Schottky diode as RE?

[Mr Evil]

Possible reasons for not seeing small-signal Schottky diodes used much: They have lower availability and higher price than junction diodes. The main advantages (low forward voltage drop, lack of reverse recovery glitches) are of low importance. The downsides (high reverse leakage current, high capacitance) can be more important.

LEDs likely behave most similarly to normal junction diodes since they are also made of a PN junction, but that's something you would have to measure yourself, as datasheets won't tell you stuff like that.

Personally I like 1N4448 for small-signal use. It's the same as 1N4148 but lower capacitance.

[Eva]

I've never tested the turn-on and reverse recovery characteristics of LEDs

Diodes in audio circuits are usually employed to generate voltage drops and to compensate for Vbe drop of bipolar transistrors. Typical >1A 40V schottky diodes produce too small a voltage drop [<200mV for small currents] so they are of little use and hardly seen on circuits except when such a small voltage drop is required

For signal clamping applications the 1N4148 is superior to 1A Schottkys since it features much smaller capacitance

Also I tried a 1A 40V Schottky for Vbe compensation some time ago and I think it showed too low temperature coefficient to be practical

[MikeB]

Hi lumanauw !

I see a very big problem using diodes instead of Re, in normal EF-type
one outputbjt is typically reversebiased while the other is heavily
conducting. This mechanism is used to reduce crossoverdistortion,
so by using diodes you would kill this mechanism...
If you have a never switchoff topology, this is different of course !

Mike

[lumanauw]

Hi, James,

My book is 1st edition, maybe it don't include that chapter. What is he writing in essence that you can quote about schottky for RE?

Hi, EVA, Mr.Evil,

The reason I ask so much about diodes, because when I first think of this, I search for the most suitable diode. I came up with schottky type, but while searching, I read much about diode types, and I just realize I know so little about diode types.
Offcourse I won't put IN4148 or LEDS for this application here, and I asked the questions to know more about diode types.

I got another "laughable" question. In diodes like IN4148, we usually note Vdrop=0V6. What happens when Vdrop is only 0V1 or 0V3? Does it fully turnoff or does it passes very little current? The graphs I can see in the datasheet starts with 0V6.

The graphs for diodes and transistors are they reverseable? Like in condition of Vdrop=0V61 for one diode, in the graph the current is 1.5A. Could we force 5A current while the drop is still the same 0V61, or the drop needs to be more?

Also, how good or how bad schottky in comparison with ordinary diode (IN4002) because schottky do not experience reverse recovery phenomena?

Hi, Mike,

I see a very big problem using diodes instead of Re, in normal EF-type one outputbjt is typically reversebiased while the other is heavily conducting. This mechanism is used to reduce crossoverdistortion, so by using diodes you would kill this mechanism...
If you have a never switchoff topology, this is different of course !

Yup, that's it ! In the picture that I attach here : ......

Figure A shows what we usually do. For simplicity, I assume the VBE of transistors is one value, 0V6. In steady condition, where the bias current is 100mA passing thru 0R5 degeneration, so the VBE multiplier value will be (0V6+0V6+0R5x100mA+0R5x100mA+0V6+0V6) = 2V5. This 2V5 will be the same value in every condition, right? So point A will be sitting at +1V25 and point B will be sitting at -1V25 if we assume the output is attached to speaker who is negative is 0V. All transistors are on in this condition.

Figure B shows when there is quite some current passing thru the speaker (in frequencies when speaker's impedance drops very low, in uS time). The current that is needed for the speaker is 10A. So the drop in RE1=10Ax0R5=5V. That makes point A is sitting at +6V2. The VBEmultiplier only gives -2V5 steady difference towards pointB, so pointB will be sitting at +3V7 instead.

Lets see the output voltage node (assume 0V) towards +3V7 in pointB. TR3 and TR4 are PNP's. Their bases is more positive than their emitors, so in this condition TR3 and TR4 is heavily reversed OFF.

This is what happens in classB amp. While one side of transistor is turning ON, the opposite side will be turning OFF/reversed. This makes ON-OFF-ON-OFF-ON-OFF-ON-OFF-ON-OFF in both sides transistors.
The ON-OFF will generate high-order distortion, because abrupt state change from ON-OFF.

You can try simple experiment like figureC. Put ordinary diode (IN4002) in place of D1-D2, then play the same music.
You will hear the trebles will have "LESS" resolution (compared when you dont have D1-D2 in that position). I think this "LESS" treble resolution is the right sound, because in normal (more) resolution, what we actually hear is not a high resolution, but it is a high order distortion being feedback to differential, regenerate by output stage(+another high order distortion), re-feedback to differential, (+another high order distortion by output stage), re-feedback to differential, and so on-so on.

When you put the diodes, the ON-OFF is reduced, feedback signal that is received by front end differential has less of this high order distortion and so on, the result is "LESS" resolution highs, but actually it is less high order distortion.

This happens in the 3 stage topology class(A)B amp that we know all this time. This is caused by 3 factors :
1=Feedback (if you built non-feedback amp you won't hear it),
2=classB output stage (if you built full classA output stage, you also wont hear it)
3=VBE multiplier (if you can get with biasing technique that is sliding perfectly) you wont hear this also.

We all cannot get too far with no.1 and 2.
NP known about this, he came up with patent #3,995,228 to deal with no.3. He also came up with fig.C. (but now he seems cannot get away with classA )
Mark-Levinson also knows about no.3, they came up with "Dynamic bias" with their biggest amp, ML-33. Krell makes "sustained plateau".
They all tries to makes all transistors not starting cycle from deep OFF/reversed but rather from always-on condition, so the high order distoriton is less.

I think putting schottky for RE like this thread also will reduce it, and we can still uses classB output stage + feedback like we usually do.

[SkyChu]

James B. use Schottky diode with R like Fig.C in Ampzilla.

http://homelf.kimo.com.tw/skychutw/a...pzilla_sch.jpg

[Elso Kwak]

Originally posted by SkyChu
James B. use Schottky diode with R like Fig.C in Ampzilla.

http://homelf.kimo.com.tw/skychutw/a...pzilla_sch.jpg

Hi, And I remove them! Bass is way better, less boomy without them.
I also do not like a non-linear element in the signal path.
The Son of Ampzilla does not have the Schottky diodes.

[SkyChu]

SAE have same circuit and call "Hypersonic Class A " .

You can refer the SAE X-25A schematic .

http://www.wardsweb.org/audio/index_audio.html

[SkyChu]

Originally posted by Elso Kwak

Hi, And I remove them! Bass is way better, less boomy without them.
I also do not like a non-linear element in the signal path.
The Son of Ampzilla does not have the Schottky diodes.

I think you're right. Because I've run SPICE to simulate GAS AMPZiLLA circuit and found the schottky diode is very non-linear .

[nemestra]

Originally posted by lumanauw
Hi, James,

My book is 1st edition, maybe it don't include that chapter. What is he writing in essence that you can quote about schottky for RE?

There is some text about the use of diodes in the first edition. It was expanded (a little) in the 3rd. In the first edition see Chapter 5, Large-signal distortion (distortion 3a), especially Figure 5.13 and the text above it on page 108.

James

[lumanauw]

Hi, James,

I found it. It is on LSN (Large Signal) distortion section. He found that putting forward diode // with RE reduce distortion for impedances <4ohm (2.6ohm in the experiment).
But he don't mentioned schottky, just junction diode for the purpose.

If we use no RE at all, just attach forward schottky in that place, will it work?