Important Update (8/16/25)
WalkPlay/TTGK is another company reacting to the issue of DRE-induced distortion associated with the CS431xx chips! A new firmware release for the TRN Black Pearl changed a key parameter related to DRE operation. With this firmware installed, the Black Pearl does not exhibit audible distortion behavior caused by DRE while retaining reconstruction filter options. An analysis based on measurements was added to the bottom of this review.
There are many portable DACs adopting the Cirrus Logic CS43131 chip, some of which have been reviewed here at ASR. Why do we want to see yet another review of the same kind? Here are some reasons:
Note that all these devices do not include an additional headphone drive stage in their design. So, they all rely on CS43131 for both D/A conversion and headphone amplification. They all adopt dual CS43131 to support balanced outputs.
The TRN Black Pearl is in fact a product designed by TTGK, an ODM (original device manufacturer) of portable DACs and dongles. It is actually the TTGK TT39518F01-Pro module housed in a black case and with CS43131 substituted for CS43198:
One thing I could not quickly figure out about this device is that it is turned on only if one of its output jacks, either 3.5mm or 4.4mm, is occupied by a headphone plug. Otherwise, it is not even recognized by a host device.
The technique of DRE increases a DAC's low-level performance and achieves a wider measured dynamic range (DR) by adjusting the chip's gain structure, depending on the signal strength, to find a good trade-off between digital and analog gain settings. DRE is not a new technology but has also been employed by other DAC chip manufacturers such as ESS and AKM. Setting aside the debatable point on whether the DR of a DRE-enabled device, measured by the AES17 standard procedure, is faithful to a theoretical definition of a dynamic range, criticism against these Cirrus Logic chips has been on the fact that the use of DRE is not officially documented by Cirrus Logic anywhere.
More importantly, DRE has been suspected to cause the anomalous distortion behavior observed from the CS431xx-based DACs. How would DRE cause this distortion? The idea of DRE is simple: whenever there is headroom in the digital input signal---say the signal's peak in a certain time window is 20 dB below full scale (dBFS)---, fill it with digital gain and decrease analog gain to achieve lower noise and hence increase the DR. However, unlike a similar technique, such as HDCD (High Definition Compatible Digital) that increases DR by applying DR compression as an offline preparation process (i.e., encoding) and reversing the process during playback (i.e., decoding), DRE operates entirely in real time. The problem is that audio signals are unpredictable when processed online. What if the headroom is abruptly decreased right after such an adjustment is made? Or what if the headroom is overestimated when the adjustment is made? For various reasons, distortion due to inadequate headroom can occur. Right, this distortion is nothing but a digital artifact.
So, DRE, by design, cannot work 100% of the time---i.e., it is destined to cause distortion even if a small % of the time---unless some pre-processing is done before input is converted. But such processing in a tiny DAC chip is unrealistic. For this fundamental reason ESS must have chosen to disable DRE by default in some of their DAC chips supporting DRE, like the ES9219 and ES9080. But DRE may still be used in such a way that it would incur distortion, say, only 0.000001% of the time. This would be accomplished by carefully choosing associated parameters. I believe Cirrus Logic engineers must have been too greedy in setting the default parameter set. For example, the maximum DRE gain in CS431xx is 24 dB, which is absolutely unnecessary. A max DRE gain of 12 dB should've been sufficient. And DRE does not need to start its action at -12 dBFS as in CS431xx. It is too high (again unnecessary). A -20 dB to -30 dB threshold should've been safe. This way there would be adequate headroom so distortion would have been much less frequent and less severe.
Thanks to an ASR member's report, it has been found that DRE is disabled when a DAC is set to the NOS (non-oversampling) mode, if a device supports such an option like the Black Pearl:
Although the absence of distortion was confirmed and reported in a post, I deemed that a full set of measurements and analyses would deserve a dedicated review.
Before presenting test results, there is one thing that needs to be made clear. Most modern DACs rely heavily on oversampling (same thing as 'upsampling' in this context) along with an interpolation algorithm to implement a reconstruction filter that removes quantization artifacts. That is, if digital audio content, created at a 44.1 kHz sample rate, is transmitted to a DAC, it is resampled within the DAC and interpolated by its built-in algorithm. When a DAC is set to 'NOS', the signal is supposed to be converted to analog audio without this reconstruction filtering. The consequence is broken soundwaves, shown in measurements as an early high-frequency roll-off and severe ultrasonic distortion. For this reason, when the NOS mode is used, the input signal should be oversampled by the host system. This is a price to pay for disabling DRE in the Black Pearl, or a similar CS431xx-based device supporting the NOS mode. Note that it should be possible to disable only DRE while retaining standard filters in a firmware design---I hear a rumor that a manufacturer is in the process of modifying firmware with this option. Fortunately, these days processing power is cheap. Computers and even mobile devices can handle this load without using too much CPU power. In measurements shown below, unless noted otherwise, the Black Pearl was tested in its NOS mode and with input signals resampled at a 192 kHz or 384 kHz sample rate, so not relying on the DAC's oversampling.
