hhmmmmmm im really concerned about the heating of the torroid.. although It has lowered in temp after replacing the output filter coils with ferrite material.....
maybe i will RE-wind ? with many strands of thinner magnet wire... aka "litz wire" way....
maybe "skin effect" contributes heating too??? smps is at 100KHz.... Eva? 😉
maybe i will RE-wind ? with many strands of thinner magnet wire... aka "litz wire" way....
maybe "skin effect" contributes heating too??? smps is at 100KHz.... Eva? 😉
RX5, peak power guessed around 200W in 2 ohms, around 80W continous (86VA trafo) Around 50W in 4 ohms and around 25W in 8ohms, these all guessed power since i have no way of measuring it.
Hi RX5, do you have a schematic of your amp and its power supply ¨? Would it be possible to publish last version of both ?
interesting thread 🙂
I'm already scrounging up parts to build me a class D amp for car use...... gonna be my third attempt.
hey raff, I see you made use of the circuit I posted. 🙂
what is the wire size that you used and how many strands? could be that the oscillator freq is too high and you're getting too much resistance from skin effect. the highest I have used is only 50kHz and often staying at about 35kHz.
I'm planning for a dual transformer SMPS doing 1kW and at only about 20-30kHz oscillator freq.
I'm already scrounging up parts to build me a class D amp for car use...... gonna be my third attempt.
hey raff, I see you made use of the circuit I posted. 🙂
what is the wire size that you used and how many strands? could be that the oscillator freq is too high and you're getting too much resistance from skin effect. the highest I have used is only 50kHz and often staying at about 35kHz.
I'm planning for a dual transformer SMPS doing 1kW and at only about 20-30kHz oscillator freq.
djQUAN said:
Hello djQUAN,
I think its OVER in freq??? osc freq TL494 is at 100KHz... maybe I have overdone it....
wire size, i think its AWG#21... it was lying around in my junk... I wound 3 strands if it, 6T+6T pri... and 24T+24T sec.....
hmmmm you say 50KHz max??? ok ill reduce mine and see if it still heats up my torroid....
post back soon 🙂
djQUAN,
im currently using Rt/Ct = 1k/10nF
datasheet says:
>>single ended output TL494 = osc freq is SAME with output freq
>> push-pull with TL494 = osc freq is divided by 2 at output
so my calculated freq WAS 100KHz devided by 2.. so I get 50KHz output freq, right??
Im going to change Rt value later to 3k3.... it would drop freq to about 60-75KHz/2...... output would be 30-37KHz....
maybe THAT would FIX the heating torroid core issue...
give it a TRY with UCD.. one hell of an AMP..
djQUAN said:
djQUAN,
are THOSE switching freq you mentioned, the OSC freq of TL494??? or the RESULTING output freq(/wc is divided by 2 /push-pull config)??
im confused....
well, I measured mine right after the push-pull output so I guess that's the osc freq/2.
50kHz is still rather high and might be difficult to do right in some cases.
three strands of #21 for the primary? I think the heating is because of that. not because of skin effect but the resistance of the pri is too high.
try using about 5-6 strands for the primary.
the Rt and Ct combination you used is also rather unusual..... I often use 6k8 for Rt and 4n7 for Ct and change it by trial and error depending on the core. I think that resulted to 35-45 kHz. 1k is a little too small for my liking. 😛
just for an example, I tested a small ferrite core (not toroidal though) at 80kHz, 12V supply and 10+10 turn for the primary and the resulting voltage at the primary is full of large spikes and draws a lot of current. I found by substituting different caps and resistors and found the best to be around 28kHz with the current draw at only 54mA (which includes the oscillator circuit ).
hopefully, I could post soon the circuit I designed to test ferrite cores to find the best Ct-Rt combination.
50kHz is still rather high and might be difficult to do right in some cases.
three strands of #21 for the primary? I think the heating is because of that. not because of skin effect but the resistance of the pri is too high.
try using about 5-6 strands for the primary.
the Rt and Ct combination you used is also rather unusual..... I often use 6k8 for Rt and 4n7 for Ct and change it by trial and error depending on the core. I think that resulted to 35-45 kHz. 1k is a little too small for my liking. 😛
just for an example, I tested a small ferrite core (not toroidal though) at 80kHz, 12V supply and 10+10 turn for the primary and the resulting voltage at the primary is full of large spikes and draws a lot of current. I found by substituting different caps and resistors and found the best to be around 28kHz with the current draw at only 54mA (which includes the oscillator circuit ).
hopefully, I could post soon the circuit I designed to test ferrite cores to find the best Ct-Rt combination.
Uhmmm 6k8/4n7 timing components results to osc freq of roughly 75-85KHz/2 = 37-47KHz(for push-pull).....
