Purifi Audio PTT6.5W04-01A midwoofer

Известное выражение "Все дороги ведут в Рим" в контексте исегодняшнего дня должно звучать "Все дороги ведут в Данию". Я не ошибусь, если скажу, что большинство бытовых акустических систем вплоть до самого высокого класса в настоящее время собираются с использованием динамиков так или иначе имеющих датские корни. Посудитете сами:

• Peerless
• Vifa
• ScanSpeak
• Danesian Audio
• Dali
• Dynaudio
• Wavecor
• Timphany
• Audiotechnology
• PointSource Acoustics

- это неполный перечень  датских компаний или компаний, уходящих корнями в датское прошлое. Вот и сегодня этот список полнился еще одним совсем юным брендом PURIFI. 

PURIFI - это объединение успешных и всемирно известных профессионалов-энтузиастов в области аудиоинженерии, целью которых является, внимание(!), преодоление существующих качественных барьеров и создание самых лучших в мире аудио продуктов. С историей компании вы можете ознакомиться здесь, а пока что просто стоит упомянуть людей, стоящих во главе этой команды:

• Bruno Putzeys, Соучредитель
  Изоборетатель UcD (Philips) технологии, вероятно, наиболее распространенный усилитель Д-класса для престижных аудиопродуктов. Изобретатель Ncore (Hypex), который общепринято считается первым усилителем Д-класса, который успешно конкурирует по звуковым характеристикам с усилителями А-класса. Разработал высококачественные АЦП и ЦАП для Grimm и Mola-Mola. Соучредитель Kii Audio. Обладает 8 патентами и многочисленными патентными открытиями
   
• Lars Risbo, Соучредитель
  Первопроходец в области прямых импульсных PCM-PWM аудио усилителей как учредитель Toccata Technology, позднее поглощенной компанией Texas Instruments. Избранный член сообщества TI (2012) и Аудио CTO (2013). Исследовательские интересы включают цифровую обработку сигналов, оптимизацию сигнальных цепей и систем, конвертеры данных с передискретизацией, преобразователи и архитектуры со смешанными сигналами. Доктор философии (Danish Technical University (1994). Обладает более, чем 30 американскими патентами
   
• Carsten Tinggaard, Совладелец (PURIFI Transducer Technology)
  Основатель и владелец PointSource Acoustics (2009). До этого Product Manager в Danish Sound Technology (2003) и CTO в Tymphany Denmark (2005), ответственный за  responsible for образование R&D Denmark производства, покрывающего Peerless, Vifa и Scan-Speak торговые марки. Инициировал перемещение производства Peerless на китайские производственные площадки. Совладелец PURIFI Transducer Technology (2017) - подразделение PURIFI
   
• Peter Lyngdorf, Соучредитель
  Владелец Lyngdorf Family Holding, Hi Fi Klubben, DALI и Steinway Lyngdorf. Группа насчитывает более 1000 сотрудников и многочисленные фабрики в Дании и Китае. Ранее владел несколькими аудио компаниями, включая NAD Electronics и Snell Acoustics. Продвигал технологию Digital Room Compensation от Snell DSP 1000 (1993) через NAD, TacT Audio, а сейчас через Lyngdorf RoomPerfect
   
• Claus Neesgaard, Совладелец, Директор
  Занимал Business/R&D Management, Product Definer и Systems Architect должности (Texas instruments), ответственные за запуск и развитие TAS/TPA серии усилителей Д-класса. Некоторое время возглавлял TI Audio DSP серию продуктов, переводя ее в системные решения, базирующиеся на стриминге. В начале карьеры разработал несколько ключевых решений в усилении Д-класса для Toccata Technology, приведшие к поглощению компанией TI в 2000. Ученая степень Master of Science in Engineering в Danish Technical University (1998). Обладатель 6 американских патентов
• Kim Madsen, Менеджер по приложениям
  Специализировался в аудио и силовой электронике. Занимал должности Applications Manager, Product Definer и Lead Systems Engineer (Texas Instrument), в последенее время сфокусировался на TI Automotive Amplifier портфолио. Разработал силовые модули военного стандарта для Ariane 5 Rocket (Alcatel) и нескольких самолетов-истребителей (Dynatech). M.Sc.EE (1991) в DTU. Обладатель 5 патентов США и нескольких патентных открытий
   