We first look at a set of measurements that do not depend on DRE. All results shown in this section were obtained in the NOS mode in which DRE was disabled.
The full-scale SINAD of the DAC's balanced output under no load (20 kOhm) is excellent:
Note that this is essentially the same performance as other comparable devices based on the same chip, measured by others: Tanchjim Space, Fosi Audio DS2, and Moondrop Dawn Pro. This result would be placed at the top of Amir's portable DAC SINAD chart:
Its unbalanced outputs' SINAD results are just as expected (i.e., about 3 dB lower than balanced outs):
Not surprisingly, this DAC is fine with the usual package of measurements using steady-state signals (i.e., full-scale multitone, jitter, SMPTE IMD vs. output, and THD+N vs frequency tests). Since these are not the focus of this review, these measurements are shown in thumbnails below:
The full-scale 32-tone test shows slight high-frequency distortions. In the jitter test, some periodic jitter (shown in spikes) and broadband random jitter (indicated by the slightly elevated noise floor) are observed. But none of these are of audibly concerning magnitude. The stepped SMPTE IMD versus output and THD+N versus frequency results are great.
An important aspect of any portable combo device is output power. The Black Pearl's headphone drive capability is decent:
Output impedance is very low:
One way to clearly see the underlying effect of DRE is to look at the DAC's noise shaping because the effect can be seen without being masked by the device's thermal noise limit. When a test tone is between 0 and -12 dBFS, there is no difference in noise performance whether DRE is enabled or disabled. Below is a comparison of the two cases when a -12 dB sine tone is fed:
Below -12 dBFS DRE starts its action, and when the signal reaches -36 dBFS, noise reduction by DRE is in its maximal state:
Stepped sine tone tests using BW of 20 Hz - 90 kHz show this transition clearly:
The full DRE gain implemented in CS431xx is 24 dB. However, the practical amount of noise reduction in human hearing range (20 Hz - 20 kHz) is substantially less because the noise level is masked by the device's analog, thermal limit (or the measurable limit).
See the same stepped sine tone tests with BW limited to 20 Hz - 20 kHz:
The measurable noise reduction of DRE in the frequency range of 20 Hz to 20 kHz is about 12 dB. This is still a considerably large difference. To get a better idea of what kind of noise performance these results indicate, we may want to look at the performance of other devices, including some well-regarded ones:
SNR at the output of 50 mV:
Though not as important, each device's dynamic range in its maximum output setting was also measured:
Some notable points of these results:
Definitely true. But ESS, despite having their own DRE technology, chose not to include it in their top DAC chips, not even as an option. It must not be simply because the DAC chips' noise performance is good enough, since ESS would still be interested in improving DAC technology whether the result is beyond the audibility threshold or not. It is because DRE has its inherent limitation as described at the beginning of this review. The Cirrus Logic CS431xx chips showcase an example.
The Cirrus hump of a CS431xx-based DAC, a distortion phenomenon shown in multitone tests, was described comprehensively in my review. These humps are gone when the Black Pearl is set to NOS and its DRE is disabled:
Note that multitone tests just happen to show part of the DRE artifacts. In fact, DRE's fundamental issue is not with its handling of steady-state test signals---all standard tests we are accustomed to seeing at ASR are based on steady-state signals. By design its problem is with dynamically changing audio content. DRE-induced distortion occurs very often during playback of real audio material, which can be shown by recording it. Actually, very frequent distortions were observed in spectrograms of recordings of my music collection. This does not mean the distortion is clearly audible in all such occurrences. More often than not, it is masked whenever the content's frequency spectrum is wide. I found that the distortion is most noticeable when audio consists of somewhat complex low-frequency signals with not much high-frequency content. The Dune soundtrack clip used in the RAA article is such an example. And an ideal example is the C Major test signal used in Part II of my review. Sill another example is the so-called 'rumble' test clip at the audiocheck website.
Below are the spectrograms of recordings of the Dune soundtrack clip played by the Black Pearl (played at -15 dBFS and normalized to -1 dBFS).
DRE enabled:
This distortion, indicated by vertical lines, is clearly audible as 'crackling' sound.
DRE disabled (in NOS mode):
Distortion is completely gone.
Next are the results of recording the C Major test clip (played at -15 dBFS and normalized to -1 dBFS).
DRE enabled:
The periodic 'clicking' sound from this test clip is very clear, making it an ideal test signal.