I replaced the hole designated for Rt with an IC PIN socket so I could quickly swap resistor values with it.... I will lower 1K to 2k2 or 3k3....
regarding strands, making it 5(compared to 3) would "LOAD" it more.. 🙂
are you using RC snubbers at primary of torroid winding/s??
I mean how are they connected?? primary of winding?? or Drain to Source of MOSFET/s??
🙂
I replaced the hole designated for Rt with an IC PIN socket so I could quickly swap resistor values with it.... I will lower 1K to 2k2 or 3k3....
regarding strands, making it 5(compared to 3) would "LOAD" it more.. 🙂
are you using RC snubbers at primary of torroid winding/s??
I mean how are they connected?? primary of winding?? or Drain to Source of MOSFET/s??
🙂
ok..... with my (Ct) 10nF, I have tried with (Rt) 2k2 and 3k3 ......
using those values REQUIRE a LARGER heatsink....
I dont know if lowering OSC freq to less than 50KHz would lessen heating of torroid??
anyone out there expert with magnetics?? need comment pls
using those values REQUIRE a LARGER heatsink....
I dont know if lowering OSC freq to less than 50KHz would lessen heating of torroid??
anyone out there expert with magnetics?? need comment pls
I think SMPS is better @ osc freq of 100KHz.... maybe FERRITE torroide IS slow?? at those speeds?? maybe ferrite E-E core would do...
You really need to know the details of the material type of your toroid, as well as its exact dimensions.
Looking at the losses vs flux. density curve, you can determine the max. allowable flux. density and from that you can calculate the primary turns ratio.
If you don't do this, you will only be guessing and your only possible approach will be trial and error.
Looking at the losses vs flux. density curve, you can determine the max. allowable flux. density and from that you can calculate the primary turns ratio.
If you don't do this, you will only be guessing and your only possible approach will be trial and error.
how would that load it more? it would result to less resistance so that all available power is sent to the load instead of wasting it as heat on the trafo.
nope. as much as possible, I don't use it. it wastes power since it heats up. if you match your transformer and oscillator well, it won't even be needed.
hmmmmm SMPS is really Ok with 100KHz osc freq (100/2 = 50KHz at mosfets)... MOSFETS DO NOT heat up.. but with Rt changed to 2k2 OR 3k3, HEATING is evident....
I guess the problem LIES with the TORROID... i think the torroid CAME from xerox machines..1.2" outer dia.... anyone could guess WHAT it (ferrite) is???
regarding snubber, resistor dont heat up.. I guess I dont really need it at 50KHz?
I guess the problem LIES with the TORROID... i think the torroid CAME from xerox machines..1.2" outer dia.... anyone could guess WHAT it (ferrite) is???
regarding snubber, resistor dont heat up.. I guess I dont really need it at 50KHz?
a way to find out if it is ferrite or iron powder.....
rub it with a rough (#100?) sandpaper. if it rubs into a smooth surface and you get very fine black powder, then it is most probably ferrite. if the surface is rough and grainy then it is iron powder.
if you don't try the comment I said about adding more strands then you will never know what is really causing heating.
rub it with a rough (#100?) sandpaper. if it rubs into a smooth surface and you get very fine black powder, then it is most probably ferrite. if the surface is rough and grainy then it is iron powder.
if you don't try the comment I said about adding more strands then you will never know what is really causing heating.
im going to wind MANY strands of much thinner magnet wire SOON.... im kinda busy with other stuff right now 😀
I hope that solves the problem.. I cant parallel more strands of #21, little space left on CORE.....
Im much happier with 100KHz osc freq, I can use really small heatsink for MOSFETs.... I have also removed the RC snubber.... they dont even give a hint of heating in actual use....
djQUAN,
go make your UCD diy now....
I hope that solves the problem.. I cant parallel more strands of #21, little space left on CORE.....
Im much happier with 100KHz osc freq, I can use really small heatsink for MOSFETs.... I have also removed the RC snubber.... they dont even give a hint of heating in actual use....
djQUAN,
go make your UCD diy now....
It'll have to wait. 🙂
I need to first build a 1kW SMPS for the class D sub amp that I'm gonna build it should do atleast 1kW RMS at 1 ohm mono. I have just bought five Ixys 60A fast recovery diodes for the output. 🙂
for the amplifier section, I'm still thinking about going self oscillating or a clocked design.
Those diodes could be overkill, or even totally pointless depending how you use them.
You can swith a good deal slower for a sub amp too dont' forget, which kind of relaxes the body diode recovery issue if you slow down your slew rate a little.
You can swith a good deal slower for a sub amp too dont' forget, which kind of relaxes the body diode recovery issue if you slow down your slew rate a little.
- Status
- Not open for further replies.