Here's the full schematic:
http://sziget.mine.nu/~danko/aramkor/testbed/16/schematic_full.png
It uses a small PCB-mounted 6VA transformer for "house-keeping".
Here's a close-up:
http://sziget.mine.nu/~danko/aramkor/testbed/16/schematic.png
Here's the PCB layout:
http://sziget.mine.nu/~danko/aramkor/testbed/16/pcb.png
WARNING!! The parts' values not same values I have soldered on the PCB. I changed them when I was experimenting. I haven't updated the schematic yet.
Oh, another important thing! "R5" is the slope-compensator resistor, but that didn't work. I used instead of that an emitter follower. I have changed the current sense resistors also. Now there are 2 paralelled 0.1R/2W resistors.
I will post an updated schematic, when I have time to see what I have changed.
http://sziget.mine.nu/~danko/aramkor/testbed/16/schematic_full.png
It uses a small PCB-mounted 6VA transformer for "house-keeping".
Here's a close-up:
http://sziget.mine.nu/~danko/aramkor/testbed/16/schematic.png
Here's the PCB layout:
http://sziget.mine.nu/~danko/aramkor/testbed/16/pcb.png
WARNING!! The parts' values not same values I have soldered on the PCB. I changed them when I was experimenting. I haven't updated the schematic yet.
Oh, another important thing! "R5" is the slope-compensator resistor, but that didn't work. I used instead of that an emitter follower. I have changed the current sense resistors also. Now there are 2 paralelled 0.1R/2W resistors.
I will post an updated schematic, when I have time to see what I have changed.
On the pictures, there's an ETD49, which I re-re-re-rewinded many many times, this is why it looks like that...
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8740.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8741.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8743.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8744.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8745.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8746.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8747.jpg
After theese steps I solder the end of the copper wires to the pins, put in the cores, and I solder the transformer onto the PCB.
Edit: I don't remember the number of the turns, and the number of wires I used to wind. One wire's diameter was 0.4mm.
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8740.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8741.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8743.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8744.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8745.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8746.jpg
http://sziget.mine.nu/~danko/aramkor/testbed/16/dscn8747.jpg
After theese steps I solder the end of the copper wires to the pins, put in the cores, and I solder the transformer onto the PCB.
Edit: I don't remember the number of the turns, and the number of wires I used to wind. One wire's diameter was 0.4mm.
you should winded your trafo like this (Fig. 17)
First you place 1/2 of primary then all secondery's and the 2/2 primary. This helps to minimaze a lot of things as can Eva tell you, making trafo perform best
First you place 1/2 of primary then all secondery's and the 2/2 primary. This helps to minimaze a lot of things as can Eva tell you, making trafo perform best
core size
You mentioned you have an ETD44 core, according to Pressman
at a freq of 100kHz this core set can provide over a Kw. I am working on HalfBridge using voltage mode control @75kHz an using 90 Volts as a low input and a bmax of 1600 the turns calculate as: Using SG3525A PWM chip
90VAC * 1.4 - 20V(for ripple and drops)
gives abot 107VDC.
Np = 107*E8/ (4*1600*75*E3*1.74) = 12 Turns
Since this will be a center tap FWB output I use this formula to calculate the secondary and I want a +/- 55 Volts @ 8A
out (aprox: 900 watts) Ipri = 10A . I plan to use IGBT's as switch
Ns = Np*Vs/Vp = 12*110/110 = 12 T
Formulas are from George Chryssis "High Frequency Switching Power Supplies Theory and Design" dtd 1984 I also have Marty Brown and Pressman as a reference.
Before winding the primary I plan on placing a shield that will be grounded to the output center tap, and then wind 1/2 primary then wind all of secondary and place another shield grounded as before, wind remainder of primary and place another shield grounded as before. A winding of three turns for aux power and another shield grounded as before. the primary will be 4 strands of AWG #19 bifilar wound, secondary will be two #17 AWG. I know wire size might be large but more than 50% of area will be used?????? I hope this will be ok. Inductors will be 140uH place in each output but wound together on a common core. Two 1000 uF caps as output filters on each rail. A type 3 compensation will be used. I was thinking about using the TL431 for isolation but the specs are for 36 volts max and I don't know how a zener in series would effect it yet (still a thought) All windinding's will be insulated with 3m yellow tape.