DRE disabled:
No hint of distortion.
Lastly, recordings of the rumble test signal (played at -17 dBFS and normalized to -6 dBFS).
DRE enabled:
Crackling is audible.
DRE disabled:
Free of distortion.
A test of the Black Pearl's PEQ function was performed with the following 5-band filter setting:
To check if it works correctly, its measured response was compared to that of Equalizer APO on Windows 11. To mimic an actual playback situation, periodic white noise was played from foobar2000 and its FFT was captured:
The measured transfer function of the TRN's onboard PEQ (green curve) closely matches that of EQ APO (red curve). There is not even a slight difference. The amplitude difference between the two measurements is not due to a difference in the PEQ setting's preamp gain---they were just separated for comparison. In fact, I found that the Black Pearl's hardware (digital) volume control takes the transfer function's peak into account and does not increase the volume beyond a clipping point.
To see if the onboard PEQ adds any distortion, a 32-tone test was performed:
The quality of the signal through the onboard PEQ is exactly the same as it is through the EQ APO.
Note that the tested PEQ setting consists of only peak filters. Although the app supports shelving filters, they currently do not work. A low-shelf or high-shelf filter is simply interpreted as a peak filter and applied as such. Shelving filters can be convenient, but PEQ with peak filters alone is fine. I personally never used shelving filters. In fact, a PEQ profile used in the second test, shown below, demonstrates that shelving filters' effects can be mimicked by peak filters of negative gain.
Another test was performed with a PEQ setting that I actually use for one of my headphones (Philips SHP9500 modified with leather pads):
All 10 bands were utilized to create this frequency response correction, which is shockingly similar to a headphone target curve of some kind.
Measurements show the TRN's onboard PEQ works correctly:
Again, 32-tone tests indicate its quality is equivalent to EQ APO's:
Conclusion
This review shows that the TRN Black Pearl in its NOS mode does not exhibit the peculiar distortion behavior associated with CS431xx. A price to pay for having this benefit is to lose some noise performance as DRE is turned off. The 85 dB SNR at 50 mV of the Black Pearl in NOS mode (i.e., with DRE disabled) still represents decent noise performance. In my experience, a signal-to-noise ratio of about 80 dB at 50 mV output is adequate even for sensitive IEMs. At that level, perhaps extremely sensitive IEMs may still produce borderline audible noise. I listened to various tracks from my collection using the Black Pearl connected to the 7Hz x Crinacle Zero:2. I noticed no increased hissing even when cranked up for softly recorded material with DRE off versus on.
Then, why did Cirrus Logic choose to enable DRE by default? Most likely to win the competition. How sweet would it be to proudly tell this tiny low-power chip achieves dynamic range even wider than desktop-grade DACs from ESS and AKM? And such performance was there even with a headphone driver included. Because of their decision, here we're dealing with this Cirrus hump distortion. Good news is that one can disable DRE and eliminate the distortion at least for some devices supporting the NOS mode (tested other devices like the FiiO KA15 and Tanchjim Stargate II). Unfortunately, this hack does not work for other devices with no such option.
The TRN Black Pearl is a budget portable DAC/headphone amp providing excellent output power, low distortion and low noise (even with DRE disabled). It even supports onboard PEQ that works correctly. What more could we ask for at this price?
Sure, there are things to be desired. Audio needs to be resampled on the host side for reconstruction filtering---Windows and UAPP (USB Audio Player Pro) on Android were tested doing this job nicely. I hope someone can contact and convince TRN Audio or TTGK to add a 'DRE off' option that retains standard reconstruction filters, which should be a straightforward firmware revision. Correctly working shelving filters would be nice, too.
Important Update (8/16/25)
WalkPlay/TTGK is another company reacting to the issue of DRE-induced distortion associated with the CS431xx chips! A new firmware release for the TRN Black Pearl, available through its Android app, lowered the DRE activation threshold, a key parameter of DRE operation. With this firmware installed, the Black Pearl does not produce audible distortion caused by DRE while retaining reconstruction filter options. Essentially, this update is identical to what FiiO did for the KA15's new 'DRE Enable' mode: it minimizes DRE artifacts by lowering the threshold for DRE engagement.
Below are the measured noise levels of different firmware options / settings:
In case you wonder about SNR at the output of 50 mV:
Dynamic range measurements:
Let's see the lowered DRE threshold's effect on distortion. Below are the measurements of total distortion plus noise (TD+N) from 32-tone tests:
With the new firmware, there is still a slight Cirrus hump (green solid line) as the DRE activation threshold is lowered to -44 dBFS. But the signal levels in that hump range are very low, resulting in the distortion much closer to the measurable noise floor. Put another way, this distortion should be inaudible, or nearly indistinguishable from noise (as the noise is unnoticeably low as well). With DRE disabled in NOS mode (blue solid line), even this hump is gone, but as shown in the preceding chart, the noise reduction for lower-level signals will be gone, too.
