Hello,
I am currently working on a prototype smps for powering up a stereo amplifier of 150w per channel. Calculated the peak power requirements at 600W and guess-timated an average power consumption with music input (high-pass filtered @ 150Hz) of around 150-200W.
For the sake of simplicity and wanting to try something new I decided for a 2 transistor forward topology with just line regulation (so i don't smoke my amp when my "dearest" energetic company is putting out 280Vac instead of the nominal 230Vac).
I have a lot of ERL-35C cores handy (slightly bigger than the more common ETD34 but smaller than ETD39) and I would like to use one of them. Designing for a maximum temperature of 90 degrees @ 40 degrees ambiental temperature I calculated the maximum power throughput from this core using regular round copper wire (no litz available here ... sad) between 210 - 250W @ 100kHz depending on winding layers and choice of wire diameter.
It would seem a good choice for my application but I am worried about the maximum design temperature of 90 degrees and I am wondering if anybody contemplated the option to attach a heatsink to the transformer. I was thinking maybe a small anodized 10K/W heatsink attached to the upper side of the core would get rid of the heat more efficiently.
I am having trouble finding relevant information about this subject on the internet, only found a study about cooling planar transformers. What is your opinion on this?
Also has anyone tried fan cooling smps transformers? Does it make a valuable difference in terms of power throughput and temperature rise?
I am currently working on a prototype smps for powering up a stereo amplifier of 150w per channel. Calculated the peak power requirements at 600W and guess-timated an average power consumption with music input (high-pass filtered @ 150Hz) of around 150-200W.
For the sake of simplicity and wanting to try something new I decided for a 2 transistor forward topology with just line regulation (so i don't smoke my amp when my "dearest" energetic company is putting out 280Vac instead of the nominal 230Vac).
I have a lot of ERL-35C cores handy (slightly bigger than the more common ETD34 but smaller than ETD39) and I would like to use one of them. Designing for a maximum temperature of 90 degrees @ 40 degrees ambiental temperature I calculated the maximum power throughput from this core using regular round copper wire (no litz available here ... sad) between 210 - 250W @ 100kHz depending on winding layers and choice of wire diameter.
It would seem a good choice for my application but I am worried about the maximum design temperature of 90 degrees and I am wondering if anybody contemplated the option to attach a heatsink to the transformer. I was thinking maybe a small anodized 10K/W heatsink attached to the upper side of the core would get rid of the heat more efficiently.
I am having trouble finding relevant information about this subject on the internet, only found a study about cooling planar transformers. What is your opinion on this?
Also has anyone tried fan cooling smps transformers? Does it make a valuable difference in terms of power throughput and temperature rise?
I have seen SMPS transformers with aluminum strips attached for heatsinking, seen transformers on the side of the PCB with one core-edge attached to the chassis, and have seen a transformer sandwiched between two heatsinks. All of these approaches are to accomplish precisely what you are seeking. Just a general improvement in performance and longevity through core temp reduction.
Unfortunately, I can't provide sources or pictures, but the application is basically the same for heatsinking any power component: More surface area or more air and overdoing beyond a calculated minimum it is merely a space and budget constraint.
Please keep us posted on your project. It sounds great and I suspect a number of DIY'ers around here would love to do something similar - myself included.
Regards,
Paul
Unfortunately, I can't provide sources or pictures, but the application is basically the same for heatsinking any power component: More surface area or more air and overdoing beyond a calculated minimum it is merely a space and budget constraint.
Please keep us posted on your project. It sounds great and I suspect a number of DIY'ers around here would love to do something similar - myself included.
Regards,
Paul
I can make you the litzwire for free. made my own for 1kw smps.
PM me with the details.
Regards,
Savu Silviu
Wow, i wasn't expecting to receive input so fast ... that's the main reason diyaudio rocks !!! (I don't want to enter politics but most of the forums in my country are full of ego-maniacs that think they know everything when in fact know nothing, just guess at everything ... )
Getting over enthusiasm now, my thoughts are summarized this way :
1) in case of planar magnetics the addition of a heatsink is said to lower the thermal resistance about 40-50% which does lead to two design perspectives : achieve a smaller temperature rise for the same power output OR increase power throughput for the given temperature rise ; any of them is fine with me.
