♫♪ My little cheap Circlophone© ♫♪

[Elvee]

Originally Posted by danielwritesbac

Can MR types be used for this?

Yes, practically any fast 1A diode will do.

[abetir]

terranigma/daniel,

It's weekend here, I'll get back on my build again and see what went wrong, thank you for the help.

terranigma, I can't do measurements because D669a heats up fast, in a few seconds I figure it will explode in smoke...


I will post updates again soon.



Sir Elvee, thank you for the updates UF series is much more easier to find here than BATs, I only found one local supplier that sells BAT85. Can 1N5818-22 qualify too?

Regards!

[danielwritesbac]

Originally Posted by abeteer

1N5818-22 qualify too?

No. The 1n5818 datasheet says rectifier type schottky with a highly variable switch on point. If at D7, those will nuke your nards. And Elvee said:

Originally Posted by Elvee

. . .Do not use schottkies!!!. . .

1N5818-22 Rectifier schottky is not a signal type.
1N5818-22 will misinform/misfire the sensor and that is not pleasant.

[danielwritesbac]

1N4148 at D6, plus BAT86 at D7, sums up to 1V. <<-most durable example
1N4148 at D6, plus BAT85 at D7, sums up to 1V.
FR at D6, plus FR at D7, sums up to 1V (check datasheet graphs to confirm).
MR at D6, plus MR at D7, sums up to 1V (check datasheet graphs to confirm).
BY at D6, plus BY at D7, sums up to 1V (check datasheet graphs to confirm).
UF at D6, plus UF at D7, sums up to 1V (check datasheet graphs to confirm).

So, these examples are approximately 1V, at low current, in most temperature conditions.

P.S.
Conversely, 1n4148 at D6, plus 1N5819 may engage at 0.7v and cause a sensor misfire that would blow the output devices eventually, especially if the diodes are warm. And, 1N5819 at D6 plus 1N5819 at D7 will blow the output devices immediately. The 1n5819 and similar rectifier schottky have a highly variable switch on point that is very, very low voltage and thus not suitable for either D6 or D7, since random operation of the sensor wouldn't be useful.

[danielwritesbac]

It seems the first thing to do for going forward with the CFP. . . is to discover the maximum practical value for C5 (The standard schematic's "RF Filter" of 330p). Some people may experience rolled off/rounded treble with 3n3. I also tried 2n7, which was borderline. 2n2 looks decent. But, don't take my word for it--this needs tested because filtering depends on your source and your potentiometer. When using the 10k potentiometer, the input loss varies and that alters the filtering. SO, the first thing to do is try to find the biggest RF filter cap you can that won't dull your treble. Please help research this?

P.S.
The following links will expire/change in time: It was possible to make Standard Circlophone and visually matching CFP Circlophone schematics, and, caution, this CFP schematic is possibly not ready for prime time (the values are probably not right)--It needs proofread, simmed, checked out. Help?

[Elvee]

Originally Posted by danielwritesbac

It seems the first thing to do for going forward with the CFP. . . is to discover the maximum practical value for C5 (The standard schematic's "RF Filter" of 330p). Some people may experience rolled off/rounded treble with 3n3. I also tried 2n7, which was borderline. 2n2 looks decent. But, don't take my word for it--this needs tested because filtering depends on your source and your potentiometer. When using the 10k potentiometer, the input loss varies and that alters the filtering. SO, the first thing to do is try to find the biggest RF filter cap you can that won't dull your treble. Please help research this?

You can also keep the original 330p and increase R19 (on your schematic) to 4.7K.

The impedance is high enough to allow for it without inconvenient.
P.S.
The following links will expire/change in time: It was possible to make Standard Circlophone and visually matching CFP Circlophone schematics, and, caution, this CFP schematic is possibly not ready for prime time (the values are probably not right)--It needs proofread, simmed, checked out. Help?

Some corrections:
C6=120p
R23=1K8
R2=47
C4=3p3
C2=390p
C12=270p

[danielwritesbac]

Are these good?

CFP Circlophone
(click image to enlarge)



Standard Circlophone
(click image to enlarge)

See how they match?

[keantoken]

I am not having any overshoot issues in simulation with the CFP version. At first I was skeptical it could work with just 45uA input pair current, as the Ft at this current should theoretically be around 8MHz. But, the Ft spec in many situations isn't very useful. Just to be sure, I've increased the input pair current to over 500uA.

This version gets -3db at 5MHz. I don't see why any feedback capacitor is necessary.

[Elvee]

Originally Posted by danielwritesbac

Are these good?


See how they match?

Looks OK, but let us examine Keantoken's contribution before going further.

Originally Posted by keantoken

I am not having any overshoot issues in simulation with the CFP version.
........
This version gets -3db at 5MHz. I don't see why any feedback capacitor is necessary.

Yes, but the compensation components you used are very different.

This compensation strategy looks much more agressive, but if it doesn't harm the performances, it could be the way to go.
A fresh look from an outsider often brings interesting new things, and I welcome it.
I was a bit stuck with a somewhat unsatisfactory scheme, and if the issues could be solved that simply, that would be nice.
Have you looked at the slew-rates?

[keantoken]

All I ever used to get compensation values was trial and error, informed by bode plots.

Here is the slew rate and an updated schematic. Redtrace is with no load, orange trace is with 6R load.

The feedback employs a capacitive divider that doesn't add any poles to the response and makes the feedback network good to the tens of MHz. This simplifies finding the right compensation values.