First.
I see almost none using ferrite bead in the gate of mosfets in amplifiers.
Nelson Pass dont use them. And what he does not know about MOSFETs there is nothing to know.
Here is a quote when Bob Cordell was asked about ferrite beads on the gates.
I see almost none using ferrite bead in the gate of mosfets in amplifiers.
Nelson Pass dont use them. And what he does not know about MOSFETs there is nothing to know.
Here is a quote when Bob Cordell was asked about ferrite beads on the gates.
Ferrite bead is a component type not to be used in audio. It is not very linear so to say.
Otherwise, I would stick with the smallest one for this purpose.
Otherwise, I would stick with the smallest one for this purpose.
But for MOSFETs it should be okay, since the currents are small and there is no DC, right?
There is another idea I had about making a Zobel out of SMD inductors and resistors...
- keantoken
There is another idea I had about making a Zobel out of SMD inductors and resistors...
- keantoken
So, Bob said not to pickup stray fields... A very reasonable paranoia:
But simple beaded wire or leg, adds no extra loop for stray flux to cross.
Might even be a lossy barrier, like a radar proof paint? Or maybe not...
For chip inductor with multiple turns, maybe pickup is worth worrying.
What do we care here non-linear? Seriously, compared to mosfet CGD?
Besides, High Q inductance would be an absolute catastrophe in this app!
A resistor alone isn't free of parasitic inductance and capacitance either,
nor does its ability to damp increase with frequency.
I do final test on Class D audio all day long. I've auditioned plenty of beads.
That is to say I've never heard or measured a simple bead ever doing anything
evil in over three years, aside from finding the occasional cracked or defective
not doing its job...
Not saying I prefer the sound of Class D at home, too much like being at work.
But its not the bead's fault. I prefer bead grid stoppers on my Tube amps too.
Also fine for Zobels, I actually see that way more often than gate stopper.
But simple beaded wire or leg, adds no extra loop for stray flux to cross.
Might even be a lossy barrier, like a radar proof paint? Or maybe not...
For chip inductor with multiple turns, maybe pickup is worth worrying.
What do we care here non-linear? Seriously, compared to mosfet CGD?
Besides, High Q inductance would be an absolute catastrophe in this app!
A resistor alone isn't free of parasitic inductance and capacitance either,
nor does its ability to damp increase with frequency.
I do final test on Class D audio all day long. I've auditioned plenty of beads.
That is to say I've never heard or measured a simple bead ever doing anything
evil in over three years, aside from finding the occasional cracked or defective
not doing its job...
Not saying I prefer the sound of Class D at home, too much like being at work.
But its not the bead's fault. I prefer bead grid stoppers on my Tube amps too.
Also fine for Zobels, I actually see that way more often than gate stopper.
Last edited:
A (carbon) gate resistor (and a decoupling capacitor) can provide appropriate protection and stability. The instability is likely to be caused by something else.
One should exercise vigilance when importing solutions from other branches of electronics into audio. In my view, it is best to avoid ferrite beads, diodes, ceramic capacitors and inductive resistors.
One should exercise vigilance when importing solutions from other branches of electronics into audio. In my view, it is best to avoid ferrite beads, diodes, ceramic capacitors and inductive resistors.
Vigilance I occasionally get to measure on an Audio Precision at my customer's lab.
In my own lab I have to settle for 2.5GHz Tek scope with active probes...
In my own lab I have to settle for 2.5GHz Tek scope with active probes...
kenpeter,
are you working on the bias circuit with output current sensing and thermal compensation to be used with any FET, should be simple for a precise function, not doing too much damage to the signal. (I can replace the diodes with resistors later).
😀
are you working on the bias circuit with output current sensing and thermal compensation to be used with any FET, should be simple for a precise function, not doing too much damage to the signal. (I can replace the diodes with resistors later).
😀
Last edited:
I think someone should try it, with their assembled prototype. We've had much sharing of views, but no one has come forth with experience with the sound or effectiveness of beads in MOSFET amplifiers. So I think we should verify at least for this amp.