2) applying a heatsink implies minimising the thermal resistance from power source to air ; this thermal resistance has two parts as i recall from a fairchild app note - Rint (the resistance between the windings and the core) and Rext (the resistance between the core and air) ; as I understand not much can be done for the Ri part , the external heatsink only lowers the Rext part? (to be further studied)
3) planar magnetics are nothing like a EE, EI, EER or ETD core (i think they have more surface area in contact with the air for the same volume ... correct me if i'm wrong). Since the core i intend to use will have a lot less exposed surface area (and contact area with the heatsink) than a planar core can I expect the improvement to be more than 5-10 percent?
DCPreamp, you said you saw this method applied before. Could you tell me to what kind of core was it applied ? Was it an ETD ... an EE ... an EI ... RM ... etc?
I will keep you posted but things move really slow around here (day job ... only the night remains free for hobby smps production but one has to sleep also ... sad).
Started this project almost a year ago and I got caught up in more stringent things, mostly work related. First attempt was to use a TOP250 which promised much but delivered little. In fact it delivered some nice smoke from the zenner clamp that almost made my neighbours call the firemen. Now I receive fire extinguishers for birthday gifts 🙂) (joking). Second attempt made this thing working really nice at about 300W continous power output but I felt lazy enough to not finish the damn thing. 2 months later it blew up when powered up, probably a stupid mistake made by me wiring the thing back up. I just wish I had more time ...
Regards,
Marius
Getting over enthusiasm now, my thoughts are summarized this way :
1) in case of planar magnetics the addition of a heatsink is said to lower the thermal resistance about 40-50% which does lead to two design perspectives : achieve a smaller temperature rise for the same power output OR increase power throughput for the given temperature rise ; any of them is fine with me.
2) applying a heatsink implies minimising the thermal resistance from power source to air ; this thermal resistance has two parts as i recall from a fairchild app note - Rint (the resistance between the windings and the core) and Rext (the resistance between the core and air) ; as I understand not much can be done for the Ri part , the external heatsink only lowers the Rext part? (to be further studied)
3) planar magnetics are nothing like a EE, EI, EER or ETD core (i think they have more surface area in contact with the air for the same volume ... correct me if i'm wrong). Since the core i intend to use will have a lot less exposed surface area (and contact area with the heatsink) than a planar core can I expect the improvement to be more than 5-10 percent?
DCPreamp, you said you saw this method applied before. Could you tell me to what kind of core was it applied ? Was it an ETD ... an EE ... an EI ... RM ... etc?
I will keep you posted but things move really slow around here (day job ... only the night remains free for hobby smps production but one has to sleep also ... sad).
Started this project almost a year ago and I got caught up in more stringent things, mostly work related. First attempt was to use a TOP250 which promised much but delivered little. In fact it delivered some nice smoke from the zenner clamp that almost made my neighbours call the firemen. Now I receive fire extinguishers for birthday gifts 🙂) (joking). Second attempt made this thing working really nice at about 300W continous power output but I felt lazy enough to not finish the damn thing. 2 months later it blew up when powered up, probably a stupid mistake made by me wiring the thing back up. I just wish I had more time ...
Regards,
Marius
found a workaround
I've been searching for information on thermal modelling of transformers but have come up pretty empty. I guess it takes a lot of cut-and-try to get a specific idea of the improvements that can be obtained.
In the meantime I managed to optimize the windings for a temperature rise of 31 degrees at 200W output. (Quite a pain cause it's an iterative process but it helps to have multiple wire gauges in the workshop) At 40 degrees ambient that's a hot spot temperature of 76 degrees. I am happy with it now so I'll just go ahead and build it. After it's finished I will also try to attach a heatsink and see if this improves anything.
Regards,
Marius
I've been searching for information on thermal modelling of transformers but have come up pretty empty. I guess it takes a lot of cut-and-try to get a specific idea of the improvements that can be obtained.
In the meantime I managed to optimize the windings for a temperature rise of 31 degrees at 200W output. (Quite a pain cause it's an iterative process but it helps to have multiple wire gauges in the workshop) At 40 degrees ambient that's a hot spot temperature of 76 degrees. I am happy with it now so I'll just go ahead and build it. After it's finished I will also try to attach a heatsink and see if this improves anything.
Regards,
Marius
- Status
- Not open for further replies.