suzyj said:
LOL!
Back in tech school when I was 17, I repaired an old Peavey PA amp that blew it's wirewound zobel resistor to open, (no fire) BURNED UP the zobel cap to a crisp, but the amp itself was OK.
2 of the 4 inputs had blown opamps, and 2 shorted caps for the opamp supply. After fixing the caps, and changing all 4 of the dual 4558 opamps, it worked fine. A blown input opamp can also kill a zobel, go figure.
Since seeing stuff like this, I like to use high rated caps so they don't burn up before the resistor goes in case of something like this.
Bob Cordell said:
OK, Bob
I would guess that an RC output Zobel is much less likely to be a superfluous inclusion in a design that uses output devices particularly susceptible to parasitic oscillation, such as power MOSFET’s - or an oscillation susceptible output stage topology such as a CFP.
Though I think in many cases an RC Zobel is included just out of habit and is not necessary in many cases – particularly so when included in the misguided belief that a high frequency load is generally necessary to keep an amplifiers open loop gain and phase margin under control.
Cheers,
Glen
well i dont thing so
It's not something that has any effect at audio frequencies, so there's no point optimising it. The actual values used can legitimately be whatever you've got to hand, within a decade or so. As long as the amp sees some reasonably low impedance at high frequencies.
to my expirience and very very recently i noticed that a mosfet amp i constructed ( a very simple one ) actually pass most of bench tests .... also like yours with square waves ...and goes on
but in real life playing especially at the upper end there was confusion and treble was kind of messy ....due to 5 m of cables ???? quality of cables ???? really cant tell ....
then i try any inductor as you said and there was audible diference high got much more clear ....after that try 3 diferent of them till i end up with the one i thing sounded better ....
at least between the 3 of them thre was a clear audible diference ....
straight result may be that my amp was ocilating before and i havent notice ??? since bench test was passed didnt really bother ...and said to my shelf ok thats the amp dont expect more from it ....
now i got more of it ....just an inductor did the trick
It's not something that has any effect at audio frequencies, so there's no point optimising it. The actual values used can legitimately be whatever you've got to hand, within a decade or so. As long as the amp sees some reasonably low impedance at high frequencies.
to my expirience and very very recently i noticed that a mosfet amp i constructed ( a very simple one ) actually pass most of bench tests .... also like yours with square waves ...and goes on
but in real life playing especially at the upper end there was confusion and treble was kind of messy ....due to 5 m of cables ???? quality of cables ???? really cant tell ....
then i try any inductor as you said and there was audible diference high got much more clear ....after that try 3 diferent of them till i end up with the one i thing sounded better ....
at least between the 3 of them thre was a clear audible diference ....
straight result may be that my amp was ocilating before and i havent notice ??? since bench test was passed didnt really bother ...and said to my shelf ok thats the amp dont expect more from it ....
now i got more of it ....just an inductor did the trick
Sakis,
Interesting that you heard differences between the inductors! I would have expected that without it would be best sounding, because often less is more.
Did you, before adding an inductor, measure with an oscilloscope that the amplifier is completely stable with the loudspeakers connected? I once made an amplifier which was okay on the test bench (also with capacitive loads) but showed some oscillation with the loudspeakers connected. After adding an output inductor // resistor combination the amplifier was stable.
I am interested to hear from you what is the type off inductor you prefer? Which inductance, wire diameter, with a resistor inside it or not, and if so, has that resistor magnetic end-caps?
What type of inductor didn’t you like?
Peter
Interesting that you heard differences between the inductors! I would have expected that without it would be best sounding, because often less is more.
Did you, before adding an inductor, measure with an oscilloscope that the amplifier is completely stable with the loudspeakers connected? I once made an amplifier which was okay on the test bench (also with capacitive loads) but showed some oscillation with the loudspeakers connected. After adding an output inductor // resistor combination the amplifier was stable.
I am interested to hear from you what is the type off inductor you prefer? Which inductance, wire diameter, with a resistor inside it or not, and if so, has that resistor magnetic end-caps?
What type of inductor didn’t you like?
