Hi Esperado,
That is very close what I've got in my simulation. You don't say THD at what power level, my simulation shows THD10k 0.006445% at 50W/8ohm.
Regarding SR my simulation shows 200V/usec, and I think that you have to measure SR with that low pass filters as thy are part of the amp not outside it and thy influence phase margin, from 65 degree with the filter on to 60 degree with no filter. Who needs higher SR??
I never said that this amp is not good, in my opinion is better than plain VSSA.
Damir
PS. I forgot to add that ULGF is quite high at 8,6 MHz, quite challenge for a layout.
Last edited:
I guess my thought was that maybe the VSSA is actually accurate at low volume and some of your other amps are weak in the bass at that level. If the Klipsch is bass heavy the VSSA may just be revealing that. I will have to do some comparisons now to see if I can duplicate your findings. I have a couple pair of nearfield set up that I can listen to as well. The Dynaudio's are pretty neutral and may give me a good read at low volume. Something to do while I wait for OS to complete his newest layouts.
Blessings, Terry
Definitively not. No and no. (With all my respect ;-)
Slew rate is an indication of the ability of a servo system (GNFb) to follow fast transients without producing IM. It has no interest else where or in a non looped system.
The first stage of the diamond is outside of the loop and can be considered as some kind of a pre-amp.
Same remark with the curve of phase to evaluate stability margin. Set this filter at a lower FC. It will not change the stability margin and, if you measure the phase with the filter in, you'll have wrong evaluation.
Nyquist apply only at closed loops ;-)
And both are the same story.
Last edited:
Thanks mate. I've visited Bob's site and read through the MOSFET amp paper. Bob was right on so many things already back in the '80s.
Your research on OPS's made me decide to give the HEC a try, guessing the OPS improvement would add to the overall performance. I used Bob's implementation as a basis, but it didn't work for my LatFET OPS using the Exicons. They have a very low Vgs treshold and so the regulating trannies of the HEC were not able to 'open' fully and get the quisent current low enough.
I moved the sense points from the LatFET bases, to the bases of the buffer transistors driving the FETs, including them into the correction as well, but most importantly it gave some voltage headroom increase allowing the regulating transistors to do their jobs.
After getting the bias range and setting right, I went to do a full power THD 20K and dayumn I was amazed! THD had dropped by almost a half! Then I went on to tune the resistors feeding the bases of the transistors driving the OPS drivers and I was amazed at the numbers I ended up with:
All into 8R:
250W : 0.000036%
100W : 0.000015%
1W : 0.000001%
At 1W, it remains at 1% across the full audio band. It's sick. On average things have improved by a factor 7 - 20.
So yea, big thanks to you and Bob
This warrants further investigation and playing with (hours of simulation fun ahead)You were after bad models?
Why would you trust unknown, unverified models over Bob Cordell's?
Apart from my earlier comments, there's the tip-off that either the developer doesn't have the confidence to put their name on it, or it's been pirated or it's copied without source or version control.
That's practically the first rule of software, incompetence here is a pretty solid clue as to the quality.
Best wishes
David
We're not building rocket ships here. All I asked was if any of you tried them. Did you? I don't know who developed them. I asked for some models to use in a sim of the little mini OPS I wanted to try while OS was gone and someone shared them with me. They worked well enough for me to run the spice model. There is no need for you to come here and start accusing folks of pirating.
Hello
Guys do you think these mosfet circuit worth a try or better to stick with the still4given latfet version
I started to draw the layout but I do have some second taught so the best it would be to ask before I do any farther progress not to waste my time or the components (in case if it blow up)
Thank you very much
Guys do you think these mosfet circuit worth a try or better to stick with the still4given latfet version
I started to draw the layout but I do have some second taught so the best it would be to ask before I do any farther progress not to waste my time or the components (in case if it blow up)
Thank you very much
Attachments
Last edited:
I don't get it ? I just tried them. I did not have many latfet models.
BTW - they also sucked...
I'm almost sick of all this ... (you know what)
.I don't even have an amp out of all this.
Why don't you give me a parting gift and share your modifications.(schema
please).
I'm considering not to post anymore.
Bye.
OS
Yes, it can be unfortunate when folks jump into a thread without making any real contribution, especially if they have the knowledge and skills to make a positive impact. If an issue is discovered it would be nice to hear why it's an issue and offer up a potential solution than to be strung up for making an error.
