JFET input, MOSFET VAS, LATERAL output = Perfect!!

[AndrewT]

FetZilla and discussion about it, is in the Manufacturers' section under AKSA

This Thread started as a discussion on an all FET non LTP input power amplifier.
Many versions have been proposed. The most talked about versions are different from what opened the Thread. A lot different.

I think both versions are valid topics in this Thread.

If one feels that their FetZilla post needs a more specific Thread then post on the AKSA FetZilla thread.

 

[mikelm]

Up until now this has been a friendly, creative thread where any design ideas can be presented and discussed - I think it would be a great pity to limit this broad ranging discussion within the bounds of one particular implementation.

I'm sure Hugh's amp sounds very nice indeed and i would love to hear it some time but it is only one possibility among many possibilities within the scope of the original idea presented here.

If people desire a much more limited discussion about a particular product surely that should be the topic for a new thread and I will follow it with great interest

cheers

mike

 

[Ian Finch]

Builders thread

This thread has run for the largest part of its 2,000 posts with both SWF and Mikelm's projects running side by side. I suggest there is only a problem for builders of SWF/AKSA Group who perhaps haven't already followed it closely.
Now that SWF has turned his attention to the next project, the focus naturally shifts to Mike, AndrewT and KTs different directions. I think there are others wanting to see how far these go and sound comparitively too.

My preference, if you like to keep things ordered for a specific project, is to start a thread entitled "Fetzilla Builders Thread" and head it up with a note explaining which Fetzilla is meant and go from there. Post essential docs and move any such posts from the AKSA forum if need be. The laws of thread gravity will sort out where the squeaky wheels are and how long it should run. What say you, all?

 

[metal]

May be other threads can be started with different versions of Fetzilla, each with a known/well recognized member name so people can easily recognize/compare these versions?

 

[keantoken]

Hey, there's no "KT Fetzilla". I just built the circuit with what I had, I have no direction...

As I have no great experience with SQ and design changes, I suppose by posting here with circuit analysis I was giving my posts more authority than they were really worth. My apologies. Maybe after experimenting more radically I will have some meaningful things to say.

- keantoken

 

[mikelm]

If you end up understanding amp design as much as you understand spice my guess is that you will build some very nice amps indeed

 

[AKSA]

Mike,

The last thing we should be doing is stifling discussion, so I propose that we head up a new thread entitled 'Fully DC coupled FetZilla'.

Would you like to start it with your latest circuit? I do sincerely find your findings fascinating, but my point was, and remains, if people would like to build your version, then there should be a dedicated thread. If there is sufficient interest, a board could be done - indeed, I'd be happy to help there - and a GB initiated.

I am not being churlish about this, simply trying to be practical and place various diverse discussions into their respective threads, nothing more.

Mike, you say 33pF does it for all, huh? That would indicate that the one with the greatest open loop gain would have the lowest phase shift, that's a find!

Cheers,

Hugh

 

[keantoken]

The miller cap, regardless of the version of circuit, will always see the same amount of AC voltage across it, and thus it will have the same current through it regardless of the circuit version. So maybe this has to do with how it loads the input FET rather than it's effect on the VAS? I've been suspecting this for a while.

- keantoken

 

[mikelm]

it's just that however much OL gain one has in the audio band it seems to drop off to about the same level at the critical high frequencies

 

[GregH2]

Hi all,

My apologies. I was unaware of all the recent replies to this thread - not getting forum notifications for some reason. I have been working on a bjt driven vertical fet version that shows great promise with a similar distortion profile to fetzilla but not one single exotic part. Lower overall distortion too. Will keep you posted.

Pawel and Kean,

Thanks for your replies.

I sorted out my spice issues. I needed to run the simulation for longer and start saving data later in the simulation. I will post an update over coming days. Kean, my spice simulations suggest that placing a larger source resistor on the p channel fet reduces second harmonic distortion. Since I don't actually mind a bit of second harmonic I am happy to put the larger resistor on the n channel device. Try simming it with cordell's models and see if your results concur. I would be interested to know.

 

[keantoken]

For the Cordell 2SK1056/J162 models, I found that .1R for the N-MOS and .33R for the P-MOS offered almost perfect current sharing, and thus second harmonic cancellation. Playing with these values could create either a positive or negative second harmonic, switching polarity depending on which way the source resistors were mismatched. I wonder what this would mean for the sonics.

I've had a sudden drop off of post notifications in the past too. Supposedly if you don't visit the site in too long it will stop sending you notifications, but I found this to happen regardless of how often I visited the site. I think it's an internal error and I check threads manually when I see a suspicious lack of notifications.

Another possibility is that, with thousands of members, the site has trouble shoving all those notifications out the internet cord and some get timed out/dropped in the process. Occasionally Email services might fail to receive a sent email, so perhaps the forum software temporarily blacklists addresses which reject emails.

- keantoken

 

[nelsonvandal]

It does something quite special with horns, guitar and piano. I really like the sound quality. It is a very fast amp, and without an input filter is 1dB down at better than 300KHz with VAS running at 13mA. Sound field is very wide, speakers simply disappear. Image depth is average, I would not rank it with the best in this quality, at which tube amps excel. But it still manages to sound very real and dynamic, with great vitality and verve to the sound, none of that twee, slightly fragile sound you hear from SETs.

