Perry-
THX for the link. I did it the hard way: deriving the equation until I saw a constant. I neglected to include the 3(Rd) because I rarely, if ever use the Discharge pin, and because 300W doesn't affect the frequency much when Rt is in the 10-20kW range.
But my calculations are confirmed by ST.
amirdelta-
For the RC values you have, I calculate you oscillator frequency at 138.6kHz, for a switching freq of 69kHz. I concur with Perry: raise your R(t) to 20kW to get a switching freq of ~35.1kHz.
Luka-
Having the equation is very handy, especially when graphs aren't available or clear. Easy to miscalculate using the graphs. Much more accurate with the equation.
Cheers,
Steve
THX for the link. I did it the hard way: deriving the equation until I saw a constant. I neglected to include the 3(Rd) because I rarely, if ever use the Discharge pin, and because 300W doesn't affect the frequency much when Rt is in the 10-20kW range.
But my calculations are confirmed by ST.
amirdelta-
For the RC values you have, I calculate you oscillator frequency at 138.6kHz, for a switching freq of 69kHz. I concur with Perry: raise your R(t) to 20kW to get a switching freq of ~35.1kHz.
Luka-
Having the equation is very handy, especially when graphs aren't available or clear. Easy to miscalculate using the graphs. Much more accurate with the equation.
Cheers,
Steve
Hi Steve
I agree, 100%, it is much easyer, but "especially when graphs aren't available or clear"... If you don't have them at all, your mistake it 100% anyway, and if they are not clear, use some other PDF, since this chip is produced by many.
Formula is more on the save side, but I use graph, it is faster
I agree, 100%, it is much easyer, but "especially when graphs aren't available or clear"... If you don't have them at all, your mistake it 100% anyway, and if they are not clear, use some other PDF, since this chip is produced by many.
Formula is more on the save side, but I use graph, it is faster
if I don't know the core type or material (like I have any other choice over here..)
I replace Rt with a 50kohm trimmer and adjust for best waveform/efficiency/performance while monitoring with a scope and then remove it (powered down, without moving the trimmer setting) and measure with a VOM. replace with a fixed resistor with closest value.
I replace Rt with a 50kohm trimmer and adjust for best waveform/efficiency/performance while monitoring with a scope and then remove it (powered down, without moving the trimmer setting) and measure with a VOM. replace with a fixed resistor with closest value.
hi guys
thanks for your answers , i appreciate.
dear friends perry ,steve ,luka and quan
everything is correct but what i did before
formula is quite right and frequency should be around 136KHz but
i placed a wrong 10n cap instead of one 1n ! (what a shame).
avo meter mesuares ic's output frequency about 7.22 KHz and formula says frequency is almost 14.5 KHz (absolutely matches each other 7.22*2~= 14.5)
dear luka ;
i didnt have the graphs.formula calculate output frequency quite acceptable.
i need about 160 w peak power and about 140w continus power , 130w maybe sufficient too.
i count the windings again today and i saw i wond the primary 6+6 with 1.3mm dia and 24+24 with 0.9mm wire for secondary.
dear steve ;
i calculated frequency about 136KHz before but i was Dubious because avo meter was monitoring around 7.5KHz.
so i find the problem " for one more time human kind made mistake"
now the question is "
witch frequency will have best performance with this conditions ? "
6+6 for primary
24+24 for secondary
+-36 for output rails
around 140w output power.
i think it should be 80KHz (40KHz in ic's output pin's)
tell me about your experiences.
best regards and wishes
excuse me for my weak english
amir
thanks for your answers , i appreciate.
dear friends perry ,steve ,luka and quan
everything is correct but what i did before
formula is quite right and frequency should be around 136KHz but
i placed a wrong 10n cap instead of one 1n ! (what a shame).
avo meter mesuares ic's output frequency about 7.22 KHz and formula says frequency is almost 14.5 KHz (absolutely matches each other 7.22*2~= 14.5)
dear luka ;
i didnt have the graphs.formula calculate output frequency quite acceptable.
i need about 160 w peak power and about 140w continus power , 130w maybe sufficient too.
i count the windings again today and i saw i wond the primary 6+6 with 1.3mm dia and 24+24 with 0.9mm wire for secondary.
