Discrete custom design of RS485 driver

[Klaus Kragelund]

Hi

The standard RS485 drivers available has a minimum voltage of 3V and a rarther large drop voltage when loaded with the defined bus load for Modbus of 54ohms, and this causes problems for our design since we have limited power available for driving the bus

So, we are thinking about designing our own driver in discrete components, so we can reduce the supply down to 2V and still comply with minimum 1.5V differential voltage into 54ohms.

We only need 115k baud, so we could use a tiny logic level FET as the output stage. Shortcircuit protection would be done with a current limit circuit along with a low value supply capacitance (to reduce peak power in the FETs)

Backfeed would need to be solved with a beefy diode to a defined clamp voltage.

So, anyone been down this road, designing your own RS485 driver?

Cheers

Klaus

 

[Klaus Kragelund]

Hi



The standard RS485 drivers available has a minimum voltage of 3V and a rarther large drop voltage when loaded with the defined bus load for Modbus of 54ohms, and this causes problems for our design since we have limited power available for driving the bus



So, we are thinking about designing our own driver in discrete components, so we can reduce the supply down to 2V and still comply with minimum 1.5Vdifferential voltage into 54ohms.



We only need 115k baud, so we could use a tiny logic level FET as the output stage. Shortcircuit protection would be done with a current limit circuit along with a low value supply capacitance (to reduce peak power in the FETs)



Backfeed would need to be solved with a beefy diode to a defined clamp voltage.



So, anyone been down this road, designing your own RS485 driver?

A rough first draft:

www.electronicsdesign.dk/tmp/RS485_Custom.pdf

 

[[email protected]]

The standard RS485 drivers available has a minimum voltage of 3V and a rarther large drop voltage when loaded with the defined bus load for Modbus of 54ohms, and this causes problems for our design since we have limited power available for driving the bus

This is not a Modbus specific issue, but rather RS-485 specific issue
with a twisted pair bus with characteristic impedance of 100-120 ohms.
In order to avoid reflections at the open ends of the bus cable,
termination resistors are typically used at both ends with the same
value as the cable characteristic impedance.

For DC, those two resistors are effectively in parallel and hence the
45 ohm total load.

However, those termination resistors are needed only to avoid the
reflections from voltage _transitions_. Thus, putting a capacitor in
series with the termination resistor(s) should reduce the idle power
consumption, when no data is being sent. Of course, without DC
continuity, the end to end signal ground conductor is essential.

There are application notes describing even more elaborate termination
methods, describing their advantages and disadvantages. You should
also look for various termination techniques used on CAN bus (which is
essentially RS-485).

 

[Joerg]

A rough first draft:

www.electronicsdesign.dk/tmp/RS485_Custom.pdf

Just a comment: Diodes are already in the FETs, in the form of body diodes.

One thought would be whether a hysteretic sync-buck IC could be pressed
into service here. I haven't needed one this low in voltage yet but they
should come for very low supply voltages (processor core supplies and such).

 

[Klaus Kragelund]

With only a 2V supply, how do you get enough drive for the P-channel

device? Do you have a more negative supply available?

The voltage rail for the FET are driven by 2V and I will generate an additional supply voltage to drive the gates, about 3V.


The majority of the power goes for the bus, driving the 54ohms load (120//120//1500 ohms in parallel, that is two termination resistors and the 32 unit load impedance).

Right now the implementation is using a standard RS485 driver running at 3Vsupply, but with 54 ohms resistance along with the driver impedance, draws90mW during transmission.

A low RDSon driver at 2V would reduce that to about 60mW

Regards

Klaus

 

[Klaus Kragelund]

Just a comment: Diodes are already in the FETs, in the form of body diodes.

Yes, I added parallel more sturdy diodes, to direct the current away from the low current body diodes.

Regards

Klaus

 

[Klaus Kragelund]

This is not a Modbus specific issue, but rather RS-485 specific issue

with a twisted pair bus with characteristic impedance of 100-120 ohms.

In order to avoid reflections at the open ends of the bus cable,

termination resistors are typically used at both ends with the same

value as the cable characteristic impedance.



For DC, those two resistors are effectively in parallel and hence the

45 ohm total load.



However, those termination resistors are needed only to avoid the

reflections from voltage _transitions_. Thus, putting a capacitor in

series with the termination resistor(s) should reduce the idle power

consumption, when no data is being sent. Of course, without DC

continuity, the end to end signal ground conductor is essential.



