Bob, do you have any experience with ferrite bead at the gate, I remember you saying something like this?
If so, could you give some guidelines on how to choose a proper one for linear mode, I mean impedance at 25Mhz, at 100MHz, number of turns or so... and any results like bandwidth, stabiity, cross-conduction?
Best regards,
Adam
If so, could you give some guidelines on how to choose a proper one for linear mode, I mean impedance at 25Mhz, at 100MHz, number of turns or so... and any results like bandwidth, stabiity, cross-conduction?
Best regards,
Adam
Re: The Importance of current slew rate capability
I don't understand. Do you assume 50mA dynamic base current? To my understanding a base current can not only stop being pushed, but also can be reverse-sucked from base at fast transients. And for drivers which may well switch to classB that's much more than 50mA, note that most darlington output stages have a capacitor between emitters of drivers. Same goes to bjt EF+mosfet SF configuration.
Regards,
Adam
Bob Cordell said:
I don't understand. Do you assume 50mA dynamic base current? To my understanding a base current can not only stop being pushed, but also can be reverse-sucked from base at fast transients. And for drivers which may well switch to classB that's much more than 50mA, note that most darlington output stages have a capacitor between emitters of drivers. Same goes to bjt EF+mosfet SF configuration.
Regards,
Adam
Re: Cgd
Hi Edmond,
These things do tend to be design-dependent and layout-dependent.
The best way to handle the increasing Ccb problem is to use a Baker Clamp that prevents the output stage from trying to swing any closer than 5V to the rail. Under these conditions, the Ccb for an IRF240, for example, will be less than about 400 pF. Baker Clamps are always a good thing for clean clipping anyway. Moreover, staying 5V away from the rail tends to better keep the relatively dirty high-current main rail better isolated from the signal path.
Cheers,
Bob
Edmond Stuart said:
Hi Edmond,
These things do tend to be design-dependent and layout-dependent.
The best way to handle the increasing Ccb problem is to use a Baker Clamp that prevents the output stage from trying to swing any closer than 5V to the rail. Under these conditions, the Ccb for an IRF240, for example, will be less than about 400 pF. Baker Clamps are always a good thing for clean clipping anyway. Moreover, staying 5V away from the rail tends to better keep the relatively dirty high-current main rail better isolated from the signal path.
Cheers,
Bob
darkfenriz said:
Hi Adam,
I did try ferrite beads a bit a long time ago and did have some success with them, but I don't have any recent experience or recommendations for them. A ferrite bead that transitions to about 100 ohms resistive at 10 MHz or so might be a good choice.
I do wonder if there are some out there who would be concerned about the sound of a ferrite bead in the signal path. Sometimes a ferrite bead in the drain will help as well.
Cheers,
Bob
Re: Re: The Importance of current slew rate capability
Hi Adam,
I am assuming a driver with a standing current of 50 mA, so 50 mA of suckout current is available, assuming the driver stays in Class-A and no inter-emitter speedup capacitor is used. You are right, the use of a speedup capacitor can help this situation.
Note that one also needs to give some thought as to where this base charge is being dumped and whether it can introduce a bit of distortion of its own into the signal path.
Cheers,
Bob
darkfenriz said:
Hi Adam,
I am assuming a driver with a standing current of 50 mA, so 50 mA of suckout current is available, assuming the driver stays in Class-A and no inter-emitter speedup capacitor is used. You are right, the use of a speedup capacitor can help this situation.
Note that one also needs to give some thought as to where this base charge is being dumped and whether it can introduce a bit of distortion of its own into the signal path.
Cheers,
Bob
Re: MOSFET pole
Enclose the mosfet in Local current feedback loop, to act as Voltage Controlled Current Source. Its highly effective when you tend to drive the Mosfets near the rails. Also configure the driver in EF Push-pull fashion for driving the gate of Mosfets. Its highly effective scheme when dealing with wide variation of Cgd w.r.t to VDS.
Kanwar
Edmond Stuart said:
Enclose the mosfet in Local current feedback loop, to act as Voltage Controlled Current Source. Its highly effective when you tend to drive the Mosfets near the rails. Also configure the driver in EF Push-pull fashion for driving the gate of Mosfets. Its highly effective scheme when dealing with wide variation of Cgd w.r.t to VDS.
Kanwar
VCCS!!
http://www.diyaudio.com/forums/attachment.php?s=&postid=1281847&stamp=1187524503
CCCS!!
http://www.diyaudio.com/forums/attachment.php?s=&postid=1202773&stamp=1178522807
A very simple Rail to Rail with push-pull EF drivers, Max rail loss is less than 1 V, NO Cgd Trouble. No Rail-sticking during severe HF Clipping.
http://www.diyaudio.com/forums/attachment.php?s=&postid=937736&stamp=1150022131
Kanwar
http://www.diyaudio.com/forums/attachment.php?s=&postid=1281847&stamp=1187524503
CCCS!!
http://www.diyaudio.com/forums/attachment.php?s=&postid=1202773&stamp=1178522807
A very simple Rail to Rail with push-pull EF drivers, Max rail loss is less than 1 V, NO Cgd Trouble. No Rail-sticking during severe HF Clipping.
http://www.diyaudio.com/forums/attachment.php?s=&postid=937736&stamp=1150022131
Kanwar
dimitri said:
Hi Dimitri
That is quite a different situation, isn't it?
At mosfet's gate dynamic current should not exceed, say 5mA at normal operation, what's more, in your example ferrite bead is not within feedback loop, so this -50dB harmonics are not so scary.
It is like comparing a bjt emitter follower, which centainly consists of a b-e diode in signal path, to a rectifier.
Re: Ferrite Bead in Spkr Line
@Dimitri:
Thanks, lesson learned: Don't use "inductors" with varying inductance vs. current, which is exactly what seems to happen with lossy ferrites, in the audio band. AF is DC current bias from the ferrite's POV. To shield RF from the amp one should only consider common mode type of lossy/nonlinear filters, and even those might be compromised.
- Klaus
@Dimitri:
Thanks, lesson learned: Don't use "inductors" with varying inductance vs. current, which is exactly what seems to happen with lossy ferrites, in the audio band. AF is DC current bias from the ferrite's POV. To shield RF from the amp one should only consider common mode type of lossy/nonlinear filters, and even those might be compromised.
- Klaus
I haven't found lossy ferrite to measure that bad - I guess there are many formulations and geometries that affect saturation distortion in lossy ferrite parts - you do have to test
My TPA6120 based headphone amp uses a ferrite smt bead core to isolate the cable C, as I recall the distortion at 8 Vrms into 32 Ohms wasn't obvious with my Juli@ card using RMAA
further, even using Dimitri's example the distortion is odd harmonic so it should scale with at least the square of the current -40 dB less current should give -80 dB less distortion for -130 dB and that is inside the feedback loop in a gate stopper application
My TPA6120 based headphone amp uses a ferrite smt bead core to isolate the cable C, as I recall the distortion at 8 Vrms into 32 Ohms wasn't obvious with my Juli@ card using RMAA
further, even using Dimitri's example the distortion is odd harmonic so it should scale with at least the square of the current -40 dB less current should give -80 dB less distortion for -130 dB and that is inside the feedback loop in a gate stopper application
That is correct. When I put 15 A rated common mode choke (two wires through the same bead) in the same amp I found the same distortion signature but 20-30dB less. Please be aware about the symmetry of common mode choke.
That is correct, jcx. But what will be the peak gate current? Will it cause IMD inside audio band? My point was to warn those of you, who trust measurements. BTW in A/B comparison, golden ears preferred bead.