Sounds like it could be a good guitaramp. Valve sound without hiss crackle and pop.

 

[AndrewT]

I use: search - advanced search, and insert my name in the user name field.
keep the default "show results as threads".
Up pops all the threads you have posted to with the newest post at the top of the list.
The threads that have a new post since my last read are highlighted in bold.

 

[keantoken]

Here's the post with the spreadsheet Andrew mentioned.

http://www.diyaudio.com/forums/solid-state/94029-when-output-inductor-needed.html#post1105724

I suspect the brain is like a second stomach, which releases energy during intense thought. I've been pounding mine with AC theory vector math for the last 2 days in order to flesh out something I discovered about oscillation. After having some success, I stepped out of my room and spontaneously did two things I thought were impossible -

1: I did a full pull-up.

2: I did it on the lintel.

Furthermore I have had nothing but turkey for the last week, a bowl a day or less.

I wonder if you guys would be willing to test if there are resonances on your rails above say 1 ohm. Fetzilla seems to be very sensitive to these and adding a 1uF+1R zobel to the rails might improve stability in some cases.

- keantoken

 

[mikelm]

Thx KT,

I noticed when I added the i/p stage ccs that the amp sounded cleaner and as I had spent a lot of time & effort building a very quiet & clean choke regulated supply I found this a bit dissappointing.

So now I am looking for the kind of thing you have just highlighted.

My current goal is to get the amp sounding as clean without the i/p stage ccs as it is with it. That, I think, would be my ideal version. It may mean resorting to separate supplies for the I/P & VAS stages that that will by last resort.

 

[DIGORA]

Do you have the last schematic with 2SK1058/2SJ162 ?

Thanks

 

[AKSA]

Interesting, Mike. I wonder why it's cleaner with the CCS on the input stage? I supply my input stage from the fb resistor on the NAKSA as well with no clarity issues whatever, that's an interesting find. Clearly with a CCS a great current variation will now pass to the fb network, that might influence proceedings.....

I'm using a simple supply with two 4,700uF caps at each rail. There are 0R15 resistors between each of the caps, but with this supply I hear absolutely no hum or noise whatever, not even with my head well inside the speaker cones. That's pretty good - best I've heard in 15 years - I would think with chokes we would be improving on perfection, however that would sound.

Cheers,

Hugh

Cheers,

Hugh

 

[keantoken]

AKSA, the .15R resistors in your power supply may be doing the service of damping any resonance on the rails between the filter caps and the film decoupling, like how I suggested to try a 1R+1u snubber on the rails. Have you approached the power supply from this perspective before? Maybe this is why it always seems best to have a little resistance in the supply?

If this is true than the .15R resistor will be serving a dual purpose, although the optimal value for snubbing and the optimal value for filtering may be different, which is suggesting compromise.

- keantoken

 

[AndrewT]

I'm using a simple supply with two 4,700uF caps at each rail. There are 0R15 resistors between each of the caps, but with this supply I hear absolutely no hum or noise whatever, not even with my head well inside the speaker cones. That's pretty good - best I've heard in 15 years - I would think with chokes we would be improving on perfection, however that would sound

I am using 4700uF first cap, then 40uH+0r15 inductor followed by 10mF second capacitor.

I can clearly see the sharp sawtooth at the first cap.
The very rounded sawtooth at the second cap is clearly missing many of the higher harmonics.
The waveform at the output drains is virtually identical to that at the second caps. Ripple when drawing 400mA is quite high. I'm thinking about adding 15mF or 22mF with a dropping resistor to create a 3stage PSU filter.

 

[mikelm]

Interesting, Mike. I wonder why it's cleaner with the CCS on the input stage? I supply my input stage from the fb resistor on the NAKSA as well with no clarity issues whatever, that's an interesting find. Clearly with a CCS a great current variation will now pass to the fb network, that might influence proceedings.....

I'm using a simple supply with two 4,700uF caps at each rail. There are 0R15 resistors between each of the caps, but with this supply I hear absolutely no hum or noise whatever, not even with my head well inside the speaker cones. That's pretty good - best I've heard in 15 years - I would think with chokes we would be improving on perfection, however that would sound.

Cheers,

Hugh

Cheers,

Hugh

I think there are two reasons why the i/p stage CCS gives a cleaner sound

1) much higher gain gives more FB & this in turn gives much better PSRR

2) the CCS ensures that the i/p stage is as clean as it can be

Funny thing with noise, which I don't pretend to understand, is that, in my experience so far, no matter how low I think the noise is already whenever I take another noise reduction measure the result is always clearly audible and sounds better.

My amp also has produced no audible noise - even with ears at the speaker - since I added the inter stage but filters but never-the-less each additional measure I have taken brings improvement.

So for me the answer to your question is "better"

I just did some spice research and in theory there really is no substitute for a separate supply for the the i/p and gain stages - so I'll try that next.