Why?
The upper mid is a 4 Ω driver -> Modulus-286 on ±36 V rails.
The lower mid is an 8 Ω driver -> Modulus-86/186 on ±36 V rails.
I'm sure my ears would have started to ring way before the woofer ever maxed out. The resulting IMD products would cause me to reach for the volume control and turn the volume down.
No amp will ever be able to provide 80.000000000 W and not a fW more. That's a statistical impossibility. I can prove it mathematically to anyone who cares.
So any amp will provide either more than 80 W or less than 80 W. The question is how much less or more is acceptable? I found 40 W to be not quite enough, but 65 W was fine. I had no issue with 130 W either as I'm a responsible person who doesn't turn the volume to the max all the time. (All power numbers assume 8 Ω). That's not to say that my experience is the only correct experience (assuming an experience can be correct anyway). I do know the 80 W figure has been floated as some holy grail, but I do suggest that there's considerable room for nuance there.If high SPL is your ultimate goal, an open baffle speaker is probably not your best choice. The controlled directivity box speakers solve the same problem as the OB types but are considerably more efficient.
Tom
Last edited:
Do I see right for 2ch 1000 USD for 686 modules?
Yep. Your eyesight is working fine. Well, actually, it's $998, but close enough.
Tom
I'm not sure I understand that point. All moving coil drivers produce significant percentage distortion, particularly when the excursion increases. All driver manufacturers quote xmax for any driver, which is the one way value (so 2xmax peak to peak). At that point even a top flight driver is producing significantly greater than 1% distortion, regardless of whether you measure harmonic, intermodulation, or anharmonic tone sequence distortions.
This is not something that loudspeaker driver, or loudspeaker manufacturers want you to know. They know these facts of driver physics perfectly well, but for clear marketing reasons do not want the buying public to know. Driver manufacturers in particular use software from Klippel Klippel GmbH - Diagnostics of Audio Systems . It is not cheap, but it enables driver manufacturers to walk the fine line of driver optimization.
For a little light bedtime reading this is the list of deeply technical papers by Kippel Papers , in particular for this one on loudspeaker distortion https://www.klippel.de/fileadmin/kl...linearities_Causes,Parameters,Symptoms_06.pdf
The point of ~80W for the LX521 bass drivers is the limit determined by xmax of the drivers at 20Hz. Of course there is nothing to stop you truncating the response with a second order high pass filter at whatever frequency you like (to mimic the roll off of a closed box speaker), which enables a higher power handling.
All loudspeakers are a compromise. The main thing is to make the best of what you have, or can afford, and enjoy listening to music.
Craig
Last edited:
I had an LX521 system and I drove the woofers in parallel from a UcD400 amplifier due to the low impedance load the two woofers in parallel create.
Those woofers have X Damage at 56mm p-p so 28mm each way. They are capable of a lot more travel before damage well beyond x-max. Those drivers are really amazing, I have had them jumping all over the place with sub 20Hz content for testing, at that sort of level the main panel is driving you out of the room because it sounds so strained.
Short of a malfunction in the equipment or a user error in programming the outputs (which could happen at any time in any active system anyway) there isn't anything to fear from having an overly powerful amplifier, but not really anything to gain either.
Those woofers have X Damage at 56mm p-p so 28mm each way. They are capable of a lot more travel before damage well beyond x-max. Those drivers are really amazing, I have had them jumping all over the place with sub 20Hz content for testing, at that sort of level the main panel is driving you out of the room because it sounds so strained.
Short of a malfunction in the equipment or a user error in programming the outputs (which could happen at any time in any active system anyway) there isn't anything to fear from having an overly powerful amplifier, but not really anything to gain either.
I think he's referring to the amplifier distortion. If using an underpowered amp causes you to turn the volume knob to the point where the amp is clipping, the resulting amplifier distortion could damage your speaker.
Yep. Your eyesight is working fine. Well, actually, it's $998, but close enough.
Actually, $449/each ($898/pair) as of this morning. I don't normally do sales, but I do occasionally have the odd intro sale. So, while we wait for the next revision '686 to arrive, you get a discount. I expect to put the next batch into production Wednesday or Thursday this week. I'm waiting on the PCBs and a batch of inductors to arrive.
Tom
The UCD400 is a good choice for driving both woofers in parallel, although it is not really specified to drive 2.5 ohms (the minimum of two woofers in parallel). If operated from 50V rails it is not too far above the 80W into 8 ohms spec for the LX521 though. I went in the direction of driving each one separately with 100W max into 8 ohms.
Of course you need speaker protection, such as Tom sells, to prevent driver fires under amp fault conditons. And I use Neutrik Speakon connectors on the amps and speakers. And colour coded XLR cables and socket plates. So one you have internal wiring and cables checked, it reduces the risk of wiring (eg) the woofer signal up the tweeter to very low indeed.
Craig
Ah - that would make sense. Hard clipping is not a good thing for speakers, not so much from the point of the actual distortion components (horrid though they sound) it is more to do with the average power of a clipped waveform being much higher than a sine.
