My "ExtremA" is Built. Whew.

[PB2]

lgreen the amp looks real nice by the way.

Just noticed that this person got fake Sankens from MCM:
Another high DC Adcom GFA-555

Anyone know of a reliable source?

[lgreen]

Great, something new to worry about -fake Sankens!

To provide an update- I am still waiting for someone else to build one of these so we can compare data. I have received a few emails off-line so I know people are working on them. I have also received emails off-line from others concerning interactions with and the personality of the original poster, but I'll not get into that here.

[Luke]

Originally Posted by SSassen

Hugh,



It does, I blame Lgreen's construction for it not behaving as many of the prototypes that were built (sorry!). As mentioned previously it will drive into class-B when it leaves the class-A setting, which is rather abrupt in terms of distortion.

This amplifier was designed and perfected using measuring equipment that far outclasses anything shown here. Particularly the AP System 2 was used to do all THD and stability testing as only that test system was capable of measuring THD that low. Ruthless testing and elimination was done to make sure this amplifier was unconditionally stable under all conditions. The fact that Lgreen's doesn't show this behaviour can in my opinion only be attributed to the 'shoe string' approach used in construction and him deviating from some of the design requirements.



I strongly disagree. The majority of this design is done by Bruno Putzeys, someone who has a long history of designing nothing but excellent performing amplifiers and other audio equipment. Bruno is someone that lives and breathes amplifier design like no one I've ever encountered, he is a prefectionist in every sense of the word. Stating that such a complex design should not be attempted is merely a sign of one's own incompetence and in no way a representation of Bruno's capabilities.

The ExtremA still is the best performing DIY class-A to date, but also one that is not for the novice builder. Lgreen has shown it can be built by the novice builder but I'd be surprised if it comes near the THD figures we measured on the prototypes as well as the final design presented in the published article.

Sorry if this reply rubs some folks the wrong way, but that's simply the way it is. Despite Lgreen's best efforts I wouldn't describe his ExtremA build as 'typical' for what the ExtremA is capable of.

Cheers,

Sander.

Sander, do you have pictures of how it should look, so as other builders can get it right?

[RSQ8P]

yeah right, images ...

[PB2]

GK posted this on another forum and I thought it might be of help:

I noticed that you resurrected the "ExtreemA" thread. I haven't the stamina to read through it all, but has anyone asked Lgreen what kind of preamp/source or signal generator he is using?

The reason I ask is that the amplifier uses shunt feedback. What this means is that if the source is disconnected from the input, the amplifiers input resistors are un-terminated, and the amplifier reverts to a closed loop gain of unity. Unless the amplifier is frequency compensated for unity closed loop gain (very unlikely) that will turn it into an instant oscillator.

Also, a lot of test bench audio signal generators have an output impedance of 600 ohms. The ExtreemA's input/feedback resistors are 1k.

600 ohms in series with that would result in a substantial closed loop gain reduction of 1.6
times. That could very well cause the oscillation that he is seeing.

He also mentioned:
There are a few things with that design that are a bit suspect; The cfp VAS buffers (making these with to-92 BC5** parts in my experience often makes an oscillator), no zobel or output inductor.

[lgreen]

Originally Posted by PB2

GK posted this on another forum and I thought it might be of help:

I noticed that you resurrected the "ExtreemA" thread. I haven't the stamina to read through it all, but has anyone asked Lgreen what kind of preamp/source or signal generator he is using?

The reason I ask is that the amplifier uses shunt feedback. What this means is that if the source is disconnected from the input, the amplifiers input resistors are un-terminated, and the amplifier reverts to a closed loop gain of unity. Unless the amplifier is frequency compensated for unity closed loop gain (very unlikely) that will turn it into an instant oscillator.

Also, a lot of test bench audio signal generators have an output impedance of 600 ohms. The ExtreemA's input/feedback resistors are 1k.

600 ohms in series with that would result in a substantial closed loop gain reduction of 1.6
times. That could very well cause the oscillation that he is seeing.

He also mentioned:
There are a few things with that design that are a bit suspect; The cfp VAS buffers (making these with to-92 BC5** parts in my experience often makes an oscillator), no zobel or output inductor.

To answer the questions-

The preamp is a Krell KRC-2 and source is Wadia 301 (through preamp or not, results are the same). Also tried with my X-CCS-BOSOZ (same results).

The test bench signal generator is a Tektronix FG-501A.

Is it thought a zobel would correct the issue? What zobel values should be tried? Would you put in two zobels to ground or one zobel from + to - output?

To summarize for everyone so you don't have to reread everthing, under normal circumstances the amp is fine, no oscillation with no input or short circuit or even any input other than one sufficient to cause the signal to approach the rail voltage. So unloaded there are no issues (despite the theory above). There are no issues unless you are very near clipping. As the rail voltages are neared an oscillation slowly forms.

[kipman725]

I wouldn't worry too much if the behavior is only at clipping as one shouldn't operate there most of the time!

As for getting distortion mesurments try using a computer! something like the creative SB24bit can be purchased for 15$ and has crazy low distortion:
Buy Creative Audigy SE | Creative Labs Online Store (UK)
Couple that with software like rightmark and a voltage divider you have a relativly easy to use high resolution mesurment card.