Dear friends,
One of the major problems of headphones amps is not the performance at very high sound levels but the performance at very low volume settings. Especially when using a source with a high output level and sensitive, low impedance headphones mechanical potentiometers all can show very noticable channel imbalances.
The only solution is to use a discrete volume control with e.g. stepped attenuators, relays or digital potentiometers.
Good stepped attenuators are generally bulky, expensive, and normally are limited to 23 steps (24 levels) with around 2 dB stepsize.
A solution with relais, as used in the CORDA CLASSIC or CORDA SOUL, allows for more and finer steps but is also expensive and technically complex.
Digital potentiometers use a resistor chain and do have a large number of electronic switches in the signal chain. This limits quality of sound.
In the JAZZ I wanted to implement a discrete volume control with a decent number of steps (32 levels) using electronic switches. Implementation is such, that at all volume settings only one switch is directly in the signal path. For this a 32:1 multiplexer IC is used.
Unfortunately multiplexer IC's have limitations with respect to their supply voltages and as a result the output voltage of the first amplification stage of the JAZZ is limited.
It is very well possible to make the output of the JAZZ not be limited by the input stage. One only has to increase the gain factor of the output stage and lower the gain factor of the input stage. The overall gain can be kept the same.
However a very low gain factor in front and a high gain factor at the output is not the best constellation for optimum sound quality. It is better to give the output amplifier moderate gain only.
So I had to choose/compromise between high output capacity and higher quality of sound.
The compromise that I made guarantees that any normal headphone can be driven to any "normal" sound level without distortion. Sure, with very inefficient headphones like a K1000, a HE-6, or a Susvara the limitations can appear, but if you drive the JAZZ into distortion with a HD600 or a T1 then you better pay a visit to your ENT-doctor, because then there is something very wrong with your hearing or listening habits.
So the limitations of output power of the JAZZ are "on purpose". Not a design flaw but a reasonable compromise towards quality of sound.
Over the last few years almost 500 JAZZ amplifiers have been sold and despite a 2 weeks return policy only a very few amplifiers have ever been returned and never any complaints have been received about output power limitations. To be honest, that tells me more than any measurement result.
And now some more comments to the comments!
"The volume control is achieved by changing the parallel feedback resistor between VolCtrl_A and VolCtrl_B."
Confirmed!
> I don't think it is a clever design as it will change the gain of the first 2 OPAMPs during volume adjustment.
??
That's exactly the purpose of the volume control!
" The whole 'active ground' theory seems like BS and complicates the output stage without any proven benefits. "
10..15 Years ago, there were quite a few discussions at the internet about the benefits of adding an active ground to headphone amplifiers. A rather well known example is the DIY M3 amp design.
At that time balanced headphones were not common yet and I started thinking of a method to pass some of the advantages of a balanced design into a single ended output. The result is the concept of "Active balanced ground".
I then build an amplifier that allowed me to switch its outputs between single ended, active ground, and active balanced ground operation.
Testing by ear revealed only small differences between single ended and active ground operation.
However, active balanced ground showed clear sonic benefits. Music came with more ease and power. Sound became more effordless.
So for me the benefits of active balanced ground are well proven and therefore both JAZZ and CLASSIC use this concept.
No need to call something BS if you haven't tested/heard yourself!
"I have no idea why he chose such high feedback values."
The JAZZ uses inverting amplification stages. The advantage is, that there is no voltage swing at the inputs of the opamp. The non-linear input capacities do have no effect on signal integrity. Generally inverting opamp stages do sound better than non-inverting stages.
The input impedance of the amp is the resistor value between input and first opamp. You don't want it to be small. The gain factor determines the feedback value. Thus higher feedback values result from the concept.
In this context it should be noted that some opamps do react adversely to low feedback resistor values, even with the gain factors kept the same. Often higher resistor values do sound better. But of course they shouldn't be extremely high because of thermal noise and other problems.