Peter
well....
it is not my practice to test amplifiers with a scope /signal genarator /distortion analyzer.... with speakers connected to them ......
i usually do this with a dummy load
BUT !!!! for this particular amplifier and also was my first mosfet that i ve been working arround for almost 2 years i actually had to "rearange" my dummy load cause strange things were happening that never happend before .......:
i own a repair shop and i had constructed a dummy load for testing with 2 wire resistors made of s spiral wire on porselan and the all thing in a metal box filled with a special kind of cement that we use and has thermo transfer abilty ...
eventhough this was with four terminals rated 8.2R @ 200W i had connected it with only three leads , bridged ground terminals ( one lead ) and use the other two for left, right
this configuration didnt work to the mosfet amplifier when you had one amplifier for testing and then connected the second one with this wiring very strange things happend like either sine wave or square wave distorted ..... probably this amp was ocilating .....with this kind of load on it ....
then rearange the configuration with 4 leads 2 for ground and two for signal and then amplifier ( stereo version ) passed all the bench tests .....
sine wave was very very clean up to 40khz and then square wave was starting to get round in the upper top something a bit more than 20k.....( that was ok for me since i added miller cap in the VAS stage that in the original schematic didnt exist )
i was happy with that and closed the case ....
till that moment no inductor ....i didnt even thing about it ( it wasnt included in the schematic )
put the amp to play listen to it and sounded ok more or less but high was confusing.... since had no previous expirience with mosfets i said to my shelf this the sound of mosftes and rest my case......
after that i spended a gozilion of mosfets money and smoke trying to make this particular amp to work in higer rails that finaly never worked ( let us not mess up with this old story now )
any way after 2 years i started to construct the amp again making it an audiophile version and try to focus on less power quality work trafos psu and so on and more safety ....
after i finished a project like that it was playing nicer than the high rail version but still the problem of high was there
one day i see a post about zobel + inductor and a bell rings that i have to listen this amp with inductor ...i take one drill 10mm and i wind 0.8 wire on it 10-12 turns ...i look for any resistor be tween 2-10R /2-5W but wasnt any !!!! in my shop .... so i find one 220 and put it on
i didnt work ...sound was diferent and after a few tests i decided to remove the resistor totally MUCH BETTER !!!!
then the other day got resistors be tween 2.2 and 10 R and started to play .....
the coils winded in 10mm diameter no matter the resistor improoved a lot but the best was after a few try i setled for 13mm diameter and 14 turns with a normal 2w 2.2R resistor ...
results : it seems all the hiss hiss gone away from high frequency is gone away .... meaning that in a bad production of greek music the percusion was excesive and very messy after the inductor all that was totaly clean ....then of course when the hiss hiss and noisy high goes away you ofcourse understand lower and mid better ....
my testing ( for listening ) includes a standar set of speakers one marantz cd player and ONLY ONE CD !!!!! that i listen to compaire with
sorry for the long post hope we all get somthing out of this
it is not my practice to test amplifiers with a scope /signal genarator /distortion analyzer.... with speakers connected to them ......
i usually do this with a dummy load
BUT !!!! for this particular amplifier and also was my first mosfet that i ve been working arround for almost 2 years i actually had to "rearange" my dummy load cause strange things were happening that never happend before .......:
i own a repair shop and i had constructed a dummy load for testing with 2 wire resistors made of s spiral wire on porselan and the all thing in a metal box filled with a special kind of cement that we use and has thermo transfer abilty ...
eventhough this was with four terminals rated 8.2R @ 200W i had connected it with only three leads , bridged ground terminals ( one lead ) and use the other two for left, right
this configuration didnt work to the mosfet amplifier when you had one amplifier for testing and then connected the second one with this wiring very strange things happend like either sine wave or square wave distorted ..... probably this amp was ocilating .....with this kind of load on it ....
then rearange the configuration with 4 leads 2 for ground and two for signal and then amplifier ( stereo version ) passed all the bench tests .....
sine wave was very very clean up to 40khz and then square wave was starting to get round in the upper top something a bit more than 20k.....( that was ok for me since i added miller cap in the VAS stage that in the original schematic didnt exist )
i was happy with that and closed the case ....
till that moment no inductor ....i didnt even thing about it ( it wasnt included in the schematic )
put the amp to play listen to it and sounded ok more or less but high was confusing.... since had no previous expirience with mosfets i said to my shelf this the sound of mosftes and rest my case......
after that i spended a gozilion of mosfets money and smoke trying to make this particular amp to work in higer rails that finaly never worked ( let us not mess up with this old story now )
any way after 2 years i started to construct the amp again making it an audiophile version and try to focus on less power quality work trafos psu and so on and more safety ....