dear steve ;
i calculated frequency about 136KHz before but i was Dubious because avo meter was monitoring around 7.5KHz.
so i find the problem " for one more time human kind made mistake"
now the question is "
witch frequency will have best performance with this conditions ? "
6+6 for primary
24+24 for secondary
+-36 for output rails
around 140w output power.
i think it should be 80KHz (40KHz in ic's output pin's)
tell me about your experiences.
best regards and wishes
excuse me for my weak english
amir
some results
i did some testing with increasing the number of mosfets and by rewinding the toroid and found much better results with no heating copper or mosfets at all...
here is my configuration:
4 mosfets each side: IRF540
Primary: 0.4mm (26awg) 10x 5 turns
Secondary: 0.4mm (26awg) 6x 9 turns
at no load: 4 amps used from a 7ah battery i am using for testing.... (is this amps good or bad at no load on the output) ?
voltages are seems to be fine....
now, here i need a little help to test this thing without putting the actual load...
is there any way to put the load on the output and check the amps usage and voltage drops?....
can i use a high wattage resistor for testing?...if yes, then what value do you guyz suggest?....
please help if you can.
Thanks again for all of you guyz.
i did some testing with increasing the number of mosfets and by rewinding the toroid and found much better results with no heating copper or mosfets at all...
here is my configuration:
4 mosfets each side: IRF540
Primary: 0.4mm (26awg) 10x 5 turns
Secondary: 0.4mm (26awg) 6x 9 turns
at no load: 4 amps used from a 7ah battery i am using for testing.... (is this amps good or bad at no load on the output) ?
voltages are seems to be fine....
now, here i need a little help to test this thing without putting the actual load...
is there any way to put the load on the output and check the amps usage and voltage drops?....
can i use a high wattage resistor for testing?...if yes, then what value do you guyz suggest?....
please help if you can.
Thanks again for all of you guyz.
4 amps at idle is too much. I'd expect well under 1 amp of current.
You stated that you pulled the core from a switching supply. Was the core being used in the transformer or in an output inductor? It appears that the core was in an inductor. An inductor core won't work.
I'm assuming that you're running at 30-40kHz.
You can use a resistor to test the power supply. You'll need a resistor rated to handle the power produced by the supply. You initially stated that you needed 28v at 7-10 amps. That's 250w+. You'll need a 2.8 ohm, 250 watt resistor.
IRF540 FETs will work for initial testing but they have too much 'on-resistance' unless you use more of them. IRFZ44s are an inexpensive FET and are generally readily available.
You should realize that the resistor will be a MUCH tougher load than and amplifier driving an equivalent load.
How much capacitance to you have across the B+ and ground supply?
You stated that you pulled the core from a switching supply. Was the core being used in the transformer or in an output inductor? It appears that the core was in an inductor. An inductor core won't work.
I'm assuming that you're running at 30-40kHz.
You can use a resistor to test the power supply. You'll need a resistor rated to handle the power produced by the supply. You initially stated that you needed 28v at 7-10 amps. That's 250w+. You'll need a 2.8 ohm, 250 watt resistor.
IRF540 FETs will work for initial testing but they have too much 'on-resistance' unless you use more of them. IRFZ44s are an inexpensive FET and are generally readily available.
You should realize that the resistor will be a MUCH tougher load than and amplifier driving an equivalent load.
How much capacitance to you have across the B+ and ground supply?
Hi perry
Thanks
This smps has a feedback regulator and I have’nt any problem with secondary winding and voltage.
Before I used to wind frafo with 4+4 in primary and 12+12 in secondary but I had a big voltage drop down from 30v to 21 v with 70w output power and a big power dissipation maybe around 70w.
So I decided to increase windings and output rails voltage to 36v in addition to atone voltage drop down.
I got good results then , and now I want to increase performance.
Remember , everything I said happened when pwm frequency was about 14KHz
I think if I switch frequency to 80KHz (2*40KHz) I will have the best performance
Is’nt it ?
Hey aliazhar
4 amps in idle condition is horrible ( my smps consumes around 90 mA in idle mode )
I use two irf540 mosfets too.
You can test just like me (with two 24v bulbs or two truck bulbs if you have had) .
You can use high wattage resistors but I’m not agree with them because it’s hard to get full load with two resistors and they will became too hot in their way to hurt you !! so be carefull.