"Strange circuit. Probably designed just to be different than the rest ?"
The natural crossfeed filter was not designed just to be different but for many people is a real asset.
Volume control by variable gain does have major advantages over conventional solutions (less noise, less distortion).
Active balanced ground, as explained on my website, not only has theoretical advantages but the differences to "normal" single ended solutions really can be heard.
And the advantages of the FF-implementation also has been confirmed by many people (also in double blind tests).
Yes, I do many things differently, but not for the sake of just being different. If we would stick with old recipies all the time no real progress would ever be made.
"Welcome to the forum Jan. And thank you for detailed explanation."
I'm glad you allow me to add my comments.
"To my knowledge ff is hard to measure. "
"Why is it hard to measure? And if so, how was the design verified?"
"If you can't measure it, most likely it is snake oil."
Design of the FF-circuitry was initiated by my observation that amplification stages with higher amplification factors did have a more dynamic bass whereas lower amplification factors resulted in better, more resolving treble. So I simply made an amplifier with frequency depending gain factor and a correcting network in front. And it worked. Of course some fine tuning was needed but the initial listening tests were already very promising.
Measurements however, did not show any major differences in distortion levels or SNR. Actually, with the FF-technique the latter is a little bit worse.
Now, I don't have the very best measurement equipment but distortions less then 0.01% should be inaudible anyway.
I even built a setup with 8 amplifications stages in line in order to "amplify" distortion and thus make the differences between standard and ff-implementation more visible. However, I did not succeed. Distortion levels were still too small for my setup.
"If you can't measure it, most likely it is snake oil."
Or you measure the wrong parameter!
There is much more to quality of sound than distortion numbers, frequency responses and SNR.
My feeling is, that the sonic advantages of the FF-technology are not explained by differences in distortion or SNR. It's more about timing-
But how do you measure timing, silkyness, effordless, .......?
You can have endless discussions on that!!
" The used logo looks like 'China Export' logo."
To be honest, before this discussion I've never heard of the "China Export" logo before. It is not an issue over here in Europe. But in the US situation probably is different as over there the CE-logo is nothing official so any "copy" like the "China Export" logo is not forbidden.
" Rf appears to be resistor 'ladder ?'
No, resistors are placed in parallel.
" Agreed that there seems to be unused potential in the output stage. It could be a 'design' philosophy from Jan that the output stage should never come anywhere near clipping level. Could also be that he did not find it was necessary to have huge output powers available that would never be used."
As explained above the latter is the case.
"Given the fact that most opamps he uses perform best at higher power supply rails and is of the opinion that opamps must be 'pulled' into class-A makes me believe that he believes that subjective evaluations and theories are more important than technical limits."
The JAZZ uses +/- 18V powerrails, which is the maximum allowed by the components used. Higher voltages do result in better sound (although the differences to 15V supply rails are relatively small.
And yes, pulling the output stage of the opamp into class-A does, to my ears, improve sound quality. To be honest, I've never compared distortion numbers with and without though. But differences can be heard.
" Given that the specs Jan provides seem to be incorrect and belonged to another amp says to me that he does not find it an important spec. "
Well, the spec is important, but I mixed up with the JAZZ-FF. The FF-technology allows for much higher overall output levels. As an example some measurements for this amp:
http://www.meier-audio.homepage.t-online.de/bilder/F0000TEK.BMP
At 100 Hz the JAZZ-FF outputs a maximum of 11.5 Vrms .
Granted, at higher frequencies the maximum output level decreases but at 400 Hz we still have 6.0 Vrms.
http://www.meier-audio.homepage.t-online.de/bilder/F0001TEK.BMP
So if you feel the JAZZ is too limited for your personal purposes, the JAZZ-FF may well be worth a consideration. Depending on the music signal you have 6..8 dB more headroom.
That's 4..6 times the power of the regular JAZZ.
Cheers
Jan