I'be uilt it according to spec and there is definatly no problem with my PCB (wiring wise)
The thing is I get +-80v off a 4:10 transformer running off the 12v rail of a computer PSU. The fets are getting fairly warm (there heatsinks so its only about 40-45c on a 0.7c/w heatsink). THe load on the secondary is the rectifier diodes, a 2200uf cap on each rail and 12k resisters to discharge them.
The Comp PSU is at around 7.8v so its being fairly heavily loaded.
Can anyone surgest what to do?
The thing is I get +-80v off a 4:10 transformer running off the 12v rail of a computer PSU. The fets are getting fairly warm (there heatsinks so its only about 40-45c on a 0.7c/w heatsink). THe load on the secondary is the rectifier diodes, a 2200uf cap on each rail and 12k resisters to discharge them.
The Comp PSU is at around 7.8v so its being fairly heavily loaded.
Can anyone surgest what to do?
the transformer
you have to go look at the equations. the turns ratio doesn't get you all the way there, it's only a first approximation.
the first approximation equation for turns is:
<b>N= 100 X SQRT (L/A sub i) </b>
for a T106-26 core which you will probably find in a computer power supply, A sub i is 900 uH/100 turns.
but you have to determine the permeability change due to DC bias, %uH,
and for this you need the coefficients for change of permeability
<b>%uH = SQRT ((A + cH + eH^2)/(1+bH + dH^2))</b>
for a type 106 core, A= 10090, B= 5.05e-3, C=13.1, D= 1.17e-3, E= 0.0212
what the above equation, %uH tells you is the amount which the permeability, A sub i, is reduced.
H, magnetizing force (oersteds) is:
<b>(0.4 pi N I /<em> l</em> )</b> (last term is magnetic path length, for a T106 core it's 6.49)
In the spirit of full disclosure: All of the above is in the Jan/Feb issue of the ham radio magazine QEX, a somewhat difficult article to follow, but invaluable.
you have to go look at the equations. the turns ratio doesn't get you all the way there, it's only a first approximation.
the first approximation equation for turns is:
<b>N= 100 X SQRT (L/A sub i) </b>
for a T106-26 core which you will probably find in a computer power supply, A sub i is 900 uH/100 turns.
but you have to determine the permeability change due to DC bias, %uH,
and for this you need the coefficients for change of permeability
<b>%uH = SQRT ((A + cH + eH^2)/(1+bH + dH^2))</b>
for a type 106 core, A= 10090, B= 5.05e-3, C=13.1, D= 1.17e-3, E= 0.0212
what the above equation, %uH tells you is the amount which the permeability, A sub i, is reduced.
H, magnetizing force (oersteds) is:
<b>(0.4 pi N I /<em> l</em> )</b> (last term is magnetic path length, for a T106 core it's 6.49)
In the spirit of full disclosure: All of the above is in the Jan/Feb issue of the ham radio magazine QEX, a somewhat difficult article to follow, but invaluable.
I tried with a new core and using the formula on the Valve Audio site. It came to be 1.5 primary turns so I made one up and had 7 secondary turns to start with. Voltage was ~65v.
I took off two turns and it came to 55v.
Another 2 and it was at 44v
I left it on for about 30 secs and C15 Blew up (lots of smoke, scared the bejesus outta me).
The second core did heat up a bit (enought to be warm) and the fets where as hot as last time.
I'm confused
any help would be great
BTW the only stats about my core it the dimensions
I took off two turns and it came to 55v.
Another 2 and it was at 44v
I left it on for about 30 secs and C15 Blew up (lots of smoke, scared the bejesus outta me).
The second core did heat up a bit (enought to be warm) and the fets where as hot as last time.
I'm confused
any help would be great
BTW the only stats about my core it the dimensions
if the only stats about your core are its dimensions then you're in trouble, This is a problem with the web, the articles aren't vetted. With the RSGB or the ARRL there are technical staffs which double check the articles for acuracy and technical competence.
It takes too much time to go into the magnetics of the transformer. Let me just say that I don't believe the author has properly characterized the core. One of the reasons which the MOSFET's are over-heating is that they are seeing too low a primary impedance. Too high a primary impedance causes other problems.
Are you using a low ESR capacitor? A junk box capacitor is going to heat, badly, very badly.
The ratings for the diodes and MOSFET's have to be much higher than the average load. The peak values depend upon the format of the converter (and take too much time to explain here.)
The MOSFET's "might" share the load equally, but I doubt it.
There should be an RFI filter on the input to the supply unless you want EMI to shut down your car's ignition or cause the roof to retract!
