This is only a preview of the October 2021 issue of Silicon Chip.
You can view 47 of the 112 pages in the full issue, including the advertisments.
Items relevant to "Tele-com – an intercom using analog phones":
Items relevant to "Two- or Three-Way Stereo Active Crossover – Part 1":
Articles in this series:
Items relevant to "SMD Test Tweezers":
Items relevant to "Touchscreen Digital Preamp with Tone Control – Part 2":
Articles in this series:
Purchase a printed copy of this issue for $10.00.
And now . . . by Allan Linton-Smith THE UT-P 2016 MEMs WOOFER! Back in May 2020 we told you about the amazing, minuscule UT-P 2017 MEMS Tweeter from USound. We mentioned that it had a “big” brother (if big is the right word!) – the UT-P 2016 “woofer” or midrange driver. This tiny device can provide full range reproduction down to 20Hz and all our tests proved that it also has a great deal of potential. A ustrian developer USound launched the UT-P 2016 at the same time as the UT-P 2017. Both are MEMS or Micro Electrical-Mechanical Systems. Identical in size, the main difference between the two is that the UT-P 2016 is intended for wide-range speaker roles while the UT-P 2017 is designed as a tweeter. These devices have the potential to be very cheap because they can be manufactured using integrated circuit (IC) fabrication and device packaging processes. And from a manufacturing viewpoint, they are also easy to mount because they can be soldered in place by reflow soldering techniques, which is how most SMD components are incorporated into commercial applications. These MEMs speakers are in fact SMD speakers! 90 Silicon Chip These little speakers can be made far more easily than conventional moving-coil miniature speakers. It has been estimated that MEMS speakers will require around one thousand times less manufacturing time to produce! USound woofer performance Listening tests with the woofer were encouraging. A variety of music was auditioned including jazz, piano, classical and hard rock and the tiny MEMS speakers performed admirably with all genres. Our Test Bed: the MEMS speaker was mounted on a small PCB with the recommended 3mm gap. This feeds directly into our Bruel & Kjaer microphone but it is slightly different from the manufacturer’s setup, accounting for slight differences in the specifications. The back pressure from the small port in the rear of the speaker allows it to “breathe”, especially at low frequencies. Australia’s electronics magazine siliconchip.com.au Particularly impressive was the very lifelike reproductions of drums, possibly because of the excellent transient response. The big advantage of a tiny item like this is that it allows a frequency response to low frequencies for in-ear or near-earware which add to the realism of rumbles, quakes and explosions. Also because it is effectively a capacitor, its impedance has no significant peaks or troughs especially at lower frequencies. These often dog conventional dynamic drivers. It is easy to drive and does not require much current. Virtually any amplifier, even a preamplifier will be OK as long as it can deliver up to 5.3V RMS (15V peak-to-peak) although, as we said in the May 2020 issue (siliconchip.com.au/Article/14441), we would be reluctant to use a Class-D amplifier. Frequency response The USound MEMS UT-P 2016 woofer is quite smooth below 2kHz at its near maximum input of 15V pk-to-pk and this is close to the manufacturer’s test data. At low frequencies (below 1kHz) there is almost no variation in the measurements which were made in a closed test setup. A “normal” dynamic speaker would fall off dramatically below 100Hz and would also have significant peaks and troughs. Most headsets using dynamic drivers would also have these peaks and troughs, so this speaks highly of the excellent engineering of these MEMS speakers. We noted that in their “Danube” +2.5V to +5.5V C3 10uF 10V B2 GAIN C6 A3 AUDIO_IN 1uF 6.3V B3 C9 1uF 6.3V GND VDD SHDN_N GAIN SW VBST VAMP IN+ OUT+ IN- OUTSGND PGND D2 GND D1 C1 A1 C7 B1 C8 A2 D3 1uF 35V +15V R3 10K MEMs_BE MEMs_TE 1uF 35V R4 10K LM48580 GND GND Fig.1: the manufacturer’s suggested circuit for driving the MEMS UT-P 2016. It uses an LM48580 IC which comes in a tiny SMD DSBGA package measuring approx 2mm x 1.5mm (intended for hearing aids). Although small size is important, unfortunately this particular chip has 10% distortion at 10kHz and we feel that there are better alternatives. siliconchip.com.au plus for movies with a lot of dinosaurs or explosions etc. The graph was produced with the recommended maximum DC bias of 15.0V and a peak-to-peak input of 15V from an Audio Precision System