Craig
Yep. The RMS value for a bipolar square wave (as it would be with a hard clipping amp) is the same as the peak value, whereas the RMS value of a clean sine wave is Vp/sqrt(2). So at hard clipping you could get up to 3 dB higher (= 2x) power dissipated in the speaker. Then add that you'd also get significant power at higher frequency due to the harmonic distortion. Clipping. Bad.
Tom
Yep. The box in the middle is full of LM3886es! There isn't a lot of packing material in those boxes either. I'm hand-delivering them, so I know how they're handled.
Lots of differences in the assembly process. The most visible one is a new one-piece bracket design. I also found some tweaks in the input section that should lower noise and distortion a bit. I'm up against physics pretty hard on that one, so no promises just yet, but it should be slightly better than Rev. 1.0. I also did away with the Molex connectors and used the JST VL-series instead. That's the same connector as Hypex uses on their amps. They're nice connectors.
Also new is that the Modulus-686 will be available as a fully assembled module only. By focusing on the fully assembled modules I was able to keep the price down for everybody. You can still DIY the chassis around it, though.
Tom
Tom and others,
Have you measured the distortion of a good chip amp (like LM3886) as the temperature warms up from room temp to a steady-state "warmed up" temp? Does such an amp show measurably worse distortion, say, just a few seconds after turn-on?
Does a normal discrete Class B amp show behaviour different from an LM3886 amp?
Have you measured the distortion of a good chip amp (like LM3886) as the temperature warms up from room temp to a steady-state "warmed up" temp? Does such an amp show measurably worse distortion, say, just a few seconds after turn-on?
Does a normal discrete Class B amp show behaviour different from an LM3886 amp?
I generally measure my amps at room temp and after a one-hour warm-up. I have not seen any degradation in THD, but the THD+N will always be slightly higher in noise-dominated amps such as the Modulus-series. Basically the THD part of THD+N is below the THD+N floor of the instrumentation, but the +N part rises with higher temperature. That's how thermal noise works.
I don't recall anything significant happening at higher temperatures with the LM3886 by itself either.
Whether a regular Class B or Class AB amp is well-behaved versus temperature depends entirely on the competency of the circuit designer.
Tom
I don't recall anything significant happening at higher temperatures with the LM3886 by itself either.
Whether a regular Class B or Class AB amp is well-behaved versus temperature depends entirely on the competency of the circuit designer.
Tom
Yep. With the bias transistors and the output transistors on the same die (as in the LM3886) the thermal coupling between them is very tight. The devices can also be much better matched than you would ever be able to accomplish in a discrete circuit. Further, the IC layout can be constructed to maximize the matching between the devices, which will lead to even better thermal stability.
Now, I have not seen the LM3886 layout, so I don't know if it was done that way. But as a former IC designer, I'd say it's the only way that makes sense, so I would fully expect it to be designed to optimize device matching.
In my view, there's a key difference between IC design and discrete design: Discrete designs rely much more heavily on the absolute value of the components. IC designs rely much more on the matching between components.
Tom
Now, I have not seen the LM3886 layout, so I don't know if it was done that way. But as a former IC designer, I'd say it's the only way that makes sense, so I would fully expect it to be designed to optimize device matching.
In my view, there's a key difference between IC design and discrete design: Discrete designs rely much more heavily on the absolute value of the components. IC designs rely much more on the matching between components.
Tom
Hello Tom
I've noted that there was a bracket revision on the 686 recently. It lit looks to be a nice simplification. Are there any circuit changes that you might have implemented with the latest batch of boards and its effect on performance?
Will there be a notification on when the sale price of the assembled 686 will end? Perhaps some kind of notification might sway some who might have been sitting on the fence.
One final question. I have a couple of spare 120V primary 500VA toroids with each one outputting 35+35 VAC secondaries. I was thinking that I could connect each primary 120V coil in series and thereby have 18 +18+18+18 VAC leads that could power the 2x 686. Do you foresee any problems connecting it this way? I'm thinking the current demand on once channel would modulate the current in the two transformers and cause cross channel issues.
If that is possible, what estimates of output power could one expect and I imagine the driving low 2ohm impedance loads would be a cakewalk with 25-26 VDC rails.
I've noted that there was a bracket revision on the 686 recently. It lit looks to be a nice simplification. Are there any circuit changes that you might have implemented with the latest batch of boards and its effect on performance?
Will there be a notification on when the sale price of the assembled 686 will end? Perhaps some kind of notification might sway some who might have been sitting on the fence.
One final question. I have a couple of spare 120V primary 500VA toroids with each one outputting 35+35 VAC secondaries. I was thinking that I could connect each primary 120V coil in series and thereby have 18 +18+18+18 VAC leads that could power the 2x 686. Do you foresee any problems connecting it this way? I'm thinking the current demand on once channel would modulate the current in the two transformers and cause cross channel issues.
If that is possible, what estimates of output power could one expect and I imagine the driving low 2ohm impedance loads would be a cakewalk with 25-26 VDC rails.
Last edited:
Any consideration for Phoenix Contact PCB terminal blocks? I've had good experiences with their spring-cage design and prefer them to rising-cage whenever possible (at least where I can use terminal blocks). Then you don't need to worry about terminating any wires or figuring out the proper mating part.