Member
Joined 2006
The fun in all of this for observers is that even the objectively built amps may not be preferred in a proper blind test versus a cheap Behringer. Subjective/objective - just go with whatever you prefer. It's all bollocks
You have a reference for that?
A linear electro magnetic motor is about as efficient as a rotating one, so this is not the cause of loudspeaker driver inefficiency. The main cause of this ineficiency is the impedance mismatch between cone and the air that is to be moved. In other words, a driver only converts part of the energy required to produce the cone movement into sound energy. If you improve the impedance matching between cone and medium, for example by applying a horn or making a coincident stack, you can go up to very high efficiencies. A large stack of horn loaded drivers can get somewhere around 50 percent iirc (certainly over 25%).
The back emf generated by a loudspeaker is quite considerable and it shows up as an increase of the driver's impedance. It is good that this happens, since the back emf starts to increase around the resonant frequency of the driver/enclosure comination, and it prevents this resonance from becoming destructive. The back EMF linearizes the cone movement around resonance.
As long as your drivers are voltage driven, there is no huge difference between the effects of back EMF in passive or active speakers, so you can discard it for your purposes.
Note that I decided to check your Blog and you just say there 'the sumiko OCOS report' with no quotes or links from that actual report. Googling OCOS suggests it is/was a coax speaker cable and nothing to do with interconnects!
Two blurred graphs with the y axis missing does not a credible input to design parameters make.
Actually, according to Sean Olive, Harman-Kardon laid off their golden ears because they didn't provide any meaningful data. He specifically mentioned a golden ear who had a talent for "voicing" speakers for the German market who was laid off. Basically, the end customers couldn't tell the difference between the speakers voiced for the German market and those voiced for the rest of the world, so the golden ear was laid off, the product line simplified, and production cost reduced.
I'm sure they still listen to their products. However, they base their design choices on DBLTs rather than the opinions of a few selected golden ears.
Makes good sense to me, actually... But then again, I'm one of those weird fact/science-based personalities.
~Tom
Last edited:
There is a discussion of the input impedance of the THAT1200 line receiver at the end of the IC's datasheet. Formula are given to allow self generating an impedance curve if that is needed. They recommend minimum 10uF for capacitor Cb to maintain minimum 1MOhm Zin at 50Hz where the primary low freq. PS noise threat exists. As you know from reading your design doc, Mod-86 uses significantly higher value than this, ensuring adequately high input impedance throughout the power band.
Mr. Soong,
I'm not sure why the input impedance of the amp is perceived as such a mystery. The impedance is set by the THAT1200 chip and associated components. The manufacturer's datasheet indicates differential input impedance it is typically 48k Ohms. Since that is set by resistors on an IC process, you likely can expect 15% tolerance. Given the topology, this is likely constant over the entire audio band and somewhat beyond.
There, one of you questions answered.
I'm not sure why the input impedance of the amp is perceived as such a mystery. The impedance is set by the THAT1200 chip and associated components. The manufacturer's datasheet indicates differential input impedance it is typically 48k Ohms. Since that is set by resistors on an IC process, you likely can expect 15% tolerance. Given the topology, this is likely constant over the entire audio band and somewhat beyond.
There, one of you questions answered.
Tom, did you try other input opamps?
the SMD type LME49990 PSRR of 144dB typical(!) not to expensive,
can this make your design even better?
The LT1028AC this one is super expensive.
Has a typical PSRR of 132dB vs 125dB LME49710.
I wonder if it has less distortion?
All in all you have a very clever design to 'isolate' the power supply,
without affecting power delivery
Thanks for the reply in post #1085
the SMD type LME49990 PSRR of 144dB typical(!) not to expensive,
can this make your design even better?
The LT1028AC this one is super expensive.
Has a typical PSRR of 132dB vs 125dB LME49710.
I wonder if it has less distortion?
All in all you have a very clever design to 'isolate' the power supply,
without affecting power delivery
Thanks for the reply in post #1085
No, I asked to see measurements. Did not think it to be a difficult think to do.
Well, I cannot say it is easy to find the right people. Normally the people I find best are the people whom understand music.Actually, according to Sean Olive, Harman-Kardon laid off their golden ears because they didn't provide any meaningful data. He specifically mentioned a golden ear who had a talent for "voicing" speakers for the German market who was laid off. Basically, the end customers couldn't tell the difference between the speakers voiced for the German market and those voiced for the rest of the world, so the golden ear was laid off, the product line simplified, and production cost reduced.
I'm sure they still listen to their products. However, they base their design choices on DBLTs rather than the opinions of a few selected golden ears.
Makes good sense to me, actually... But then again, I'm one of those weird fact/science-based personalities.
~Tom
The AP does not measure impedance directly. I'll have to set up a test jig to perform this test. Then I need to debug the test jig to ensure that it actually measures the input impedance. This is time consuming. I have other priorities right now. Besides, you've already said that my amp isn't up to your (unspecified) criteria, so what incentive do I have to spend a day delivering data that you obviously don't care about?
The THAT1200 is intended for use in professional audio gear. Its input impedance is 48 kOhm. If you have a look at the equivalent circuit in the data sheet, it is intuitively obvious that the input impedance is set by resistors. A generic BiCMOS process will offer an absolute tolerance on the resistors of about +/-15 %. If the process THAT uses offers a thin film resistor module, this tolerance could be significantly tighter.
The input impedance of the MOD86 and PAR86 is that of the THAT1200.
Right now, my first priority is to wrap up the documentation for the Parallel-86, complete the Parallel-86 web site, and hopefully generate some sales.
~Tom
Last edited:
Yes for some reason they actually developed speaker cables, which I did not feel was the best approach. But the report talked about interconnects. I have shown measurements of some interconnects I have sitting around (not individually labeled). The impedance is not flat. So for most people using unbalanced input configuration, it is not a good idea to have flat inputs.
If posting the report is allowed here, please let a moderator say so. Otherwise, people whom are interested can PM me. It should be easy for me to find it.
This question was discussed early on in this thread.
My requirements were:
- Low THD
- Low noise
- Highest AVOL at 20 kHz (--> high UGBW)
- Must support +/-15 V operation
- Must be available in DIP and reasonably affordable
I've looked a few times since I designed the MOD86 and the LME49710 is still the best candidate. Sure, you can find an op-amp that beats the LME49710 on one spec, but it will typically be considerably worse on another spec. Overall, the LME49710 is the best candidate for a composite amp that I can find.
The MOD86 and PAR86 are intended as a plug-n-play solution. Assemble according to the BOM, plug in, get world-class performance.
~Tom
I understand you have your set of priorities like anyone else. I missed the window to assemble the Mod 86, so it gets a backseat.