First, to take off the gate stoppers to see what happens, and then add the beads and have a listen.
Mouser has carbon comp. resistor kits for around $25, I got myself one. (my 2.2R from that kit are actually, .5W instead of .25W, maybe they're NOS?)
- keantoken
First, to take off the gate stoppers to see what happens, and then add the beads and have a listen.
Mouser has carbon comp. resistor kits for around $25, I got myself one. (my 2.2R from that kit are actually, .5W instead of .25W, maybe they're NOS?)
- keantoken
Paranoia from ancient bad experiences with saturating cores in loudspeaker crossovers should not be freely associated with the general use of any ferrite materials, particularly RF grades used on Mosfet gates. If you read Bob Cordell's comments in his book about why ferrite beads should not be used, you'll see it is more not to anger some readers, such as those who associate component behaviours with little understanding of physics and how such problems arise.
There are plenty of components such as wire-wound resistors that may be more "harmful" to your precious audio electrons in some locations. Even so, I would be doubtful of the use in some BJT output bases where peak currents can be quite high, by several orders over Mosfet gate current.
I still construct and modify the occasional Mosfet amp and always include RF grade ferrites on the gates so as to allow low value stoppers without G-D zobels as Bob advocates, presumably as an alternate strategy. I could not detect any sonic effect, but it is certainly cleaner on the 'scope and more reliably stable than using high value stoppers alone. That is a reasonable concern for equipment to be used by other people, I think.
There are plenty of components such as wire-wound resistors that may be more "harmful" to your precious audio electrons in some locations. Even so, I would be doubtful of the use in some BJT output bases where peak currents can be quite high, by several orders over Mosfet gate current.
I still construct and modify the occasional Mosfet amp and always include RF grade ferrites on the gates so as to allow low value stoppers without G-D zobels as Bob advocates, presumably as an alternate strategy. I could not detect any sonic effect, but it is certainly cleaner on the 'scope and more reliably stable than using high value stoppers alone. That is a reasonable concern for equipment to be used by other people, I think.
Ferrite bead is an interesting idea.
Its inductive impedance at audio frequencies should be below an ohm, if your worried about it you can still put another resistor in series with it.
Just recommended the Fair-Rite sampler kit on another thread.
Various materials suitable for a few hundred ohms to nearly 100Mhz.
Impedance could become significant at amps cross-over frequency if above 1Mhz. As mentioned peaking should be checked and eliminated by choice of material even a parallel resistor may be helpful.
Looking forward to hearing what becomes of this.
You guys are amazing
Thanks
-Antonio
Its inductive impedance at audio frequencies should be below an ohm, if your worried about it you can still put another resistor in series with it.
Just recommended the Fair-Rite sampler kit on another thread.
Various materials suitable for a few hundred ohms to nearly 100Mhz.
Impedance could become significant at amps cross-over frequency if above 1Mhz. As mentioned peaking should be checked and eliminated by choice of material even a parallel resistor may be helpful.
Looking forward to hearing what becomes of this.
You guys are amazing
Thanks
-Antonio
Probably not what you were wanting, but whats on my screen right now.
Current limits kick in at 6A, but one side still limiting with harsher knee.
Damn single ended PP output stage always a pain to current limit with
any degree of symmetry. Replacing load with 0.1R to test limits.
I think I got it stable now with -3db at 96KHz, and soft clipping before
the amp can do anything too stOOpid. Offset mostly fixed. Devil always
resides in the details...
There has to be an easier way to trick offset of J1 to exactly cancel
the offset of Q5?
Attachments
Last edited:
Ken, with a 10k input resistor, 1N4148's should do fine for the input limit diodes. The fat ones you're using may have hundreds of pF of nonlinear, lossy capacitance.