• Søren Poulsen, Разработчик
  Занимал должности Product Definer и Lead Systems Engineer (Texas Instruments), особенно фокусируясь на оптимизации производительности и надежности микросхем аудио усилителей Texas Instruments средней/высокой мощности. Обладал описанием и архитектурой  разработки самого мощного и производительного в мире микросхемного усилителя. Доктор философии в DTU (2004). Обладатель 5 патентов США и нескольких публикаций
   
• Morten Halvorsen, Разработчик (PURIFI Transducer Tech.)
  Работал в качестве Senior Acoustic Engineer в PointSource Acoustics. Со-разработчик громкоговорителей и звуковых устройств в  Montana Sound и SoundBy PointSource продуктах. Специализировался в электромагнитных симуляциях, измерениях и процессах оптимизации. Исследовал эффект модуляции силового фактора в динамиках, разработал измерительное оборудование и модели для симуляции. Результаты представлены на  AES 138. Master of Science in Engineering в Danish Technical University в области акустики (2014)

В мае 2019 года компания дебютировала в Мюнхене (Германия), представив свои первые продукты, впечатляющие своими характеристиками - усилительный модуль Д-класса  1ET400A и низкочастотный динамик PTT6.5W04-01A. Как правило, любая компания стартует со своего самого сильного продукта, чтобы продемонстрировать свой технический уровень и потенциал, в противном случае может получиться смерть на взлете. Сегодня ее мидвуфер уже добрался до нас. Что ж, посмотрим, с чего стартует компания.

Я хотел бы привести отрывок из беседы с Lars Risbo - одним из основателей и разработчиков компании PURIFI Transducer Technology, затрагивающей конструктивные и технические аспекты нового мидвуфера:

Yevgeniy Kozhushko:  Lars, what does the logo of Purifi brand mean?
Lars Risbo:  Three droplets, being gradually purified down. As sketched by Bruno, the leaf initially pointed the other way, forming a P. Then a real graphics designer intervened.
The name itself is Putzeys-Risbo-Fidelity.

 

Yevgeniy Kozhushko:  I was very glad to see paper as a cone material in the PTT6.5W04-01A. I like paper membranes because their natural timbre, smooth and rich sounding.  The PTT speakers sound exceptionally well. I would characterize them as very dynamic, clear, transparent and natural sounding speakers.

Lars  Risbo: Great to hear that you like paper - we all love paper as cone material here - its an amazing material. Carsten Tingaard has build up so much experience in selecting fiber mixes and fiber lengths. Unfortunately, there seems to some folks who consider paper as less good simply because it also happens to be quite cheap. There is not much appreciation for how much optimization of the geometry means for the performance - exotic materials seems to be a more sexy story.  We have a client who insisted on an alu cone and the prototype has actually more distortion than the paper cone (plus a more peaky response).  We are of course keeping an eye open for new cone materials we plan to continuously experiment.  But good point you made about our documentation not mentioning paper.
Our marketing challenge is that only few people seem to care about measured performance and often judge quality just on the looks and materials. Our driver has all the expensive parts (except the special surround) hidden inside the motor but it looks quite normal seen from the outside and then its even using paper as cone material. There are also people who dismiss our driver due to its approx. 26 g Mms and 4 layer coil because they think it is 'slow' - that changes of course when they measure or listen to it. Our motor construction and use of very thick copper rings gets inductance down below a normal 2 layer coil so the bandwidth is not affected. Actually, a higher cone mass can allow a stiffer construction so higher mass can actually give higher bandwidth (at the expense of sensitivity).