After reading Brown and comparing his calculations for xfmr
on page 123 the calculation for the primary he recommends the highest expected input voltage .
therefore my xfmr primary turns would be
Npri = 368 V *E8/(4 *75*E3* 2800G *1.74) = 26 T
(2 times the copper) and the secondary would be
Nsec = 1.1(111*25)/252*.95 = 12 T ok
and Pressman uses 259VDC as Vmin therefore
Npri = 130E8/(4*1600*75E3*1.74) = 16T
Nsec = Np*VS/130 = 16*110/130 = 14 T
and I have noticed that a value of 160 is used quite often in this forum therefore:
Npri = 160E8/(4*1600*75E3*1.74) = 20 T
Nsec = 20*110/160 = 14 T
Can you help with the correct numbers?? I like Browns book but has quite a few error's from the printer I think (math values ) in examples......and then again maybe I need to read it again.
You mentioned you have an ETD44 core, according to Pressman
at a freq of 100kHz this core set can provide over a Kw. I am working on HalfBridge using voltage mode control @75kHz an using 90 Volts as a low input and a bmax of 1600 the turns calculate as: Using SG3525A PWM chip
90VAC * 1.4 - 20V(for ripple and drops)
gives abot 107VDC.
Np = 107*E8/ (4*1600*75*E3*1.74) = 12 Turns
Since this will be a center tap FWB output I use this formula to calculate the secondary and I want a +/- 55 Volts @ 8A
out (aprox: 900 watts) Ipri = 10A . I plan to use IGBT's as switch
Ns = Np*Vs/Vp = 12*110/110 = 12 T
Formulas are from George Chryssis "High Frequency Switching Power Supplies Theory and Design" dtd 1984 I also have Marty Brown and Pressman as a reference.
Before winding the primary I plan on placing a shield that will be grounded to the output center tap, and then wind 1/2 primary then wind all of secondary and place another shield grounded as before, wind remainder of primary and place another shield grounded as before. A winding of three turns for aux power and another shield grounded as before. the primary will be 4 strands of AWG #19 bifilar wound, secondary will be two #17 AWG. I know wire size might be large but more than 50% of area will be used?????? I hope this will be ok. Inductors will be 140uH place in each output but wound together on a common core. Two 1000 uF caps as output filters on each rail. A type 3 compensation will be used. I was thinking about using the TL431 for isolation but the specs are for 36 volts max and I don't know how a zener in series would effect it yet (still a thought) All windinding's will be insulated with 3m yellow tape.
After reading Brown and comparing his calculations for xfmr
on page 123 the calculation for the primary he recommends the highest expected input voltage .
therefore my xfmr primary turns would be
Npri = 368 V *E8/(4 *75*E3* 2800G *1.74) = 26 T
(2 times the copper) and the secondary would be
Nsec = 1.1(111*25)/252*.95 = 12 T ok
and Pressman uses 259VDC as Vmin therefore
Npri = 130E8/(4*1600*75E3*1.74) = 16T
Nsec = Np*VS/130 = 16*110/130 = 14 T
and I have noticed that a value of 160 is used quite often in this forum therefore:
Npri = 160E8/(4*1600*75E3*1.74) = 20 T
Nsec = 20*110/160 = 14 T
Can you help with the correct numbers?? I like Browns book but has quite a few error's from the printer I think (math values ) in examples......and then again maybe I need to read it again.
Chas1-
I have read all of the books you mentioned. Indeed, all together (Chryssis, Pressman, and Brown) they form the "bible" I refer to whenever doing SMPSs. I have not done a flyback yet (almost always half- & full-bridges & push-pulls) but after re-reading the transformers section of Chryssis' 2nd edition (1989), your numbers seem correct.