In the same recording condition as used in my TRN Black Pearl review, the spectrogram of recording of the Dune soundtrack clip (w/ Fast-PC filter) shows no clear distortion:
If you zoom in, you may see very light vertical lines when the signal is decreasing in the later part of the clip. But I could not hear any effect of them in my listening tests. Interestingly, I found that FiiO KA15's new "DRE Enable" mode also shows a tiny hint of DRE artifacts in the same recording when the spectrogram gain is increased, but it is not visible in the same plotting condition. I have no idea why the distortion is slightly higher for the Black Pearl (also shown in the 32-tone test above). But in any case, the level of distortion is too low to be audible.
Recording of the C Major test clip with the new firmware and Fast-PC filter:
Again, if you zoom in, you can see very light vertical lines during transition. But I still couldn't hear these effects.
Recording of Audiocheck's 'rumble' test clip:
Clean.
This new firmware essentially makes the TRN Black Pearl free of audible DRE artifacts while retaining the option of selecting a standard reconstruction filter. It also keeps DRE's low-level noise performance when the signal level is very low (e.g., < -50 dBFS), although there shouldn't be noticeable benefit of this in practice.
We have another CS431xx-based device with no audible DRE artifacts. The TRN Black Pearl, as shown in the review, provides excellent performance as well as 10-band onboard PEQ in a portable form factor. The WalkPlay/TTGK team has been very responsive to my request and feedback in the process of making this update available.
WalkPlay/TTGK is another company reacting to the issue of DRE-induced distortion associated with the CS431xx chips! A new firmware release for the TRN Black Pearl changed a key parameter related to DRE operation. With this firmware installed, the Black Pearl does not exhibit audible distortion behavior caused by DRE while retaining reconstruction filter options. An analysis based on measurements was added to the bottom of this review.
Introduction
(Picture from TRN-Audio website)
(Picture from TRN-Audio website)
There are many portable DACs adopting the Cirrus Logic CS43131 chip, some of which have been reviewed here at ASR. Why do we want to see yet another review of the same kind? Here are some reasons:
- This review will demonstrate, for the first time, that a CS431xx-based DAC can be completely free of the peculiar distortion associated with the chip, reported at Reference Audio Analyzer , by some audio enthusiasts, and in my previous review. Yes, completely free of the distortion, not even the problem shown in Part II of the review.
- This review will assess, for the first time again, the noise performance of a CS431xx-based DAC with its DRE (dynamic range enhancement) function disabled.
- The TRN Black Pearl is relatively inexpensive (< $50 at AliExpress) and supports onboard 10-band parametric equalization (PEQ).
| Tanchjim Space | Fosi Audio DS2 | Moondrop Dawn Pro | TRN Black Pearl |
|---|---|---|---|---|
| USB Bridge | Comtrue CT7601 | Savitech SA9312L | Comtrue CT7601 | TTGK CB5100 |
| NOS mode | Supported | Not supported | Supported via button? | Supported |
| Onboard PEQ | Not supported | Not supported | Not supported | Supported (10 bands) |
| Connection | 3.5mm, 4.4mm | 3.5mm, 4.4mm | 3.5mm, 4.4mm | 3.5mm, 4.4mm |
| Measured by | ASR, L7Audio | ASR | ASR (Unofficial) | This review |
Note that all these devices do not include an additional headphone drive stage in their design. So, they all rely on CS43131 for both D/A conversion and headphone amplification. They all adopt dual CS43131 to support balanced outputs.
The TRN Black Pearl is in fact a product designed by TTGK, an ODM (original device manufacturer) of portable DACs and dongles. It is actually the TTGK TT39518F01-Pro module housed in a black case and with CS43131 substituted for CS43198:
One thing I could not quickly figure out about this device is that it is turned on only if one of its output jacks, either 3.5mm or 4.4mm, is occupied by a headphone plug. Otherwise, it is not even recognized by a host device.
Why Disabling DRE Matters
The technique of DRE increases a DAC's low-level performance and achieves a wider measured dynamic range (DR) by adjusting the chip's gain structure, depending on the signal strength, to find a good trade-off between digital and analog gain settings. DRE is not a new technology but has also been employed by other DAC chip manufacturers such as ESS and AKM. Setting aside the debatable point on whether the DR of a DRE-enabled device, measured by the AES17 standard procedure, is faithful to a theoretical definition of a dynamic range, criticism against these Cirrus Logic chips has been on the fact that the use of DRE is not officially documented by Cirrus Logic anywhere.
