♫♪ My little cheap Circlophone© ♫♪

[danielwritesbac]

@terranigma Thanks man. I think that Elvee's Circlophone is very fun amplifier, so easily tunable and then there's the amazing dynamics. Have you tried optional gain settings? Aren't the dynamics just huge that way? I think it sounds a lot like a real concert.

@Ken Your soft clip device is amazing! Has anyone else tried it? Wow, enormous soundfield size boost! Although it doesn't seem to work as expected, I do really like it. Thank you!!!

[kenpeter]

Other than expected in what way?

[terranigma]

Originally Posted by danielwritesbac

@terranigma Thanks man. I think that Elvee's Circlophone is very fun amplifier, so easily tunable and then there's the amazing dynamics. Have you tried optional gain settings? Aren't the dynamics just huge that way? I think it sounds a lot like a real concert.

Agree Daniel, That's why circlophone became my prior diy occupation. Want to try different drivers? power transistors? bias servo transistors?.. no struggle to find any pnp/npn complementaries.. just read specs, put it if fits, then listen... This is really fun. Altering caps, pcbs even resistors is also fun. I didn't alter nfb resistor value yet but it is one of things in trial list also. Circlophone seemed to me a template of an ideally specified elegant amplifier rather than something immutable.

[danielwritesbac]

Originally Posted by kenpeter

Other than expected in what way?

Acutance, with the dramatic high resolution, high fidelity benefits plus the unfortunate apparent frequency response caveats that can come with it if not precisely calibrated. It is a similar topic to anti-compensation, feed-forward, r-comp, and square wave test. The unfortunate part happens when the acutance halo size, aka duration, is set too large. If we can't dial it back, the feature is self defeating after leveling apparent response via the power circuit, in that two wrongs won't make a right. If we can shrink the halo size (reduce noise), aka decrease duration, then some mild power circuit tuning that does very little, can recoup the apparent frequency response. . . resulting in a very top class high fidelity amplifier.

Right now, the effect is a "toe over the line" into noise. Let's try to adjust this effect in some way.

In print the topic is that you must have a tighter focus to make a larger print, and the larger the print, the tighter focus you must have to avoid amplifying blur (falls apart is the print term for when amplifying the size causes suddenly too much blur). Magnifying the print size is amplification, as for example a 4 foot photo goes beyond the means of a 35mm negative. Precisely sized acutance removes the blur and makes a sharp overlarge print possible. The concept also works for audio. Of course it is more complex than that, but it is surprising that your simple soft clip device has some control over this necessary aspect of amplification.

Yes, you pushed the sharpen button quite well, but now we must set the gate size smaller so the edges aren't exaggerated. The amplitude seems perfect but the duration an overdo. How to adjust only that?

[Elvee]

Originally Posted by danielwritesbac

there are some alternate parts listed as attractively as possible. I'd like suggestions on making this look even better and see if the parts are optimal selections.

Although it seems that parts availability problems have been defeated (awaiting your approval of the parts listing)

I missed something important yesterday: 1N5819 is not a suitable type for D7, it is a power diode with a very low Vf at low currents, and there is a very real risk it clips too early, fooling the bias servo into believing the current in Q10 is too low.
This will have disastrous consequences, as the servo will send the maximum current into both OP devices, destroying them in a matter of milliseconds.
Only signal schottky's should be used there, with a maximum current rating of 300mA or less.

[danielwritesbac]

Originally Posted by Elvee

I missed something important yesterday: 1N5819 is not a suitable type for D7, it is a power diode with a very low Vf at low currents, and there is a very real risk it clips too early, fooling the bias servo into believing the current in Q10 is too low.
This will have disastrous consequences, as the servo will send the maximum current into both OP devices, destroying them in a matter of milliseconds.
Only signal schottky's should be used there, with a maximum current rating of 300mA or less.

1N5819 starts at 0.12v (onsemi's most popular signal schottky)
That dodgy little thing will blow up my output devices? Oh no!
BAT86 starts at 0.12v (static resistant, Vishay and NXP)
BAT85 starts at 0.09v (low availability)
BAT83 starts at 0.09v (low availability)

That all seems possibly accident prone? Perhaps an ordinary 1n914 is a safer choice than a schottky? So, what part is the safest/best choice for D7? I don't know, but had to ask.

[Elvee]

Originally Posted by danielwritesbac

1N5819 starts at 0.12v (onsemi's most popular signal schottky)

The 1N5819 is NOT a signal schottky, it is categorized as schottky rectifier.

