The ZETA has been redesigned, but it still has 4-5V of rail loss with IRFP250N at full load. It is a compromise, to get around this rather small rail loss i would have to add a small gain in the output stage itself (as some of the rail loss comes from the gain stage). I think that would complicate the nice and very simple ZETA construction too much compared with the small problem it would solve.
I do recognize that the simple SIEMENS construction, now modified by you in this thread has lower rail losses. (Only about 1.5 V in a comparable +/- 60V setup). So the difference is 2.5 - 3.5V.
But IMO on all other points the ZETA is far better
After all it is my baby ..
I do recognize that the simple SIEMENS construction, now modified by you in this thread has lower rail losses. (Only about 1.5 V in a comparable +/- 60V setup). So the difference is 2.5 - 3.5V.
But IMO on all other points the ZETA is far better
After all it is my baby .. 
Lars Clausen said:The ZETA has been redesigned, but it still has 4-5V of rail loss with IRFP250N at full load. It is a compromise, to get around this rather small rail loss i would have to add a small gain in the output stage itself (as some of the rail loss comes from the gain stage). I think that would complicate the nice and very simple ZETA construction too much compared with the small problem it would solve.
I do recognize that the simple SIEMENS construction, now modified by you in this thread has lower rail losses. (Only about 1.5 V in a comparable +/- 60V setup). So the difference is 2.5 - 3.5V.
But IMO on all other points the ZETA is far better
Hey ZETA mamma[mother of Zeta]
Ok If I want to build your Zeta with wattage of around 1200W at 2 ohms and with your saying it has rail loss of 4-5 volts, then according to ohms law if 25A is flowing from the bank of output mosfets the total WASTE DISSIPIATION will become around 5 X 25 = 125Watts of PURE ANALOG SONICALLY APPRICIATED HEAT DISSIPIATION would welcome me . Then what is the benefit gained from using mosfets at outputs, only inefficiency is obtained.
Sir, mine modification of that amp resulted in Rail loss of around less than 1 volts at 2 ohms loading. In that case the dissipiation is only 1 X 25 = 25 Watts.
There exist a way by which your baby could be made much more efficient.
AT your website you somewhere stated that you donot do compromises in the circuit, you only design REFERENCE level audio gear, then why are you compromising here.
Cheers,
Kanwar
richie00boy said:Maybe he's too busy
Anyway, I thought he said in the Zeta thread how he achieved low rail loss. Maybe something to do with cascodes?
No he is not that much busy, because he discusses several important topics with us so well and i really appriciate him for that.
The rail loss has nothing to do with the CASCODES, either you use a CASCODE or CASCADE it will remain same unless you do something such as to increase the VAC supply voltage or bootstrap the upper mosfets driver resistor with an elevated supply or use what we have developed......, but this must be a secret telling if i tell this then , i would be on streets someday with no livlihood.
richie00boy said:
If its in a Class-A design then its ok, but its an Class-AB design which is known for its efficieny first and sonic clarity afterwards.
Its not that the ZETA is not intended for it , its is the fact that lies in its efficiency alone, if one doesnot get efficiency with Class-AB , then for what purpose one is going to build it, why not switch it to Class-A otherwise.
Kanwar: You are right, the rail losses of a Zeta are controlled by the difference in VGS from idle to max current. With the new output setup, which will be posted soon, the voltage amplifiers rail loss is close to 0.
Looking at the datasheet for IRFP250N it can be established that the typical VGS increase from 25 mA to 5A per device = total 25 A is not 4-5V but only 1.5V. So it's not quite so bad as i had anticipated.
Looking at the datasheet for IRFP250N it can be established that the typical VGS increase from 25 mA to 5A per device = total 25 A is not 4-5V but only 1.5V. So it's not quite so bad as i had anticipated.
Lars Clausen said:
Lars: I really like your attention towards the rail loss is increased manifolds.