More importantly, DRE has been suspected to cause the anomalous distortion behavior observed from the CS431xx-based DACs. How would DRE cause this distortion? The idea of DRE is simple: whenever there is headroom in the digital input signal---say the signal's peak in a certain time window is 20 dB below full scale (dBFS)---, fill it with digital gain and decrease analog gain to achieve lower noise and hence increase the DR. However, unlike a similar technique, such as HDCD (High Definition Compatible Digital) that increases DR by applying DR compression as an offline preparation process (i.e., encoding) and reversing the process during playback (i.e., decoding), DRE operates entirely in real time. The problem is that audio signals are unpredictable when processed online. What if the headroom is abruptly decreased right after such an adjustment is made? Or what if the headroom is overestimated when the adjustment is made? For various reasons, distortion due to inadequate headroom can occur. Right, this distortion is nothing but a digital artifact.
So, DRE, by design, cannot work 100% of the time---i.e., it is destined to cause distortion even if a small % of the time---unless some pre-processing is done before input is converted. But such processing in a tiny DAC chip is unrealistic. For this fundamental reason ESS must have chosen to disable DRE by default in some of their DAC chips supporting DRE, like the ES9219 and ES9080. But DRE may still be used in such a way that it would incur distortion, say, only 0.000001% of the time. This would be accomplished by carefully choosing associated parameters. I believe Cirrus Logic engineers must have been too greedy in setting the default parameter set. For example, the maximum DRE gain in CS431xx is 24 dB, which is absolutely unnecessary. A max DRE gain of 12 dB should've been sufficient. And DRE does not need to start its action at -12 dBFS as in CS431xx. It is too high (again unnecessary). A -20 dB to -30 dB threshold should've been safe. This way there would be adequate headroom so distortion would have been much less frequent and less severe.
Thanks to an ASR member's report, it has been found that DRE is disabled when a DAC is set to the NOS (non-oversampling) mode, if a device supports such an option like the Black Pearl:
Although the absence of distortion was confirmed and reported in a post, I deemed that a full set of measurements and analyses would deserve a dedicated review.
Implication of Using NOS (non-oversampling) Mode
Before presenting test results, there is one thing that needs to be made clear. Most modern DACs rely heavily on oversampling (same thing as 'upsampling' in this context) along with an interpolation algorithm to implement a reconstruction filter that removes quantization artifacts. That is, if digital audio content, created at a 44.1 kHz sample rate, is transmitted to a DAC, it is resampled within the DAC and interpolated by its built-in algorithm. When a DAC is set to 'NOS', the signal is supposed to be converted to analog audio without this reconstruction filtering. The consequence is broken soundwaves, shown in measurements as an early high-frequency roll-off and severe ultrasonic distortion. For this reason, when the NOS mode is used, the input signal should be oversampled by the host system. This is a price to pay for disabling DRE in the Black Pearl, or a similar CS431xx-based device supporting the NOS mode. Note that it should be possible to disable only DRE while retaining standard filters in a firmware design---I hear a rumor that a manufacturer is in the process of modifying firmware with this option. Fortunately, these days processing power is cheap. Computers and even mobile devices can handle this load without using too much CPU power. In measurements shown below, unless noted otherwise, the Black Pearl was tested in its NOS mode and with input signals resampled at a 192 kHz or 384 kHz sample rate, so not relying on the DAC's oversampling.
Measurement Setup
- AD converter: E1DA Cosmos ADCiso Grade A in Mono Mode, and Cosmos Scaler as a buffer.
- Software: Room EQ Wizard (v5.40 beta 85 and 95).
Performance Not Affected by DRE
We first look at a set of measurements that do not depend on DRE. All results shown in this section were obtained in the NOS mode in which DRE was disabled.
The full-scale SINAD of the DAC's balanced output under no load (20 kOhm) is excellent:
Note that this is essentially the same performance as other comparable devices based on the same chip, measured by others: Tanchjim Space, Fosi Audio DS2, and Moondrop Dawn Pro. This result would be placed at the top of Amir's portable DAC SINAD chart:
Its unbalanced outputs' SINAD results are just as expected (i.e., about 3 dB lower than balanced outs):
Not surprisingly, this DAC is fine with the usual package of measurements using steady-state signals (i.e., full-scale multitone, jitter, SMPTE IMD vs. output, and THD+N vs frequency tests). Since these are not the focus of this review, these measurements are shown in thumbnails below:
The full-scale 32-tone test shows slight high-frequency distortions. In the jitter test, some periodic jitter (shown in spikes) and broadband random jitter (indicated by the slightly elevated noise floor) are observed. But none of these are of audibly concerning magnitude. The stepped SMPTE IMD versus output and THD+N versus frequency results are great.
