Hi Norman,Purifi PRs Go Boom
Part 1 – Fi12v1 DSP bi-amplification
On forums like diyAudio and YouTube things rarely go wrong, following my impression that human nature leads us to more quickly post successes while silently hurrying past failures. Perhaps less so on YouTube, I’ve watched lots of hot rodders grenade engines with excessive supercharger boost. Today I will add my first Purifi driver failures brought about by the hi-fi equivalent of excessive supercharger boost. Not quite melted cylinder heads or connecting rods through blocks but fair warning, cone abuse lies ahead. Also ahead are glimpses of what heretofore-unexplored deep bass capabilities are present in the Purifi driver architecture.
As background the following posts earlier in this thread detail other experiences where I pushed the PTT6.5 woofer and matching PRs to their limits.
Finding the limits of Fi16v1
https://www.diyaudio.com/forums/mul...purifi-woofer-speaker-builds-post6244079.html
PTT6.5W04 and the evil twins PTT6.5PR want to explode your speaker box
https://www.diyaudio.com/community/...er-speaker-builds.352063/page-24#post-6383932
For this tale each bass system is the Purifi powered speaker I call the Fi12 the ‘12’ denoting enclosure size 12 liters loaded with one PTT6.5W and dual PTT6.5PRs. Tweeter GR Neo3 horn loaded. Crossover and EQ (more about that EQ later) Danville Signal audio DSP setup as bi-amp into four channels of ACG 1000 amplification. Crossover and EQ set for 6 dB/octave crossover @ 1800 Hz 0 dB tweeter attenuation, used with analog attenuation on tweeter feed. Woofer EQ +10 dB @ 20 – 60 Hz and +3 dB @ 20 – 200 Hz. Tweeter EQ +1 dB @ 10 – 20 kHz.
Yes, these early days down in the foothills of the DSP learning curve we have a configuration that has a big old bunch of bass EQ. Digital and analog gain match the horn loaded planar tweeter to the PTT6.5 and addresses a 1 dB droop between 10k and 20k Hz. The +3 dB 20 – 200 Hz is intended as baffle step correction. The +10 dB boost across 20 – 60 Hz is a deep bass extension experiment. Not unexpectedly this DSP program is bass heavy compared to flatter response configurations.
Any experienced listener would recognize the bass heavy tuning of this DSP load. And many listeners would recognize it as very much big time fun. The little PTT6.5W and PRs got to it sounding like the fat muscular sound of an Altec or JBL 12-15” woofer in a refrigerator sized reflex box. I was also reminded of the typical too high bass levels used when subwoofers are first added to an audiophile grade HiFi rig. During the new toy honeymoon period extra deep bass after which the sub’s gain is adjusted down to a flatter level more in tune with a goal of balanced HiFi reproduction. Nonetheless I lingered in the bass fat land of the lotus-eaters.
With typical music and video sources the EQ boosting frequencies below the driver/enclosure system tuning added explicit first octave bass like what one gets with subwoofers or very large enclosure volume 3- or 4-way speakers.
Graph below shows in-room response of the Fi12v1 thus configured. The 30-50 Hz boost is a room mode, mentally subtracting that room mode see the –3 dB is 20 Hz. The weeks (yes, so much fun for weeks) this configuration was in place in my listening room that response to 20 Hz made its presence known while playing music with organ and synth we audiophiles catalog containing sub-40 Hz bass. Playing movies, the low frequency effects in action adventures and the very low frequency tones added to enhance tension lit up the room with power.
View attachment 1262719
Part 2 – Edge condition spoils Thursday HiFi night.
In engineering we refer to edge conditions and the need to account for them. Devices have a range of conditions they are expected to operate across. These often fall into a classic bell curve with 90 to 99% of those conditions in the fat middle of the curve, however beware the 1%, 5% or 10% of conditions that live on the trailing edges of the curve. The classic example in audio engineering is an amplifier’s stability, getting the feedback loop correctly compensated. Everything is fine with resistive loads, then add some capacitance and/or inductance and oscillations appear. On that fateful night Fi12v1s encountered an edge condition known as Igor Stravinsky.
Specifically, the recording Stravinsky: The Rite of Spring performed by Minnesota Orchestra conducted by Eiji Oue on Reference Recordings catalog number RR-70 (https://referencerecordings.com/recording/stravinsky-the-rite-of-spring/).
During Thursday HiFi night my friend Michael requested the Eiji Oue Rite of Spring.