There are application notes describing even more elaborate termination

methods, describing their advantages and disadvantages. You should

also look for various termination techniques used on CAN bus (which is

essentially RS-485).

Yes, but to conform to the Modbus standard, the termination resistors are added without diodes

Cheers

Klaus

 

[Klaus Kragelund]

Just use a cmos quad xor gate; two paralleled sections for one phase, twofor

the other, with maybe 3.3 volt supply and 30 ohm source terminations. There's no

need to use discrete fets.



We recently did this:



https://dl.dropbox.com/u/53724080/Circuits/ESM/Line_Drivers.pdf



The basic line driver is a couple of tiny-logic gates driven from complementary

FPGA outputs. The downstream junk is selectable line driver equalization,to

partially correct for CAT5 cable losses. This runs up to 125 MHz.

Maybe a good point, if I can find a logic device that has low RDSon at 2V.

The ones I have found have 10ohms RDSon (NC7SZ74), but could parallel some of those to bring down the RDSon to the 2-3 ohms range

Regards

Klaus

 

[Tim Williams]

Yes, I added parallel more sturdy diodes, to direct the current away
from the low current body diodes.

Are you expecting huge common mode transients? MOSFET diodes have been
rated at, or above, the channel current for ages. FDV301N says 0.29A
diode, 0.22A channel (both I'm sure depend on thermal resistance, it's
only an SOT-23). I've never used external diodes in an inductively loaded
inverter and never found any reason to: the body diodes do a fine job.
They just aren't good at hard switching (slow recovery).

Have you considered BJTs for this? They tend to be easier to drive at
lower voltages. With Vceo as low, you can easily find fast transistors
with high hFE, so even with saturated operation, you don't have to worry
about switching speed or error in the current source. You may still need
a bootstrap (using all NPNs, or a negative bootstrap for the PNP pair),
but only one at least.

The TL431 as shown clamps about 5V, which is way more than your supply --
are you sure about this? If it's for ESD, it's only clamping 100mA, and
takes a moment to respond. A zener TVS would be a bit sloppier (a 3.3V
rated device might break down at 5V and carry a heavy load at, say,
8V...), but much faster and more robust. You could also use a diode back
to the +2V supply, which is probably as transient-resistant.

Tim

 

[Klaus Kragelund]

Are you expecting huge common mode transients? MOSFET diodes have been

rated at, or above, the channel current for ages. FDV301N says 0.29A

diode, 0.22A channel (both I'm sure depend on thermal resistance, it's

only an SOT-23). I've never used external diodes in an inductively loaded

inverter and never found any reason to: the body diodes do a fine job.

They just aren't good at hard switching (slow recovery).

Yes, the RS485 line is subjected to hot swapping, termination resistors inserted "live" and must be tested against surges/bursts. I am also worried about injected DC voltages from user wrongful installation.

The big diodes is used to divert current to the clamp using the 1ohms resistor to allow for the external diodes to draw the biggest portion of the current.

Have you considered BJTs for this? They tend to be easier to drive at

lower voltages. With Vceo as low, you can easily find fast transistors

with high hFE, so even with saturated operation, you don't have to worry

about switching speed or error in the current source. You may still need

a bootstrap (using all NPNs, or a negative bootstrap for the PNP pair),

but only one at least.

Yes, could be a good idea, just need to add circuitry to draw the carriers out of the base to switch them off fast.

Regards

Klaus

 

[Joerg]

Yes, I added parallel more sturdy diodes, to direct the current away from the low current body diodes.

Usually they are about as sturdy as the channel in the FET, can take a
similar current.

Diverting current away from those only works (to some extent) if you
hang a Schottky of sigifnicant size in parallel. The best method would
be to steer the FET conductive while some massive surge current shows up
for some reason.

 

[[email protected]]

Yes, but to conform to the Modbus standard, the termination resistors are
added without diodes

What diodes ? I was suggesting using capacitors.

What Modbus "standard" ?
The closest that I can think as electric Modbus standard is the
http://www.modbus.org/docs/Modbus_over_serial_line_V1_02.pdf
"MODBUS over serial line specification and implementation guide V1.02"

Look at page 28

Line termination may be a 150 ohms value ( 0.5 W ) resistor.

A serial capacitor ( 1 nF, 10 V minimum ) with a 120 Ohms ( 0.25 W )
resistor is a better choice when a polarization of the pair must
be implemented (see here after).

Polarization = "Fail safe termination" in RS-485 speak.