 

Yevgeniy  Kozhushko: The membrane is made of paper composite with some synthetic long fibers added?
Lars  Risbo: It's a mix of cellulose with some other fibres and then hard pressed. The actual material of the fibres is part of the "secret sauce" but the precise mix of fibre lengths even more so. Coupled with an optimised profile this yields breakup behaviour that we think is rather more useful than, say, a metal cone.

 

Yevgeniy  Kozhushko: I think the peripheral circular rib at the back edge of the cone is made for resolving surround-edge-of-cone resonance problem?
Lars  Risbo: Yes, this is to strengthen the cone rim and thereby to suppress unwanted modes.

 

Yevgeniy  Kozhushko: What materials are the basket and surround made of?
Lars  Risbo: The basket is diecast aluminium, which tends to use recycled aluminium so the exact alloy varies a bit. To be honest, being able to use recycled rather than recyclable materials was a strong argument in favour of aluminium. The surround is made of NBR rubber so pretty much standard. Again, it is it's special geometry that makes the difference.

 

Yevgeniy  Kozhushko: For what purpose is the base of the basket not made solid, but with slots between the spikes of the basket? I have not seen this anywhere before.
Lars  Risbo: It wouldn't make much difference in any other speaker but when you manage to get such constant Le(x) it's worth considering even such minutiae as the shorting from eddy currents caused by the basket. The change in Le(x) is measurable for a solid base basket so we got this special basket designed and tooled.

 

Yevgeniy Kozhushko: What is kind of material the coil former is made of?
Lars  Risbo: Glass fibre composite.

 

Yevgeniy  Kozhushko: I see that the voice coil has variable winding pitch. Is it correct? I think it is one of the reasons of very linear BL(x) and L(x) behavior.
Lars  Risbo: That's correct when it comes to BL(x), but it does very little to solve Le(x). That trick is hidden in the magnet system geometry. We'll discuss it when the secret outs but we should stress that we managed to flatten Le(x) down to DC.  Shorting/Faraday rings aren't the secret for constant Le(x) because even the most solid Faraday rings won't help much below 100Hz. We use very thick and long copper rings (the by far most expensive components in the design) but this is mainly to control the hysteresis distortion from the iron and to reduce Le at mid frequencies.

 

Yevgeniy Kozhushko: Your speaker forced me to extended my of measurements and to add IMD ones because its are the measurements where PTT6.5W04-01A really excells.  I have to add this type of measurements to my standart set and will be publishing them at individual speaker's measurement pages. 

Lars  Risbo:  It is a great idea to include IMD plots since it probably tells a lot about more the sound quality. Whereas a 2nd harmonic might sound warm then there is nothing nice about the atonal IMD2 components coming from the very same2nd-order  distortion source. I have done some two tone experiments and found that IMD products can be 20 or even 30dB lower than for a benchmark driver.
Our goal is indeed to drastically reduce the IMD and e.g. let bass and mids to come from the same driver. I suspect that a conventional half roll driver exhibits even more distortion/IMD when put in a small box due to the position dependent change of the radiating area Sd(x). This modulates the box air compliance - this does not happen in free air infinite baffle. So the improvements may be even bigger in a real life box.

Yevgeniy Kozhushko: I did a listening test - to a musical recording with female vocal I added 30 Hz sine signal and listened to a quality of midrange reproduction. PTT6.5W04-01A had undistorted sounding at much more 30 Hz excursions than many other renowned high quality speakers I have on hands. The PTT is able to find its place in the most sophisticated two way loudspeakers. 

Lars  Risbo: The two tone IMD is ultimately limited by the Doppler distortion (increasing proportionally with the voice tone frequency). The big question if of course how sensitive our human ears are to the Doppler distortion. The fact that we seem to hear a dramatic sound improvement with our new 6.5 Driver suggests that Doppler may not be so audible - more investigation is needed of course.