I was not that familiar with the performance so when it started very quietly, I increased the volume to bring the level up to 70-80 dB. It’s a wonderful recording and all was well for the first 9 minutes, then the orchestra, including massive drum whacks, goes from quiet to triple fortissimo the level instantly jumping many 10s of dBs as the percussion section wails on their drums. Before it was humanly possible to reduce levels the Fi12s were making very un-musical noises of extreme distress. Got playback stopped and surveyed damage. No voice coil rubbing or other damage to the woofers, amazing. Passive radiators did not get away similarly unscathed. All four PRs had suffered creased cones. Later disassembly found one PR had such extreme excursions the inner spider detached from the mass slug.
View attachment 1262720
View attachment 1262721
Not a happy result. Purifi makes very rugged drivers however there is a limit. The photo of the front side creasing makes the cone look flattened, that is not the case they still have their original dish profile. The extreme excursion creased the cones but they did not tear. Using medium thickness CA glue the one detached spider was reattached. While the creases activate my audiophile obsessive tendencies truth is with the one PR’s spider repaired and the four PRs placed back in service the Fi12s sound fine.
Part 3 – Post crash analysis.
With crystal clear 20/20 hindsight it is easy to see my attempt to turn the Purifi PRs inside out resulted from both operator error and design oversight. Operator error playing dynamic material at too high a gain setting. Design oversight implementing EQ boost below Fs without complementary dynamic frequency specific limiting.
I have been around the audiophile music block more than a few times thus should have questioned what the dynamic range of a Reference Recordings recording of a large classic war horse might contain. Further investigation shows RR-70 is truly extraordinary in this regard. The following two images are the track loaded into Adobe Audition with the dynamic range shown on the waveform display.
View attachment 1262722
This display shows all 21 minutes of the movement, notice the first 9 minutes average –30 dB with louder sections at –20 dB. Then at 9 minutes 20 seconds all hell breaks loose and levels jump to –3 dB!
View attachment 1262723
The second Audition screen capture zooms in to the couple of minutes before and after the dramatic jumps in levels. Playing the track in Audition the music before 9:20 is averaging –25 to –35 dB then when the triple fortissimo with drums hits at 9:20 level jumps to –3 dB. A 32 dB step change in output in the bass into a system with 10 dB bass boost using an amp capable of a kilowatt into 4 ohms, what could possibly go wrong?What I find so very impressive is the woofers took such thermal and mechanical abuse on the chin and survived. The PRs, well like a race car a pit stop after wall contact was required and we have some dings but they too race on.
As an example of just how the much an edge case the Eiji Oue RR-70 Rite is here is an Audition capture of Stravinsky conducts Stravinsky on Sony Classical.
View attachment 1262724
With the quiet vs loud parts of the piece at different times it is apparent this is a different arrangement. Here the point is that the quiet parts average -15 dB, at least 10 dB louder than the Reference Recordings version by Oue. The Sony recording has good dynamic range, the RR truly exceptional in this regard.
Part 4 – Advance to 2024.
This incident occurred in Summer 2022, quite a bit of water under the speaker design bridge since then. One takeaway was that whoever with experience in DSP controlled active speaker systems was first to use a PTT6.5 driver with a dynamically controlled bass boost design would blow minds rather than drivers. And along comes Mads Buchardt & teams’ Anniversary 10 (https://buchardtaudio.com/collections/active-speakers/products/anniversary-10). And the result I envisioned. I find their expressions and reactions typical after a good demo for someone new to Purifi drivers.
View attachment 1262726
The video is 70 seconds of fun, the coming review will be an interesting add to those already linked on the A10’s product page.
Sorry to hear this. Everything has a limit of course. Boosting at the PR tuning frequency or below is rather dangerous. We are always interested in learning about such issues and preferably receive the parts for failure analysis. In this way we can learn and improve.
All the best,
Lars
A follow up to my post “Purifi PRs Go Boom“ (see page 122 this thread).
Comment by jmpsmash “And yes, very impressively, the driver is bomb proof. I don't use the 6.5PR but surprising to see how fragile it is.”
No no no please no. It was never my intent to imply anything approaching fragility on this set of Purifi components. Rather I invite all my fellow Purifi fans and staff to laugh at me and with me at my carelessness that led to what I consider a demonstration of Purifi drivers ‘bomb proof’ nature. Stated another way in engineering teams we often comment “we can make it idiot resistant, not idiot proof”. To which another engineer will add “yes, the idiots can be so ingenious”.
A good point to paste a sentence from my original post that is key to this discussion. About the DSP system driving this system I documented its settings. “Crossover and EQ set for 6 dB/octave crossover @ 1800 Hz 0 dB tweeter attenuation, used with analog attenuation on tweeter feed. Woofer EQ +10 dB @ 20 – 60 Hz and +3 dB @ 20 – 200 Hz. Tweeter EQ +1 dB @ 10 – 20 kHz.” If you are new to speaker design or missed it in my verbiage on first read that is what is technically known as a ‘crap ton of bass EQ’.
