TMUX4827: Request for suggestions in selecting a suitable part.

The TMUX4827 has a −3 dB bandwidth of 18.3 MHz. Any additional resistance or capacitance will reduce the bandwidth further.

Please show a detailed schematic that includes all components on the RS-422 lines. Also specify the length and capacitance of any cables.

Hey Sameep,

Just reiterating Clemens point here. The insertion loss of the TMUX4827 at 290kHz should be similar to the insertion loss at 120kHz. Looking at the insertion loss curve, you really shouldn't see the impact of the TMUX4827 bandwidth until roughly maybe 5MHz when it starts to decay and at 18.3MHz you'll hit that -3dB point. 
However, that doesn't mean that the overall bus isn't near it's limits here. Doubling what Clemens has said, a schematic and the length and capacitance of the relevant cables would be useful here to identify the issue.
Additionally, have you run the tests bypassing the mux and seeing the results?

Thanks,
Rami

HI Rami ,

HI Clemens,

I have attached the image of detailed schematic in this post.

"Just reiterating Clemens point here. The insertion loss of the TMUX4827 at 290kHz should be similar to the insertion loss at 120kHz. Looking at the insertion loss curve, you really shouldn't see the impact of the TMUX4827 bandwidth until roughly maybe 5MHz when it starts to decay and at 18.3MHz you'll hit that -3dB point."

---Does this means that the output voltage(across D1 and D2) should be equal to that of the input Voltage(across S1A and S2A / across S1B and S2B) across a frequency range of 10Khz to 5Mhz?

However, that doesn't mean that the overall bus isn't near it's limits here. Doubling what Clemens has said, a schematic and the length and capacitance of the relevant cables would be useful here to identify the issue.

---We are currently prototyping the board using Dot matrix board. For all RS422 lines we are using 26 AWG twisted pair wire with length not exceeding 15cm.I am not sure on what exactly "capacitance of the relevant cable" means and how to find it.

Additionally, have you run the tests bypassing the mux and seeing the results?

--When the mux is bypassed and only one channel is connected directly to the optocoupler (receiver end), It works fine and is having no issues. We have tested till 5Mhz, and it is working fine.

Also, In the data sheet of TMUX4827 "Figure 6-22. Maximum Signal Swing" Sayas that the maximum swing after 50khz is 2V. What does this maximum swing mean? is this the maximum voltage swing at output?

Regards,

Sameep

R253/R254 would be needed only for a bidirectional bus; remove them.

Termination must be at the end of the bus, so R255/R256 are at the wrong place. Remove them, and add termination right in front of the optocoupler.

HP's Optocoupler Designer's Guide suggests the following RS-422 receiving circuit for high-speed applications:

Your application probably does not need the second optocoupler and flip-flop; ignore them.

The 1.62 kΩ resistor limits the DC current. Adjust it so that you get the required forward current for your ACPL-K74T (10 mA?).

The 20 pF capacitor counteracts the capacitance of the cable and the LED by increasing the current during switching. You can try adjusting it if the edges are too slow, but it should not be larger than about 100 pF.

R1 and C1 terminate the differential line. R1 is the termination impedance, about 120 Ω (but might need to be somewhat higher because it is in parallel with the optocoupler input); C1 prevents DC currents. For details about AC termination, see section 4.3 of the RS-422 and RS-485 Standards Overview and System Configurations application report; it suggest C1 = 1000 pF, but you can try to make it smaller if the edges are too slow. You could also try completely removing R1 and C1.

The capacitance of twisted pairs is very roughly 40 pF/m. Your cable should have a datasheet.

Hi Clemens,

Your swift response is much appreciated.

Ill implement the changes what you have said and update the results here.

But I am having few confusions.

