A Transconductance Amplifier

[Duo]

To start, I'm not trying to steal Mr. Pass's idea for a transconductance amplifier. Of course, my implementation is far different from his to begin with.

I've begun developing a transconductance amplifier for my own personal use and thought I'd like to talk about it.

I'm using a three-stage design with error amplifier, so this does have feedback, and is more like a typical three stage voltage amp than anything. The major difference is that the feedback loop is now with a speaker in place of the feedback resistor, so that it senses current drawn through the speaker.

For the prototype, I'm using an op-amp to drive a push-pull output stage. This is done through an emitter follower which is loaded by a current source to drive the output transistors. Nothing new here.

My desire to do this was merely out of wanting to see how a transconductance amp acted. I'd never used one for driving loudspeakers before.

To say the least, the results are absolutely astonishing. As I have studied, the effects of a lot of internal speaker limitations are more or less taken care of. It really is something else.

If anyone would like to see what I've done exactly, I suppose I could post a schematic.

[fab]

Originally posted by Duo
...I've begun developing a transconductance amplifier for my own personal use and thought I'd like to talk about it.
...If anyone would like to see what I've done exactly, I suppose I could post a schematic.

I suppose I would like to see it...

Do you have any test waveform results?

[latala]

I suspect this will have been done before i can think of a few examples but i would like to see a schematic values dont have to be shown!
Then i will be prepared to pass opinions / comments
regards trev

[lineup]

Originally posted by Duo

The major difference is
that the feedback loop is now with a speaker in place of the feedback resistor,
so that it senses current drawn through the speaker.
.

About 15 years ago, in the beginning of the 90-ies, I think,
Elektor had such an amplifier.
I could find this article and post a copy of schematic.
I have all them old Elektor magazines stored here.

The basic and rather simple circuit can be seen in my attachment.
I have drawn this schematic from memory.
It shows the Current Feedback princip.

The Gain in this case is: LSP 8 Ohm / 1 Ohm = 8


Regards
lineup

[hacknet]

ive done that with an lm3875..

the results were pretty good, super bass but i found the mids to be extremely hard and the there seemed to be a rising highs, i think it is because of the rising impedance curve of the single driver.

all it required abit of a filter and it was pretty nice.

[hacknet]

there is a more efficient way by having 2 feedback loops and the sense resistor at the output, that way you wont have a floating output problem.

[lineup]

Originally posted by hacknet
there is a more efficient way by having 2 feedback loops and the sense resistor at the output, that way you wont have a floating output problem.

Yes,
this is probably the way to do it, hacknet.

My attachment is not a complete design circuit.
As I told, it is a simplified diagram - to show the basic princip.

Actually this is what I would call A True Current Feedback Amplifier.
CFA - In the correct sense.

Because gives an output current proportional to input voltage.


One thing about woofers/speakers to use with this.
You may have to rolloff the rising impedance of the speaker.
Using some impedance equalizing filter.
Similar in effect to the Zobel Network.

Say the impedance would rise to 30 Ohm in higher freq.
The gain would be also 30!!


Regards, lineup

[kelticwizard]

Originally posted by lineup


One thing about woofers/speakers to use with this.
You may have to rolloff the rising impedance of the speaker.
Using some impedance equalizing filter.
Similar in effect to the Zobel Network.

Say the impedance would rise to 30 Ohm in higher freq.
The gain would be also 30!!


Regards, lineup

Don't know that much about amps, but I think I might have caught a small error here.

If the normal impedance of the speaker is 8 ohms, and it gradually rises to 30 ohms, wouldn't the gain be 30 / 8 = 3.75?

Which would still be considerable, and very hearable, but not as high as if the gain were 30.

[lineup]

Originally posted by kelticwizard


If the normal impedance of the speaker is 8 ohms, and it gradually rises to 30 ohms, wouldn't the gain be 30 / 8 = 3.75?

Which would still be considerable, and very hearable, but not as high as if the gain were 30.

See my attachment:
http://www.diyaudio.com/forums/attac...amp=1170003209

The gain is: LSP-Z / R3
R3 is in this case 1 Ohm.
So, Nominal gain is: 8/1 = 8

At Z=30 Ohm ...... 30/1 = 30


This is why we need to watch out, that speaker impedance stays within a reasonable level.
Nelson Pass, for his First Watt amplifiers, suggested simply a resistor in parallel with the Loudspeaker.
It will work.
--------------------------------------------------------------------

But I prefer use a Impedance Equalizing Network:



Le is the inductance of your driver's voice-coil (in henries).
Re is the D.C. resistance of the voice-coil (ohms).

The design equations for this type of filter are:
Capacitor = Le/Rc2
Resistor (Rc) = 1.25 X Re
Websource:
Zobel Loudspeaker Impedance Correction Circuit

[latala]

I have seen something like this in the 1970,s I think a company called servosound did this allthough in that case it was a mixture of both voltage and current feedback !
I have also seen H+H amplifiers use a degree of current feedback with very good results
Rod Elliot The Ausie has an article that relates to this on his web site !
Always worth a visit
Just remember when dealing with the amplifier spk combination a moving coil speaker is not only a varying impedance but can also be a generator ie a negative impedance such as at times when the cone/coil overshoots and for this reason I believe your ampifier may excert better control
regards Trev