Headphone Output Noise:
Tascam DP-008, DP-24SD and DR-40

Oct 20, 2020

This article provides results of headphone-output noise level measurements for three Tascam recording devices: In this first section, an Analog Discovery dual channel oscilloscope was used to monitor the noise level of each channel for each headphone output. For each recorder, the output headphone level was set at maximum. (The headphone-output noise results were essentially identical at minimum and maximum settings.) A FiiO E17 very low noise headphone amplifier with a gain of 12 dB (voltage gain of 4) was used to amplify the headphone-output noise level of each Tascam recorder above the scope measurement background noise (~ 800uV RMS). [NOTE: These scope measurements are considerably higher than the actual total noise values within a 20 kHz bandwidth as the scope has a bandwidth of 10 MHz. However, it does give a semiquantitative picture of the relative noise values. See below for accurate noise measurements for the DP-24SD].

Reference Noise Level with 12 dB amp and splitter cables/box in place (5mV/div)

(noise level of testing setup is just below scope input noise of ~ 800uV)

DP-008 Recorder Headphone Output Noise with 12 dB amp and splitter cables/box in place (5mV/div)

(noise level is about 50% higher than scope noise with some channel difference)

DR-40 Recorder Headphone Output Noise with 12 dB amp and splitter cables/box in place (20mV/div)

(noise level is about 10X higher than scope noise with some channel difference)

DP-24SD Recorder Headphone Output Noise with 12 dB amp and splitter cables/box in place (20mV/div)

(noise level is 10-20X higher than scope noise with strong channel difference. The suprisingly high background noise for the tested DP-24SD (compared to the DP-008 recorder) is audible particularly in quiet music sections. It is not known if it originates in the headphone-output analog driver circuit. Only one DP24-SD was tested.

DP-24SD Recorder Stereo Out and Monitor Out Noise with 12 dB amp and splitter cables/box in place (10mV/div)

The noise measurements below for the Stereo Out (~ 1.5mV RMS) as compared to the Monitor output (~ 8mV RMS) and Headphone Out under the same conditions as above (with 12dB amp in line) might suggest that the noise originates from the two output amps driving the Monitor and Headphone outputs. However, more careful measurements (see below) and inspection of the schematic diagram shows that these output amps don't add significantly to the noise but simply amplify the noise originating at the output of the DACs. The block diagram of the DP-24SD below includes integrated noise values at each output and the corresponding gain of each section separately (from the schematic):

Listening Test: DP-24SD Headphone Out vs DP-24SD [Stereo Out + FiiO E17 headphone amp]

For a listening test, to compare the DP-24SD Headphone out signal and noise level with that of the [Stereo Out + FiiO E17 headphone amp], a track was played back and the Track level and Stereo sliders were adjusted with the Monitor Level set at 12:00. Headphones were Sennheiser HD 380 Pro (~ 50ohm). This was a fairly loud listening level using DP-24SD headphone out with these phones. Next the playback level was checked with the same headphones through the [Stereo Out + FiiO E17 headphone amp] output. The gain of the FiiO E17 was adjusted to get the same listening level which was determined to be almost 0dB (voltage gain of 1) so in this case, the FiiO E17 essentially acts as a unity gain buffer amp between the Line Out and the headphones. With these signal playback level adjustments, the high listening levels were the same.

The audible noise levels were then checked. As noted above, the background "hiss" noise is faint but distinctly audible in the DP24SD headphone output. The noise level from the [Stereo Out + FiiO E17 (at 0dB)] is completely inaudible. This demonstrates that the Stereo Out of the DP-24SD combined with a low noise buffer amp (even with only unity voltage gain) can provide a very easy workaround, if the DP-24SD headphone-out noise is distracting.

DP-24SD Headphone Out Noise Attenuator

The Tascam DP-24SD and DP-32SD are designed with a fairly high headphone drive power capability (70mW/channel into 32 ohm headphone loads). Since the Monitor level adjustment is digital and occurs before the DAC driving the Monitor and Headphone outputs, the analog noise generated after these output DACs is constant in amplitude, independent of the Monitor level setting. This contributes to the Headphone noise floor discussed above. The opamp driving the Monitor output has a gain of ~ 3.5 (11dB). This is in series with an additional opamp with gain ~ 4.9 (14dB) driving the headphone output jack and the analog noise is cumulative. Since there is abundant headphone output analog drive cabability, a very simple method for lowering the noise floor at the headphone output makes use of a simple output external resistive voltage divider network. Examination of the DP-24SD schematic reveals that the output NJM3414AL opamp has a "source" output impedance of 54 ohm (two 27 ohm series resistors R1 and R2 at opamp output). This is easy to verify by simply playing back a simple tone track without headphones connected, measuring the output voltage at the headphone jack using any DVM.. Then insert resistive loads of say 54 ohm at each stereo output pin between the pins and ground and measure the output voltage again. The measured output voltage will drop by half. Of course other resistor values will change the output voltage as in any resistive voltage divider network, which is the key to the noise-floor reduction. To reduce the output noise-floor level, simply put 2 identical resistors in an external circuit between the each channel output and ground. A very compact box with input and output jacks (either 1/4" or 1/8" stereo). These extra shunt resistors (Rs in diagram below), one for each channel, will be in parallel with your headphone impedance and this parallel combination with form a simple voltage divider network with the 54 ohm internal circuit resistance of the DP-24. This will lower the constant output-floor voltage as well as the playback signal level obviously (note that this voltage divider does NOT change the Monitor output level). The Monitor level can then be simply increased to compensate for the level drop at the headphones without raising the noise (this will also increase the Monitor output level). The diagram below shows a very simplified view of the headphone output circuit for the purposes of this discussion. The value of the shunt resistances Rs depends obviously on the headphone impedance but it is fairly easy to determine suitable values depending on how much noise-floor reduction is desired. The equation below is just the voltage divider network between the opamp output voltage Vo and the headphone voltage V_HP:

