Hi All
This is for those of you who are good with magnetics. I have picked out a Ferroxcube toroid, TX58/41/18. It is 58mm OD, 41mm ID and 18mm H. This translates to an effective area of 152mm(sq).
I can get them made of 3C90 ferrite material, good for switching below 200KHz.
My question is this:
How much power can I get through this? Originally I would have thought about 250W at 40KHz or so, but at 100KHz I should get a lot more surely?
I need at least 400W, should I look at other cores perhaps? I am heavily limited by vertical height - wound the trafo may not stand more than +-21mm absolute max above the PCB.
Perhaps an E core on its side sans bobbin? Or a horizontal ETD49 but I cant find any height specs for these anywhere.
Its for a push-pull +12V to +-35V SMPS.
Thanks!
Gareth
This is for those of you who are good with magnetics. I have picked out a Ferroxcube toroid, TX58/41/18. It is 58mm OD, 41mm ID and 18mm H. This translates to an effective area of 152mm(sq).
I can get them made of 3C90 ferrite material, good for switching below 200KHz.
My question is this:
How much power can I get through this? Originally I would have thought about 250W at 40KHz or so, but at 100KHz I should get a lot more surely?
I need at least 400W, should I look at other cores perhaps? I am heavily limited by vertical height - wound the trafo may not stand more than +-21mm absolute max above the PCB.
Perhaps an E core on its side sans bobbin? Or a horizontal ETD49 but I cant find any height specs for these anywhere.
Its for a push-pull +12V to +-35V SMPS.
Thanks!
Gareth
That core is easily enough for 400 watts at 40kHz. I use a core that's slightly smaller for an amp that will produce ~800 RMS.
When designing the supply you have to determine how the amp is going to be run. If you want it to hold up to a continuous sine wave output into the toughest rated load at full power, it's going to need a larger transformer than an amp that you simply want to be able to drive to full power intermittantly with music.
When designing the transformer, you'll need to decide on the acceptable temperature rise above ambient. If you use 400 circular mils per amp as a guide, the transformer will run hotter than if you use 700cm/amp. With the core you have, you can easily go with 700cm/amp (if it will fit into the alloted space).
If you figure that the amp will draw 40 amps, you will need 28,000cm of copper (at 700cm/a). That's roughly equivalent to 7 strands of 14AWG solid wire. If you were to go with 400cm/amp, you could get by with 4 strands of 14g wire. If you are not building the amp to run a sine wave at full power continuously at maximum operating temperatures, the 16,000cm of copper (4x14g) would be plenty.
Most all manfacturers run their supplies at 25-40kHz. It's easier to switch in this range of frequencies and Litz wire is not needed (IMHO, Litz wire is a major headache). You'll also get more core loss at higher frequencies.
As Steve mentioned, when you wind the core, interleave the windings so you get good coupling between the primary and secondary. If you don't get good coupling, you'll have problems with ringing/switching transients. If you wind the transformer well, there is almost no need for a snubber. If you wind it poorly, a snubber will not be able to control the ringing. The ringing will show up as noise (generally inaudible) on the audio output.
When designing the supply you have to determine how the amp is going to be run. If you want it to hold up to a continuous sine wave output into the toughest rated load at full power, it's going to need a larger transformer than an amp that you simply want to be able to drive to full power intermittantly with music.
When designing the transformer, you'll need to decide on the acceptable temperature rise above ambient. If you use 400 circular mils per amp as a guide, the transformer will run hotter than if you use 700cm/amp. With the core you have, you can easily go with 700cm/amp (if it will fit into the alloted space).
If you figure that the amp will draw 40 amps, you will need 28,000cm of copper (at 700cm/a). That's roughly equivalent to 7 strands of 14AWG solid wire. If you were to go with 400cm/amp, you could get by with 4 strands of 14g wire. If you are not building the amp to run a sine wave at full power continuously at maximum operating temperatures, the 16,000cm of copper (4x14g) would be plenty.
