THS4551: Driving a differential cable

Well Rohit

Normally you will also terminate your cable in a differential input impedance equal to the cable characteristic Zo. I assume you are talking twinax and not coax. 

Then, you would take that Zo, divide by 2 and put a build out R at that value on each output to drive the cable. You will get a -6dB insertion loss but better response shape with impedance matching. 

Hi Michael

It is a shielded twisted pair of AWG26.

Why I will have -6db loss? Is there any solution to prevent that?

So only cable characteristic impedance matters not the capacitance of cable?

Regards,

Rohit

Well you can do an active output match like they do in DSL. Here is a very good discussion for FDA's while DSL does this active match in a similar but not identical fashion, 

But really first, what frequencies are you talking about - 

Thank you for the document, it is very helpful.

My signal bandwidth is 0 -270KHz.

Regards,

Rohit

Pretty low, why are you using a 130Mhz part for a 270kHz signal? slower options (save power) include the 

1. THS4561

2. THS4531A

3. OPA1637

and then - 

a. What Vpp differential do you need at the end of the cable

b. What characteristic Zo is that twisted pair (I seem recall typ in the 120ohm region)

c. What do you care about in the signal - is it a comm signal where HD might be important, is it a pulse type signal (dc coupling suggests that) where settling time is more important? 

Hi Michael,

THS4551 we are using from some old design.

a. VPP is 0-5V. with 2.5V common bias.

b. Cable details are

Impedance (differential)                             80 Ohm

Capacitance                                                68 pF/m (C/C)

                                                                     98 pF/m (C/S)

Insertion loss                                                0,5   dB/m at 100 MHz

                                                                     0,65 dB/m at 200 MHz

                                                                     0,85 dB/m at 400 MHz

                                                                      1,1   dB/m at 600 MHz

c. Signal is from hall sensor which is differential and it will not be a pulse signal, more important is phase lag.

Regards,

Rohit 

Hello Rohit,

   The document Michael linked will help calculate the necessary resistors for output impedance matching. For preventing the 6dB gain loss, I would follow the first part of the document where it discusses upping the gain of the FDA to counteract this effect; since you have the BW margin you wouldn't need to use the synthesized output impedance matching approach. Depending on the ADC and output loading that you brought up in the last thread, you could use the lower BW part like the THS4561 if later on power is a concern.  

Thank you,

Sima

Hi Sima,

Thank you for your suggestions.

I will increase the gain of FDA for 6dB loss.

I have updated the cable details.

Please suggest me how to model the cable in TINA. to simulate the THS4551 with cable model.

Regards,

Rohit 

Hello Rohit,

  Sorry for the delay in response. Here are Tina-TI simulations you can work with to determine how much you want to compensate for insertion loss. I included source impedance, but you can take that out if compensated already. The first file has THS4551 model with datasheet specified 1k load, the second file just has the 80 ohm load which will differ from datasheet specs. Important figures to look for your capacitive and resistive load would be figures: 4,5,6,8,10, and 16. 

Thank you,

Sima

THS_OutputImpedance_1kLoad.tsc

THS4551_OutputImpedance_80OhmLoad.TSC