Also, you mentioned you are using SG3525AN as your control. How are you driving IGBTs? IR2110/2113? Driver xfmr? If you were to use the MC33025 (the '3525's current-mode cousin), you could just follow the design example in Brown's book for a 280W (28V@10A) Half-bridge supply. Just scale the magnetics up to the ETD44, plug in your desired output voltages in place of the 28V given, and let 'er rip.. Also follow the section that covers feedback compensation, using a TL431 and Optocoupler for isolation. I have used this example as boilerplate for many AC-DC and DC-DC converters I've done over the years, with very good results.
Danko-
If you can get any one, or better yet, all three of these books, you will have some very powerful "tools" in your design toolbox.
Just my humble opinion.
Steve
I have read all of the books you mentioned. Indeed, all together (Chryssis, Pressman, and Brown) they form the "bible" I refer to whenever doing SMPSs. I have not done a flyback yet (almost always half- & full-bridges & push-pulls) but after re-reading the transformers section of Chryssis' 2nd edition (1989), your numbers seem correct.
Also, you mentioned you are using SG3525AN as your control. How are you driving IGBTs? IR2110/2113? Driver xfmr? If you were to use the MC33025 (the '3525's current-mode cousin), you could just follow the design example in Brown's book for a 280W (28V@10A) Half-bridge supply. Just scale the magnetics up to the ETD44, plug in your desired output voltages in place of the 28V given, and let 'er rip.. Also follow the section that covers feedback compensation, using a TL431 and Optocoupler for isolation. I have used this example as boilerplate for many AC-DC and DC-DC converters I've done over the years, with very good results.
Danko-
If you can get any one, or better yet, all three of these books, you will have some very powerful "tools" in your design toolbox.
Just my humble opinion.
Steve
Gate Drive
hey Steve
thanks for conformation, below is what I have so far.
I am using gate drive transformer wound on a ferroxcube toroid
846t250-3c8 has an AL 1650mh/1000T Npri = 16T (inductance of 420uh) and sec of 12 T, I pri works out 12*6.6us/422uh = 185ma
energy = 1/2 * 422uh(185)^2 = 7uJ and this core can store 16uJ
I put a 2.5 ohm resistor in series with pri of transformer, I might increase this to 4.5 ohm. My large problem is closing feedback loop using the TL431, can't seem to find decent app note for this and after crunching numbers in Browns examples on pg 180 - 183
simulating them along with my bode plots I can't get close to what he has in fig 4-24 (note: using his exact numbers as an example for proof) therefore there must be a cockpit problem or I
don't understand bode plots.
hey Steve
thanks for conformation, below is what I have so far.
I am using gate drive transformer wound on a ferroxcube toroid
846t250-3c8 has an AL 1650mh/1000T Npri = 16T (inductance of 420uh) and sec of 12 T, I pri works out 12*6.6us/422uh = 185ma
energy = 1/2 * 422uh(185)^2 = 7uJ and this core can store 16uJ
I put a 2.5 ohm resistor in series with pri of transformer, I might increase this to 4.5 ohm. My large problem is closing feedback loop using the TL431, can't seem to find decent app note for this and after crunching numbers in Browns examples on pg 180 - 183
simulating them along with my bode plots I can't get close to what he has in fig 4-24 (note: using his exact numbers as an example for proof) therefore there must be a cockpit problem or I
don't understand bode plots.
SG3525A vs MC34025P
Steve
Your are correct about the MC34025P being the best choice for my design I like a lot of people have about 10 SG3525A's handy and rush to use them in a new design and since I had been away from this for a while I started my design which is to power a 250 Watt AB amp designed with the newest transistors from onsemi who were kind enough to send me two rails as samples (NJL1302-D & NJL4281-D). By using the MC34025 it will also save PCB real estate and money even though the cost of chip is 3.00 dollars at Newark and I have not past the point of no return in this project I had the PCB layout up to the control circuit and sense all components are the same I can finish PCB before chips arrive, many, many thanks for the tip when I finish I will be glad to share it with all DIY. Nice forum!!!!
