My little beast is getting ready..

[johnny1]

I can't resist... A few comments as the amp has some minor flaws.

1, as jaycee already said, a basestopper to q6 is mandatory, it's sad how many published schematics forgot this one. Leaving it out can destroy your speakers...Will check it out

2, Connect r7 to the other side of r8, this reduces DC-offset as the amps gets better balanced. You are right. This is the correct topology. Nevertheless, the amp has only 3mV ofsset

3, c8 must be a typo, you can't really put a permanent capacitive load to the amp ? This will ask for resonances in combination with L1. Well, it is not! This cap is usually in series with an 22 Ohm resistor
L1 = 7uH ? Very big ! Use a much smaller value here (<1uH) and place a Zobel on the other side of the coil, skip c8...
The actuall design had 6.8uH. I put 7uH cause i couldn't find Jensen coil with that value. With these values, the -3dB limit is at 70KHz with a 4 Ohm load. I know, it's quite low, haven't decided if this is the final design.

4, a 2 diode ccs is nearly the worst ccs possible, consider a zener or led here. (or replace r4 with another ccs)

Mike

Mike and jaycee,
I really apriciate your comments.
Please find my answers in red.


Before I decide to built this amp I had auditioned it (made by a friend 2 years ago) and like it very much.
The only flaw i found on the bench, is that the negative side clips 1 Volt earlier than the positive.

 

[johnny1]

Re: amplifier nice?

I asked for the layout to Johnny 1 but I don't need it, I have it
from the Ultra-D, thanks.

Glad you found it, cause i don't have it in electronic form.

 

[jmateus]

Ultra-D amplifier

Here are some transcriptions from the description of the amplifier:

"The four paralleled 1.5 ohm emitter (?) resistors for each CFB transistor pair are there to help to stabilise the quiescent current.
They also slightly improve the frequency response of the output stage by adding local current feedback."

"Note however, that there are no intrinsic means in the circuit for ensuring even current sharing between Q13 and Q114 and between Q15 and Q16. What current current sharing there is will depend on the inherent matching (or lack of it) between the transistors"

And finally:

"By the way, we did try the effect of small emitter resistors for each of the power transistors but these had the effect of worsening the distortion performance - so we left them out. Note that the current
and power ratings of the output transistors are such that even if the current sharing is quite poor, it won't cause any problems."

There are some other considerations about the output coil and capacitor serving as a load, but I think they are of no interest. We
all agree (I think) that the resistor in series with the cap will be
the way to go. As per the size of the coil, that depends of the taste.
There are some discrepancies about the assertions on this description one being calling the 1.5ohm resistors "emitter" resistors
and then stating they tried emitter resistors but they worsened
the frequency response. A bit odd, don't you think?

In any case, if I decided to build this amp, I'm not sure if I should
consider some modifications proposed by this thread.

I mentioned these as a way to cause some reaction and clarify
what could be the modifications to implement.

One thing that comes to mind is to invert the position of the
outputs, placing the NPN's on the top part of the amp and the
PNP's on the lower part. This way they will have the emitter
resistors (4x 1.5 ohm) tyied to the emitters...Would this be a valid
way to do it? I think so.....

 

[ilimzn]

Re: Ultra-D amplifier

[/quote]
There are some other considerations about the output coil and capacitor serving as a load, but I think they are of no interest. We all agree (I think) that the resistor in series with the cap will be the way to go.

Definitely. A relatively low resistance (10 ohms or less) will do fine here.

As per the size of the coil, that depends of the taste.

Well, not only on taste - a large L will limit the slew rate for low impedance loads - in fact, it is even possible it gets as low enough to affect the audio band. This is often missed with current dumping configurations like Quad 405 and similar!

There are some discrepancies about the assertions on this description one being calling the 1.5ohm resistors "emitter" resistors and then stating they tried emitter resistors but they worsened the frequency response. A bit odd, don't you think?

No, no discrepancy - excpt they mentioned it worsened distortion.
A CFP behaves like a BJT polarised according to the driver (first) transistor in the CFP, with much higher gain. In that sense, the emitter of the 'driver' is also the collective emitter as far as the circuit is concerned, regardless of the fact that it is acyually the collector of the output transistor that connects to that point. Hence, the term 'emitter resistor' is correct.
The second point could be debated. On one side, current sharing between multi-emitter (RET, LAPT etc) transistors is unexpectedly good even without outside emitter resistors, compared to regular paralleled transistors only. This is because these transistor topologies have a sort of built-in emitter resistor. This also means that you can get very good current sharing by using relatively small emitter resistors (0.1 ohm). In theory, this would reduce OLG of the CFP since it is emitter degeneration, but improve linearity, so it is difficult to use a blanket statement that it would increase distortion. WHat does change, is biasing: such a CFP will need a different bias current for minimum distortion, compared to the one without emitter resistors in the outputs.
Keep in mind, however, that this does NOT mean the collective CFP 'emitter' resistors can be removed!!! (the 4x1.5 ohms). From outside of the CFP, the emitters of the output transistors are 'invisible', hidden in the l100% ocal NFB loop of the CFP itself.

One thing that comes to mind is to invert the position of the outputs, placing the NPN's on the top part of the amp and the PNP's on the lower part. This way they will have the emitter resistors (4x 1.5 ohm) tyied to the emitters...Would this be a valid way to do it? I think so.....

If by that you think of modifying the output from CFB pairs to EF 'darlington' (which of course means a different topolofgy, not just switching BJTs around, as the latter will not work!), it is certainly possible. That being said, the bias servo transistor then needs to be mounted on the heatsink of the output transistrs, whereas with the CFP it does NOT, but needs to be monted on a common heatsink with the drivers - unless the drivers and outputs are on a common heatsink. You may also need to modify the values of the resistors in the Vbe multiplier around the bias servo transistor, because CFP needs 2xVbe plus voltage drop on the Re's, while EF needs 4xVbe plus drop on Re's.

 

[jaycee]

The CFP while good for an amp that only needs a single pair of output transistors, is a nightmare when you need more than one. I much prefer the triple darlington in that situation.

The usual output LC is 10 ohms 2W, and then wind your inductor around the body of the resistor with enamelled wire. I don't know what inductance that is exactly (around 2uH i think) but it's worked well enough for me.

C8 on it's own has to be wrong - it looks like it should have formed a Zobel with a 10 ohm 1W resistor in series. While you see a lot of amps where the Zobel is before the LC network, you do see them after it too (usually on the speaker terminals, as recommended by Leach).

(edit - i read that you worked that out.. need more sleep..)

This schematic reminds me of an Elektor amp.. I can't recall the name of it at the moment, but it seemed very similar.

 

[johnny1]

You were right about Q6. It saturates when the amp is close to clipping.
A base stoper helped it.

After some simulations, I'm considering to run the earlier stages (up to the Vbe Multip.) with a different supply with +5V higher rails (around 58 to 60Volts).
What do you think?

 

[jaycee]

If this was a MOSFET amp it would be worth it. For BJT's it's not really worth it, unless you were planning to use a seperated, regulated voltage. I personally wouldn't bother.

 

[johnny1]

A separate 200VA toroid with 2x10.000uF per channel is what i have in mind.......

I believe that regulation is going to "slow down" the amp