after i finished a project like that it was playing nicer than the high rail version but still the problem of high was there
one day i see a post about zobel + inductor and a bell rings that i have to listen this amp with inductor ...i take one drill 10mm and i wind 0.8 wire on it 10-12 turns ...i look for any resistor be tween 2-10R /2-5W but wasnt any !!!! in my shop .... so i find one 220 and put it on
i didnt work ...sound was diferent and after a few tests i decided to remove the resistor totally MUCH BETTER !!!!
then the other day got resistors be tween 2.2 and 10 R and started to play .....
the coils winded in 10mm diameter no matter the resistor improoved a lot but the best was after a few try i setled for 13mm diameter and 14 turns with a normal 2w 2.2R resistor ...
results : it seems all the hiss hiss gone away from high frequency is gone away .... meaning that in a bad production of greek music the percusion was excesive and very messy after the inductor all that was totaly clean ....then of course when the hiss hiss and noisy high goes away you ofcourse understand lower and mid better ....
my testing ( for listening ) includes a standar set of speakers one marantz cd player and ONLY ONE CD !!!!! that i listen to compaire with
sorry for the long post hope we all get somthing out of this
lumba .....
the point of such a long post ..... the explanation about the dummy load detail.... and the all procedure shows simply that for reasons that i dont really know or understand this particullar amplifier performed the best on bench but with cables connected to speakers show some ocilation .....
in this case what do you do ????
through the amplifier away or use inductor ..... the benefit of the inductor i think is the best option ....
i have constucted more than 300 bjt amps ( not design by me ) and this procedure made me understand many things and or reevaluate the way this mosfet amplifier is working ....
also from personal expirience i ve never seen a mosfet amplifier
( i mean comercially made ) without an inductor
the point of such a long post ..... the explanation about the dummy load detail.... and the all procedure shows simply that for reasons that i dont really know or understand this particullar amplifier performed the best on bench but with cables connected to speakers show some ocilation .....
in this case what do you do ????
through the amplifier away or use inductor ..... the benefit of the inductor i think is the best option ....
i have constucted more than 300 bjt amps ( not design by me ) and this procedure made me understand many things and or reevaluate the way this mosfet amplifier is working ....
also from personal expirience i ve never seen a mosfet amplifier
( i mean comercially made ) without an inductor
G.Kleinschmidt said:
Hey, let's face it, the AES paper was not an amplifier condtruction project. A lot of stuff was left out of the schematic, including many transistor types, no protection, etc.
I don't think I have ever built a solid state amplifier without a Zobel, and in my opinion only a fool would do so. Here is why. Even if the amp is stable on the bench with no load or a resistive load, you don't know what the complex nature of the load might be when speaker cables and a speaker are connected. The presence of the Zobel network largely guarantees the absence of a high-Q resonance on the output node of the amplifier.
Cheers,
Bob
thanks bob ...
for join in !!!!!
all your input is a bless !!!
i am with you 100% the only thing that the big mess is about the inductor mostly ....
to the amplifier i worked ( the specific one ) and all others i constructed capacitor and resistor in the output always existed .....
the inductor though ..... is another story .... i was playing with rods P3 ( the simplest thing you can ever imagine ) which i like very much and seems that works without an inductor without any problems ....
you see.... i am runing both amps in the same system ....same speakers ...same cables ....same source ....same cd ....
and this is what really makes sense .... that my mosfet amp actually needs a thing like that .... ( inductor)
of course i could place a schematic of the amp ....hmmmm and pcb also so real experts like you and others could comment why this amp is sensitive to ocilation related to cables and others
i will not ...last time i did that was a war about it and since i feel that i make so many good friends here i dont want want to loose any of them ...especially the fat ones ( carlos he he he )
for join in !!!!!
all your input is a bless !!!
i am with you 100% the only thing that the big mess is about the inductor mostly ....
to the amplifier i worked ( the specific one ) and all others i constructed capacitor and resistor in the output always existed .....
the inductor though ..... is another story .... i was playing with rods P3 ( the simplest thing you can ever imagine ) which i like very much and seems that works without an inductor without any problems ....
you see.... i am runing both amps in the same system ....same speakers ...same cables ....same source ....same cd ....
and this is what really makes sense .... that my mosfet amp actually needs a thing like that .... ( inductor)
of course i could place a schematic of the amp ....hmmmm and pcb also so real experts like you and others could comment why this amp is sensitive to ocilation related to cables and others
i will not ...last time i did that was a war about it and since i feel that i make so many good friends here i dont want want to loose any of them ...especially the fat ones ( carlos he he he )
Bob Cordell said:
Great! Now we are down to name calling. The few oscillations free amps I have put togeather without bothering with an RC Zobel obviously make me a fool then!