Most all manfacturers run their supplies at 25-40kHz. It's easier to switch in this range of frequencies and Litz wire is not needed (IMHO, Litz wire is a major headache). You'll also get more core loss at higher frequencies.
As Steve mentioned, when you wind the core, interleave the windings so you get good coupling between the primary and secondary. If you don't get good coupling, you'll have problems with ringing/switching transients. If you wind the transformer well, there is almost no need for a snubber. If you wind it poorly, a snubber will not be able to control the ringing. The ringing will show up as noise (generally inaudible) on the audio output.
Hi Perry
You reckon it will be ok? It just seems a little small to me. I was originally going to switch at 50K but I think I will push it up to 100K with this core - this is still withing limits I would say.
I would like the PSU to be able to handle max continuous power if it is required. By my calcs, if the amps (n=75%) put *out* 350Wrms, then the PSU must supply a good 460 or so and if you take the 80% efficiency of the SMPS into account, thats more like 580W
I was going to use 0.2mm wire strands to make composite windings - I dont know where to find Litz wire here in South Africa. There must be somewhere, we have a thriving electronics industry here and most everything is available pretty cheaply if you can find it.
I have ordered the cores nevertheless but I have also ordered an ETD44 and an ETD49 of the same material (3C90) to see if I can maybe fit them in on their sided with cut-down bobbins or with home-made bobbins. These cores have a greater effective area.
What exactly is a circular mil? My trafo could shift in excess of 50A by my calcs, I would like to make sure there is enough copper but I cant make it too big or it will hog space
I wish my heatsink extrusion housing was a bit higher
Thanks for the feedback!
Gareth
You reckon it will be ok? It just seems a little small to me. I was originally going to switch at 50K but I think I will push it up to 100K with this core - this is still withing limits I would say.
I would like the PSU to be able to handle max continuous power if it is required. By my calcs, if the amps (n=75%) put *out* 350Wrms, then the PSU must supply a good 460 or so and if you take the 80% efficiency of the SMPS into account, thats more like 580W
I was going to use 0.2mm wire strands to make composite windings - I dont know where to find Litz wire here in South Africa. There must be somewhere, we have a thriving electronics industry here and most everything is available pretty cheaply if you can find it.
I have ordered the cores nevertheless but I have also ordered an ETD44 and an ETD49 of the same material (3C90) to see if I can maybe fit them in on their sided with cut-down bobbins or with home-made bobbins. These cores have a greater effective area.
What exactly is a circular mil? My trafo could shift in excess of 50A by my calcs, I would like to make sure there is enough copper but I cant make it too big or it will hog space
I wish my heatsink extrusion housing was a bit higher
Thanks for the feedback!
Gareth
I've never used that particular core/material but I have used a 44916TC made of 'P' material from mag-inc.com in an 800 watt amp. The core is (IIRC) 49mmx33mmx16mm with an effective area of 116mm^2. To run at full power sine wave, it needs a fan (but the amp is fan cooled). To play music at full power, a fan is not needed. Unless your core has more loss than this one, you will be OK. Look up the core on mag-inc's site and compare it to the specs for your core.
A circular mil is equivalent to a single round copper conductor that is 1/1000 of one inch in diameter. If the round conductor is 10mils in diameter, the conductor has 100cm of copper. 10AWG wire is ~100mils (0.1") in diameter and it has ~10.000cm. If you're using mm^2 as a unit...
0.0005mm^2 = 1 circular mil.
1mm^2 = 1973cm.
If you're going to make a cover for your amp, you may be able to make it step up from the mounting points so that it allows a little more space than a simple flat cover.
A circular mil is equivalent to a single round copper conductor that is 1/1000 of one inch in diameter. If the round conductor is 10mils in diameter, the conductor has 100cm of copper. 10AWG wire is ~100mils (0.1") in diameter and it has ~10.000cm. If you're using mm^2 as a unit...
0.0005mm^2 = 1 circular mil.
1mm^2 = 1973cm.
If you're going to make a cover for your amp, you may be able to make it step up from the mounting points so that it allows a little more space than a simple flat cover.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.