Krill - The little amp that might...

[GK]

Hey Glen,
Are we not getting an idea though? The flat area?

The "flat area" has to be shown to be zero volts to demonstrate non switching. You need to acctrately look at the voltage developed across the added resistor down to a few tens of mV.

My function generator is not battery powered (grounded IEC). I don't have an isolation transformer for my scope either.

Well you're stuffed then

 

[MJL21193]

The "flat area" has to be shown to be zero volts to demonstrate non switching. You need to acctrately look at the voltage developed across the added resistor down to a few tens of mV.

See, here we have the stumbling block (from my perspective). How can the flat area be at zero volts when there is bias? The quiescent current has the outputs "on" and conducting, even without a signal. My understanding of all this is limited, admittedly, but I'm working with a definition of class A that says the output devices will conduct 100% of the signal, not 65%.
When the flat area appears, the other device is supplying that section of the signal, correct? That would be a loose meaning for switching, no?

 

[mannycc]

Hi Jkeny,

not, it would be gracious of those who banged on about this for so long now apologise to Steve...

Believe this will not happen, it's not natural human nature, if you were not made to and dont come to a realization people would think that the world is circling around them, and they are at the center of the universe. A classic example, when Adam and Eve were confronted by God for eating the forbidden fruit, Adam tried to blame God first then Eve for what he did, likewise Eve blamed the serpent for what she did, natural people always try to find ways to escape instead of saying sorry... same nature we inherited from the beginning.

Regards,
mannycc

 

[GK]

See, here we have the stumbling block (from my perspective). How can the flat area be at zero volts when there is bias? The quiescent current has the outputs "on" and conducting, even without a signal.

The quiescent current (bias) in a typical class B output stage is only there to ensure both P and N devices are conducting during the crosover region. Once the load current exceeds ~2*Iq the opposing output device gets cut off.

In a true class A output stage each transistor not only conducts for the full 360 degrees, but conducts signal current for the full 360 degrees as well - there is no "flat area".

Cheers,
Glen

 

[AndrewT]

but I'm working with a definition of class A that says the output devices will conduct 100% of the signal, not 65%.
When the flat area appears, the other device is supplying that section of the signal, correct? That would be a loose meaning for switching, no?

I think most of us are agreed that this is not a ClassA output stage.
The current discussion is whether the output stage is non-switching.
My reading of that ripple is a very high even order distortion has been detected. The flat does not look completely flat it seems to turn off and turn on very slowly.
This slow turn on/turn off eliminates most of the high frequency artifacts that are usually generated when the device stops/starts conducting abruptly.

If my reading of that curve/flat is correct, then the slow change at near turn off should be very audible and that would prove that SD's circuit is achieving a higher quality of ClassAB operation.

Who can do the diff measurement more accurately so that we can properly read what is happening?
Or find another test that helps us read the situation.

 

[GK]

The flat does not look completely flat it seems to turn off and turn on very slowly.
This slow turn on/turn off eliminates most of the high frequency artifacts that are usually generated when the device stops/starts conducting abruptly.

It actually looks like no better or worse than other properly working class B bipolar output stage.

 

[AndrewT]

Hi Glen,
can you post a couple of example scope views so that we can see what optimal ClassAB looks like?
And maybe a couple that are deliberately non optimal for comparison.

 

[jkeny]

Can we also get some more detail on MJL's "Krill facsimile" - schematic, devices used, CCS current, etc?

 

[MiiB]

PH 104

I have used 3 different pairs and even mixed transistor types without problems. The Krill output works with a voltage amplifier ranging from Steve Dunlap’s design, through an op amp (ostripper), a Borbely-designed front-end to a vacuum tube (the only variation among these 4 I have not personally built).

Would it be possible to ask wich combination you use in your 150 W version an also state the values for yor support resistors and trimmer values...I would like to build this.. and it would be nice with a good starting point...

I do PCB..layouts in eagle..and would not mind to post files...

michael

 

[MJL21193]

... same nature we inherited from the beginning.

Lecture over?


My Krill is shown by Steve in post #1460 - officially sanctioned by the man himself. Devices? I used the ones in the schematic (a bit of a waste to use all of those 2SA1837/4793 and the 2sC5200/1943 for drivers.)
Precisely as the schematic shows (barring front end and supply bypass / decoupling).

I even made a board (getting a layout here is like pulling teeth). For us real DIY guys (you know, the ones who actually build things), why not post a pdf of the board layout? Free design my eye.