1.When we bypassed the MUX (We didn't change any resistors. All were mounted) and connected the output of the driver to the opto-coupler ,It was working fine up to 5MHz.

a. We didn't remove the mux from the circuit. We tapped the connection from S1A and S2A pins using a small, twisted pair wire (less than 10cm) and connected it to the second channel of the optocoupler (ACPL-K74T)

b. At the First channel output of ACPL-K74T (Input is given through MUX) it was observed that the transactions were happening up to 290KHz.Above which the voltage was dropping considerably such that the opto-coupler was not able to recognize it as a high (LED was not turning on)

c. At the second channel output (mux bypassed), there were no issues. In this case it is working fine across frequency up to 5Mhz. (We haven't verified above 5Mhz as it was not in our area of concern).

d. I wonder as you mentioned the issue in our design is all about terminations and capacitance, how bypassing the MUX (Not removing it from the circuit. Just tapping the input) would provide a proper output from the opto-coupler?

2.As asked in the previous post can you let me know " In the data sheet of TMUX4827 "Figure 6-22. Maximum Signal Swing" Sayas that the maximum swing after 50khz is 2V. What does this maximum swing mean? is this the maximum voltage swing at output?"

3. Meanwhile, Ill rig up the new circuit along with the changes you have suggested and come with the results I observed.

Regards,

Sameep

I have no explanation what the physical reason for the limits in figure 6-22 is. But your observation confirms that the TMUX4827 does not support high voltages at high speeds (worse than indicated by the specified bandwidth).

Why did you choose the TMUX4827? A simpler switch like the TMUX136 would not have such a signal swing problem.

HI Clemens,

I have no explanation what the physical reason for the limits in figure 6-22 is. But your observation confirms that the TMUX4827 does not support high voltages at high speeds (worse than indicated by the specified bandwidth).

I am testing TMUX4827 as a standalone device by giving input a frequency range of 10KHz to 5MHz with 5Vpp and observing the Corresponding output. I am Using few decoupling caps. Other than that, no passives are being used. I'll keep this thread updated with my observation.

Why did you choose the TMUX4827?

Our final objective is select a mux/switch or a solid-state relay which shall be used in a 2:1 configuration.

There is no solid point with me to say why we choose TMUX4827 in first place. But here are few things we observed in the datasheet of TMUX4827 when we selected this part.

a.  As we were having 5V generated on the board, " Supply range: 1.8 to 5.5 V " was suitable for us.

b. As we were using this to switch between 2 RS422 signals, "Beyond the supply signal range: -12 V to 12 V" property gave us some Voltage room to play around with.

c. It was having Low Ron

d. it was having 1A of output current capability.

e. Moreover, under its application section, it specifies that it can be used for "RS485/232" applications.

A simpler switch like the TMUX136 would not have such a signal swing problem

As you suggested I just gone through TMUX136 datasheet once. But I'm having fer questions here...

a. TMUX136 has a maximum VCC Voltage of 4.8V.I don't have any 4.8V rail in my board.

b. Thus to use TMUX136 I must add one more regulator to generate 4.8V. Which I think is not the best solution.

c. Also, in datasheet, under recommended Operating condition its specified that "Input-output DC voltage" (WRT GND) can be up to 3.6V. As in RS422 Voltages Can be up to 6V. Can you please clarify the same.

d. I'm having 3.3V rail in my board. Is it ok to power TMUX136 using 3.3V rail and to connect IO with RS422 signals? Are the inputs being tolerant to be powered more than VCC voltage?

If TMUX136 is not the Best fit, can you please let me know of any possible solutions available That can be a direct fit as a RS422 Switch/Mux. (Working at 2.5Mhz)? 

Is there any Training material available with TI which helps me better understand about RS422 protocol and the necessary termination strategies. If available, please provide us the link for the same.

Regards,

Sameep 

Oops, the TMUX136 was marked as supporting 5 V in the search function, but it is actually a 3.3 V device. A switch that does support 5 V would be the TMUX1072.

The TMUX4827 or the lower-bandwidth TMUX2889 are the only switches that allow voltages beyond the supply rails. I do not know how much of a voltage offset you will see between your boards.

There is lots of documentation about RS-422, but it always assumes a normal (high-impedance) receiver; there is not much information about optocouplers in this situation. Maybe consider using an isolated differential receiver like the ISOW1432.