Aleph-X builder's thread.

[flg]

1" or so of 2 twisted Teflon wire wrap wires will make a nice small Teflon cap. Cut to taste

[Eric]

William, Here are some pictures of the clipping behavior without the feedback caps in place.

1kHz clipping - no cap:
1kHz clip no cap.JPG

5kHz clipping - no cap:
5kHz clip no cap.JPG

10kHz clipping - no cap:
10kHz clip no cap.JPG

With the 10pF caps in place, clipping was a very neat and orderly event. The tops and bottoms of the sinewaves just flattened.

flg: Ha - I always feel like a chump when I realize I have paid for something I could have done myself... I twisted a few wires together and was quickly able to make something that measured 5pF, the trouble is that its value kept changing. Whatever I did, it wasn't stable over time. Thus, my feeling of being a chump quickly faded and I'm more than happy to wait for my ordered parts to arrive

[flg]

Yes with all the efforts to isolate sound and vibration from the equipment in your system this could also be an issue. It may be possible to rectify that situatuion with silicone goop???
Regarding flat tops to the clipped waveform, transistor circuits are famous for behaving like that. That would essentially be creating many odd harmonics. This would likely be considered a harsh sounding nasty event. More ideal would be something that rounded or compressed more gradually. It is said tubes do such things and they sound very good when used in amplifier stages.
Can't say much about the radical clipping waveform other than my stuff does that sometimes too Even below clipping but that is probably for different reasons. At the very high end usually. You want to be compensating or rolling off before you get into that stuff.

[Magura]

Just dip the twisted wires in superglue before you solder them in.
That takes care of the values being all over the place.


Magura

[Nelson Pass]

Originally Posted by Eric

William, Here are some pictures of the clipping behavior without the feedback caps in place.

You will be interested to compare the differential output with
the absolute.

[Eric]

My new assortment of caps arrived yesterday and I spent some time comparing behavior of the amp with 1pF, 3pf, and 5pF caps in the feedback loop and made some composite images of the resulting square waves.

First up, I compared bandwidth:
0pF bandwidth = 240kHz (posted earlier)
1pF bandwidth = 210kHz
3pF bandwidth = 140kHz
5pF bandwidth = 120kHz
10pF bandwidth = 80kHz (posted earlier)

The next comparison shows square waves for 1pF, 3pF, and 5pF at 10v peak-to-peak output level across a range of frequencies:
10vp-p pF Comparisons_sm.jpg

The square waves below show various RMS levels of output at 1kHz for these three caps:
1kHz 1pF 3pF 5pF values_sm.JPG

The square waves below show various RMS levels of output at 10kHz for these three caps:
10kHz comparison_sm.jpg


The 1pF cap provides the widest bandwidth, but at the cost of ringing from 5kHz and higher frequencies, so I'm thinking this is not enough capacitance. The 3pF cap provides better corners than the 5pF cap, but at the 5v, 10v, and 15v RMS output level it also exhibits some ringing. The 5pF caps shows squares that are little more rounded than the 3pF caps, but the 5pF caps seem to show better control of ringing behavior.

If I changed the function generator to produce sine waves instead of square waves, all of the above measurement points look clean and undistorted.

In my unenlightened view, the 5pF caps looks like the best option with the least ringing and bandwidth to 120kHz. Any other opinions of these curves?

Nelson: I'll explore the differential vs absolute output in greater detail next - thanks for the suggestion!

[wuffwaff]

Hi,

somehow both halves of the bridged amp seem to have different clipping behaviours.

William

[Eric]

William, I noticed the non-symmetric clipping, too. It seems most obvious in the 20v RMS traces in the last set of 10kHz waves. I'm not sure what is causing it. I'm wondering if the amp was not fully warmed up before I made these measurements. I did not check absolute DC offset before I began. I will double check this and see if the pattern persists.

If this is still present when the amp is fully warmed up, I am not sure what to check as I worked hard to make sure that both sides of the amp were well matched right down to each resistor...

This seems like a great time to follow up on Nelson's suggestion to check absolute vs. relative output. This will at least narrow things down to one side of the amp for me.

Eric

[Eric]

I made more measurements last night looking for the source of the non-symmetric clipping behavior. I think I found it: I was using the wrong setting on my scope - Ch2 had been inverted (told you I was a noob... ) After correcting the scope setting (like before, the amp has been fine all along), I got a nice trace like this one for symmetric clipping behavior:
clipping.JPG

Nelson, you are a genius for coming up with super-symmetry! Also, thanks for your hint - it led me to realize I had the scope set incorrectly (but you already knew that...). I was really surprised when I compared the absolute output of each half of the amp with the differential output of both sides together! Each side by itself is not too pretty in the image below, but when you put 'em together, the result is quite nice! What a great design!
output.JPG

An finally, another composite image of square waves generated at 7.0vRMS (10v p-p) at 1kHz, 10kHz, and 20kHz comparing capacitance values in the feedback loop:
7vRMS 4ohms square waves sm.JPG

I think I like the square wave produced by the 3pF value the best. 5pF is too rounded and the 0pF and 1pF have bigger ripples...

Eric

[Magura]

Time for some music


Magura