The issue I see with J1 compensating Q5 is that Q4 has no Vce. It will have much less dissipation not to mention Early effect. Maybe you should try cascoding Q5 at 0V Vce? Also, try making Q4 a current mirror (with output to Q5 emitter), to exactly cancel the current through the feedback loop. This will need to be a cascoded current mirror (not wilson, to maintain the diode connection of Q4). I hope all your transistors are well-matched or you have some well-placed trimmers, and a clear method of adjustment.
- keantoken
The issue I see with J1 compensating Q5 is that Q4 has no Vce. It will have much less dissipation not to mention Early effect. Maybe you should try cascoding Q5 at 0V Vce? Also, try making Q4 a current mirror (with output to Q5 emitter), to exactly cancel the current through the feedback loop. This will need to be a cascoded current mirror (not wilson, to maintain the diode connection of Q4). I hope all your transistors are well-matched or you have some well-placed trimmers, and a clear method of adjustment.
- keantoken
http://www.fairchildsemi.com/ds/MB/MBR0540.pdf
I see cap curve. 80-100pF as you spotted, would be very audible.
LT's model obviously didn't roll at 10KHz, another case of trusting
a little too much.
But I want to tailor a soft clip threshold that won't knee too early.
I don't think 1N4148 will give right curvature or glitchless recovery.
Probably demands a small signal Schottky, I just need to pore over
some spec sheets to find one with less capacitance. At least its not
phase shifting or clipping inside the feedback loop.
As for the current mirror and cascode obsessions, yeah whatever.
I could throw 30 transistors at that, but I'd rather knock the count
down than up. Too much clutter added already. Besides, something
in this non-ideal mirror is cancelling a lot of 2nd H distortion...
I see cap curve. 80-100pF as you spotted, would be very audible.
LT's model obviously didn't roll at 10KHz, another case of trusting
a little too much.
But I want to tailor a soft clip threshold that won't knee too early.
I don't think 1N4148 will give right curvature or glitchless recovery.
Probably demands a small signal Schottky, I just need to pore over
some spec sheets to find one with less capacitance. At least its not
phase shifting or clipping inside the feedback loop.
As for the current mirror and cascode obsessions, yeah whatever.
I could throw 30 transistors at that, but I'd rather knock the count
down than up. Too much clutter added already. Besides, something
in this non-ideal mirror is cancelling a lot of 2nd H distortion...
Last edited:
Why hunt the soft clipping.. why not just decrease gain so you never go there even when taking the full input voltage-sving....??
I can understand the need for this arrangement in commercial and professional gear.. but for DIY....I don't get it....
Oh yes for the circuit vise exercise...🙂
I can understand the need for this arrangement in commercial and professional gear.. but for DIY....I don't get it....
Oh yes for the circuit vise exercise...🙂
Because it sounds better. No such thing as an amp that never clips.
I don't want to listen to diodes, but its better than listening to a
a feedback loop try to recover from the rail...
An amp I can't stand to listen at nearly full volume is worthless.
Real music has incredible dynamic range to challenge any amp.
Do you own a turntable in your home? Ever hear a click or pop?
How quick does your amp get back to normal?
I don't want to listen to diodes, but its better than listening to a
a feedback loop try to recover from the rail...
An amp I can't stand to listen at nearly full volume is worthless.
Real music has incredible dynamic range to challenge any amp.
Do you own a turntable in your home? Ever hear a click or pop?
How quick does your amp get back to normal?
Last edited:
Hey guys, I just fixed my signal generator... I plugged it in. 
I am seeing very slight ringing at 50MHz, but I don't know how to tell if this is just bad cable termination (which is none) or if there is actually something wrong. Can anyone recommend a good online tutorial for transmission line basics? (EDIT: I think it's termination because I switched to a longer cable and it moved down to 30MHz)
MiiB, you never know, Kenpeter may get rich off his ideas and make fools out of us all.