Yevgeniy Kozhushko:You mention in PTT6.5W04-01A specification about low magnetic hysteresis distortion...
Lars  Risbo: We discovered that the magnetic hysteresis distortion is a problem both for speakers and class D amplifiers (when using a ferrite cored inductor). The same goes for cross over inductors with iron/ferrite cores.

Yes it is a bit of a challenge to measure since it just shows up as odd harmonics when you drive a sine wave. For a more complex signal (like music) the memory effects get exposed (this most likely why it is so bad for the sound). Bruno did a simple experiment where the drivers response to a short tone burst changes depending on what the speaker played back a minute back in time - this change was in the order of 0.1%.

But even with a sine, we can see a telltale sign: if you plot harmonic distortion ratio for a sine versus level then you see a region where the HD ratio (both for the 3rd and 5th harmonic) increases slower or equal to 1st order. A memoryless 3rd order nonlinearity would show a 2nd order slope of this curve. This means that there is relatively more odd harmonics at low level from the hysteresis. You can also see it from the distortion vs frequency of the current when driving a voltage tot he speaker. The distortion is high at low frequencies (Kms and Bl etc) but as the frequency increases then the excursion drops very fast (2nd order with frequency). This means that the distortion drops very fast but then hits a minimum level and then rises with frequency again. This increase at high frequencies is due to the hysteresis (or rather magnetic domains flipping around). It increases in level with frequency since the induced voltage in the coil is proportional to the derivative of the flux in the coil. Our driver has its 3rd harmonic current distortion down around  the -80dB level (there are graphs in the data sheet). We may in the future release improved models with even lower hysteric distortion - we have the technology but we are looking into how to manufacture it in volumes. The dynamics, staging, space and  ‘black canvas’ are very much affected by this hysteresis distortion. Our driver is between 10 and 50dB better than what we have compared it to. Underhung drivers are prone to lots of hysteric distortion since there is so much iron around the coil.

It was first when Bruno discovered the Preisach model that we understood how complex this distortion is. There are some animations of Preisach models on the web. Meanwhile, the slide from the presentation I sent you explains quite a bit (see the slide below). The plot is measured using two windings on a ferrite core. The first winding is driven by an amplifier and the voltage induced on the 2nd winding is recorded (blue trace) along the current in the primary winding (red trace). The white trace shows the traditional BH-hysteresis loop (the B field is found by integration of the induced voltage in the 2nd winding).

Normally you see the BH hysteresis loop for a sinewave with just one max and min current (H-field). Bruno got the idea to mix in a 3rd harmonic to the current so that we get more local minimums and maximums (points 1 2 3 5 6 and 8). The ferrite core remembers the minimum at point 2 and when the current swings up again towards a maximum at point 3  then the slope of the BH curve changes and we get a sub BH loop ending in point 4 where the current passes the prior minimum from point 2. At this point the BH curve shanges slope suddenly and we get a step in the induced voltage (blue). This voltage step is even bigger at point 7 when the current passes the past minimum at point 5. This voltage step is equivalent to a voltage step out of the amplifier and will be played back by the speaker. The ferrite will remember an infinity of past local maximum and minimum current and produce step voltages whenever these points are passed again. For a sinewave, we just get odd harmonics but for a complex music signal, we get an extremely annoying  complex distortion that our ears are very sensitive to. Its not random noise (for which our ears are forgiving to) and it is also not a simple nonlinearity without memory. The response of the speaker depends on what it played moment ago (or even yesterday). When you reduce this distortion then our experience is that we get a much more dynamic sound. A transient passes all these remembered past maxima and minima so transients get very distorted compared to a steady sine wave.

Для тех, кому интересно наблюдать за обсуждением мидвуфера PTT6.5W04-01A - приглашаю посетить англоязычную ветку на известном ресурсе audiosciencereview.com

Также интересен обзор двухполосной демо-акустической системы на базе рассмотренного мидвуфера и твитера Mundorf AMT на ресурсе audiophilestyle.com