Differential buffer/driver for long cable.

Hi Mikael,

I expect that a pair of high voltage, or power operational amplifiers could be configured in a differential-in, differential-out arrangement. But before I can give this more thought there are questions that I will need answered:

1. What supply voltage do you have available? I assume dual supplies are being used for the preamp.
2. When you mention a differential signal of +/- 20 V does that mean that one differential output will be at +20 Vpk, while the other differential output is at -20 Vpk? Or is that 20 Vpk-pk appears between outputs?
3. Is the cable always 1.5 km in length?
4. Can you tell me the cable number? If not, what is the capacitance (pF/m) between the two conductors and from each conductor to the shield, if used?
5. What is the frequency range that the buffer must pass without attenuation?
6. Does the load appear as a high impedance at all frequencies? Is it capacitive?

Regards, Thomas

PA - Linear Applications Engineering

Hi Thomas,

Thanks for answering!

1. The base case is a a single 28-30 VDC supply.

2. Each of the differential lines swings between 0 V and 20 V.

3. The cable length is fixed for each product, but may vary between 1300 and 1500 m in each case.

4. The capacitance is 44pF/m between the two conductors. The capacitance to the shield is not specified.

5. The frequency range is DC to 100 kHz.

6. As far as we know the load is resistive 100k for the whole signal frequency range.

Mikael

Hello Mikael,

Actually, this is a surprisingly challenging application. I have a solution that covers the requirements. It is preliminary at this point and some refinement would be required if you decided to use it.

Even though the line is essentially open-circuited it has significant capacitance that must be charged and discharged as the applied signal transitions from one peak to the opposite peak. This requires that the amplifier have sufficient current drive capability, slew rate and bandwidth to support that signal. Since this is a differential output application with a 40 Vpk-pk requirement each output must swing 20 Vpk-pk, or 10 Vpk. Using the 100 kHz maximum frequency the minimum slew rate, assuming a sine wave:

SR = 2 x pi x Vpk = 6.28 x  10⁵  x 10 = 6.3 V/us

Using a conductor-to-conductor capacitance of 44 pF/m, 22 pF/m for conductor to shield, and a 1.5 km cable length, results in a differential load capacitance of 66 nF and 33 nF off common-mode capacitance. Then, each output sees about 100 nF of load capacitance.

The capacitive reactance which is lowest at 100 kHz is:

X_C = 1 / (2 x pi x  C_L ) = 1 / (6.28 x 10⁵ x 100 10^-9 ) = 15.9 Ohms

The output current required to drive this reactive impedance is:

I_OUT = Vpk / X_C = 10 V/15.9 = 0.63 Amps

These are approximations but they get us in the ballpark in understanding what the amplifier must deliver.

Using this information and the details of what you provided I developed a differential-in, differential-out line driver. It is based upon two of our OPA564 power operational-amplifiers. You can see the circuit below. A a larger image it available in the attachment.

The OPA564 has the slew rate, bandwidth and current output sufficient for the application. Its maximum power supply of 26 V so you would have to reduce the voltage from 28 V to 30 V down to 24, or 25 V. A 24 V, 3 terminal regulator should be good. Also, the amplifier would require a midpoint reference voltage of half the supply V+ level.

No serious attempt has been made to frequency compensate the circuit and there is some high frequency peaking our beyond a Megahertz. Those are the sorts of things that would have to be addresses in a more in-depth design. I am only presenting my circuit approach at this point.

Regards, Thomas

PA - Linear Applications Engineering

Hi Thomas,

Thank you for working out this solution.

I would like to do some testing on this but I couldn't find an evaluation board option for the OPA564.

Making an evaluation board from scratch will take some time. Especially now during the summer as a lot of companies have summer vacation.

Do you have any suggestions on how to do a practical evaluation on this or an equivalent circuit?

Best regards,

Mikael

Hi Mikael,

I don't have an evaluation module for the OPA564, unfortunately. However, there is an adapter which takes OPA564AIDWP HSOP-20 package and breaks it out to a double-wide DIP pin-out. The underside of the adapter has a tinned, copper layer with no solder mask. There are vias from the topside where the OPA564 PowerPad attaches. They conduct heat from the underside PowerPad, to the underside copper layer. There might be enough underside copper area to provide adequate heat dissipation, but if not a strip of aluminum can be mounted to the screw holes integral to the adapter. The adapter can be plugged into a white board. Although a white board isn't the best way to test a circuit idea, especially when a wideband amplifier is involved, it might be useable to at least get an idea if the circuit will do what you need.

If you would like some adapters with OPA564AIDWP operational amplifiers mounted on them I can have them assembled and sent to you.

Regards, Thomas

PA - Linear Applications Engineering

Hi Thomas,

Yes, it would be helpful to have some circuits for testing. Could you organize, say four of them, then I have spares in case something happens?

Best regards,

Mikael

Hi Mikael,

I will have four OPA564AIDWP devices assembled on the DIP adapter boards and sent to you. Let's go ahead and take this off the E2E forum so that you can provide me your address information privately. My TI email address is: kuehl_tom@ti.com

Regards, Thomas

PA - Linear Applications Engineering