Half-Nelson Amp

[wrenchone]

At Burning Amp this year, I brought with me a Zen variant I christened the "Half-Nelson", as the skeleton started out as an early member of the Zen series, with some bows and ribbons added.

The schematic is shown below. The LM317 is configured as a current source to bias both the input P-channel follower and the output P-channel FET. The current source is shunted by an RC network to provide opposite drive on the P-channel output to increase the possible outpout current. Bias was set at about 1A, and the pot is used to adjust the output centering. The amp was auditioned at Burning Amp, and did ok with a pair of Metronome speakers using an 8" hemp driver.

I've recently spun a second board with some improvements. Hopefully Ill have a chance to populate it over the holidays and bring it up, but I thought it would be good tio present the original concept. Pictures of the amp will follow.

[Nelson Pass]

That's really great. Have you had the opportunity to pull some
numbers off of it?

[wrenchone]

It hasn't seen an analyzer as of yet, though I guess I could haul it in to work after the holidays and subject it to the Audio Precision analyzer there - despite its nice rep, not the most user friendly of instruments. In initial simulations, the amp looked pretty Zen-ish in character, with low levels of higher harmonics and a similar gradual rise in distortion with output power. One thing I was proud of (and relieved about) is that the quiescent current is stable, even though I didn't put the bias FET and the upper output device near each other. The extra buffering makes the amp pretty fast - square wave rise time was only a few microseconds, and perfectly damped.

I sort of held off on this guy for a while after Burning Amp, as I've been messing with my "Shrine" tube amp and looking into making some more sensitiive speakers than the ones I have now. Also in the works is repackaging my preamp to include a JFET follower active crossover for biamping (and improve the cosmetics). The RIAA preamp uses a single ended JFET topology with a few special features, the result of a year or two dinking around with simple single-ended circuits using Amurrrican FETs. I'll be starting a thread on that one when I get the time.

Any way, I'll take a couple of pictures of the amp tomorrow and post them. If I have enough energy to drag it down to the basement, I'll also take some pictures of the square wave response. Have a nice Christmas....

[wrenchone]

Attached is a photo of the Half Nelson piglet. Those who attended Burning Amp might have noticed it slouching on one of the side tables. The unusual manner of packaging was a matter of expedience, as I didn't have time to wrap any sheet metal around it before the big do. However, with a little tarting up, this kind of open packaging could be made to look pretty nice. Possible combinations: black anodized heatsink, input/power panels and toroid cover with oak base, or gold anodize with black piano finish base. If I can get the thing to perform to my satisfaction, maybe I'll get it some dress clothes

[wrenchone]

A side view, showing the ins and outs - nothing special, really.

[wrenchone]

My scope wouldn't save the waveforms (needs a new floppy drive), but output rise and fall times were ~ 1 usec.

[wrenchone]

For amusement and edification, I ran a set of simulations on the amp after tweaking the centering resistor values for higher impedance and gain. The JFET buffering allows the use of large values for centering resisors without affecting the bandwidth substantially. The first picture shows the altered schematic with quiescent bias values. Bias resistors were changed to reflect the on threshold voltages of the parts in the library. Gain is around 13X with the new values.

[wrenchone]

Attached are the simulated distortion values with 10mV input.

[wrenchone]

Here are the results for 100mV input.

[wrenchone]

Finally, results for 1V input.