Ken, look at Cordell's website for a list of what I'm sure you'll find to be highly accurate models (barring inadequacies of the model type). He has diodes, LEDs, and most of the popular bipolars of modern times. Also Jfets, MOSFETs... You should really check it out... (he has a Schottkey model for the BAT85, 10pF@1V)
And, just a suggestion, this is also a good place to look:
Simulation Tools
- keantoken
I am seeing very slight ringing at 50MHz, but I don't know how to tell if this is just bad cable termination (which is none) or if there is actually something wrong. Can anyone recommend a good online tutorial for transmission line basics? (EDIT: I think it's termination because I switched to a longer cable and it moved down to 30MHz)
MiiB, you never know, Kenpeter may get rich off his ideas and make fools out of us all.
Ken, look at Cordell's website for a list of what I'm sure you'll find to be highly accurate models (barring inadequacies of the model type). He has diodes, LEDs, and most of the popular bipolars of modern times. Also Jfets, MOSFETs... You should really check it out... (he has a Schottkey model for the BAT85, 10pF@1V)
And, just a suggestion, this is also a good place to look:
Simulation Tools
- keantoken
Last edited:
I really do hope KenPeter get's rich from his ideas....as well as I see you picked up by a good R/D section making amplifiers....
And for sure turntable is another story, as the transients there can get quite big,,, but then there is always the possibility of making something with higher rails....🙂
And for sure turntable is another story, as the transients there can get quite big,,, but then there is always the possibility of making something with higher rails....🙂
I am never getting rich working 70+hrs at this CM.
Ramen noodle salary, Kean can have it when I leave.
Cut back 22% since 2.5 yrs ago. While the workload
expectations for this sad paycheck more than triple.
The ships not sinking. A great comfort that must be
to the men drowning in the engine room. Wonder if
Foxconn's is still hiring?
Chief test monkey pays nothing, its pure slave labor.
Ideas rarely taken seriously by degreed "engineers".
You won't get anywhere without the magic paper...
----
No matter how high the rails, or Amps, or Watts, you
should feel unsatisfied to never have full use of them.
All due an occasional transient we elect not to handle
gracefully in some manner.
The objective is to reproduce music, not sine waves...
Ramen noodle salary, Kean can have it when I leave.
Cut back 22% since 2.5 yrs ago. While the workload
expectations for this sad paycheck more than triple.
The ships not sinking. A great comfort that must be
to the men drowning in the engine room. Wonder if
Foxconn's is still hiring?
Chief test monkey pays nothing, its pure slave labor.
Ideas rarely taken seriously by degreed "engineers".
You won't get anywhere without the magic paper...
----
No matter how high the rails, or Amps, or Watts, you
should feel unsatisfied to never have full use of them.
All due an occasional transient we elect not to handle
gracefully in some manner.
The objective is to reproduce music, not sine waves...
Last edited:
this is not a quest for right and wrong....I genuinely like your idea of non switching output-stage...that maybe more than the the idea of transient limiting in the input of an amplifier..that latter has some notes of compression built into it....where as the non-switching relieves the feedback from dealing with nasty spikes...Not sure how the output stage will sound with more sand in the signal path...but for sure an interesting approach....!
At least it's an energy efficient way to improve on the class B shortcomings..
the two together could form a basis for an efficient good sounding soft clipping professional amplifier..that potentially could compete with class D designs...
At least it's an energy efficient way to improve on the class B shortcomings..
the two together could form a basis for an efficient good sounding soft clipping professional amplifier..that potentially could compete with class D designs...
Mike,
How big do you think those switching spikes really are?
Like you, I really like the non-switching idea, but I'm no longer convinced the spikes amount to much at all from a listening POV.
Ken,
Great to hear you say, 'because it sounds better'. Long may this approach prosper.... I think ears are actually very sensitive.
Hugh
How big do you think those switching spikes really are?
Like you, I really like the non-switching idea, but I'm no longer convinced the spikes amount to much at all from a listening POV.
Ken,
Great to hear you say, 'because it sounds better'. Long may this approach prosper.... I think ears are actually very sensitive.
Hugh
- Status
- Not open for further replies.