The rail loss you depicted through the datasheet only confronts only when the reference gate drive is elevated a step higher than the output rail or only when you use floating gate driver supply. in datasheet the case is presented in a switching phenomenal drive which is at least around 10 volts to get aroung saturation.
The bandgap mode will not alow you to saturate the output mosfet with drivers at same rails unless you try some bootstrapping or floating drivers.
What you are doing now is what we have done 5 years ago to study a lot of best available alternative methods to get negligible rail loss.
our commercial amp feature rail loss of less than 0.75 Volts at 2 ohms using 10 Pairs of IRF250 devices alone.
Kanwar: This is very good. And i am sure rail loss is a key priority for a PA amplifier design. However the rail loss was not the top priority with the ZETA. I know it was shown in a PA size version, with 1000W but that was just to start from the top down. I see the typical ZETA as a single MOSFET amplifier for home use. In sizes of 100 - 400 W which is enough for most home use.
Key features are simplicity, high sound quality, stability, good efficiency and low cost, so anybody can afford to build one.
Compared with other well known MOSFET amplifiers the ZETA is vastly efficient, and has much lower rail loss. But OK maybe not compared with your super high efficiency PA amplifier. Good for you!
Another thing is that you can reduce the rail losses of the ZETA to below 1V by using a logic level FET instead. Such as the FQA19N20L. This will not increase the circuit complexity.
Key features are simplicity, high sound quality, stability, good efficiency and low cost, so anybody can afford to build one.
Compared with other well known MOSFET amplifiers the ZETA is vastly efficient, and has much lower rail loss. But OK maybe not compared with your super high efficiency PA amplifier.
Good for you!Another thing is that you can reduce the rail losses of the ZETA to below 1V by using a logic level FET instead. Such as the FQA19N20L. This will not increase the circuit complexity.
Lars Clausen said:Kanwar: This is very good. And i am sure rail loss is a key priority for a PA amplifier design. However the rail loss was not the top priority with the ZETA. I know it was shown in a PA size version, with 1000W but that was just to start from the top down. I see the typical ZETA as a single MOSFET amplifier for home use. In sizes of 100 - 400 W which is enough for most home use.
Key features are simplicity, high sound quality, stability, good efficiency and low cost, so anybody can afford to build one.
Compared with other well known MOSFET amplifiers the ZETA is vastly efficient, and has much lower rail loss. But OK maybe not compared with your super high efficiency PA amplifier.
Another thing is that you can reduce the rail losses of the ZETA to below 1V by using a logic level FET instead. Such as the FQA19N20L. This will not increase the circuit complexity.
Yes it is good in PA amps, i know it.
It is Ok if the Zeta is intended for low wattage say 100 to 400W application+heat.
With whom are you comparing your Zeta! ............with other DIY amps or some branded ones.
Our Class-AB amp is only super efficient when it is driven near its full output only.
If one has to use Those Logic level FETs , then where is the aspect of DIYING lies, since these switching mosfets are easily available then logic ones.
By the way i really love your folded cascodes very much , as they eliminate Mr.Millers great influence on the amp high frequency response................, but they have the tendency to turn the mids in to harsh sounding material if not driven properly.......as Rajeev once said that Zeta is harsh at mids ........
Hey Kanwar .. I have used the FC also in other amplifiers, and i didn't find the midrange to be harsh.
So if somebody found the ZETA to be harsh in the mids i don't thing it's owing to the FC's.
Did someone (other than myself) try to build it in a single MOSFET version (only with one set of output devices) ?? Please post your results...
So if somebody found the ZETA to be harsh in the mids i don't thing it's owing to the FC's.
Did someone (other than myself) try to build it in a single MOSFET version (only with one set of output devices) ?? Please post your results...
Hey Lars .. I have used the FC also some amplifiers, and i did find the midrange to be harsh.
So if somebody found the ZETA to be harsh in the mids i think it's owing to the FC's.
