Hi Chris !
Yes, eddy current seem the only explanation to me...
Will try bypassing, if still heating up, than it is definitely eddy current.
Or i put some other piece of metal at this place and check if it heats up !
(Like a non magnetic coin)
This circuit starts frightening me ! I am waiting for this amp to open
a tear in the timespace-continuom...
Mike
Yes, eddy current seem the only explanation to me...
Will try bypassing, if still heating up, than it is definitely eddy current.
Or i put some other piece of metal at this place and check if it heats up !
(Like a non magnetic coin)
This circuit starts frightening me ! I am waiting for this amp to open
a tear in the timespace-continuom...
Mike
Hi Mike,
Heh, yeah, fun stuff or what? I think they (class d)have to be more exciting doing it yourself than other types of amps, there's so much more going on. Seems to make it more of a challenge though too.
It is strangely amusing/interesting that your fuse holders would heat like that, but I fully agree with you eddy current is a very good explanation for it.
I just took a look at the pic of your board again, seeing where your fuses are placed, at either end of the coil, makes alot of sense.
Also I take it those are your filter caps between the coil ends and the fuses? They must be taking an EMI beating as well, I'm wondering what kind of effect that'd have on the sound.
Might be worth it to try standing that coil on end.
Best Regards,
Chris
Heh, yeah, fun stuff or what? I think they (class d)have to be more exciting doing it yourself than other types of amps, there's so much more going on. Seems to make it more of a challenge though too.
It is strangely amusing/interesting that your fuse holders would heat like that, but I fully agree with you eddy current is a very good explanation for it.
I just took a look at the pic of your board again, seeing where your fuses are placed, at either end of the coil, makes alot of sense.
Also I take it those are your filter caps between the coil ends and the fuses? They must be taking an EMI beating as well, I'm wondering what kind of effect that'd have on the sound.
Might be worth it to try standing that coil on end.
Best Regards,
Chris
MikeB said:
That will be the first encounter of Class D aliens!I'm very pleased to see so much explanation about class D amps in this thread: you are very good boys, as Carlos said.
Maybe I am getting interested into this: would you post some board layout when you do it? And some complete schematics and parts list? What are the power specs of the amp?
Thanks in advance!
Hmm, maybe if i build more amps and place them in a circle, i get a stargate ?
I just ordered some Amidon-rings (106-2 & 106-6)...
I was not sure which material, so i ordered red and yellow...
Giaime, i will post PCB if i make one...
Poweroutput is somewhere at 60watts into 8 ohm.
If working perfect it should be able to give 160 watts into 4ohm,
but i doubt that...
As the edges got steeper, EMI might have increased, and that might
be the reason for decreased soundquality ?
Mike
I just ordered some Amidon-rings (106-2 & 106-6)...
I was not sure which material, so i ordered red and yellow...
Giaime, i will post PCB if i make one...
Poweroutput is somewhere at 60watts into 8 ohm.
If working perfect it should be able to give 160 watts into 4ohm,
but i doubt that...
As the edges got steeper, EMI might have increased, and that might
be the reason for decreased soundquality ?
Mike
MikeB said:Hmm, maybe if i build more amps and place them in a circle, i get a stargate ?
I just ordered some Amidon-rings (106-2 & 106-6)...
I was not sure which material, so i ordered red and yellow...
...
Mike
Hi!
Stargate is the best series all over the wolrd!
I also make a selfresonating class D amp... I've used 106-40 as a toroidal core... It gets very hot even without any load... so I also ordered some 106-2, which is referenced in this paper, and as some people in this forum also suggested...
I hope it will be good
Regards, lkadar.
MikeB,
You may want to consider skin effect in the inductor wire as well. I once had a flyback SMPS operating at ~~300kHz that produced a voltage drop of 5V @1A in about 18 turns of 0.8mm diameter wire, on that core this is about 35cm length. There were 2 such windings so the added ~~10W heat from the inductor was not a good thing
You might need to wind the inductor with foil, or use several thinner wires together.
If you have a coreless inductor, you may want to do two of them and connect in series, locate longitudinally next to each other (take care that magnetic field chains between them, not cancels). This way you will have a much lower stray fielad at right angles to the winding axis. Winding on a toroid made out of insulation is also great but it is a pain to do as there are a lot of turns and not much space, usually. A friend of mine used to wind such inductors on those wooden or plastic rings you use as curtain holders sometimes using two rings stacked together to get larger cross-sections.
Regarding the loss of sound quality, it may be because now there is no cross-conduction but a real dead time.
This is the great compromise on NRZ class D, cross conduction is conceptually similar to high bias on class AB, dead time to low or no bias. In your previous version with the 9540N, you had cross-conduction and therefore at least one dead time (transition from P to N conducting) was missing. Theoretically, a minimum amount of cross conduction may even be desirable, but it is very difficult to get it controlled and always keep it at that minimum.
Regarding nagging
Nagging = always complaining about something even when there is no reason for it
You may want to consider skin effect in the inductor wire as well. I once had a flyback SMPS operating at ~~300kHz that produced a voltage drop of 5V @1A in about 18 turns of 0.8mm diameter wire, on that core this is about 35cm length. There were 2 such windings so the added ~~10W heat from the inductor was not a good thing
You might need to wind the inductor with foil, or use several thinner wires together.
