This morning we’re glad to resume production of C5D! UPS delivered parts at 10am and I’ve been listening to C5D for more than two hours with a smile on my face.
In case you missed the drama, we caught a tiny error in the first batch of C5D and temporarily halted production. Our terrific engineers, George Boudreau and Ken Mathews, collaboratively resolved the mistake in less than 48 hours. We owe them many thanks.
A few C5D’s shipped on Monday and Tuesday without the correction. We’re emailing affected customers today and will offer an easy exchange.
C5D shipments to distributors resume next week:
Headphone Bar / Canada
Munkong Gadget / Thailand
Headsound Audio / Germany
Kingsound Audio / Hong Kong
Noisy Motel / Australia
C5D samples also ship to reviewers by Monday.
If you’re curious, C5D is a complex device that makes use of 12 highly regulated, onboard power supplies.
Prior to the production fix, the first batch of C5D exhibited an extremely low level hum under USB data transfer. Hum vanished while music was not playing over USB, so it was easy to miss while listening. And by nature of its presence, it could never appear in Noise measurements, and was simply a tiny component of THD measurements (all excellent). This type of problem could only be found by an engineer actively looking for it. Conclusion: Subjective results matter as much as Objective measurements!
Working backwards, we disabled individual power supplies in C5D to isolate the issue. We found that when the USB controller placed its heavy load during data transfer (0.5W), it brought the primary +7V rail to 60% capacity. The +7V rail has been stable in C5, and the amplifier and DAC measure well independently. While powering both amp and DAC, the higher power draw pushed the rail slightly out of regulation. Line ripple increased by a factor of 10.
The solution was shockingly simple–we added a larger capacitor to the +7V rail (this capacitor will be hand soldered on the first batch of C5Ds). Ripple decreased more than 10x, to levels better than ever. Performance is restored. Low level hum is gone.
On a related note, we pushed a major site update to JDSLabs.com last night. Please call, email, tweet, or reply below if you encounter any browsing issues.
C5D Production Status
Design: 100% Complete
Benchmarks: 100% Complete
Production: Ships by November 22
C5D is entirely complete and ships immediately! We temporarily paused production to make a final tweak. Shipments resume no later than November 22.
C5D adds an outstanding PCM5102A DAC and extra bass boost level to our C5 headphone amplifier. Both C5 and C5D are built for portable users who demand exceptionally low noise, sufficient output power, and a super fine volume control to handle sensitive headphones and IEM’s.
Our goal for C5D is simple: Merge a reference grade DAC with C5, valuing performance and compatibility over superfluous features. C5D works natively with iPhone, iPad, and all devices and operating systems which support UAC1.
USB Audio Class 1 (UAC1) is the widely compatible standard for transmitting digital audio over USB. UAC2 is required to go beyond 24/96 audio, but UAC2 support remains limited and requires special drivers for Windows XP/7/8, etc. In other words, connecting a UAC2 DAC becomes more involved and potentially buggy.
It’s easy to understand why audiophiles develop specification wish lists such as 32/384kHz PCM via UAC2, or DSD, or asynchronous operation. The numbers and algorithms look really impressive. But ultimately, you can’t utilize 32/384 audio when your music collection is the bottleneck. It makes perfect sense from a marketing standpoint to enable the latest features on a new device. Fortunately, we’re engineers and not marketers.
C5D’s hardware actually supports DSD and 32-bit audio. We disabled both. UAC2 breaks compatibility with many portable devices, and C5D needs to work out of the box with phones and tablets. Plus, transparency is achievable through UAC1 and fully utilizing 24-bit depth can be unrealistic.
So instead of giving C5D compatibility limiting UAC2 features, it’s configured for awesome performance under UAC1. And we still managed some interesting characteristics–galvanic isolation, asynchronous operation, and a low latency jitter filter.
Reference D/A Conversion
C5D’s new DAC circuitry fits in previously unused space beneath the battery, so size remains identical to C5. The PCM5102A DAC and SA9027 controller pack incredible performance in this small space.
The large chip next to the USB jack is an Analog Devices ADuM3160. This IC serves two functions:
Enhanced ESD protection at the USB jack
Also next to the USB jack is a new toggle switch. Flipping this switch right allows C5D to charge. Flipping the switch left makes use of the ADuM3160’s air core transformer technology to operate C5D in self-power mode*. That is, the DAC runs from its own battery when connected to a portable USB audio player. This is known as galvanic isolation, and it cleverly prevents the DAC from being subjected to noise on the USB +5V rail.
Self-power mode also gives C5D maximum flexibility with portable devices since most smartphones and tablets disable USB devices that consume too much power.
* Full isolation is utilized with low power USB devices. C5D enters a hybrid self-powered mode when connected to strong USB devices, only consuming extra power. DAC performance is identical in all power modes.