Thanks
This smps has a feedback regulator and I have’nt any problem with secondary winding and voltage.
Before I used to wind frafo with 4+4 in primary and 12+12 in secondary but I had a big voltage drop down from 30v to 21 v with 70w output power and a big power dissipation maybe around 70w.
So I decided to increase windings and output rails voltage to 36v in addition to atone voltage drop down.
I got good results then , and now I want to increase performance.
Remember , everything I said happened when pwm frequency was about 14KHz
I think if I switch frequency to 80KHz (2*40KHz) I will have the best performance
Is’nt it ?
Hey aliazhar
4 amps in idle condition is horrible ( my smps consumes around 90 mA in idle mode )
I use two irf540 mosfets too.
You can test just like me (with two 24v bulbs or two truck bulbs if you have had) .
You can use high wattage resistors but I’m not agree with them because it’s hard to get full load with two resistors and they will became too hot in their way to hurt you !! so be carefull.
Sorry about the confusion. I didn't look at the user name.
aliazhar:
4 amps is too much at idle. You need to find the reason your supply is drawing so much current before you do any testing.
Does anything get hot?
amirdelta:
What aspect of performance do you want to improve?
You stated that you are using regulation. What's the regulated output voltage?
aliazhar:
4 amps is too much at idle. You need to find the reason your supply is drawing so much current before you do any testing.
Does anything get hot?
amirdelta:
What aspect of performance do you want to improve?
You stated that you are using regulation. What's the regulated output voltage?
yeah thats what i thought too....4amps is a lot on idle...
(when i say idle, it means that everything is running including sg3525 but, no load at the outputs)
i am running it on 37Khz...
i think i need to change this core....i will be receiving etd core today...so, after winding it and testing it, i will post the results.
initially, i need 24v at 7amps....but, i have changed my design a little bit and now, i need 42v 15amps....
is the number (4x4 mosfets) right for this power?
i will change the mosfets to IRFZ044's.....
Thanks again for your help...
(when i say idle, it means that everything is running including sg3525 but, no load at the outputs)
i am running it on 37Khz...
i think i need to change this core....i will be receiving etd core today...so, after winding it and testing it, i will post the results.
initially, i need 24v at 7amps....but, i have changed my design a little bit and now, i need 42v 15amps....
is the number (4x4 mosfets) right for this power?
i will change the mosfets to IRFZ044's.....
Thanks again for your help...
First test:
i have removed the transformer to check the current draw...
current draw=240 milliamp
its definitly is the toroid !
what is the best setup for 42v at 15amps ?
i am gona use ETD49 (since thats the only one i can find)
turns/gauge/number of mosfets ?
need your advice from your experiance.
Thanks again.
i have removed the transformer to check the current draw...
current draw=240 milliamp
its definitly is the toroid !
what is the best setup for 42v at 15amps ?
i am gona use ETD49 (since thats the only one i can find)
turns/gauge/number of mosfets ?
need your advice from your experiance.
Thanks again.
Second Test:
ok, i have just received ETD49 core...
after winding it with:
primary: 0.4mm (26 awg) 12x 4 turns
secondary: 0.4mm (26 awg) 7x 15 turns
output voltages are 42.5vdc (no load)
with no load, amp usage is 0.295 milliamp !!!
now, the next step is using IRFZ44...but, i still like to have the answer to the question above about the winding according to the load specified?
Thanks again everybody.
ok, i have just received ETD49 core...
after winding it with:
primary: 0.4mm (26 awg) 12x 4 turns
secondary: 0.4mm (26 awg) 7x 15 turns
output voltages are 42.5vdc (no load)
with no load, amp usage is 0.295 milliamp !!!
now, the next step is using IRFZ44...but, i still like to have the answer to the question above about the winding according to the load specified?