If you have a coreless inductor, you may want to do two of them and connect in series, locate longitudinally next to each other (take care that magnetic field chains between them, not cancels). This way you will have a much lower stray fielad at right angles to the winding axis. Winding on a toroid made out of insulation is also great but it is a pain to do as there are a lot of turns and not much space, usually. A friend of mine used to wind such inductors on those wooden or plastic rings you use as curtain holders
sometimes using two rings stacked together to get larger cross-sections.Regarding the loss of sound quality, it may be because now there is no cross-conduction but a real dead time.
This is the great compromise on NRZ class D, cross conduction is conceptually similar to high bias on class AB, dead time to low or no bias. In your previous version with the 9540N, you had cross-conduction and therefore at least one dead time (transition from P to N conducting) was missing. Theoretically, a minimum amount of cross conduction may even be desirable, but it is very difficult to get it controlled and always keep it at that minimum.
Regarding nagging
Nagging = always complaining about something even when there is no reason for it
Indeed, mostly done by members of the female sex, the recipient being the male with which the female has chosen to live with, or even marry. Situations that are usually prone to induce nagging are electronic parts spread across the kitchen table. The odor of soldering tin and flux and of course the repeated testing of amps, speakers, CD-players, etc. at various sound pressure levels for extended periods of time. Now if you can find a fix for that I'd gladly pay top-dollar.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
"In normal use the outputcoil of the UcD180 stays below 60-65oC"
Okay, no 80°, but still too warm to touch. But, what is "normal use" ?
Here is the posting:
http://www.diyaudio.com/forums/showthread.php?postid=602508#post602508
I think the heatup should be not very load-dependent, the HF-currentswing
should be always the same ?
Mike
Okay, no 80°, but still too warm to touch. But, what is "normal use" ?
Here is the posting:
http://www.diyaudio.com/forums/showthread.php?postid=602508#post602508
I think the heatup should be not very load-dependent, the HF-currentswing
should be always the same ?
Mike
Unless you are using no load. That is NOT the wisest thing to do as without a load you have an undamped resosnant circuit formed out of the output filter L and C. Being a series filter, it is essentially a short circuit (or rather the sum of series DCR of L and C) at it's resonant frequency. being you are feeding this with largely square waves, you can expect there to be a armonic that will hit the resonant frequency of the filter.
Otherwise, it's the DCR and the core losses that heat up the inductor (I am assuming no saturation, of course), DCR mostly being affected by skin effect. Again, keep in mind you have a steep edge waveform, there is a lot of VHF content there, and you want your inductor to be inductive and not resistive at those frequencies.
Otherwise, it's the DCR and the core losses that heat up the inductor (I am assuming no saturation, of course), DCR mostly being affected by skin effect. Again, keep in mind you have a steep edge waveform, there is a lot of VHF content there, and you want your inductor to be inductive and not resistive at those frequencies.
Well, i'm still waiting for the Amidon-rings to arrive, i hope replacing
the coil remove most of my problems... (Like fuses heating up)
I'm still strying to understand the mechanism of the distortions created
by ClassD, hopefully finding a fix. Adding a 10nf to GND before coil
eliminate them ? (In sims) But this might cook the amp...
Mike
the coil remove most of my problems... (Like fuses heating up)
I'm still strying to understand the mechanism of the distortions created
by ClassD, hopefully finding a fix. Adding a 10nf to GND before coil
eliminate them ? (In sims) But this might cook the amp...
Mike
Hi,
That cap would reduce your slew rate and possibly cause some malfunction (haven't put much thought into it).
What I've found helped with noise is not to drive the mosfet any harder than you have to. That's why I'm using 100 ohm gate resistors, which both reduce slew rate (without the detrimental possibilites though) and limit gate drive voltage due to the nature of the discrete drivers. According to sim it may only be 8 volts.
Not quite current controlled but almost, can't argue with what stays cool though huh?
Aside from how cool they stay, a lower Ron really brings on the noise, might be more apt to ringing, wish I could tell ya more but that's all I could tell without a scope.
Cheers
That cap would reduce your slew rate and possibly cause some malfunction (haven't put much thought into it).
What I've found helped with noise is not to drive the mosfet any harder than you have to. That's why I'm using 100 ohm gate resistors, which both reduce slew rate (without the detrimental possibilites though) and limit gate drive voltage due to the nature of the discrete drivers. According to sim it may only be 8 volts.
Not quite current controlled but almost, can't argue with what stays cool though huh?
Aside from how cool they stay, a lower Ron really brings on the noise, might be more apt to ringing, wish I could tell ya more but that's all I could tell without a scope.
Cheers
Hi Mike, I think it could toast your MOSFETs depending on how much dead time the circuit has. It could switch more efficiently until the volume is cranked up. Then, the energy balance in the choke could become uneven enough to prevent zero voltage switching in one direction and cause hard switching into the capacitive load (10nF) when the MOSFET in is the high-loss zone.MikeB said:
Hi classd4sure !
I think this cap reduces slewrate mainly during deadtime, making
the squarewave more syncron to the current-triangle through the coil ?
I had no noise yet, but my mosfets does not have the lowest Rds-on.
The irf540n had 50mohm, but as long as the p-channel has such higher
value, this problem might not occur.
You really should borrow a scope, it helps much with ClassD !
Mike
I think this cap reduces slewrate mainly during deadtime, making
the squarewave more syncron to the current-triangle through the coil ?
I had no noise yet, but my mosfets does not have the lowest Rds-on.
The irf540n had 50mohm, but as long as the p-channel has such higher
value, this problem might not occur.
You really should borrow a scope, it helps much with ClassD !
Mike
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.