Two generator. This generator has an output impedance of 30Ω and is not used to drive bigger speakers without using a power amplifier. What this means is that amplifiers for the MEMS speakers can be preamplifiers because the current demands are low. Distortion measurements C5 1uF 25V U2 C2 AMP_ENABLE D1 4.7uH C4 1uF 10V GND spectacle kits, USound have used a tweeter and woofer along with a DAC which doubles as an electronic crossover set at approximately 3.5kHz. We believe this is a reasonable setup for the MEMS speakers but because the woofer peaks at 3.3kHz with an SPL of 104dB we think that a crossover point of 1-2kHz might be better if the UT-P 2017 tweeter was incorporated. The speaker had no problem with an SPL of 83dB at 20Hz, which is very good indeed and would definitely be a L2 C3 GND The specifications for the UT-P 2016 show that the parameters are really tiny compared to bigger woofers . . . and any other speaker is bigger than this one! Remarkably the tiny size is really not a disadvantage because the membrane can easily respond below 20Hz for earware. Australia’s electronics magazine Although the distortion figures look high, it is not unusual to see THD+N figures of 20% or higher even in dedicated subwoofers. This is partly because the levels of sound at 20Hz, for example, are very low and the higher harmonics and noises which are generated at the higher frequencies (like “whooshing” or “huffing” noises) are reproduced more efficiently and increase the amount of THD+N at low frequencies. This little microspeaker having a very flat response means it is relatively low in distortion at 20Hz and this is a definite advantage. October 2021 91 FREQ RESPONSE UT-P-2016 MINI SPEAKER P-P INPUT FREQ RESPONSE UT-P 2016 MINI SPEAKER 15V P-P15V INPUT +50 THD+N VS FREQUENCY USOUND U-TP 2016 17V P-P INPUT 100 +40 50 +30 +20 d B r 20 + 10 0 % 10 A -10 5 -20 -30 2 -40 -50 20 30 40 50 70 100 200 300 400 500 600 Hz 800 1k 2k 3k 4k 5k 6k 8k 10k 20k Hz USound woofer practical applications All sorts of innovations come to mind when you can have a thin woofer and mount it on a flat surface. The obvious one is for earphones, earbuds and headphones but there are many other novel uses and this particular unit can be used virtually NΩ /2$' NΩ Ω +] 92 Silicon Chip )5(48(1&< 20 30 40 50 70 100 200 300 400 500 600 800 1k 2k 3k 4k 5k 6k 8k 10k 20k Hz Fig.2: frequency response of the USound MEMS woofer is quite smooth below 2kHz at its near maximum of 15V p-p and is close to the manufacturer’s test data. Zero dBr was set at 1 Pascal which represents a sound pressure level of 94dB so the peak is an SPL of 104dB. The speaker had no problem in reproducing 83dB at 20Hz! The graph was produced with the recommended maximum DC bias of 15.0V. Naturally, the distortion level is higher at low input levels because when the voltage drops below about 1V, the SPL is almost inaudible and the resultant signal-to-noise ratio is also low. As the input approaches the maximum of 15V peak-to-peak the distortion level drops to around 2-3% which is not bad for any speaker. There are no microphones which do not have their own distortion. Ours contributes about 0.3-0.4% so the best measurement using this system is about 1-1.5% because the microphone actually multiplies the distortion – it doesn’t simply add to it. 0.9 Fig.3: this plot of THD+N vs frequency shows that our prototype is as the manufacturer designed and has a fairly low distortion in the 2-3kHz range. Naturally this speaker would use a low pass filter at 3kHz or higher to be in its “happy” range and would mate well with the UT-P 2017 tweeter. anywhere you have restricted space and power, and require close proximity stereo or surround. An example is audio visual and virtual reality glasses. For this, USound market a two-piece unit (left & right) to act as “near ear” speakers, complete with a tiny MEMS woofer and tweeter. These have a crossover point of 3.54kHz and promise excellent performance down to 20Hz. These can also be obtained from Digi-Key for around A$700 per pair. This device is called the “Danube” with respect to its German origin. It is designed to fit in a spectacle frame. The sound travels directly into the ear and the speaker “cabinet” is a dipole design. They do have inbuilt DAC and audio amplifiers but require a power supply and a 16-pin connector as well as a Bluetooth receiver. USound also market ready-to-go spectacles in their “Fauna” range and these come in a variety of styles and include a microphone for connecting to a phone. We were not able to get hold of a pair for testing but were able to obtain a similar Asian product on ebay for $75 including GST. These are remarkable products and the stereo effect is quite stunning. The sound seems to be coming from a distance – your brain tells you that N+] it must be out there Australia’s electronics magazine because you can hear background sounds as well. When you turn your head, the distant sounds seem to follow – it’s quite an experience! You don’t need to constantly pull out those annoying earbuds or remove a headset to hear someone talking to you either. Overall it’s a very pleasant and comfortable arrangement. I didn’t try it for VR but I am sure the experience would be enhanced with the type of freedom the spectacles add. Of course, if you don’t normally wear glasses, you can always get them as sunnies or simply tinted. Your local optometrist can easily arrange to have your personal prescription lenses fitted. In fact, our local guy is fitting up our $75 unit as this article is being written. Conclusion No doubt tremendous advances have been made to create such a tiny speaker with excellent performance characteristics. The current and potential applications will no doubt increase over time. The capability of manufacturing billions of these little devices using integrated circuit technology and the ability to install them on electronic devices using flow soldering techniques will inevitably reduce prices in the usual fashion. These devices are currently very expensive but so far, it has been a monopoly for USound. We just have to wait for more players to enter the market! Editor’s note: the UT-P 2016 has been recently obsoleted and replaced with the similar UT-P 2018, which we have not yet tested. siliconchip.com.au ; UPDATE: CUI DEVICES MODEL CDS-13138-SMT IMPEDANCE VS FREQUENCY 8 OHM MEMS 9.8 9.6 Speaker Impedance (ohms) While strictly speaking this tiny speaker from CUI Devices isn’t a true MEMS device, we include it as a possible alternative. It is about the same price as the USound MEMS speakers and is available from Digi-key China (Cat 102-3536-2-ND). This particular speaker measures 13 x 13 x 4mm; about five times larger than the tiny USound 2016 but has a much poorer performance. However, the larger size does make it much easier to handle and solder. That makes it more useful for DIY projects. It is a purely dynamic speaker and requires no power supply which would make it useful for miniature devices. From the frequency response curve and the distortion data of the CUI MEMS speaker, we would conclude that it would be adequate for voice reproduction and frequencies above 1kHz. It would be useful where a range of high frequency tones are required such as instrument SC keyboards, small computers etc. (Actual size) 9.4 9.2 9.0 8.8 8.6 8.4Ω 8.2 20 50 100 200 500 1k 2k 5k 10k 20k Frequency (Hz) Graph 1: the CUI has a resonance of 800-900Hz because it is a dynamic speaker, as opposed to the USound 2016 MEMS which is an electrostatic speaker. The resonance shows that this speaker will not reproduce much below 1kHz. THD+N VS FREQUENCY 8 OHM MEMS FREQUENCY RESPONSE 8 OHM MEMS Additional Resources: Product Page | 3D Model | PCB Footprint date 01/20/2020 page 1 of 5 d B MODEL: CDS-13138-SMT │ DESCRIPTION: SPEAKER r A FEATURES • • • • SMT (surface mount) speaker reflow solder capable wide operating temp range compact size Graph 2: it is pretty obvious that this speaker is very poor compared to the USound devices. It has a huge peak around the 7kHz point and has virtually nothing below 1kHz, as you might expect from the impedance data. Graph 3: the distortion is acceptable from 2kHz to 20kHz but there is high distortion in the low frequency range due to the poor frequency response. This creates a high signal-to-noise ratio below 1kHz. SPECIFICATIONS The CUI Devices CDS-13138SMT mounted on a small piece of perfboard to allow easier connection (shown life size). siliconchip.com.au parameter conditions/description input power maximum power: IEC-60268-5, filter 60s on/120s off, 10 cycles at room temp impedance at 1.5 kHz, 1.0 V resonant frequency (Fo) at 1.0 V frequency response output SPL ±10 dB Fo sound pressure level at 0.7 W, 0.1 m ave, at 1.0, 1.6, 2.0, 3.2 kHz at 1.0 W, 1.0 m ave, at 1.0, 1.6, 2.0, 3.2 kHz 84 67 distortion at 2.0 kHz, 0.7 W buzz, rattle, etc. must be normal at sine wave between Fo ~ 20 kHz dimensions 13 x 13 x 4.0 magnet Sm2Co17, Ø6.0 x 1.0 mm material LCP cone material mylar terminal surface mount, Au plating min typ max units 0.7 1.0 W 6.8 8 9.2 Ω 840 1,050 1,260 Hz storage temperature -40 yes dB dB 5 % V 1.1 -40 RoHS Hz 90 73 mm Australia’s electronics magazine operating temperature yes 20,000 2.37 weight washable 87 70 g October 2021 93 85 °C 85 °C