yes but isn't that the case on any sensitive part of the modulator?
source is 6k8 (rin of the integrator) see above- obviously you can use whatever opamp you like for a one off; my company builds several hundred at a go so we need repeatability, which this gives.
Basically its all down to layout and the old phrase SIMSO (SH*t in more SH*t out)
source is 6k8 (rin of the integrator) see above- obviously you can use whatever opamp you like for a one off; my company builds several hundred at a go so we need repeatability, which this gives.
Basically its all down to layout and the old phrase SIMSO (SH*t in more SH*t out)
EMI spike cross-talk in self oscillating amplifiers manifests itself as an increased noise floor, it sounds like "frying" noise. In a good stereo design you have to make one channel play full power music into a dummy load and attach a loudspeaker to the other channel to perceive this, and the noie is not loud at all. There are no "birds", this can only happen when a fixed clock modulator gets disturbed.
Self oscillating amplifier channels suffering from EMI tend to mutually synchronize when their frequencies are close enough. Also, synchronization in clocked designs only masks the EMI problem, it does nothing to reduce the distortion resulting from stray magnetic field pickup, it only masks the problem when the amplifier is idle
On the other hand, self oscillating amplifiers tend to exhibit a better rejection to EMI and stray magnetic fields than clocked designs, both due to their delta-sigma nature, because the modulators are simpler and smaller and because they are easily made balanced.
Balanced triangle generators, comparators and feedback are a *must* for class D because external interferences tend to cancel themselves. Unbalanced modulators are too sensitive to EMI and stray magnetic fields. For example, balanced clocked designs would probably no longer require synchronization to get the "birds" below the noise floor.
In one of my latest prototypes I have a (balanced) self oscillating modulator idling at 126Khz and a 132Khz TNY266 flyback controller whose transformer has its windings on the same spatial plane as the modulator PCB (silly mistake, I know). To make it worse, this flyback is providing floating +/-15V for the modulator and +18V for gate drive, so the grounds are unavoidably joined and shared. However, there are no audible crosstalk effects, there was some faint frying noise at the beggining but it disappeared when I increased the value of the "Y" capacitor of the flyback from 220p to 2.2n (the fast turn off from the flyback was producing too much common-mode "ground bouncing" EMI).
PD: Please continue hating and discarding self oscillating moduladors without knowing them in depth. Please continue using only unbalanced clocked modulators for the professional market. This is going to be great, I will not have any competitors to worry about
Self oscillating amplifier channels suffering from EMI tend to mutually synchronize when their frequencies are close enough. Also, synchronization in clocked designs only masks the EMI problem, it does nothing to reduce the distortion resulting from stray magnetic field pickup, it only masks the problem when the amplifier is idle
On the other hand, self oscillating amplifiers tend to exhibit a better rejection to EMI and stray magnetic fields than clocked designs, both due to their delta-sigma nature, because the modulators are simpler and smaller and because they are easily made balanced.
Balanced triangle generators, comparators and feedback are a *must* for class D because external interferences tend to cancel themselves. Unbalanced modulators are too sensitive to EMI and stray magnetic fields. For example, balanced clocked designs would probably no longer require synchronization to get the "birds" below the noise floor.
In one of my latest prototypes I have a (balanced) self oscillating modulator idling at 126Khz and a 132Khz TNY266 flyback controller whose transformer has its windings on the same spatial plane as the modulator PCB (silly mistake, I know). To make it worse, this flyback is providing floating +/-15V for the modulator and +18V for gate drive, so the grounds are unavoidably joined and shared. However, there are no audible crosstalk effects, there was some faint frying noise at the beggining but it disappeared when I increased the value of the "Y" capacitor of the flyback from 220p to 2.2n (the fast turn off from the flyback was producing too much common-mode "ground bouncing" EMI).
PD: Please continue hating and discarding self oscillating moduladors without knowing them in depth. Please continue using only unbalanced clocked modulators for the professional market. This is going to be great, I will not have any competitors to worry about
Eva...Alaways on deffensive! I never said that I hate self oscilating design! I just prefered something that I can predict performance...I found that's there too many variable in self oscilating design...I test UCD, IcePower and some chinees crap and I didnt like it. When someting didnt perform properly at first test, I have doubt. I'm at the base an aerospace engineer and I'm very award about EMI and disturbing stuff. I'm used to work with micro Volt in NAVAID guidance system at hundreds MHZ, so dont worry about my EMI tolerance!
Go with selfoscilating design! I'm just asking why Crest, Crown, QSC, Powersoft and Lapgruppen use clocked design....Oh...I think I know why! In big setup, with many SEPARATE AMPLIFIER, they have proved that they are less sensitive to beating, bird and frying than self oscilating design! Just make a 3 channel with Ice power and look carefully what happen....But please, dont use 88db tweeter, use a 110Db 2'' driver....You will be surprise...Same as with UCD, even with 3 different PSU, switching or not!
