Yes..you need a higher signal level...
You have to cut the center leg of the core.Use a rotational stone cutter..be carefull ferrite is breaking easy.
Last edited:
10000/470=~21
So you need ~21 volts..but i used r7=82kohm so i need around 5.72 volts rms..anyway this is testing only..i will search for the lowest stable value for r8 and highest for R7
Last edited:
I can not stop 50hz hum with gapped core..with ungapped core i was able to reduce it to very low levels..
With gain 10x also white noise is very smal but then it is necesary to add ultra low noise input buffer with gain 3x..
But I have many more problems..amp is not souding good at all...please try to listen with 0.22 ohm load at low volume..it sounds distorted or not?..does output level decrease with added load or not?
With gain 10x also white noise is very smal but then it is necesary to add ultra low noise input buffer with gain 3x..
But I have many more problems..
amp is not souding good at all...please try to listen with 0.22 ohm load at low volume..it sounds distorted or not?..does output level decrease with added load or not?With smaller impedance distorsion is bigger. 4ohms load has bigger distorsion of 8ohms load. This is normal for all kind of amlifiers. Ofcourse you have lot of distorsion on 0.22 ohms load.
P.S I dont know why do you have a noise. Probably because you still use test board and bad grounding. Whit PCB and good grounding I don have noise, and I use gaped EE core. Well, I dont know, maybe little when I'm very close to speaker, but many comercial AB and G amps have it. Main problem with noise is in input part, grounding and gain (that was problem with my desing)
Regards
P.S I dont know why do you have a noise. Probably because you still use test board and bad grounding. Whit PCB and good grounding I don have noise, and I use gaped EE core. Well, I dont know, maybe little when I'm very close to speaker, but many comercial AB and G amps have it. Main problem with noise is in input part, grounding and gain (that was problem with my desing)
Regards
Last edited:
In some kind of good PA AB amps, I have seen two GND's. One gnd is power gnd (central wire from transformer and capacitors) and second gnd is input gnd ( from signal source).
This two GND's are not the same wire on PCB. All input stage of amp is input gnd and finaly power GND is conected to input GND accros 10 ohms 2W resistor on PCB.
Try this one, maybe helps.
This two GND's are not the same wire on PCB. All input stage of amp is input gnd and finaly power GND is conected to input GND accros 10 ohms 2W resistor on PCB.
Try this one, maybe helps.
I have 4 grounds(Speaker Gnd,Capacitor Gnd,Digital Gnd and Signal Gnd) they all go to star Gnd .. i tried all kinds of gnd versions and wire positions..they all have noise,feeadback wire (even shielded) picks it up from my gapped toroid and oscillation bus...anyway i measure it with mai ear glued to the speaker..i am comparing to my receiver which sounds completly dead,like unplugged....
I will try the resistor to see if it changes anything.....
Thanks Dzony988 for your help..if i have better results or more questions i will post again.
,feeadback wire (even shielded) picks it up from my gapped toroid and oscillation bus...anyway i measure it with mai ear glued to the speaker..i am comparing to my receiver which sounds completly dead,like unplugged....I will try the resistor to see if it changes anything.....
Thanks Dzony988 for your help..if i have better results or more questions i will post again.
Last edited:
what is percent of THD 10V befere clip? What is supplay voltage and what is imedance of load?
What bridge do you use? If you use fast diodes, you will never kill 50Hz noise ( some of harmonics will pass accros diode). for main bridge you have to use slowest diode because the most slowest diode on the world is stll very fast to distroy 50Hz noise. You have to slow your bridge with 4x 100nF in parallel with all 4 diodes in bridge
What bridge do you use? If you use fast diodes, you will never kill 50Hz noise ( some of harmonics will pass accros diode). for main bridge you have to use slowest diode because the most slowest diode on the world is stll very fast to distroy 50Hz noise. You have to slow your bridge with 4x 100nF in parallel with all 4 diodes in bridge
I already have 2x220uf, 1x470uf and 6x100 nF on each rail (4x100nF and 470uf cvlose to mosfets)
I also tried conecting coil far for the board..
I do not understand what is wrong..now i have distorsions starting 10 volt before clipping point..
Could you send the exact schematic and pcb layout you have? And if you can the photo of the test setup! I am sure i can give lots of ideas.
What type of core do you use? Gapped toroid in my opinion is worse then a gapped EE, ETD or RM ferrite. Iron powder (micrometals material -2) toroids are usually good, and has the pro that magnetic field is same all around. In a gapped toroid the B function has a shift around the gap, leaking radiation.
For example try some of the Coilcraft shielded flat-wire wound inductors. They are extremely good, two of VER2923-103K connected series will give you 20uH 5mOhm, and >30A saturation current.
Use one big ground plane! If you can, use a two-sided board, even if you have no wire on top, use top layer as a big ground plane! See Bruno's ideas about star-ground vs ground plane.
Only if necessary separate ground planes as told in IR's manual (analog input, V- the common of IR2110, and power otuput ground)
My experiments have shown much reduced noise and EMI with one big ground plane.
Are you still useing the old IRF540 (no N no Z)? Even with the schottkys, it will produce EMI.
I reccomend NOT to use that core for class D audio.
Core losses will be high at the switching freq, saturation current is not high enough. This core material is not so suitable for this use.
Use micrometals -2 or -6 instead, if iron powder you want. Or use gapped ferrite. RM cores are amongst the best, since they provide shielding, low PCB space. The previously given coilcraft one was an only an example, but it is one of the bests for this use.