It takes too much time to go into the magnetics of the transformer. Let me just say that I don't believe the author has properly characterized the core. One of the reasons which the MOSFET's are over-heating is that they are seeing too low a primary impedance. Too high a primary impedance causes other problems.
Are you using a low ESR capacitor? A junk box capacitor is going to heat, badly, very badly.
The ratings for the diodes and MOSFET's have to be much higher than the average load. The peak values depend upon the format of the converter (and take too much time to explain here.)
The MOSFET's "might" share the load equally, but I doubt it.
There should be an RFI filter on the input to the supply unless you want EMI to shut down your car's ignition or cause the roof to retract!
Well I think in the process of testing that transformer my fets died.
The most info I could find on the core is the demensions and the fact thats its an L15 type. I've googled but found nothin.
I'm going to keep trying to find the number of primary turns that works
I'll make enquiries with the supplier of the core about specs but I wouldn't be hopeful.
All caps are low ESR (cept the 220uf one since its not in the path of the transformer which shouldn't amke a difference). resisters are 1% metal film except the 2w ones in the snubber network which are 5%.
If my problems to low independence I should be able to increase the turns on the primary (and secondary to keep in the ratio) until the Fets cool, right?
I h8 when things don't work the first time, learning is learning I spose...... Stupid hard way
The most info I could find on the core is the demensions and the fact thats its an L15 type. I've googled but found nothin.
I'm going to keep trying to find the number of primary turns that works
I'll make enquiries with the supplier of the core about specs but I wouldn't be hopeful.
All caps are low ESR (cept the 220uf one since its not in the path of the transformer which shouldn't amke a difference). resisters are 1% metal film except the 2w ones in the snubber network which are 5%.
If my problems to low independence I should be able to increase the turns on the primary (and secondary to keep in the ratio) until the Fets cool, right?
I h8 when things don't work the first time, learning is learning I spose...... Stupid hard way
can you borrow an impedance bridge?
even high schools and community (2 year technical) colleges have impedance bridges -- it's pretty simple to "characterize" a coil -- simply measure the inductance with 100 turns of wire -- this is Ai (I refer to it as "A sub i" as they do in the patent office.) I have plans for an impedance meter, just haven't the time to put them on the web.
You also have to know where the coil will saturate -- when it saturates the impedance goes down to squat.
Rather than invest in more FET's -- take a look at all the tutorials which Nat Semi, Texas Instruments (Unitrode) , Linear Tech, have on the web. International Rectifier has a tutorial on "Flyback Transformer Design" -- which is not 100% germane, but good for basics.
The SG3524 does have "over-current" sense -- this isn't current control as in the Unitrode PWM chips, but perhaps you could employ it
even high schools and community (2 year technical) colleges have impedance bridges -- it's pretty simple to "characterize" a coil -- simply measure the inductance with 100 turns of wire -- this is Ai (I refer to it as "A sub i" as they do in the patent office.) I have plans for an impedance meter, just haven't the time to put them on the web.
You also have to know where the coil will saturate -- when it saturates the impedance goes down to squat.
Rather than invest in more FET's -- take a look at all the tutorials which Nat Semi, Texas Instruments (Unitrode) , Linear Tech, have on the web. International Rectifier has a tutorial on "Flyback Transformer Design" -- which is not 100% germane, but good for basics.
The SG3524 does have "over-current" sense -- this isn't current control as in the Unitrode PWM chips, but perhaps you could employ it
An easy SMPS for your car:
I have a few hundred of thes On-Semi MC31466 switcher chips -- you should be able to get 40V, 3 to 4 amps out of them. I built a distributed power supply as described.
The circuit below is for an "off-line" switcher, meaning that it takes the mains current, rectifies it filters it, etc then uses the two MJE13005 transistors as a 50kHz self-oscillating multivibrator. You could employ something like this for an auto audio system power supply -- should have an RFI filter,etc. All the design equations are on On-Semi's website.
I have a few hundred of thes On-Semi MC31466 switcher chips -- you should be able to get 40V, 3 to 4 amps out of them. I built a distributed power supply as described.
The circuit below is for an "off-line" switcher, meaning that it takes the mains current, rectifies it filters it, etc then uses the two MJE13005 transistors as a 50kHz self-oscillating multivibrator. You could employ something like this for an auto audio system power supply -- should have an RFI filter,etc. All the design equations are on On-Semi's website.
Attachments
Another transformer tutorial
the PDF on the Linear Tech LT1683 (slew rate controlled PP DC converter) has a good discussion of the magnetics, beginning on about page 16:
http://www.linear.com/pdf/1683fa.pdf
the PDF on the Linear Tech LT1683 (slew rate controlled PP DC converter) has a good discussion of the magnetics, beginning on about page 16:
http://www.linear.com/pdf/1683fa.pdf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.