Hi

If you configure those two rail as voltage monitoring instead of current monitor, what are the reading from GUI?

Regards

Yihe

Hello Yihe,

Many thanks for coming back with that suggestion.  I've removed these inputs as current monitoring inputs, and reconfigured them as voltage rail monitors, as shown below:

3.3 Volts:

For the 3.3 Volt current monitoring rail, I'm correctly reading out approximately 0.2 Volts from the Fusion Designer Monitor section:

This correctly ties in with the 0.2 Volts DC level that I'm monitoring on the oscilloscope, but I'm still seeing the 100µs period pulse on the waveform:

12 Volts:

I'm seeing the waveform read from about 0.24 to 0.30 Volts in the Fusion Monitor:

This largely ties in with my oscilloscope measurement:

Thoughts

It's interesting that the period of the 3.3 Volt signal spike is still at 100 µs (10 kHz), even though current channels are sampled at 200 µs intervals, and voltage channels at 400 µs intervals.  I would expect the period of these pulses to change depending on whether this is set as a current or a voltage monitor input on the UCD90320.

Many thanks for your continued assistance with this!

Hi

There are many things impact. ground noise, V33A ripple,

Have you checked V33A voltage? 

can you adjust the INA so that the voltage reading could be larger?

Regards

Yihe

Hi Yihe, thanks for the suggestions.  I've done some tests with the changes to the circuit highlighted in red, and the tests in black headings below.

R431 and R432 disconnected

• The UCD90320 is completely disconnected from the INA190A1 output (R431 and R432 removed).
• The ferrite beads between ground and REF, FB4 and FB5 are replaced with wire links

Check 3.3 Volt current amplifier output

The 10 kHz period spikes disappear from the 3.3 Volt current amplifier, shown in Figure 1.

Figure 1: 3.3 Volt current sense amplifier at TP24 with R431 and R432 disconnected, 10 mΩ sense resistor

In Figure 1,

• a DC voltage of 211 mV is measured on the scope
  • 1 mV is measured on a DMM to confirm
• This corresponds to a 840 mA current
  • through a 25V/V amplifier and 10 mΩ sense resistor
• The bench supply output current reads 670 mA
  • A DMM current sensor reads 667 mA to confirm
• The amplifier is 25% out

Check 12 Volt current amplifier output

Figure 2: 12 Volt current sense amplifier at TP25 with R431 and R432 disconnected

In Figure 2,

• a DC voltage of 272.2 mV is measured on the scope
  • 5 mV is measured on a DMM to confirm
• This corresponds to a 175 mA current
  • through a 25V/V amplifier and 62 mΩ sense resistor
• The bench supply output current reads 190 mA
  • A DMM current sensor reads 187 mA to confirm
• A Utility current of 15 mA is drawn upstream of the sense resistor
  • A 175 mA current is expected through the sense resistor
• The amplifier is correct

Increase the 3.3 Volt sense resistor

As suggested by Yihe, try increasing the sense resistor on the 3.3 Volt INA190A1 current sense amplifier.  Increase from 10 mΩ to 166 mΩ.

Figure 3: 3.3 Volt current sense amplifier at TP24 with R431 and R432 disconnected, 167 mΩ sense resistor

• a DC voltage of 2.83 V is measured on the scope
  • 825 V is measured on a DMM to confirm
• This corresponds to a 677 mA current
  • through a 25V/V amplifier and 166.8 mΩ sense resistor
  • Sense resistor was measured through Kelvin Sense probes
• The bench supply output current reads 690 mA
  • A DMM current sensor reads 685 mA to confirm
• The amplifier is now 1% out

Check the noise on the 3.3 Volt power supply to INA190

Yihe suggested checking the noise and ripple on the 3.3 Volt power supply to the current sense amplifiers.  This is measured at the decoupling capacitor of the INA190 using an oscilloscope probe with a ground blade reference.  The measurement is shown in Figure 4.

Figure 4: 3.3 Volt power supply to INA190 current sense amplifiers, measured at decoupling capacitor C750

Figure 4 shows a mean measurement of 3.267 Volts and a peak to peak measurement of 9 mV.  The power supply is in specification, and the peak to peak measurement of voltage ripple is not a cause for concern.

Fit R432 to reconnect 12 Volt Current Sense Amplifier

R432 is fitted to connect the 12 Volt current sense amplifier back into the UCD90320

Figure 5: No 10kHz pulses present on the 3.3 Volt current amplifier output, TP24

Figure 6: 10kHz pulses present on 12 Volt current amplifier output, TP25

Figure 7: Zooming in on one of the pulses in Figure 6

Fit R431 to connect 3.3 Volt Current Sense Amplifier to UCD90320

Now, both the 3.3 Volt and the 12 Volt current amplifier outputs are connected to the UCD90320 through 200 Ohm resistors.

Figure 8: No 10kHz pulses present on the 3.3 Volt current amplifier output, TP24

Figure 9: 10kHz pulses present on 12 Volt current amplifier output, TP25

Questions

• Why does increasing the shunt resistor make such a big difference to the 3.3 Volt PCIe current sense amplifier’s output?
  • As far as I’m aware in the datasheet, the lower limit for shunt resistor is to stay above 0 Volts in the configuration I’m running
  • I therefore believe that the amplifier is running in-specification with both 10 mOhm and 167 mOhm sense resistors
• Should I consider using a bigger gain INA190 instead of a larger shunt resistor?
  • I’d rather not inflict too large a voltage drop on the 3.3 Volt PCIe rail
• When I connect both of the current amplifier outputs back up to the UCD90320 inputs, why has the switching moved from the 3.3 Volt to the 12 Volt amplifier?
  • The only thing that has changed is the 3.3 Volt current sense shunt resistor

Many thanks for your continued support with this case, it's much appreciated!

Hi

As for the shunt resistor or INA related question, you may need reach out to the INA team to get better understanding. We are not familiar with the part.

What I asked is for the V33A feed to the UCD90320 since it is the source for the analog circuity inside the UCD.

When you have R431 and R432 disconnected, the ripple on the V12 is still present. but at this moment, UCD is totally separated with your INA

Regards

Yihe

Hi Yihe,

Thanks for this feedback.  I'll set up another forum post for the INA190A1 as you've suggested.

As requested, here is a trace of the 3.3VA pin.  It's measured on one of the UCD90320's decoupling capacitors with a ground blade on the scope probe.

Here's a measurement of the 3.0 Volt external reference to the UCD90320 for VREFA+.  It's measured at R234/C247 with a ground blade on the scope probe.



Many thanks,

Pete

Hi

Thank you for the capturing, the ripple of  V33A seems ok but below 10mV is better though i don't think it caused problem. 

I10mV on VREF is also fine since it is less than 0.5%

Regards

Yihe