Just weeks before we approved C5D for production, I received word that a code update could convert C5D from adaptive to asynchronous operation. Features are always second to performance at JDS Labs, so we repeated all benchmarks.
C5D jitter already measured extremely well in adaptive mode. We want to see a sharp signal in this test, with minimal sidebands (especially near the signal). Keeping the sum of matched components below -100dBFS prevents an audible impact. C5D in adaptive mode far surpasses this reference goal at -111dBFS:
Running asynchronously, jitter improves little more than the measurement’s margin of error:
Jitter measures slightly better at -112dB in async mode, and is the only standard test that revealed any difference from adaptive mode on C5D. All other benchmarks returned identical results. Thus, C5D ships in asynchronous mode.
Asynchronous mode and galvanic isolation together make C5D a rare UAC1 DAC. These features make it highly self-reliant, generating its own clock and power.
+/- 0.14 dB
IMD 19/20kHz, -7dBFS
> 109 dB A-Weighted
Linearity Error, -90dBFS 24/96
USB Jitter, Marked Sum
DAC measurements were obtained by hardwiring a line-output jack to C5D’s PCM5102A output for connection to our dScope Series III audio analyzer.
Frequency response is excellent, with negligible rolloff of 0.1dB in the final octave of human hearing.
THD+N measures well below our reference goal of 0.005% at all frequencies:
A-Weighted noise exceeds expectations with components at -110dBu, and overall noise better than -100dBu:
Dynamic range measures quite well at 109dB:
The DAC’s line output crosstalk measures at -86.1dB, outperforming our reference requirement of -80dBFS. Note that crosstalk is limited by 3.5mm cables in actual use (still excellent).
The IMD CCIF test checks DAC performance during simultaneously playback of 19kHz and 20kHz tones. C5D returns excellent results here, with minimal blurring between the high frequency signals (noise below -120dB). Total distortion measures well at a very low 0.0013%:
Low Latency Filter
The PCM5102A DAC used in C5D features a configurable low latency filter. In testing, we’ve observed no significantly audible difference. C5D ships with its Low Latency Filter set High.
C5D’s firmware is released freely under the Creative Commons BY-SA 3.0 license. Refer to line 55 if you wish to experiment with the PCM5102A’s LLF feature. Note that a programmer and pogo pins are required for DIY tinkering.
C5D’s amplifier and supporting power circuitry is identical to that of C5, with the exception of an additional bass boost level and smaller output resistors. Output impedance of C5D has improved to 0.62Ω. This specification change minimally impacts overall performance, and ensures neutral operation with low-impedance balanced armature IEMs.
+/- 0.1 dB
THD+N (20-20kHz, 150 Ω)
THD+N (20-20kHz, 32 Ω)
Crosstalk @ 150 Ω
Inter-channel Phase @ 1kHz
+/- 0.01 degrees
+/- 0.55 dB
Max Output @ 600Ω
Max Output @ 150Ω
Max Output @ 32 Ω
Battery Run Time
0°C to 60°C
0 to 85% Relative
99.5 x 61.5 x 14.0 mm
Triple Bass Boost
C5D’s bass boost has three positions: Off / Medium / High. The High position is identical to C5’s bass boost, with the Medium level residing audibly halfway between off and high. Below are C5D’s bass boost curves in low gain:
These curves relax at high gain, in effect producing four unique bass boost curves:
Camera Connection Kit and iOS7
Camera Connection Kit
ROM and OS must support UAC1
Mac OS X
We considered developing C5D for fully native operation with Android, and discovered the goal is presently futile. Even a DAC designed for native functionality via Android’s Open Accessory Protocol remains limited to 16-bit, 44.1kHz operation. And even then, support is not 100% guaranteed across all Android devices! Only an app like USB Audio Recorder Pro unlocks full 24-bit digital audio, by utilizing alternative drivers.
C5D works with every Android device we’ve tested under USB Audio Recorder Pro. We met a few Android users at the 2013 CanJam who successfully used C5D natively (i.e., with any app). Some Android phones and tablets output UAC1 natively. Others require special ROMs or apps.
Guaranteeing DAC operation with all Android devices is currently not realistic. Since Android is opensource, it’s definitely possible to enable 24-bit digital audio output on any Android device. Hopefully Google will make UAC1 output standard in future Android updates to simplify the user experience.
The good news: C5D is self-powered, so its power consumption is not a limitation. You’ll only need to enable digital audio output on your device if it’s not already available.
iPad and iPhone
C5D works out of the box with iPad and iPhone! Apple has finally enabled UAC1 output as of iOS7. You simply need a Camera Connection Kit cable. C5D is self-powered, so power consumption is of no concern.