Exactly how detrimental are these high-Q resonances, which at all but high frequencies well beyond the audio band, are well damped by the low output impedance of the amplifier?
How about a basic amplifier design without a particularly high slew rate to excite such resonances?
Cheers :-;
Glen
And another thing. I am now working on a webpage for my K10A design here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=119409&pagenumber=5
This design uses TO-3 output devices, which are available with fT’s in the range of 1 to 4MHz.
It does not have an RC Zobel to ground directly at the output node and I have elected to have the output coupling network mounted off the board.
The recommended output coupling network will be of the type I cited earlier in this thread, but consisting of a 1uH inductor in parallel with a 6.8 ohm resistor, with a 100nF capacitor directly across the speaker output.
This network provides the combined function of inductively isolating the load capacitance and that of an RC Zobel, providing a low impedance high frequency load to the amplifier output.
I will recommend that this network be mounted on a bit of tagstrip behind the actual speaker terminals/binding posts. The output capacitor also acts as a direct shunt to RFI, which is why I like to have it on the speaker terminals.
The design is of the single rail type with a 4700uF board mounted electrolytic output coupling capacitor, electrically between the amplifier output and the load coupling network.
I have measured the bog standard 4700uF 25V electrolytic capacitor that I will be using in the first PCB prototype and its impedance is in the order of 5 ohms at 50MHz.
Between 1 and 10MHz (where a TO-3 output device is likely parasitically oscillate, if at all) it is well beyond an order of magnitude lower than the minimum equivalent high frequency “Zobel” impedance provided by the output network (or any RC Zobel with typical component values).
In comparison the short length of hookup wire between the PCB and the output network also represents a totally negligible impedance at these frequencies.
Please explain, with technical reasons, why I am a “fool” for not sticking an RC Zobel directly at the output node of this design on the PCB.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=119409&pagenumber=5
This design uses TO-3 output devices, which are available with fT’s in the range of 1 to 4MHz.
It does not have an RC Zobel to ground directly at the output node and I have elected to have the output coupling network mounted off the board.
The recommended output coupling network will be of the type I cited earlier in this thread, but consisting of a 1uH inductor in parallel with a 6.8 ohm resistor, with a 100nF capacitor directly across the speaker output.
This network provides the combined function of inductively isolating the load capacitance and that of an RC Zobel, providing a low impedance high frequency load to the amplifier output.
I will recommend that this network be mounted on a bit of tagstrip behind the actual speaker terminals/binding posts. The output capacitor also acts as a direct shunt to RFI, which is why I like to have it on the speaker terminals.
The design is of the single rail type with a 4700uF board mounted electrolytic output coupling capacitor, electrically between the amplifier output and the load coupling network.
I have measured the bog standard 4700uF 25V electrolytic capacitor that I will be using in the first PCB prototype and its impedance is in the order of 5 ohms at 50MHz.
Between 1 and 10MHz (where a TO-3 output device is likely parasitically oscillate, if at all) it is well beyond an order of magnitude lower than the minimum equivalent high frequency “Zobel” impedance provided by the output network (or any RC Zobel with typical component values).
In comparison the short length of hookup wire between the PCB and the output network also represents a totally negligible impedance at these frequencies.
Please explain, with technical reasons, why I am a “fool” for not sticking an RC Zobel directly at the output node of this design on the PCB.
lets keep ....
a low profile about all this ...i think we give our opinions about all these things Glen .....
also statistics talk here ..... 99.9% of the amplifiers seen in this forum include at least rc network in the out ....
on the other hand i think that bobs comment was generall and not personal ....
from personal expirience though i never constructed or operated single rail amplifier (yet i am 42 years old ) so i have no epirience on the particular design .
still as far as i know the same rules apply for single rail amps ....rc should be there to protect the out ....
a low profile about all this ...i think we give our opinions about all these things Glen .....
also statistics talk here ..... 99.9% of the amplifiers seen in this forum include at least rc network in the out ....
on the other hand i think that bobs comment was generall and not personal ....
from personal expirience though i never constructed or operated single rail amplifier (yet i am 42 years old ) so i have no epirience on the particular design .
still as far as i know the same rules apply for single rail amps ....rc should be there to protect the out ....
G.Kleinschmidt said:
I apologize if my characterization hit a nerve.
High-Q resonances at the output node of an amplifier can cause parasitic oscillations, and these are not good. One has very little control over the high-frequency nature of the reactive load formed by a random combination of loudspeaker cable and a loudspeaker. You want that Zobel there to cover the bases and assure that there is always some lossy element on that node at high frequencies.