An important aspect of any portable combo device is output power. The Black Pearl's headphone drive capability is decent:
This performance is as good as it gets for a CS431xx-based device with no additional op-amps, equaling the Fosi Audio DS2 and slightly more powerful than the Tanchjim Space.
Output impedance is very low:
Output | Left Ch. | Right Ch. |
|---|---|---|
3.5 mm | 0.23 Ohm | 0.22 Ohm |
4.4 mm | 0.12 Ohm | 0.12 Ohm |
Noise Performance: DRE On vs. DRE Off
One way to clearly see the underlying effect of DRE is to look at the DAC's noise shaping because the effect can be seen without being masked by the device's thermal noise limit. When a test tone is between 0 and -12 dBFS, there is no difference in noise performance whether DRE is enabled or disabled. Below is a comparison of the two cases when a -12 dB sine tone is fed:
Below -12 dBFS DRE starts its action, and when the signal reaches -36 dBFS, noise reduction by DRE is in its maximal state:
Stepped sine tone tests using BW of 20 Hz - 90 kHz show this transition clearly:
The full DRE gain implemented in CS431xx is 24 dB. However, the practical amount of noise reduction in human hearing range (20 Hz - 20 kHz) is substantially less because the noise level is masked by the device's analog, thermal limit (or the measurable limit).
See the same stepped sine tone tests with BW limited to 20 Hz - 20 kHz:
The measurable noise reduction of DRE in the frequency range of 20 Hz to 20 kHz is about 12 dB. This is still a considerably large difference. To get a better idea of what kind of noise performance these results indicate, we may want to look at the performance of other devices, including some well-regarded ones:
SNR at the output of 50 mV:
Device | SNR @ 50 mVRMS |
|---|---|
TRN Black Pearl w/ DRE on (3.5mm out) | 96.5 dB |
RME ADI-2 DAC FS (IEM out) | 95.9 dB |
E1DA 9039S | 89.9 dB |
Qudelix 5K (3.5mm out) | 87.6 dB |
RME ADI-2 DAC FS (Low Power) | 86.0 dB |
TRN Black Pearl w/ DRE off (3.5mm out) | 85.2 dB |
Hi-Max CB1200AU | 76.8 dB |
Though not as important, each device's dynamic range in its maximum output setting was also measured:
Device | Dynamic Range | Maximum Output |
|---|---|---|
TRN Black Pearl w/ DRE on (4.4mm out) | 132.7 dB | 4.06 VRMS |
E1DA 9039S | 127.8 dB | 3.49 VRMS |
Qudelix 5K (2.5mm out) | 123.3 dB | 4.10 VRMS |
TRN Black Pearl w/ DRE off (4.4mm out) | 120.5 dB | 4.06 VRMS |
Hi-Max CB1200AU | 103.3 dB | 1.01 VRMS |
Some notable points of these results:
- Noise performance measured by this setup (E1DA Cosmos stack) is usually a few dBs better than measured by the APx555B.
- Except for the Black Pearl, none of the devices under comparison employ DRE.
- The two top rankers have obvious reasons for such outstanding performance, which makes comparing the other devices with them unfair:
- TRN Black Pearl relies on DRE. In fact, an SNR of 96.5 dB @ 50 mV is unrealistically good, which would place it at the very top of Amir's 50 mV SNR chart.
- RME's IEM output uses a buffer of negative gain (-10 dB) with its maximum output being only 0.55 Vrms.
- It is impressive that the E1DA 9039S achieves an SNR of 89.9 dB at 50 mV while having a balanced gain stage providing its maximum output of 3.5 Vrms.
- The Black Pearl's 85.2 dB SNR @ 50 mV with no DRE is still decent. We should not hear noticeable noise or hissing even from sensitive IEMs connected to the Black Pearl in NOS mode.
- One may think the Hi-Max's 50 mV noise performance (76.8 dB) is horrible. Not at all. I think some of us are spoiled by today's DAC technology. Although this dongle's max output is only 1 Vrms, its noise level in itself is by no means poor. For instance, this SNR @ 50 mV is on par, if not slightly better, with the performance of the ODAC + O2 amp combo available a little over a decade ago. With this dongle, noise only in softly recorded content is borderline audible. In fact, without careful attention, I do not hear much difference in noise b/w this dongle and better measured devices.
Definitely true. But ESS, despite having their own DRE technology, chose not to include it in their top DAC chips, not even as an option. It must not be simply because the DAC chips' noise performance is good enough, since ESS would still be interested in improving DAC technology whether the result is beyond the audibility threshold or not. It is because DRE has its inherent limitation as described at the beginning of this review. The Cirrus Logic CS431xx chips showcase an example.