Rajeev(other than yourself) try to build it in a single MOSFET version (only with one set of output devices) he will post the results...
So if somebody found the ZETA to be harsh in the mids i think it's owing to the FC's.
Rajeev(other than yourself) try to build it in a single MOSFET version (only with one set of output devices) he will post the results...
Lars Clausen said:What they do however is to replace the high Z of the collectors from the gain stage ( 1 Meg ) with say 10k. This gives the amplifier 100 times lower Natural Damping Factor. It also lowers the feedback factor, while maintaining the voltage amplifier's pole intact. And thus improves the stability of the amplifier. I will add in most cases, just to avoid any attacks from the tg's
Basicly it gives a lower feedback factor, and thus a higher THD, and in the same time improves the sound quality. Try it out in your own amplifier .....
In fact only one resistor is needed, but might give a slight DC offset.
You mean HIGHER open loop damping factor. This turns the amp to be realy good voltage amplifier with OL.
Yeah its right, i absolutely agree with lars, the open loop damping factor is thus increased and thus upon applying negative feedback the closed loop damping factor also increase alot, thereby making the sonic perspective of acoustic bass notes more controlled and focused. but their is a slight increase in THD also.....
hello Greg,
Nice to see your way of talking or writing!
Do a simple experiment.
Take an open loop amp
Measure its damping factor and listen it with music having bass content.
then apply the collector to ground Resistors at its VAS and then measure the damping factor and again listen to the same music.
You will definately see an improvement in damping factor with resistors at VAS.
This experiment has been done by us 5 years ago to validate the fact.
Donot tend to belive my ears or my saying of damping improvement, but try experiment and try to believe your ears only.
One more thing Crown amplifiers have damping factors of over 1000 and even 15000 with some of their models. So what are your views about the factors affecting damping factor.
So according to you is it really a worth or just a jest in rest.
respect & gratitude to guru of amplifiers
Nice to see your way of talking or writing!
Do a simple experiment.
Take an open loop amp
Measure its damping factor and listen it with music having bass content.
then apply the collector to ground Resistors at its VAS and then measure the damping factor and again listen to the same music.
You will definately see an improvement in damping factor with resistors at VAS.
This experiment has been done by us 5 years ago to validate the fact.
Donot tend to belive my ears or my saying of damping improvement, but try experiment and try to believe your ears only.
One more thing Crown amplifiers have damping factors of over 1000 and even 15000 with some of their models. So what are your views about the factors affecting damping factor.
So according to you is it really a worth or just a jest in rest.
respect & gratitude to guru of amplifiers
hi guru
it is simple:
ccs loaded VAS has output impedance of megaohms. Darlington outputs do not change voltage very much, but they amplify current. So output impedance of an amp eqals:
ccs loaded VAS output impedance/product of betas
(open loop of course)
even tripple darlington of modern transistors won't give you more than ,.. say 2milion product of betas.
So damping factor of many amps is so big mainly because of feedback. And thus it is so depending on freq. (assuming no output L).
With some kiloohms from Vbe multiplier to ground (or to both reguleted rails) open loop damping factor is many times higher, and so feeling more 'natural' as Lars has observed.
Hope I didn't mess something and have been clear.
Do you agree with my view about it?
it is simple:
ccs loaded VAS has output impedance of megaohms. Darlington outputs do not change voltage very much, but they amplify current. So output impedance of an amp eqals:
ccs loaded VAS output impedance/product of betas
(open loop of course)
even tripple darlington of modern transistors won't give you more than ,.. say 2milion product of betas.
So damping factor of many amps is so big mainly because of feedback. And thus it is so depending on freq. (assuming no output L).
With some kiloohms from Vbe multiplier to ground (or to both reguleted rails) open loop damping factor is many times higher, and so feeling more 'natural' as Lars has observed.
Hope I didn't mess something and have been clear.
Do you agree with my view about it?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.