Comment by Lrisbo “Hi Norman, Sorry to hear this. Everything has a limit of course. Boosting at the PR tuning frequency or below is rather dangerous. We are always interested in learning about such issues and preferably receive the parts for failure analysis. In this way we can learn and improve.”
Thank you Lars for continuing to watch this thread, commenting, and sharing. Generous offer to do a failure analysis. If such a failure occurred with typical flat or even 2-3 dB bass boost EQ in a correctly aligned implementation, I would be emailing for an RMA number. In this case with 13 dB total bass EQ I held the drivers faultless and marveled the woofers emerged unscathed. To me the analogy is if I reprogrammed my car’s engine control computer adding 2000 RPM to rev limiter, rev’ed to 9500 RPM thus stuffing a valve into a piston it’s not Ford’s fault.
Honoring the high technical level we enjoy on diyaudio.com lets dive into it a bit more. The following can be summarized as Lars wrote “Everything has a limit of course”.
This system used early production paper cone PTT6.5 woofers and passive radiators, woofers 4 ohm models. Each speaker used one woofer and dual PRs in 12 liter enclosures no added mass on PRs. Let’s look at the simulations of that system in VituixCAD for normal and overdriven scenarios. For these sims I did not include the DSP bass EQ, rather looking at the acoustic engines’ capability and then mentally add the extra stress added by so much EQ.
First used within proper conditions. Set at 150 watts input we are looking good, notice in the upper left SPL frequency response graph everything is below the red ‘redline’ trace. On the lower left excursion graphs the PRs are below max all the way down to 10 Hz. A 150 watt input below 30 Hz crosses woofers’ Xmax. I interpret that as good to go with 95% of all music in domestic settings, keep away from discos or the trip wire explosion in Sam Mendes’ ‘1917’. Pause and consider, it’s a 6.5”/165mm class driver absorbing 150 watts and replying “OK here’s your 107 dB output”.
Now overdriven. Recall in this bi-amped prototype I used ICEpower 1000ASP amps rated 1000+ watts into 4 ohms. Combine the deep bass on the recording, excessive setting on the preamp that night, and bass boost and it’s a guess just how big a pulse hit the drivers. My guesstimate for this discussion 550 watts. First problem is that 113 dB output, well above 109 dB max for this system. Next look at the excursion plot, I like how the woofers excursion line is trying to go vertical. In VituixCAD I read that as peaking at 85 mm @ 10 Hz! We also see the passive radiators’ going over its max limits. And if we speculate at failure the system was seeing more like a kilowatt that sim (not shown here) shows the PRs over excursion at all frequencies below 50 Hz. That is no bueno.
Final summary, this is why I want to get my hands on a few PTT10.0s!
Comment by jmpsmash “And yes, very impressively, the driver is bomb proof. I don't use the 6.5PR but surprising to see how fragile it is.”
No no no please no. It was never my intent to imply anything approaching fragility on this set of Purifi components. Rather I invite all my fellow Purifi fans and staff to laugh at me and with me at my carelessness that led to what I consider a demonstration of Purifi drivers ‘bomb proof’ nature. Stated another way in engineering teams we often comment “we can make it idiot resistant, not idiot proof”. To which another engineer will add “yes, the idiots can be so ingenious”.
A good point to paste a sentence from my original post that is key to this discussion. About the DSP system driving this system I documented its settings. “Crossover and EQ set for 6 dB/octave crossover @ 1800 Hz 0 dB tweeter attenuation, used with analog attenuation on tweeter feed. Woofer EQ +10 dB @ 20 – 60 Hz and +3 dB @ 20 – 200 Hz. Tweeter EQ +1 dB @ 10 – 20 kHz.” If you are new to speaker design or missed it in my verbiage on first read that is what is technically known as a ‘crap ton of bass EQ’.
Comment by Lrisbo “Hi Norman, Sorry to hear this. Everything has a limit of course. Boosting at the PR tuning frequency or below is rather dangerous. We are always interested in learning about such issues and preferably receive the parts for failure analysis. In this way we can learn and improve.”
Thank you Lars for continuing to watch this thread, commenting, and sharing. Generous offer to do a failure analysis. If such a failure occurred with typical flat or even 2-3 dB bass boost EQ in a correctly aligned implementation, I would be emailing for an RMA number. In this case with 13 dB total bass EQ I held the drivers faultless and marveled the woofers emerged unscathed. To me the analogy is if I reprogrammed my car’s engine control computer adding 2000 RPM to rev limiter, rev’ed to 9500 RPM thus stuffing a valve into a piston it’s not Ford’s fault.
Honoring the high technical level we enjoy on diyaudio.com lets dive into it a bit more. The following can be summarized as Lars wrote “Everything has a limit of course”.