Distortion Performance: DRE On vs. DRE Off
The Cirrus hump of a CS431xx-based DAC, a distortion phenomenon shown in multitone tests, was described comprehensively in my review. These humps are gone when the Black Pearl is set to NOS and its DRE is disabled:
Note that multitone tests just happen to show part of the DRE artifacts. In fact, DRE's fundamental issue is not with its handling of steady-state test signals---all standard tests we are accustomed to seeing at ASR are based on steady-state signals. By design its problem is with dynamically changing audio content. DRE-induced distortion occurs very often during playback of real audio material, which can be shown by recording it. Actually, very frequent distortions were observed in spectrograms of recordings of my music collection. This does not mean the distortion is clearly audible in all such occurrences. More often than not, it is masked whenever the content's frequency spectrum is wide. I found that the distortion is most noticeable when audio consists of somewhat complex low-frequency signals with not much high-frequency content. The Dune soundtrack clip used in the RAA article is such an example. And an ideal example is the C Major test signal used in Part II of my review. Sill another example is the so-called 'rumble' test clip at the audiocheck website.
Below are the spectrograms of recordings of the Dune soundtrack clip played by the Black Pearl (played at -15 dBFS and normalized to -1 dBFS).
DRE enabled:
This distortion, indicated by vertical lines, is clearly audible as 'crackling' sound.
DRE disabled (in NOS mode):
Distortion is completely gone.
Next are the results of recording the C Major test clip (played at -15 dBFS and normalized to -1 dBFS).
DRE enabled:
The periodic 'clicking' sound from this test clip is very clear, making it an ideal test signal.
DRE disabled:
No hint of distortion.
Lastly, recordings of the rumble test signal (played at -17 dBFS and normalized to -6 dBFS).
DRE enabled:
Crackling is audible.
DRE disabled:
Free of distortion.
Tests of PEQ Performance
A test of the Black Pearl's PEQ function was performed with the following 5-band filter setting:
To check if it works correctly, its measured response was compared to that of Equalizer APO on Windows 11. To mimic an actual playback situation, periodic white noise was played from foobar2000 and its FFT was captured:
The measured transfer function of the TRN's onboard PEQ (green curve) closely matches that of EQ APO (red curve). There is not even a slight difference. The amplitude difference between the two measurements is not due to a difference in the PEQ setting's preamp gain---they were just separated for comparison. In fact, I found that the Black Pearl's hardware (digital) volume control takes the transfer function's peak into account and does not increase the volume beyond a clipping point.
To see if the onboard PEQ adds any distortion, a 32-tone test was performed:
The quality of the signal through the onboard PEQ is exactly the same as it is through the EQ APO.
Note that the tested PEQ setting consists of only peak filters. Although the app supports shelving filters, they currently do not work. A low-shelf or high-shelf filter is simply interpreted as a peak filter and applied as such. Shelving filters can be convenient, but PEQ with peak filters alone is fine. I personally never used shelving filters. In fact, a PEQ profile used in the second test, shown below, demonstrates that shelving filters' effects can be mimicked by peak filters of negative gain.
Another test was performed with a PEQ setting that I actually use for one of my headphones (Philips SHP9500 modified with leather pads):
All 10 bands were utilized to create this frequency response correction, which is shockingly similar to a headphone target curve of some kind.
Measurements show the TRN's onboard PEQ works correctly:
Again, 32-tone tests indicate its quality is equivalent to EQ APO's:
Conclusion
This review shows that the TRN Black Pearl in its NOS mode does not exhibit the peculiar distortion behavior associated with CS431xx. A price to pay for having this benefit is to lose some noise performance as DRE is turned off. The 85 dB SNR at 50 mV of the Black Pearl in NOS mode (i.e., with DRE disabled) still represents decent noise performance. In my experience, a signal-to-noise ratio of about 80 dB at 50 mV output is adequate even for sensitive IEMs. At that level, perhaps extremely sensitive IEMs may still produce borderline audible noise. I listened to various tracks from my collection using the Black Pearl connected to the 7Hz x Crinacle Zero:2. I noticed no increased hissing even when cranked up for softly recorded material with DRE off versus on.
Then, why did Cirrus Logic choose to enable DRE by default? Most likely to win the competition. How sweet would it be to proudly tell this tiny low-power chip achieves dynamic range even wider than desktop-grade DACs from ESS and AKM? And such performance was there even with a headphone driver included. Because of their decision, here we're dealing with this Cirrus hump distortion. Good news is that one can disable DRE and eliminate the distortion at least for some devices supporting the NOS mode (tested other devices like the FiiO KA15 and Tanchjim Stargate II). Unfortunately, this hack does not work for other devices with no such option.
The TRN Black Pearl is a budget portable DAC/headphone amp providing excellent output power, low distortion and low noise (even with DRE disabled). It even supports onboard PEQ that works correctly. What more could we ask for at this price?