Thanks again everybody.
hi perry !
it's me ! amirdelta ! (lol)
i think i have a huge power dissipation because with a simple mathematical calculation we can observe a bad power dissipation in the circuit
see:
output power= (+30V*1.5A)+(-30V*1.5A) = 90 W
but
input power = 12V* 12A =144W
dissipation = 144 - 90 = 54w ! ! !
with a 85% efficiency rate, power dissipation should be around
15% * 90w ~= 14w but real dissipation is 54 w ! ! ! !
so efficiency is lower than 50% ! ! !
i'm sure that low core frequency makes a dissipation like that or core acts like a short circuit in higher loads !
i pulled the 10nf (wrong polyester) cap out and placed a 2nf ceramic disk cap instead.it seems that all circuit runs better but when i measured ic's output pins frequency i saw it runs at 3.5KHz ! !
i dont know why !
unfortunately i'm not able to put load on output rails now to see the result but i think overall performance is not acceptable yet !
i think switching pwm frequency to 70KHz (35KHz in output pins) will solve the problem.
what is your idea about ?
aliazhar
i dont think ETD49 can handle 630w power continual .
according to the irf540's datasheet at 100 degree centigrades it can handle 21A continual so 21A*12V = 252w means you can use four irf540 transistors with a easy cooling condition at full loads with any problem .its very easy on the paper ! ! but in real world . . . ? ? ?
and about winding i think 5+5 for primary and 22+22 for secondary
with 12v voltage drop down in full load will be a good winding.
you will recieve about 54v but my experience says you will have a
big voltage drop down.
without any dissipation you have to use a super quality battery to give you 50A continual to make 630w real !
i dont think so !
it's me ! amirdelta !
(lol) i think i have a huge power dissipation because with a simple mathematical calculation we can observe a bad power dissipation in the circuit
see:
output power= (+30V*1.5A)+(-30V*1.5A) = 90 W
but
input power = 12V* 12A =144W
dissipation = 144 - 90 = 54w ! ! !
with a 85% efficiency rate, power dissipation should be around
15% * 90w ~= 14w but real dissipation is 54 w ! ! ! !
so efficiency is lower than 50% ! ! !
i'm sure that low core frequency makes a dissipation like that or core acts like a short circuit in higher loads !
i pulled the 10nf (wrong polyester) cap out and placed a 2nf ceramic disk cap instead.it seems that all circuit runs better but when i measured ic's output pins frequency i saw it runs at 3.5KHz ! !
i dont know why !
unfortunately i'm not able to put load on output rails now to see the result but i think overall performance is not acceptable yet !
i think switching pwm frequency to 70KHz (35KHz in output pins) will solve the problem.
what is your idea about ?
aliazhar
i dont think ETD49 can handle 630w power continual .
according to the irf540's datasheet at 100 degree centigrades it can handle 21A continual so 21A*12V = 252w means you can use four irf540 transistors with a easy cooling condition at full loads with any problem .its very easy on the paper ! ! but in real world . . . ? ? ?
and about winding i think 5+5 for primary and 22+22 for secondary
with 12v voltage drop down in full load will be a good winding.
you will recieve about 54v but my experience says you will have a
big voltage drop down.
without any dissipation you have to use a super quality battery to give you 50A continual to make 630w real !
i dont think so !
what if i reduce the pwm frequency to like 150hz (i saw one circuit of 750watts here in this thread)
and then use more copper in the core to get that much power?
is it possible then?
what is the best choice - toroid or ETD core?....
if toroid then which one?....any links where i can buy it?
Thanks again guyz.
and then use more copper in the core to get that much power?
is it possible then?
what is the best choice - toroid or ETD core?....
if toroid then which one?....any links where i can buy it?
Thanks again guyz.
amirdelta:
You need to check your frequency counter to make sure it's working properly. If you have an oscilloscope, calculate the frequency using the timebase and the number of divisions/cycle.
aliazhar:
150Hz is too low. You need to begin with an operating frequency and get the supply working properly. Then you can tweak it to get precisely what you want. Most all commercial amplifiers operate their power supplies at or very near 30kHz. If you can't get it to work properly at 30kHz, there is a problem.
cwsbytemark.com has suitable toroidal cores.
You need to check your frequency counter to make sure it's working properly. If you have an oscilloscope, calculate the frequency using the timebase and the number of divisions/cycle.
aliazhar:
150Hz is too low. You need to begin with an operating frequency and get the supply working properly. Then you can tweak it to get precisely what you want. Most all commercial amplifiers operate their power supplies at or very near 30kHz. If you can't get it to work properly at 30kHz, there is a problem.
cwsbytemark.com has suitable toroidal cores.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.