 

[PH104]

originally posted by MiiB
Would it be possible to ask wich combination you use in your 150 W version an also state the values for yor support resistors and trimmer values...I would like to build this.. and it would be nice with a good starting point...

Hi Michael --
I'm away from home for a couple more days but will post resistor and trimmer values when I get back. I'll post that information in the Krill builder's thread, probably on Tuesday.

MJL -- Depending on how listening tests go, I might design an output board for a 150W - 200W amp. If so, I will post it.

Phil

 

[Steve Dunlap]

I even made a board (getting a layout here is like pulling teeth). For us real DIY guys (you know, the ones who actually build things), why not post a pdf of the board layout? Free design my eye.

All the info I have on the board is available to anyone that E-mails me and asks for it. I have mentioned this before and many people have been sent the info.

The original computer files are somewhere in storage. I do not have the ability to dig through the things that were put in storage for our last move.

A number of people have done layouts of their own. They have my permission to post them if the want to.

 

[MJL21193]

MJL -- Depending on how listening tests go, I might design an output board for a 150W - 200W amp. If so, I will post it.

Phil

Hi Phil,
Too late. I like doing everything on a real board, not the mess I had slung together before. I went ahead and did a quick layout of the output stage (board shown on the previous page).
This was just a learning experience - I have 150 - 200watts amps and amp deigns already.

 

[MJL21193]

Hi Steve,
Don't take it the wrong way - not meant as a slight on you. If it's a free design, board layout should be readily available. If it's not available, that's fine. Some references (photos) would be helpful for those of us that are willing to put a little effort.
It's not like many here are able (or willing) to make their own boards, so no threat to the boards you are selling. I don't buy boards (only one exception) so I'm not your market.

 

[bear]

Hi Steve,
Don't take it the wrong way - not meant as a slight on you. If it's a free design, board layout should be readily available

Should be? What's your theory on this??
Who is responsible for this?

If it's not available, that's fine. Some references (photos) would be helpful for those of us that are willing to put a little effort.
It's not like many here are able (or willing) to make their own boards, so no threat to the boards you are selling. I don't buy boards (only one exception) so I'm not your market.

You might want to post this request for a layout over in the builder's thread??

Also the output stage is fairly simple, so is there really a need for a layout for just the output stage?

MLJ, if you bothered to actually read the thread(?), Steve is not "in the market" to sell boards. Talking like this, because you have obviously not read much of the thread, is somewhat irritating and obnoxious. If you have/had read the thread, then you'd be simply rude and deliberately nasty.

Your complaints are not appropriate.

_-_-bear

 

[MJL21193]

_-_-bear [/B]

There is much I'm itching to say but that would get me banished. I'll bite my tongue.

 

[Steve Dunlap]

Hi Steve,
Don't take it the wrong way - not meant as a slight on you. If it's a free design, board layout should be readily available. If it's not available, that's fine. Some references (photos) would be helpful for those of us that are willing to put a little effort.
It's not like many here are able (or willing) to make their own boards, so no threat to the boards you are selling. I don't buy boards (only one exception) so I'm not your market.

Most of the response I have received has been from people that made their own boards. Many here are both willing and able.

There have been photos posted of bare boards and boards in the process of being stuffed. There have been photos of completed boards including one of the inside of one of my 100W amps. Every piece of information I have available to me at this time has been made available to anyone that has asked. I found print copies of the boards I have and have made those available. They show front and back copper and stuffing guide. I have provided a schematic and a BOM WITH the Mouser and Digi-Key part numbers. How much more do you want for free?

 

[MJL21193]

How much more do you want for free?

Is that how it looks to you? That I want stuff for free?

 

[MJL21193]

Enough nonsense. Onward.
What can be the cause of this?

Radical overshoot on the downstroke.
A magnifaction:

Here's the top magnified:

A 20k squarewave. Fairly low input (~5Vpp) - source is my FG. Yes, it gives a clean precise squarewave.

 

[PH104]

John -- What voltage gain stage are you using for your amp? I've seen something similar too, and always on the negative half of the square wave.

Bypassing right next to the drivers helps. I also found that a resistor between the voltage stage and output helped with other than Steve's voltage gain stage. I started at 1K and, for me, 4.7K worked great. I also want to see the effect of a small cap (say 10 - 100pF) and the junction of that resistor and the bases of the two input resistors to the output.