 

[Joerg]

And generally not safe to use for repetitive pulses. ...

Why that? They are often used as regular power current paths. The
current rating is roughly the same as the FET itself, usually.

... Some discrete
FET's I was using at Zarlink came with a Schottky in the same package.

With RF stuff all bets are off, RF transistors can be like the princess
on the pea. 3V reverse Vbe ... poof ... gone. Or gaoan, as they'd say at
one client.

[...]

 

[Klaus Kragelund]

What diodes ? I was suggesting using capacitors.

I meant to write capacitor, sorry

What Modbus "standard" ?

The closest that I can think as electric Modbus standard is the

http://www.modbus.org/docs/Modbus_over_serial_line_V1_02.pdf

"MODBUS over serial line specification and implementation guide V1.02"



Look at page 28

Yes, the Modbus standard defines that, but the widespread industry standard is 120 ohms and no capacitor. (adopted from the RS485 standard)

Cheers

Klaus

 

[Klaus Kragelund]

Do you need to tri-state the driver? If so, Larkin's suggestion

doesn't work. Even with tri-state you have to watch out for "kick"

above/below rails.

Yes, I need to tristate the driver, since it is a 2 wire system, half duplex.

Regards

Klaus

 

[Klaus Kragelund]

Why not? Use tri-state tiny-logic drivers.





Even with tri-state you have to watch out for "kick"




Kick? Logic chips can't drive transmission lines? Add some protection

if you expect lightning bolts.

We would need to add protection in any case to reduce the currents, the driver IC would suffer from latchup problems if not.

Cheers

Klaus

 

[Tim Williams]

Yes, could be a good idea, just need to add circuitry to draw the
carriers out of the base to switch them off fast.

And that's not even that big of a deal, really -- if you run 2N4401/3 kind
of hot (~20mA Ic, 2mA base drive, 680 ohm B-E resistor), you'll see edges
under 100ns and storage time under 300ns (storage causes skew, but it's
symmetrical in an H-bridge, so it causes shoot-through and delay).
115kbaud gives you almost 10us between edges, so there's tons of time for
switching.

The average switching transistor (like the little complementary gate drive
things, or just a plain old ZTX651 or etc.) is even beefier, maintaining
hFE > 100 at rated collector current. So, even saturated (where hFE is
lower and stored charge piles up), they don't take much drive current at
all, relative to what they're doing. They start looking like low Vgs(th)
MOSFETs, with an input diode thrown in for convenience.

Tim

 

[Joerg]

But resistive, since it's a current path from the back of the die.

What's the difference? Whether 0.7V drops across a more or less
resistive path, who cares? All that counts is total dissipation and that
it's not too localized.

Go ahead... I always enjoy your sound effects >:-}

I mis-spoke... make that Synaptics.

With RF stuff all bets are off, RF transistors can be like the princess
on the pea. 3V reverse Vbe ... poof ... gone. Or gaoan, as they'd say at
one client.

[...]

Here today, China tomorrow ;-)

I sure hope we can at least keep high-end semiconductor engineering and
processing in the country for a while, now that Obamacare is smothering
much of the med device investment climate. We can't afford to lose such
leadership positions but obviously that doesn't sink in on the hill :-(

 

[rickman]

I sure hope we can at least keep high-end semiconductor engineering and
processing in the country for a while, now that Obamacare is smothering
much of the med device investment climate.

I would love to hear some sort of rational explanation of how that is
happening.

Rick

 

[Joerg]

[...]

... Some discrete
FET's I was using at Zarlink came with a Schottky in the same package.

I mis-spoke... make that Synaptics.

With RF stuff all bets are off, RF transistors can be like the princess
on the pea. 3V reverse Vbe ... poof ... gone. Or gaoan, as they'd say at
one client.

[...]
Here today, China tomorrow ;-)

I sure hope we can at least keep high-end semiconductor engineering and
processing in the country for a while, now that Obamacare is smothering
much of the med device investment climate. We can't afford to lose such
leadership positions but obviously that doesn't sink in on the hill :-(

Except for the expert witness stuff (3 weeks in Dallas Federal
District Court starting around January 7 , all of my work is now
west of Honolulu.

Interesting. Mine is still mostly on US soil, or all of it right now.
But medical has nearly evaporated from a consulting point of view, from

50% a few years ago down to <10%. Plenty of work here but all

industrial, oil, gas, aerospace and so on. That does not bode well for
med-tech in our country.