Sorry, I am late here.
Dear opor, With reference to the thread no. #1094. Is the PCB incomplete? because when I opened it in protel it shows two unrouted connections.
Last edited:
they do, but not as an end drivers. if they were, you would need to feed the base with current, which would run out before the end of the period, unless your upper boost capacitor was big. But then IR would produce too much heat. It has to be IGBT or FET for end transistor and even then, input capacitance can't be off the charts
I have new problems🙁 and i need your help once again😀.
a)When using both my amplifier modules I have huge rf interference resulting in high frequency noises.Isolating suplies doesn't help.I have to efectively put the amplifiers 15 cm apart for them to work good and that is not acceptable.They are at least 30 khz different in switching frequency.
b)Also i can not use one heat sink since this increases ten fold these perturbations and i can see huge common mode noise.
Syncroning switching frequencies by injecting square wave signals into negative input thought a capacitor and resistor solves some first case interferences but increases distorsions alot especialy at higher modulation so this approach can not be used🙁 and anyway it doesn't solve case b).
In case b) the heat sink injects hight frequency noise into ground because there is at least a 70pf capacitance between one transistor and heat sink and also acts like emiting antenna fueld by any diffrence betwhen switching outputs between chanels since these will never be perfectly sincronized even with exactly the same frequency and phase because input signals are not identical.But i see lots of solutions with only one heat sink and multiple channels and they work ok so i am missing something.
Thanks for any ideea you may have...
a)When using both my amplifier modules I have huge rf interference resulting in high frequency noises.Isolating suplies doesn't help.I have to efectively put the amplifiers 15 cm apart for them to work good and that is not acceptable.They are at least 30 khz different in switching frequency.
b)Also i can not use one heat sink since this increases ten fold these perturbations and i can see huge common mode noise.
Syncroning switching frequencies by injecting square wave signals into negative input thought a capacitor and resistor solves some first case interferences but increases distorsions alot especialy at higher modulation so this approach can not be used🙁 and anyway it doesn't solve case b).
In case b) the heat sink injects hight frequency noise into ground because there is at least a 70pf capacitance between one transistor and heat sink and also acts like emiting antenna fueld by any diffrence betwhen switching outputs between chanels since these will never be perfectly sincronized even with exactly the same frequency and phase because input signals are not identical.But i see lots of solutions with only one heat sink and multiple channels and they work ok so i am missing something.
Thanks for any ideea you may have...
I would bet on EMI EMI EMI, something to think about and hope others will be more helpful then my post
I tried my best to design the board to minimize EMI using short traces and ground planes but at this point i have no ideea what i did wrong🙁
hi cpx
my old design the same problem found. what i did was 10ohm connect heatsink with GND main and all the gnd connected to main GND(signal&spk)and +&-supply through 10uH coil
RGDS
MANOJ
my old design the same problem found. what i did was 10ohm connect heatsink with GND main and all the gnd connected to main GND(signal&spk)and +&-supply through 10uH coil
RGDS
MANOJ
Im have seen irs2092 modules with 1 or 2 chanels with one heat sink and even no ground plane like irdaamp7 which work ok,how is that posble,is
irs2092 or pre filter feedback less susceptible to rf noise?
irs2092 or pre filter feedback less susceptible to rf noise?
Last edited:
hi cpx
10E with 2KPf parallel connected to GND and use small ferrite with 3turns connected from main GND to signal gnd noise will reduce.
cpx my old thread i mentioned ground plane is creating problem for me but nobody supported.avoid ground plane on MOSFET & Filter side OR do not use ground plane,its my opinion other side i don't know.
RGDS
Manoj
10E with 2KPf parallel connected to GND and use small ferrite with 3turns connected from main GND to signal gnd noise will reduce.
cpx my old thread i mentioned ground plane is creating problem for me but nobody supported.avoid ground plane on MOSFET & Filter side OR do not use ground plane,its my opinion other side i don't know.
RGDS
Manoj
hi cpx
my new design (you said its BAD PCB Design) i used yellow/white core with 30Turns 16SWG wire and 3x IRFP240, core temp. is 35-38deg.c MOSFET Temp. 38-40deg.c low EMI
tested voltage with +42V,-42V DC load 5E 50W resistor stead @full volume for 1hr.
i never taken THD measure,DC offset is -18mV to -24mV.
everyone said IRFP240 & yellow/white core not good for class D. My curiosity is practically what happen in this components
IRF240 is Qg is affordable rate clamp diode not good solving this connect good diode externally,my way like this
RGDS
Manoj
my new design (you said its BAD PCB Design) i used yellow/white core with 30Turns 16SWG wire and 3x IRFP240, core temp. is 35-38deg.c MOSFET Temp. 38-40deg.c low EMI
tested voltage with +42V,-42V DC load 5E 50W resistor stead @full volume for 1hr.
i never taken THD measure,DC offset is -18mV to -24mV.
everyone said IRFP240 & yellow/white core not good for class D. My curiosity is practically what happen in this components
IRF240 is Qg is affordable rate clamp diode not good solving this connect good diode externally,my way like this
RGDS
Manoj
hi manojtm
greetings how does your yellow core not heat just finished my dclass 3 pairs
90n20 have to use 2 cores together sendust coolmu cores yellow cores dont work
warm regards
andrew lebon
greetings how does your yellow core not heat just finished my dclass 3 pairs
90n20 have to use 2 cores together sendust coolmu cores yellow cores dont work
warm regards
andrew lebon
