Acoustic output of AO0 via USB-6216 BNC is 8.4 dB softer (compared to PCI 6221)

[Fuh]

I am a researcher who presents acoustic sounds to adults and infants. For adults, a system in my lab is used. For infants, a system on a cart with wheels is used, so that I can move this cart around to the infants. 

For the system in my lab, the setup is as follows: 

1. A desktop computer contains a PCI 6221 card and the LabVIEW software
2. A custom-written LabVIEW program reads in a .wav sound file. This .wav file is a pre-recorded speech utterance and has a duration of 150 ms. For a typical experiment, this .wav file will be presented repeatedly for a total of 8000 times.
3. The signal goes through a PCI 6221 card (that is inserted inside this desktop computer)
4. The signal goes through a BNC-2090A that is mounted on a 19" rack
5. The signal goes through the AO0 channel of the BNC-2090A
6. The signal goes to an insert earphone (Etymotic ER-3A, 50 ohms)

For the system on the cart, its setup is as follows:

1. A laptop computer contains the LabVIEW software
2. The same custom-written LabVIEW program reads in the same .wave sound file
3. The signal goes through a USB-6216 BNC (this device is connected to the laptop computer via a USB cable)
4. The signal goes through the AO0 channel of the USB-6216 BNC
5. Ths signal goes to the same insert earphone (Etymotic ER-3A, 50 ohms)

As I am calibrating the sound pressure levels of the acoustic outputs generated by these two systems, I connected the output of the insert earphone to a 2 c.c. coupler and measured its dB SPL by using a sound level meter (Extech 407732, dB A, fast, Hi). Surprisingly, the sound coming from PCI 6221 reads 76.7 dB SPL, while the sound coming from USB-6216 BNC reads 68.3 dB SPL. 

To make sure both the PCI 6221 ND USB-6216 produced the same voltage, I connected the Ao0 channel of each system to an oscilloscope. Their waveforms looked identical and both had a baseline-to-peak amplitude of 5 volts. 

The thing that confuses me is that, with the same baseline-to-peak amplitude and the same waveform morphology on an oscilloscope, how come the sound pressure level is 8.4 dB softer when delivered through USB-6216 BNC (compared to when delivered through PCI 6221)?

I do not have a background in engineering, and maybe this is the reason that I am confused. 

Any comments or suggestions will be highly appreciated!!!

[santo_13]

The reason is the difference in the output drive current of each of the cards.

PCI 6221  has a drive strength of 5mA whereas the USB 6216 has a drive strength of just 2mA.

USB-6216 output specification

PCI-6221 output specification

[santo_13]

BTW, as per Ohms laws, your amplitude cannot go beyond 0.25Vpk for 6221 and 0.1Vpk for 6216 (V = I*R), if you try to measure the 5Vpk amplitude with the speaker connected, you will not be able to see the 5Vpk as 50ohm is a quite large load for the instrument to effective drive.

[Fuh]

Hi Santhosh,

Thank you SO MUCH for pointing out the facts. Greatly appreciated!!! 

I have to admit that this is a new setup that I am current trying. I am glad that I have tested and asked for clarifications before using the new setup in actual experiments. Please allow me to reiterate, just to make sure that I have understood the concepts correctly.

1. The ratio between the output current drives of USB-6216 and PCI 6221 is the cause of the dB difference, which can be computed as 20 * log(2 mA / 5 mA)  =  8 dB. Correct?
2. Does this mean that the new setup (i.e., connecting a 50-ohm insert earphone directly to the Ao0 channel of PCI 6221 or USB-6216) has a fatal flaw and should not be used? Or is the new setup still legit as long as the baseline-to-peak amplitude is kept within 0.25 V for PCI 6221 and within 0.1 V for USB-6216? 
3.  When a 50-ohm insert earphone is used, the maximum baseline-to-peak amplitude will be 0.25 V for PCI 6221 and 0.1 V for USB-6216. In other words, if I want to accomplish a 5 V baseline-to-peak amplitude, I will need a output current drive of I = V / R = 5 / 50 = 0.1 Amp = 100 mA. Correct?
4. If an output current drive of 100 mA is needed, what would be the best solution? In the previous setup of my lab, I have a Tucker Davis Technology Headphone Buffer (TDT HB6) in place to boost the current drive before the signal goes into the insert earphone. Should I put the headphone buffer device back into the system? Are there other alternatives?

