Buffer amplifier for DAC8734EVM

Hi Nicola,

The DAC8734 can deliver a maximum current of ±3mA. With an output requirement of ±7V and 50Ω load, you will need to deliver ±7V/50Ω = ±140mA of current.

Nicola Dalla Pozza said:

I know you can use a buffer to perform an impedance matching, but I got lost looking for the right op amp to use.

You are indeed correct, this can be done using a buffer. Based on the requirements you described, I would recommend using the OPA552.

• Output Current = 200mA
• GBW = 12MHz
  • Unity gain bandwidth ≈ 3MHz (NOT 12MHz b/c of a 2 pole system)
• Supply Range = ±4V to ±30V
  • Swing @ Room Temp = Vpos - 3V | Vneg + 3V
  • Minimum supplies for this application = ±10V

I need to double check with the amplifier team if there is anything that I am overlooking in this amplifier. It may take a couple of days, but I am pretty sure it should be fine.

If you can share a schematic or a block diagram of the system I may be able to make better suggestions. Also, if you have any other requirements for your application then let me know and I can look for a better solution.

Hi Eugenio,
thanks for the quick reply. While I was looking for an opamp to perform the impedance matching, I indeed ordered a few free samples of OPA552 (PDIP package), but I still got problems

When this opamp is used in a buffer configuration, i.e., with a short circuit between the output and the "in-" input, it generates a sinusoidal disturb at 3MHz with an amplitude of about 4 V peak to peak that is added to the input signal.

Other guys in the lab told me that probably the opamp is not internally compensated, and indeed there is a section in the datasheet that talks about using the OPA552 in low gain. In particular, figure 11 shows an inverting configuration (gain equals to -1) with a feedback network that allows to compensate the opamp, but it restricts the bandwidth of 956kHz (in addition, it is not clear to me how this value has been evaluated).

However, this is not the main problem. Setting the feedback network eliminates the disturb, but at the output of the opamp I cannot reach the correct swing. I mean that if provide as input a sine with 10V peak to peak with frequency well below the bandwidth limit, I do not have the same amplitude in the output if I measure it with a 50 ohm load.

Again, it seems to me that the configuration cannot provide the right amount of current. Looking carefully the datasheets, it is said that the maximum output current of 200mA is actually package dependent, but in the text I cannot find which package can actually drive 200mA. I would prefer to use PDIP package because it is easier for me to set the circuitry on a breadboard.

Am I correct in the analysis? Can you please suggest other alternatives to the OPA552 or a compensation circuit with a greater bandwidth?

Thanks,

Nicola

Nicola,

Nicola Dalla Pozza said:

OPA552 in low gain

I missed the fact that the OPA551 is the unity gain stable version of the OPA552. It will provide the buffering capability without the stability issues that you are observing on the OPA552.

Nicola Dalla Pozza said:

Setting the feedback network eliminates the disturb, but at the output of the opamp I cannot reach the correct swing. I mean that if provide as input a sine with 10V peak to peak with frequency well below the bandwidth limit, I do not have the same amplitude in the output if I measure it with a 50 ohm load.

In order to debug this issue I may need some more information.

• What power supplies are you using?
• What type of feedback configuration are you using? (What are the resistor values?)
• What amplitude are you observing? Can you share an oscilloscope capture?
• Keep in mind that a ±10V sine wave with a 50Ω load will strain the device and cause it to heat up significantly.

Nicola Dalla Pozza said:

Looking carefully the datasheets, it is said that the maximum output current of 200mA is actually package dependent, but in the text I cannot find which package can actually drive 200mA. I would prefer to use PDIP package because it is easier for me to set the circuitry on a breadboard.

The note about the output current being package dependent is in place because the different packages have different temperature dissipation coefficients. It means that some packages will heat up  much quicker than others because they can't cool down fast enough. This is important because the device has a protection shut-down mode when a critical temperature is reached (160°C). 

Note that for reliability the junction temperature should not exceed 125°C.

To calculate this please refer to page 10 in the datasheet.

I did some calculations based on your required output on the DIP package.

Junction Temperature Calculations

%thetaJA SO-8 SM = 90
%thetaJA DIP-8 = 100
%thetaJA DDPak-7 = 65

thetaJA = 100;  %Juction to Air Thermal Resistance (°C/W)
Iq = 8.5e-3;       %Quiecent Current of the Device (A)

TA = 40;             %Ambient Temperature (°C)

Vs = 10*2;         %Total Supply Range (V)
Vo = 7;               %Ouput Voltage (V)
RL = 50;            %Load (Ω)

Io = Vo/RL;                          %Current Output (A)
PD_int = Iq*Vs;                  %Internal Power Dissipation (W)
PD_out = Io*(Vs/2-Vo);     %Output Power Dissipation (W)
PD = PD_int + PD_out     %Total Power Dissipation (W)
Tj = TA + PD*thetaJA        %Juction Temperature (°C)

PD = 0.59W
Tj = 99°C

This is below the 125°C threshold, but note that the datasheet suggests using the DDPak-7 for applications where the power dissipation is continuously greater than 0.5W.

Dear Eugenio,

with the Electronics guys in the lab I have set up the scheme of Figure 11 in the OPA552 datasheet correctly, and now it works.

However, we have found out that the slew rate of this opamp is too low for our purposes. This is shown for instance in the picture that I attach, where the green line corresponds to the signal at the input and the yellow line correspond to the output signal. As you can see the yellow curve seems more a triangular wave rather that a sine.

I wonder if you have alternatives to the OPA552 that I could use (consider the specification of the previous email, and add the constraint on the slew rate that have to be greater than 7V * 1MHz * 2 pi = 42 V/us).

I will try the OPA551 as soon as I got it, but I think I will found the same issue on the slew rate.

Thanks for your time and patience,

Nicola

Nicola,

I moved this to the amplifier forum since they will be able to better assist you with your request.