "W" impedance units and/or what internal chip resistance I should be able to measure would be greatly appreciated. Thanks!

 

Speculation: The printer of the datasheet did not have the ability to reproduce the "Ω" symbol and used a "W" instead, in every place the Ω appeared. Whoever's job that was didn't catch it or didn't care enough to correct it.

 

In the Microsoft Word Symbol font a capital W on the keyboard translates to the Ω symbol in the document.
So somehow that translation didn't occur.

 

Cool, I thought it might be something silly like that. Thanks!

Ok, so on to the follow up question then. Regarding the A and B input stages in the TI datasheet screenshot, they appear to have 54k path through R2 and R3 to ground. But, if I measure resistance from either input to ground on our good, working boards, it reads over 2M. I'm guessing this is because the RS485 chip has an internal signal "ground" that's separate from the power ground at pin 5. Does this sound right? If not, any other idea why I read such high resistance? Are the inputs capacitively coupled perhaps? Actually, now that I've thought of that, I kind of prefer that explanation.

 

Curiously, "capital" omega appears properly in the figures, possibly because the figures were created in Illustrator which would default to PostScript fonts. I've run into a few docs where W appears where Ω should be.

The input won't be capacitively coupled. I thought perhaps what shown as simple resistors are in fact FETs with very high channel resistance that aren't turned on unless the IC is powered. The datasheet actually shows higher input current for the unpowered state, but the currents are just weird. I can make no sense whatever of what Figure 4 and Figure 5 are trying to present. What is "bus input current?" - the driver is disabled? These are supposed to be 1/4 and 1/8 unit load transceivers, so the bus loading should be about 48k ohms and 96k ohms, respectively. How do they get currents of several tens of milliamps?!

Figure 6 seems to be showing driver output voltage of over 4 V for a part operating on 3.3 V.

TI has produced some absolute abominations for datasheets. The LM317 sheet is probably the worst piece of trash I've ever seen that actually tries to present data.

 

The input won't be capacitively coupled.

In retrospect, I should've known better than to think that. The inputs wouldn't be left in an indeterminate state like that, with only the state changes being recognized. My thinking was rushed then - sometimes I need to slow down!

These are supposed to be 1/4 and 1/8 unit load transceivers, so the bus loading should be about 48k ohms and 96k ohms, respectively.

I've read several articles on RS485 bus architecture, including unit load standards. I should've known that the actual input impedance in use is well defined. I just forgot. I'm curious now to measure the input impedance with the chip powered. Presumably it'll be somewhere between the 48k spec and the 54k I see in the schematic.

TI has produced some absolute abominations for datasheets. The LM317 sheet is probably the worst piece of trash I've ever seen that actually tries to present data.

This just makes me laugh!

 

TI inherited (bought) all that from National Semiconductors isn't?