[santo_13]

1. Yes, since the resistance is constant, taking the voltage ratio or the current ratio does not matter, both result in -7.958 dB
2. It is still legit as long as you don't exceed the 0.25V or 0.1V limitation based on the model (even if you try to exceed it will clamp to these values by ohms law due to load resistance and drive current)
3. Correct, to achieve 5Vpk, you need at least 100mA drive capacity for a 50 ohm load
4. You definitely need a buffer circuit, this could be a simple OPAMP circuit or the HB6/7 amplifier you're talking of. No alternative, either live with the limited amplitude or add an amplifier or change the card to a sound and vibration card that has higher drive strength (like PXI-4463, there are other USB or C-series cards too).

[Henrik_Volkers]

If the headphone ER-3A has the same power than then the ER-3C , mean

102.5 dB SPL in HA-2 coupler at 0.2 Vrms (50 Ohms)

and you build your own buffer, please think of an attenuator 😄 (just like the HB6/7)

(increase the signal to noise too 😉  )

[Fuh]

To confirm what I just learned, I went ahead and did the following three measurements in my lab:

1. The AO0 channel of PCI 6221 did produce a maximum amplitude of 10Vpk, when a 50-ohm earphone was not connected.
2. When a 50-ohm earphone was connected, the AO0 channel of PCI 6221 produced a maximum amplitude of about 7 Vpk.
   1. As expected, the voltage was reduced significantly from 10 Vpk. However, it was not clamped at 0.25 Vpk. Why?
   2. Is it possible that the BNC 2090A terminal block (that is part of the PCI 6221) is providing extra strength to this system?
3. I then placed a Tucker-Davis-Technology Headphone Buffer (HB6) between PCI 6221 and the insert earphone, the output voltage coming from the Headphone Buffer was measured to be slightly below 7 Vpk.
   1. The Headphone Buffer actually decreased the output voltage slightly. Why?

I must have done something wrong... 

Any comments and suggestions will be greatly appreciated!!!

[santo_13]

It is rather overperforming by driving 7Vpk into a 50-ohm earphone (~140mA), please verify the earphone impedance using a DMM.

BNC 2090A in noway provides additional drive strength because it is a straight pass-through for AO signals as described in its manual

Please share the scope waveforms you're observing.

[Henrik_Volkers]

the earphone has a complex impedance (valid assumption: inductive) so with increasing frequency the impedance migth increase 🙂

Your amplifier has a input impedance of about 10 kOhm , and a output impedance of 4 Ohms ( if I remember the spec rigth )

if you connect a load to the output, the voltage level will drop a bit.  (all Ohms law :D)

[Fuh]

Thanks a lot for the comments. Greatly appreciated!!!

This afternoon, I did the following measurements:

First, I measured the impedance of the ER-3A insert earphone. The DMM reads 52.8 ohms. Please note that, in order to reduce potential electromagnetic interferences, I have used some alloy metal to wrap around the transducer box and taped it with some red tapes. I have also replaced its cable with a shielded BNC cable. This may be the reason that the impedance is slightly higher than 50 ohms.

Second, I measured the maximal amplitude of the Ao0 output voltage from PCI 6221. For clarity, I presented a series of electric pulses (pulse width = 100 microseconds), by using a custom program written in LabVIEW (which generates a series of pulses to the Ao0 channel). When the 50-ohm insert earphone is not connected. The oscilloscope reads exactly 10 Vpk and the pulse width is exactly 100 microseconds. So far, everything is good.

Third, I connected the 50-ohm insert earphone, through a BNC T connector, to the Ao0 channel of PCI 6221. The oscilloscope now reads an amplitude of about 7 Vpk. The waveform is bit distorted, but the pulse width is still exactly 100 microseconds. This is where I get confused... The waveform is not clamped to 0.25 Vpk. Why?

My lack of engineering background is really a problem here, and I apologize for that. However, I really want to get this figured out and get it to work in my lab. 

Any comments and suggestions will be greatly appreciated!!!