This system used early production paper cone PTT6.5 woofers and passive radiators, woofers 4 ohm models. Each speaker used one woofer and dual PRs in 12 liter enclosures no added mass on PRs. Let’s look at the simulations of that system in VituixCAD for normal and overdriven scenarios. For these sims I did not include the DSP bass EQ, rather looking at the acoustic engines’ capability and then mentally add the extra stress added by so much EQ.
First used within proper conditions. Set at 150 watts input we are looking good, notice in the upper left SPL frequency response graph everything is below the red ‘redline’ trace. On the lower left excursion graphs the PRs are below max all the way down to 10 Hz. A 150 watt input below 30 Hz crosses woofers’ Xmax. I interpret that as good to go with 95% of all music in domestic settings, keep away from discos or the trip wire explosion in Sam Mendes’ ‘1917’. Pause and consider, it’s a 6.5”/165mm class driver absorbing 150 watts and replying “OK here’s your 107 dB output”.
Now overdriven. Recall in this bi-amped prototype I used ICEpower 1000ASP amps rated 1000+ watts into 4 ohms. Combine the deep bass on the recording, excessive setting on the preamp that night, and bass boost and it’s a guess just how big a pulse hit the drivers. My guesstimate for this discussion 550 watts. First problem is that 113 dB output, well above 109 dB max for this system. Next look at the excursion plot, I like how the woofers excursion line is trying to go vertical.
In VituixCAD I read that as peaking at 85 mm @ 10 Hz! We also see the passive radiators’ going over its max limits. And if we speculate at failure the system was seeing more like a kilowatt that sim (not shown here) shows the PRs over excursion at all frequencies below 50 Hz. That is no bueno.Final summary, this is why I want to get my hands on a few PTT10.0s!
@Norman Tracy , i did not laugh at you, but smiled remembering my own massive overload cases in the past.
Your simulation of the case is quite clearly picturing what happened, the power assumption you made also indicates it actually could have been more.
The occured damage (very symmetrical as shown in the pictures) also indicates to me the high quality of the materials used and of the assembly of the units.
For tweeter and midrange i use a series resistor to blow first (and recently happened when measuring room curves with acourate), but for woofers not.
Your simulation of the case is quite clearly picturing what happened, the power assumption you made also indicates it actually could have been more.
The occured damage (very symmetrical as shown in the pictures) also indicates to me the high quality of the materials used and of the assembly of the units.
For tweeter and midrange i use a series resistor to blow first (and recently happened when measuring room curves with acourate), but for woofers not.
At first sight this is correct. But the port resonance only takes over a small portion, furthermore it is still triggered by the driver and below port-res it is not damping at all, thus large cone movements. So taking all this into account i do not believe this statement is as true as it sounds.
I was trying to write 40Hz
Anyway, common sense should not be ignored. Bass performace of small drivers is always only modest. With higher spl or lower F, excursion skyrockets which easily breaks the active driver or passive BR unit. Also distortion skyrockets then. Sealed driver with eq and high power burns it's voice coil...
Purifi drivers are among the best, but physics cannot be fooled!
Well said Juhazi. I will point out ignoring common sense is one of my specialties the PTT6.5 invites one to lean into.
The intent was never an open loop system tuned thusly with such radical fixed EQ. Rather testing the water for use in a closed loop system (DSP or analog EQ with active limiting at frequencies prone to large excursions). While I was off doing other conventional open loop designs the Buchardt Audio Anniversary 10 Active Wireless Speaker showed how such a closed loop design can exploit the considerable strengths of the PTT6.5's capabilities.
Well, who needs subwoofer with Purifi?Subwoofers that needs high pass filters, are too small
Always subwoofersWell, who needs subwoofer with Purifi?
Purifi is a great driver, but it can't magically solve acoustics...
Your choise, not mine.Always subwoofers
Purifi is a great driver, but it can't magically solve acoustics...
Exactly, there is no replacement for displacement. Sub 30hz will always need large drivers in order to accurately reproduce the full range, unless you are in a small room where room modes boost the bass.Always subwoofers
Purifi is a great driver, but it can't magically solve acoustics...
And I actually tried that. 4 x 6,5 Purifi woofers can really work hard and produce impressive amounts of clean volume. But as we get lower in frequency and really want to feel the air move and considering our hearings lower sensitivity to deep bass, in contrast to the higher frequencies. Then I have never truly found any other way, then to go up in surface area of the woofer.
We do need to think about what we want with our own HIFI system. If the speaker has to be small, then we have to work around that compromise, and Purifi do offer something special here.
It's DIY, there is mostly free play within each of our budgets and other personal restraints, like neighbors, partners, time...