How can the PCI-5401 Fgen Appear to have Higher Resolution than the PCI-5402?

[rdtech]

We have multiple test stations assembled with a PCI-5401 and PCI-6034e card each. The Fgen is used to set a voltage to an amplifier  then coil and the DAQ card measures the current by the use of a shunt resistor. The test system adjusts the voltage out to achieve a certain current.

I recently assembled a new test station with a PCI-5402 card because the 5401 is obsolete. The system is not always able to achieve the correct current. I've written a simple program to manually adjust the voltage out to see the measured current. I find that the resolution of the 5402 is clearly visible and the current changes with about 1.5 mV pk-pk change in voltage out. The older 5401 card displays a more gradual, smaller change in current with small, 0.5mV pk-pk, changes in voltage out.

I thought our problem was due to the DAQ measurement card with 14 bits, until I realized it was 16 bits. One solution was to change the shunt resistor to 10 Ohms from 1 Ohm. Now I find that this is not acceptable.

Is there any solution to the Fgen with a higher resolution appearing to have a lower resolution?

[jaced]

Hi,

I believe I can help clear up some of the confusion. While the 5402 does offer a higher DAC resolution than the 5401, that does nottranslate into Amplitude resolution. Meaning that through the use of analoggain you can achieve steps of .004dB where as with the 5401 you can achievesteps of .001dB. This is due to the amplitude resolution being dependent on theresolution of the reference voltage provided to the DAC. There is however aworkaround for this. It is possible to adjust the digital data through the useof digital gain, which will give you finer resolution which based on yourdescription is exactly what your application needs. The Digital Gain propertycan be found under the output attributes category. You can then essentiallymultiply your data by something like .9999 to decrement the data in smallsteps. So now the same analog gain is being applied to smaller data. Throughthe use of this method you are able to leverage the 16 bits of the DAC towardsamplitude resolution. I hope this clears things up please let me know if youhave any questions or if anything is unclear. 

Message Edited by jaced on 09-03-2008 05:02 PM

[rdtech]

jaced,

Thank you very much for your reply. This problem has been confounding me and others for about 3 weeks. Your solution is helpful, but does not give us the resolution that we need.

I found in the specs for the 5401 and 5402 amplitude resolution measured in dB. For future reference, is this dB of the usable range, usally +/- 5V?

[jaced]

One small correction to my previous post the 5402 is a 14 bit card, not a 16 bit as I said.

[jaced]

The Amplitude resolution is easiest to think about in step size within ranges. For example. you have a 3dB step between analog attenuators, within this 3dB you have steps of .004dB. Which is why we also spec it as a percentage on the 5402, it is relative to the range that you are using. .004dB of a 5V range is larger than .004dB of a 1V range. I am a little surprised that the use of digital gain did not give you the resolution that you require. What resolution do you need and can you elaborate on how you used digital gain, I am intereted in what your gain values were. One thing to note is that the 5406 does have a 16 bit dac and thus finer resolution when using digital gain so this may be an option. I would like to explore the 5402 a bit more to ensure that we are on the same page. 

[rdtech]

I tried the gain set from minimum, 0.5, to maximum, 1.5. Higher gain actually seemed to be moving in the right direction, but I believe the waveform begain to be distorted. This was based on the secondary voltage of our test setup.

The actual voltage needed is hard to pinpoint. The fgen is used to supply current via an amplifier through a test coil. Our magnetic product is placed in the coil. I assume the coil then acts as an inducter and the required voltage to achieve the amps setpoint may change. The final problem lies in the resolution of current as measured by voltage drop across a 1 Ohm resistor. The minimum current steps at the target value are too great to do our measurements. 38.14 mA +/- 0.125%. Changing the resistor value would help, but I'm told I can't do that.

Is it possible that it actually has something to do with the power output of the 5402 and not just voltage?

[jaced]

Thank you for the explanation. I think the problem is that the steps you are using for digital gain are too large. I apologize for notmaking that clear, but something on the order of thousandths rather than tenths I believe would give you more resolution. I suggest trying something on the order of .001 step size .The reason you are seeing distortion with a digital gain greater than 1 is because you are multiplying your digital data such that it is beyond the range of the DAC and you will begin to clip. The solution to this is to set the analog gain or in this case the Amplitude to the highest value you will need and then use digital gain to move within that range. A digital gain of .5 would be the midpoint of the range. At this point I will not rule out anything but I am leaning toward resolution rather than power output. Just to be certain though could you provide the load impedance that the 5402 is connected to? Also if you know the current levels you are expecting/ requiring the 5402 to produce that would be helpful.  I hope this helps! 

Message Edited by jaced on 09-04-2008 11:50 AM

[rdtech]

The 5402 is connected to a Krohn-Hite model 7500 power amplifer. The manual says it has 5000 ohms input impedance.

The required current to do our testing is 38.14mA. The 5402 signal is routed through the amplifier, then Krohn-Hite matching transformer. I don't have a way to measure the 5402's current before reaching the amplifier. The use of a sampling resistor seems to just load it down.

I think I understand the digital gain approach to reaching a very fine resolution. The problem is that I need to keep these test stations has much the same as possible. Is there a way to determine if digital gain adjustment is supported programmatically? I believe the 5401 will not throw an error if you try to adjust the digital gain, but it won't do any good either.

Thanks for your help

[jaced]

There is a property that you can read that will determine which card you are using and then you could use a case structure to switch between methods(digital vs. analog gain). The property is shown below.

 

Message Edited by jaced on 09-04-2008 05:40 PM

[rdtech]

Jace,

Thanks for all your help. The solution was to modify the program to adjust the digital gain when the current measurement was within two percent of the target value.