Combined RS232/485 Interface Chips

MAXIM has a couple of them, MAX3160-3162 IIRC, but they are not cheap either. Your two chip solution seems to be the less expensive variant.

HTH, Jens

Remember that many RS-485 protocols require very timely switching of the bus-enable line, and this is often hard to control in software, so make sure you pick a solution that has hardware-timed bus-enable control based on a time-out for the start pulse of the next serial character.

-Robert Scott Ypsilanti, Michigan (Reply through newsgroups, not by direct e-mail, as automatic reply address is fake.)

I've hacked a DS26LS32A+DS3691N pair from RS232 (or 423 really) into RS485.

Note though that I've had trouble with some recent, 5v-only, RS485 receivers (on the customer's equipment). It seems that they don't really like input voltage swings that go much below 0v.

Thanks, but the DS26LS32A seems to be ~$29 (National)! I didn't check the other one... :)

I did toy with the idea of connecting RS232 signal levels into an RS485 tranceiver. What with the RS232 current limiting and the RS485 supposed tolerance of voltages "outside the rails".

That's how everybody else does it.

Just curious, but just how much time do you have after you activate the TX_EN line before you have to send the start pulse? Are we talking mS, uS, pS, nS or other?

Dennis.

Dennis Hawkins n4mwd AT amsat DOT org (humans know what to do)

"A RECESSION is when you know somebody who is out of work. A DEPRESSION is when YOU are out of work. A RECOVERY is when all the H-1B's are out of work." To find out what an H-1B is and how they are putting Americans out of work, visit the following web site and click on the "Exporting America" CNN news video:

The critical parameter is turning OFF the transmitter after the last stop bit has been sent. This is often in order of microseconds, before the opposite side starts to send the response.

Paul

SP334 from Sipex. IMO pretty nice chip. Here, in Poland, it costs about $6.80. The problem is that Sipex distributors can sell you a lorry of those chips off-hand but if you want to buy 5-10 pieces for a prototype, you have to wait 2 months for samples. :-( It is rumored that Sipex has such a sales policy.

Regards, /J.D.

I had a 555, Enabling the Tx on the Start bit and originally timed to Disable the Tx about 20uS after the Stop (@ 9600). But the remote device would send its Start so quickly (they do an echo, in hardware) that my Disable had to be re-timed to be half way through the Stop. Bummer. :(

How do you detect the start bit with a 555 ?

While the 555 can be retriggered by the start bit ("0"), it is also retriggered by any 0-bit in the data or even by the parity bit, if it happens to be 0.

How do you detect the stop bit without actually counting the bits after the _true_ start bit. Paul

Just some background. This was a hack of a card doing RS485 half-duplex, with the Enable/Disable of the line driver originally done in software. But the high speed remote device could start transmitting the echo-byte before we Disabled and the first bits of the echo got mangled.

So I strapped on a 555 monostable. It was triggered by the leading edge of the Start bit at the input to our line driver, and in this state Enabled the output of the line driver. 1146uS later (see below) the 555 timed-out and Disabled the line driver.

Crude, but when you are desperate...........

No, a 555 (in monostable mode) doesn't retrigger.

This was 9600 with a fixed 8E1 protocol, which is 1146uS from leading edge of Start to trailing edge of Stop. We have pullup/down resistors that pull the line into the same state as Stop as well, so it is possible to Disable the line driver anywhere between the end of Parity and the end of Stop.... about 1042uS to 1146uS.

I am not sure how you would use RS-232 to receive RS-485 since one is single ended and the other is differential. The main point to using RS-485 or RS-422 is the added noise immunity and high drive currents which results from using low impedance drivers and rejecting common mode noise. How would you do this with RS-232 devices regardless of voltage levels? I guess an RS-485 receiver could work with an RS-232 xmitter, but how would RS-485 drive into RS-232 properly?

As to the original question, I have found multiprotocol parts from Sipex, Maxim and Linear Tech. All cost upwards of $4 compared to about $1 for either RS-232 or RS-485/422. I use them because of space issues, although they are starting to put RS-232 in MLF packages which start to errode the size advantage of the single chip.

It wouldn't. The RS232 chip, or portion of the chip, would be used when the connection is RS232.

The application is a piece of industrial test equipment. At this stage it looks like it will be used with RS232 (connected to a PC). But in the future we might want to network them with RS485, for example. If there is a low cost, simple way to make the interface software selectable between RS485 and RS232 that would be nice.

It looks like the "combined chip" solutions are all quite a bit more than the cost of two separate chips. The cost alone isn't really that much of an issue, but it also seems that the "combined" solutions would involve relatively obscure devices (compared to MAX232 and

Thanks, rickman, Grant, and everyone else too :)

Perhpas I misunderstood. What I was refering to was haveing both RS-232 and RS-485 drivers and receivers hooked to the same pins. Then you just enable wichever set you want to use.

What do protocols have to do with it?

Yes, I misread the posts. I have considered using both types of parts since the newer RS-485 parts will tolerate a wider voltage range than before. Size is also an issue, but I am finding new MLF packaged RS-232 parts which helps a lot. I think I have not found a good RS-485/422 part yet. Most already have the driver and receiver connected on the bus side and so will not support full-duplex operation.

That is the term they use for a chip supporting mulitple interface standards.

My experience is that the industrial customers want both RS-232 and RS-485 offered, but nobody buys the RS-485 version or is willing to obey a half-duplex protocol (which is an absolute necessity with RS-485). This experience is gathered during nearly 20 years of process instrumentation.

In the newest design, there are pads for both RS-232 and RS-485 buffers but only RS-232 chip is in. Nobody has requested an instrument stocked with the

Tauno Voipio tauno voipio @ iki fi

My solution is to put two sockets (MAX232 or equivalent, 75176 or equivalent), 4 capacitors and two connectors in. Then either of the two chips is plugged in, or both could be soldered in using tiny packages and the receiver input selected using some method (a resistor and the tristate output on the RS-485 chip might be okay). Crude, but it does the trick. If you're adding RS-485 to an existing device that has only RS-232, watch the worst-case current draw on the drivers, it can be quite significant.

Best regards, Spehro Pefhany

Hi Tauno,

Yes, you are right (although this is early days yet for the product), this is already starting to sound familiar!