Sure, there are things to be desired. Audio needs to be resampled on the host side for reconstruction filtering---Windows and UAPP (USB Audio Player Pro) on Android were tested doing this job nicely. I hope someone can contact and convince TRN Audio or TTGK to add a 'DRE off' option that retains standard reconstruction filters, which should be a straightforward firmware revision. Correctly working shelving filters would be nice, too.
Important Update (8/16/25)
New Firmware (V0.3) with Lower DRE Activation Threshold
WalkPlay/TTGK is another company reacting to the issue of DRE-induced distortion associated with the CS431xx chips! A new firmware release for the TRN Black Pearl, available through its Android app, lowered the DRE activation threshold, a key parameter of DRE operation. With this firmware installed, the Black Pearl does not produce audible distortion caused by DRE while retaining reconstruction filter options. Essentially, this update is identical to what FiiO did for the KA15's new 'DRE Enable' mode: it minimizes DRE artifacts by lowering the threshold for DRE engagement.
Below are the measured noise levels of different firmware options / settings:
In the new firmware TTGK lowered DRE's activation threshold to -44 dBFS (green solid line) from the original firmware's -12 dBFS (blue solid line) when a reconstruction filter is engaged. As shown in the review, DRE is disabled in the NOS mode (red solid line). That is, DRE is turned on when the signal level is below -44 dBFS when a standard reconstruction filter is selected. This is a very conservative approach and minimizes DRE artifacts. This mode will still increase its dynamic range, by 10 dB, measured with a -60 dBFS tone. Also, it will lower the noise floor when low-level tones (e.g., < -50 dBFS) are played. This is an excellent trade-off.
In case you wonder about SNR at the output of 50 mV:
Condition | SNR @ 50 mVRMS |
|---|---|
3.5mm out w/ old firmware | 96.5 dB |
3.5mm out w/ new firmware or in NOS mode | 85.2 dB |
4.4mm out w/ old firmware | 93.5 dB |
4.4mm out w/ new firmware or in NOS mode | 82.2 dB |
Dynamic range measurements:
Condition | Dynamic Range | Maximum Output |
|---|---|---|
3.5mm out w/ old firmware | 129.3 dB | 2.03 VRMS |
3.5mm out w/ new firmware | 127.4 dB | 2.03 VRMS |
3.5mm out in NOS mode | 117.4 dB | 2.03 VRMS |
4.4mm out w/ old firmware | 132.7 dB | 4.06 VRMS |
4.4mm out w/ new firmware | 130.6 dB | 4.06 VRMS |
4.4mm out in NOS mode | 120.5 dB | 4.06 VRMS |
Let's see the lowered DRE threshold's effect on distortion. Below are the measurements of total distortion plus noise (TD+N) from 32-tone tests:
With the new firmware, there is still a slight Cirrus hump (green solid line) as the DRE activation threshold is lowered to -44 dBFS. But the signal levels in that hump range are very low, resulting in the distortion much closer to the measurable noise floor. Put another way, this distortion should be inaudible, or nearly indistinguishable from noise (as the noise is unnoticeably low as well). With DRE disabled in NOS mode (blue solid line), even this hump is gone, but as shown in the preceding chart, the noise reduction for lower-level signals will be gone, too.
In the same recording condition as used in my TRN Black Pearl review, the spectrogram of recording of the Dune soundtrack clip (w/ Fast-PC filter) shows no clear distortion:
If you zoom in, you may see very light vertical lines when the signal is decreasing in the later part of the clip. But I could not hear any effect of them in my listening tests. Interestingly, I found that FiiO KA15's new "DRE Enable" mode also shows a tiny hint of DRE artifacts in the same recording when the spectrogram gain is increased, but it is not visible in the same plotting condition. I have no idea why the distortion is slightly higher for the Black Pearl (also shown in the 32-tone test above). But in any case, the level of distortion is too low to be audible.
Recording of the C Major test clip with the new firmware and Fast-PC filter:
Again, if you zoom in, you can see very light vertical lines during transition. But I still couldn't hear these effects.
Recording of Audiocheck's 'rumble' test clip:
Clean.
This new firmware essentially makes the TRN Black Pearl free of audible DRE artifacts while retaining the option of selecting a standard reconstruction filter. It also keeps DRE's low-level noise performance when the signal level is very low (e.g., < -50 dBFS), although there shouldn't be noticeable benefit of this in practice.
We have another CS431xx-based device with no audible DRE artifacts. The TRN Black Pearl, as shown in the review, provides excellent performance as well as 10-band onboard PEQ in a portable form factor. The WalkPlay/TTGK team has been very responsive to my request and feedback in the process of making this update available.
Last edited:
)