Seems to me it's better to keep focused and try to keep the thread's S/N ratio under control. There's lots of good stuff here but it gets harder and harder to find in all the flotsam.
I say keep trucking and don't stop to feed the trolls! They will get hungry and go forage for fresh fodder elsewhere in due time.
He could of just given me better models , instead of ripping me for
unprofessional behavior.
Magicbox's contribution is more tangible (HEC FB points).
(below) is a new IPS that I let "rest" for a while.
Ohhhhh .... ICzzzz (eyesee).
Has a servo , but no servo in the loop .
Has a higher Z than a diamond buffer with fet inputs (in the chip).
Has CURRENT FB !!! Ohhhh.
If you "wrap your head around it" , It is the chip version of the GONAD.
IC's output is the CFB node , Vcc/Vee sink the cascodes.
The chip used does not matter... (R6/7/17) change bias for offset/VAS Ic.
Tried 741/ 5532 / LTxxxx .
UG and THD20 performance varied with IC's. Slow IC's gave .05% thd20 ,
fast IC's gave 20ppm.
ALL were stable with good PM. ALL were 5 ppm 1k.
So , what you "hear" ... would be the internal structure (design) of the IC.
PS - FFT 's show distinct differences in distortion products between
devices. (below 2 -5532 dual)
OS
unprofessional behavior.
Magicbox's contribution is more tangible (HEC FB points).
(below) is a new IPS that I let "rest" for a while.
Ohhhhh .... ICzzzz (eyesee).
Has a servo , but no servo in the loop
.Has a higher Z than a diamond buffer with fet inputs (in the chip).
Has CURRENT FB !!! Ohhhh.
If you "wrap your head around it" , It is the chip version of the GONAD.
IC's output is the CFB node , Vcc/Vee sink the cascodes.
The chip used does not matter... (R6/7/17) change bias for offset/VAS Ic.
Tried 741/ 5532 / LTxxxx .
UG and THD20 performance varied with IC's. Slow IC's gave .05% thd20 ,
fast IC's gave 20ppm.
ALL were stable with good PM. ALL were 5 ppm 1k.
So , what you "hear" ... would be the internal structure (design) of the IC.
PS - FFT 's show distinct differences in distortion products between
devices. (below 2 -5532 dual)
OS
Attachments
Peace Terry.
I don't think you pirated them. In earlier posts I noticed the way you stood by what you heard and in fact I think you are unusually honest.
To stand for what you believe is true, even when there is social pressure, takes a certain kind of honesty that is not so common.
So I assumed you would want to know if the models were likely to be accurate. My observation was intended to help spot suspect models.
Of course you can use whatever models you want.
Spice models are usually "freebies" anyway, probably these are merely amateurish.
I was just surprised that OS looked like he intended to use them, when Bob Cordell's models are a better option that is free, readily available, well documented and definitely above board.
Best wishes
David
You could of pointed me in that direction. I am a FET "novice" , I did not
know Cordell had them on his site. I only DL'ed his HEC papers , never
explored the site ( a google link to the PDF only).
I DO use Cordell's fairchild/ON models for all these designs.
OS
I am sorry if you read my post that way, that was not the intention.
I had taken the time and effort to explain to Christophe a little about FET models and why I think Bob's FET models are better than the ones Terry asked about.
Now that this thread has moved to discuss FET OPS I expected that would be a useful contribution.
So when your response to Terry implied that you intended to use the inferior models, I really didn't understand, and still don't. Perhaps I misread your post.
Best wishes
David
OS. I did specifically mention the Cordell models in my response to Christophe, I would say that is a point in the direction.
But I do now better understand how the confusion has arisen.
Last edited:
Hi David,
I don't know you from Adam. The first I see you, you are saying that the models I asked about are no good. I think you suggested Cordell models but to be honest I couldn't find a Cordell model for 2SK1058. I didn't think much about it until later when you seemingly attacked OS for using them. I checked your profile and it is empty. As far as I know you are just one more guy who has read a lot and now thinks he is an authority on all things spice. This thread has been going along for quite a while and a lot of good work has been accomplished. I would hate to see that ruined by a bunch of "I know better" guys. If you think you have some constructive advice, there are nice ways to go about sharing it. Let's not spoil it for those of us who are trying to learn.
Blessings, Terry