Maybe I dont have the same knowledge about PCB layout and resonance in to-220 leg, but I have experiment over 200 prototype of class d amplifier, UCD style, hysterisys, sigma delta, pre-post feedback design, phase modulation, switched rail psu, floating rail, etc....I learn a lot in what to do and what not to do...
I really admir that you can do a first try good PCD and design, but just stop to attack each other that design with real prototype, modify, optimise, try new approach or new idea....Reality is not just in book, it's in lab too. That's funny to built and try something, I like that and I will continue to try new way and approach. I built pro amp for over 10 years. My first serie was full of bug and bad conception ( as like Peavey and crown stuff in that time), but I still keep in production my HVI serie for the past 4 years, with minimal modification inside...Some PCB re-routing, modification in ground plane, optimisation of protection, re-design of the case, etc, but nothing major in the electrical design. I still use IXYS and APT power device, some STW in smaller model, same old LM319 and same old opto mosfet drivers. As for now, nothing better found.
That's all, I just take it a bit personal that you walk on design of other people that way, specialy that's not the first time..
Continue your good work, to be honnest I learn a lot with your contribution in this forum, I just dont like your attitude sometime..I will try to continue to contribute in my way too.
Have a nice day!
Fredos
Go with selfoscilating design! I'm just asking why Crest, Crown, QSC, Powersoft and Lapgruppen use clocked design....Oh...I think I know why! In big setup, with many SEPARATE AMPLIFIER, they have proved that they are less sensitive to beating, bird and frying than self oscilating design! Just make a 3 channel with Ice power and look carefully what happen....But please, dont use 88db tweeter, use a 110Db 2'' driver....You will be surprise...Same as with UCD, even with 3 different PSU, switching or not!
Maybe I dont have the same knowledge about PCB layout and resonance in to-220 leg, but I have experiment over 200 prototype of class d amplifier, UCD style, hysterisys, sigma delta, pre-post feedback design, phase modulation, switched rail psu, floating rail, etc....I learn a lot in what to do and what not to do...
I really admir that you can do a first try good PCD and design, but just stop to attack each other that design with real prototype, modify, optimise, try new approach or new idea....Reality is not just in book, it's in lab too. That's funny to built and try something, I like that and I will continue to try new way and approach. I built pro amp for over 10 years. My first serie was full of bug and bad conception ( as like Peavey and crown stuff in that time), but I still keep in production my HVI serie for the past 4 years, with minimal modification inside...Some PCB re-routing, modification in ground plane, optimisation of protection, re-design of the case, etc, but nothing major in the electrical design. I still use IXYS and APT power device, some STW in smaller model, same old LM319 and same old opto mosfet drivers. As for now, nothing better found.
That's all, I just take it a bit personal that you walk on design of other people that way, specialy that's not the first time..
Continue your good work, to be honnest I learn a lot with your contribution in this forum, I just dont like your attitude sometime..I will try to continue to contribute in my way too.
Have a nice day!
Fredos
There is one particular self oscillating product that not long ago was regarded as "high tech" by a very few. It happens to sound like a flock of migrating birds escaping a hundred year forest fire, only the skies are in flames as well. Much as I try to visualize happy little birds going about their day, I can't help but hear them crying out in agony as they burn alive.
You based your prototypes off an already proven concept, one that beat out the rest and took a generous lead by several years before you found it. They're not all equivalent.
Other than all those who blazed the trail you only now walk on.
when i designed my clocked comercial BD solution it was the best in field, this was 2001; now much better silicon, and techniques are available and i am open to new ideas- hence my emails to eva- sorry i have not emailed for 3 months!
I like self oscillation principles, indeed they have the best potential, I was just sharing my knoloedge practically as one who has commercially developed this type of product, and yes Fedos is right, dividing a clock reduces jitter, plus this crystal is very common, therefore cost effective! a 400khz crystal is $4+, 11.2896 +hc4060 <$1
a note to all, regardless of amplification method, poor layout=poor emc = problems
I sit pretty as my guys can test amplifiers with an am radio on talk sport ( a UK thing) less than 250mm away, with no lid on- give me a better practical emc test!!!!
BTW no other commercial module i know can we do this with- and we have tried quite a few-
for the emc concious, think of a fet, add an insulator, place this on a heatsink, and expose this to the world- you have a capacitive coupled radiator, it only radiates HF though- oops
regards,
I like self oscillation principles, indeed they have the best potential, I was just sharing my knoloedge practically as one who has commercially developed this type of product, and yes Fedos is right, dividing a clock reduces jitter, plus this crystal is very common, therefore cost effective! a 400khz crystal is $4+, 11.2896 +hc4060 <$1
a note to all, regardless of amplification method, poor layout=poor emc = problems
I sit pretty as my guys can test amplifiers with an am radio on talk sport ( a UK thing) less than 250mm away, with no lid on- give me a better practical emc test!!!!
BTW no other commercial module i know can we do this with- and we have tried quite a few-
for the emc concious, think of a fet, add an insulator, place this on a heatsink, and expose this to the world- you have a capacitive coupled radiator, it only radiates HF though- oops
regards,
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.