Although many of these effects are out of the audio band, we must still pay attention to them when there are devices with fts of 10's or hundreds of MHz attached to those nodes. We cannot ignore the many possible transmission line effects of the speaker cables, even though they are out of the audio band, since we are worrying about parastic oscillations which can sometimes be in the VHF range. Ignore these seemingly improbable possibilities at your own risk.
I'll go a step further. If I was a loudspeaker manufacturer, I would always put an R-C Zobel right across the input terminals of my loudspeaker to reduce far-end unterminated line effects of the speaker cables.
Some loudspeakers need this more than others. Often, loudspeakers with a resistive tweeter pad directly across the speaker input at high frequencies (i.e., just through one or more series HPF capacitors) will need this less, especially if there is quite low inductance in series in that path. The idea is that above 20 kHz or so, you REALLY want the loudspeaker to look resistive, with a resistance probably between a few ohms and 50 ohms. You do not want to allow an undamped, unterminated line effect.
Obviously, it is impossible to properly terminate any random speaker cable in its characteristic impedance, but accurate termination is not necessary in order to avoid the problems I am referring to.
I agree that 90% or more of the time you may get away without a Zobel on the output of your amplifier, but why take the chance? If you are designing just for your own system, maybe that's OK. If you are designing a real product that you want to be tolerant of real-world conditions, then it may be less OK. In some cases, these sorts of things may account for situations where we hear about finiky component matching that is otherwise difficult to understand.
Cheers,
Bob
just repaired
a Sansui AU-X 501 !!!!!
nice device !!!! very powerfull .....
for this amplifier inductor is located in the main board no more than 8-10 cm away from output transistors
then this amplifier includes A-B selector switches so the rest of the filter is located in the speaker conector board, after cables and switches and actually you have one set of filters for speaker A and another set for speaker B .....
in conclusion this brings us much more closer to what bob just said ....
the inductor is made of 1.0 mm wire 12 turns on 12 mm diameter next to it there is a 3.3 R 1W resistor classic type flame proof .....
the filter in the out is made with 100nf mkt and 3W 4.7 R resistor also included a ceramic cap 100pf before the R-C network
thanks all for the input .....
and to stand to bob's side for a minute i consider my shelf a fooll for not puting an inductor to my amp either from safety point of view or from sound point of view ....
a Sansui AU-X 501 !!!!!
nice device !!!! very powerfull .....
for this amplifier inductor is located in the main board no more than 8-10 cm away from output transistors
then this amplifier includes A-B selector switches so the rest of the filter is located in the speaker conector board, after cables and switches and actually you have one set of filters for speaker A and another set for speaker B .....
in conclusion this brings us much more closer to what bob just said ....
the inductor is made of 1.0 mm wire 12 turns on 12 mm diameter next to it there is a 3.3 R 1W resistor classic type flame proof .....
the filter in the out is made with 100nf mkt and 3W 4.7 R resistor also included a ceramic cap 100pf before the R-C network
thanks all for the input .....
and to stand to bob's side for a minute i consider my shelf a fooll for not puting an inductor to my amp either from safety point of view or from sound point of view ....
my research
about this mosfet amplifier that created this thread actually ...is about to come to an end ....
it will be very nice some real experts might be able to tell us or advice us on why this amplifier likes to ocilate depending on the load connected to the out ( lines and speakers )
i am going to be able to put schematics , pcb layout and some history about it any day now ....
if you belive that this should be discussed here in this thread and also if you nice people willing to analyze this further please let me know .....
thank you all so far
best regrads sakis
about this mosfet amplifier that created this thread actually ...is about to come to an end ....
it will be very nice some real experts might be able to tell us or advice us on why this amplifier likes to ocilate depending on the load connected to the out ( lines and speakers )
i am going to be able to put schematics , pcb layout and some history about it any day now ....
if you belive that this should be discussed here in this thread and also if you nice people willing to analyze this further please let me know .....
thank you all so far
best regrads sakis
Bob Cordell said:
Previously I this thread I plainly stated that I am not advocating that anyone out there stops using RC Zobel networks, as even if one isn’t really required, it isn’t going to do any harm. If you are not sure, then it is better to be safe than sorry.
The only thing I have questions is some of the technical reasons put forward for including one. In a few designs for a specific purpose I have not incorporated an RC Zobel, for the simple reason that it just wasn’t required.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.