That dodgy little thing will blow up my output devices? Oh no!
BAT86 starts at 0.12v (static resistant, Vishay and NXP)
BAT85 starts at 0.09v (low availability)
BAT83 starts at 0.09v (low availability)

Where do you see that diodes "start" somewhere? They have a logarithmic voltage/current characteristic.
Here is a comparison between various types for a current swept between 0 and 3mA.
At 1.5mA, the Vf should not be lower than 250mV. The actual danger zone begins at 220mV.
Diodes like the BAT54 (300mA) just fit. The 1N5819 is completely on the wrong side of the fence

That all seems possibly accident prone? Perhaps an ordinary 1n914 is a safer choice than a schottky? So, what part is the safest/best choice for D7? I don't know, but had to ask.

You should not try to second guess all the design decisions, there are excellent reasons for not using a 1N914, as there there are equally good reasons for not using power schottky's

[danielwritesbac]

Originally Posted by Elvee

You should not try to second guess all the design decisions, there are excellent reasons for not using a 1N914, as there there are equally good reasons for not using power schottky's

Well, I apologize. I didn't want to second guess or guess at all; however, the BAT diodes are in short supply except for BAT86, BAT85, BAT54 and a few similar. So, that topic is about parts availability and avoiding the hindrance of discontinuances or the expense of rarities. Also, BAT54, although highly available, the SMD part doesn't fit through hole board easily.

Originally Posted by Elvee

That's OK, the value in KΩ ~= 0.9 times the total AC supply.

??? I could not make that work in all conditions, for all transformers, with all power supplies, in all locations, so I changed the constructor's chart to real DC measures because managing transformer variety was too far beyond the scope of a little chart.

How do I calculate R21 from DC? I've used your guidelines posted here: ♫♪ My little cheap Circlophone© ♫♪ for some of the examples and then attempted to extrapolate the rest.

So, here is the chart by itself (temporarily large), along with a new question of how much DC voltage can we use when specifically 8 ohm speakers?
Apparently a fun 75 watt version of the original has appeared with the MJ15015 being the higher voltage version of 2N3055. But, for some of the higher voltage selections I worry about overheating some of Circlophone's resistors. Which chart selections aren't suitable?

[kenpeter]

Originally Posted by danielwritesbac

Acutance, with the dramatic high resolution, high fidelity benefits plus the unfortunate apparent frequency response caveats that can come with it if not precisely calibrated. It is a similar topic to anti-compensation, feed-forward, r-comp, and square wave test. The unfortunate part happens when the acutance halo size, aka duration, is set too large. If we can't dial it back, the feature is self defeating after leveling apparent response via the power circuit, in that two wrongs won't make a right. If we can shrink the halo size (reduce noise), aka decrease duration, then some mild power circuit tuning that does very little, can recoup the apparent frequency response. . . resulting in a very top class high fidelity amplifier.

Right now, the effect is a "toe over the line" into noise. Let's try to adjust this effect in some way.

In print the topic is that you must have a tighter focus to make a larger print, and the larger the print, the tighter focus you must have to avoid amplifying blur (falls apart is the print term for when amplifying the size causes suddenly too much blur). Magnifying the print size is amplification, as for example a 4 foot photo goes beyond the means of a 35mm negative. Precisely sized acutance removes the blur and makes a sharp overlarge print possible. The concept also works for audio. Of course it is more complex than that, but it is surprising that your simple soft clip device has some control over this necessary aspect of amplification.

Yes, you pushed the sharpen button quite well, but now we must set the gate size smaller so the edges aren't exaggerated. The amplitude seems perfect but the duration an overdo. How to adjust only that?

Bizarre rant makes no sense whatsoever, stealing my stolen thunder...

---------

Its simple:

A soft clip is reversible. Information is distorted. No information is lost.
If you went to enough trouble, the original signal can be reconstructed.
Maybe our brains do some of this processing intuitively.

Hard clipping is irreversible. Information is smashed flat against a limit.
The limit can be voltage or current, quantization, or any other arbitrary.
We only guess what was destroyed where the information used to exist.
Not an attenuated hint remains to help us fill in the blank.

Even at similar THD, the latter distortion sounds a lot worse (to me).

When we wish to play music with a larger dynamic range than that fits
our amplifier, we have to make choices how best to handle impossible.
Else the default choice will be be made for us, probably to no good.

[odysseybmx414]

Kenpeter, have a link for your soft clipping device? I'd like to take a look at it.