Using a Unity Gain Buffer (Voltage Follower) with a DAQ Device

[FrankHS]

This question refers to the document Using a Unity Gain Buffer (Voltage Follower) with a DAQ Device which is at:

https://www.ni.com/en/support/documentation/supplemental/12/using-a-unity-gain-buffer--voltage-follo...

In figure 5 the schematic shows a +/- 15 volt power supply powered by the NI card's 5V supply.

Can anyone recommend a power supply. I tried the Murata MEA1D0515SC module with disappointing results. Inductors and capacitors were used as specified in Murata's data sheet.

http://power.murata.com/data/power/ncl/kdc_mea.pdf

The op amps were Burr-Brown OPA4277 (the 4 unit version of the one suggested in the white paper).

Noise was very high and occasionally the power supply module got hot and stopped working. These problems do not happen with an external power supply.

There were 3 PCB's made and all had these problems.

 Can anyone recommend a power supply that will convert the 5 volts from the card to +/- 15 V to drive the op amps?

[Liquefactionist]

I think there is something wrong with your design or PCB layout.  The OPA4277 should be using a couple milliamperes and not taxing the Murata supply (33 mA rated output).  If noise is high, what are you doing for decoupling capacitors around the op-amps? 

A layout of the PCB can help here (schematic too).  But the Murata should not be getting hot unless somewhere you are shunting a lot of current that you should not be.

[FrankHS]

Thanks for the reply.

There are 4 OPA4277 chips. They pull a total of 13.5 ma. when operating normally, well within the range of the Marata. The heating only happens occasionally, most of the time it runs correctly with only the noise issue. I have seen the heating malfunction happen only 3 times over several days of testing, making it difficult to locate the cause. The first time I saw it, i thought there was a short on the board. When it happened on a second board I knew there was a problem. I am hoping to find a different power supply module.

Each op amp has a 20 meg resistor to ground on the input and a 50 ohm resistor from the output to the NI card. The output shouldn't add any load considering the high input impedence of the NI card. The connections are exactly as shown in figure 5 of Using a Unity Gain Buffer (Voltage Follower) with a DAQ Device. in differential mode.

https://www.ni.com/en/support/documentation/supplemental/12/using-a-unity-gain-buffer--voltage-follo...

[Henrik_Volkers]

About the heating:

What comes directly into my mind is a HF oscillation in one or more of the OPs. Just had that, 6MHz on a usually 200kHz bandwidth amp. Just noticed a minor shift in the  transferfunction. Looking with a high bandwidth (analog) scope showed it clearly.   And using quad versions wasn't a good idea IMHO. Migth save some space, but is prone to crosstalk.

[Henrik_Volkers]

Well after looking at the white paper and the DC/DC spec:

Since the noise is much lower with an external DC supply, the DCDC seems to be the root cause for the noise. This could be of two reasons: output ripple of the +-15V and input current ripple into the AGND.   Try an external 5V supply and ONE connection from the DCDC output filter GND to AGND. Better noise?    A input current filter (CLC) maybe will help.

A complete schematic and PCB layout would help.  (Insert 10 to 47 Ohm from the 15V rails to the OP decoupling Cs ... )

(and NI: Why there is not comment field below the WP??)

I would like to dispute the AGND to power the DCDC, instead use the DGND power and reference the (isolated) DCDC output GND to AGND.

The switching freq of the DCDC is 85kHz ... what frequency range do you want to monitor??    You say the input is noisy, have you sampled the noise at the highest possible samplerate and made a FFT? 

[Liquefactionist]

The op-amps claim 130 dB PSRR.  I am wondering about the ground/common issue.  The Murata states that the outputs are isolated from the inputs, so the ground there is not the common.  Not sure this is an issue, but could be something.

[Henrik_Volkers]

---

@Liquefactionist wrote:

The op-amps claim 130 dB PSRR.  I am wondering about the ground/common issue.  The Murata states that the outputs are isolated from the inputs, so the ground there is not the common.  Not sure this is an issue, but could be something.

---

I vote for EMC noise due to AGND current spikes from the DCDC converter. Whether input or output noise... depends on the schematic AND PCB layout.