It is easy to breadboard the layout for testing purposes. The headphone output levels are high and good shielding isn't really required. My test setup is shown below:

As an example, with my Sennheiser HD 380 Pro phones (with R_HP=54ohm), two examples are:

- Rs 33ohm (Rs||R_HP) = 20.5ohm and so V_HP = 0.275Vo which is about 5dB lower than without Rs with these phones
- Rs 15ohm (Rs||R_HP) = 11.7ohm and so V_HP = 0.18Vo which is about 9dB lower than without Rs

With these phones, the noise floor is reduced substantially with 33ohm resistors, and with 15ohm, the noise is completely inaudible to my ears and the Monitor level was easily raised to compensate for the gain loss.

Another simple method of lowering the headphone output noise floor by 6dB is shorting the Monitor out jacks. As shown in the very simplified partial schematic below, the BA4580RF Monitor output opamp has two 100 ohm output resistors in series before the Monitor out jack. These resistors are split in the middle at Va which drives the NJM3414AL Headphone opamp. Normally the Monitor jack is open or is loaded by a Monitor connection of 2kohm or higher. This means that Va is almost identical to Vs, the output of the Monitor opamp. However if the Monitor out jack is shorted, a 6dB voltage divider results lowering the signal and noise floor voltages at Va, the input of the Headphone opamp by 1/2. This also means that the load seen by the Monitor opamp is now only 200ohm. At a high 2V signal level, this means the Monitor opamp will deliver 10mA current which is well within the drive capability of the BA4580RF:

Accurate Noise Floor Measurements

The scope results above show the relative noise characteristics of the Stereo out, Monitor out and Headphone out. However the noise values are considerably higher than the true integrated noise values in an audio bandwidth because the scope has a much broader noise measurement bandwidth. (Introducing a band filter at the scope input is possible but this approach has other configuration issues that must be understood.) An accurate integrated audio-bandwidth noise measurement using an extremely low noise Creative X-Fi Elite Pro sound card provides a fairly easy way to confine the measurement BW to the audio region (up to potentially 96 kHz). These measurements were carried out to verify noise output values at each stage of the audio output section and perform a simple check on the S/N spec characteristics (not A weighted here). The maximum Stereo Out signal level Smax was verified to be 2Vrms (+6dbV) measured with a full digital amplitude stereo 16bit/44.1 kHz sine wave at 500Hz (Stereo level at 0dB; Track level at 0dB ... just at clipping). The various outputs of the DP-24SD were connected directly to the X-Fi Elite Pro sound card line-in (which has a noise floor of approximately 10 uVrms in a 20kHz noise bandwidth) and the WaveMon application was used to sample and compute the RMS noise voltage level.

en=8uVrms: Connected to DP-24SD (Turned Off). Testing Noise Floor

en=27uVrms: Stereo Out

en=63uVrms: Monitor Out

en=300uVrms: Headphone Out (no load)

It was observed that the noise levels were not affected by any digital controls (sliders, gain settings, Stereo, Monitor levels). No tracks were armed for recording (which would introduce input noise). No preamp was used in this measurement. The results show the total RMS noise voltage in a 20kHz noise bandwidth and are corrected for the small sound card noise-floor of 10uV:
which are consistent with the specs of "90dB or more". The noise on the Effect Sends was the same as the Line Out result, which is expected based on the almost identical circuit and amp in the Effect Sends section.

Block Diagram with Integrated Noise Measurements and OpAmp Gain Values

Looking at the Stereo Out section, with a mid-band gain of Av=1.67 (from detailed schematic), if we assume for simplicity that the noise output is uniformly gained, then the equivalent noise INPUT to the opamp there is ~ 25/1.67 = 15uV. This represents roughly the noise voltage after the DAC and it is reasonable to assume that since the DAC circuits look identical, then the DAC output noise of each audio output section should be similar. Therefore, since the DAC in the Monitor out section drives an opamp with gain of Av=3.5 (from schematic), we can expect a noise voltage at the Monitor out of about 15uVx3.5 = 52uV which is reasonably close to the measured value of 63uV, considering the assumptions. Furthermore, The MEASURED Monitor output voltage is further cascade amplified by the final NJM3414AL amp with an additional gain of 4.9 (from schematic) which predicts a headphone output voltage (with no headphone load) of 63uVx4.9 = 308uVrms, in agreement with the measured headphone output under no load. From these measurements and considering the amps gain design, modification of the output opamps will not reduce the ambient output noise levels as it is sourced at the DAC stage at the start of the analog output stages.