Steve
Your are correct about the MC34025P being the best choice for my design I like a lot of people have about 10 SG3525A's handy and rush to use them in a new design and since I had been away from this for a while I started my design which is to power a 250 Watt AB amp designed with the newest transistors from onsemi who were kind enough to send me two rails as samples (NJL1302-D & NJL4281-D). By using the MC34025 it will also save PCB real estate and money even though the cost of chip is 3.00 dollars at Newark and I have not past the point of no return in this project I had the PCB layout up to the control circuit and sense all components are the same I can finish PCB before chips arrive, many, many thanks for the tip when I finish I will be glad to share it with all DIY. Nice forum!!!!
Chas-
Have ONsemi "Sample" the MC33025's as well.
You may notice that I repeatedly use 33035 instead of 34025 beacuse the 33 has the industrial temperature range of -40 +105 Centigrade as opposed to the 34's temp range of 0-70 Centigrade. ON will sample either the 33 or 34, you specify. I go with the 33 because, well, it's better! 
Nice of them to sample them. These are the best bipolars, IMHO, to use for our applications.
Have ONsemi "Sample" the MC33025's as well.
You may notice that I repeatedly use 33035 instead of 34025 beacuse the 33 has the industrial temperature range of -40 +105 Centigrade as opposed to the 34's temp range of 0-70 Centigrade. ON will sample either the 33 or 34, you specify. I go with the 33 because, well, it's better!
Nice of them to sample them. These are the best bipolars, IMHO, to use for our applications.
Luka,
Can yo usend a jpg ir gif? I don't have MS Schedule, and that is the program it opens in. Thanks.
Chas-
I would DEFINITELY go with the 34035. The SG3525 is a fine chip, and many SMPSs have been designed around it giving good results, BUT, the MC33025/34025 is that much better. Cycle-by-cycle current-limiting, wider gain-bandwidth product for the error amp, a higher speed oscillator (1MHz), and 2A totem-pole outputs. I like! I like!
Almost the same pinout as the '3525, compensation pin moved next to the erroe inputs, and pin 9 is now the input for the current-limit, and pin 7 goes from being a deadtime control to ramp input. It's how this pin is connected that determines whether it is run in voltage- or current-mode. Connect it to pin 5 (Ct) like in the '3525, and you have voltage-mode operation. Connect it to pin 9 (current-limit) via a filter, and voila! You have current-mode control. Pretty versatile chip.
Can yo usend a jpg ir gif? I don't have MS Schedule, and that is the program it opens in. Thanks.
Chas-
I would DEFINITELY go with the 34035. The SG3525 is a fine chip, and many SMPSs have been designed around it giving good results, BUT, the MC33025/34025 is that much better. Cycle-by-cycle current-limiting, wider gain-bandwidth product for the error amp, a higher speed oscillator (1MHz), and 2A totem-pole outputs.
I like! I like!Almost the same pinout as the '3525, compensation pin moved next to the erroe inputs, and pin 9 is now the input for the current-limit, and pin 7 goes from being a deadtime control to ramp input. It's how this pin is connected that determines whether it is run in voltage- or current-mode. Connect it to pin 5 (Ct) like in the '3525, and you have voltage-mode operation. Connect it to pin 9 (current-limit) via a filter, and voila! You have current-mode control. Pretty versatile chip.
34025 vs 3525
Steve
I have the datasheet and along with Marty Browns book and your input I have redesigned (that's not correct I have copied and scaled the 280 watt HB on page 132 and added my tansformers. I have the parts on the way from Digi-key the two 2uf voltage divider caps are indicated as electrolytic but I am sure they are polyester film capacitors so I ordered ECG 250VAC caps What do you think ? I would have rather had Sprague 430P's but Digi-key does not stock these. I have one other question before I finish laying out my board the grounding, I ground the input power and the output is grounded using the CT of transformer and the analog is grounded in a star simular to audio with a copper flood. The layout is single sided. In the schematic it has a GND1 which is connected to the power ground and earth ground
connected to output which when boxed would invite ground loops I think, also I am going to regulate from the postive rail and use a zener to lower the voltage to the tl431, what do you think?
again thanks for input
Chas
Steve
I have the datasheet and along with Marty Browns book and your input I have redesigned (that's not correct I have copied and scaled the 280 watt HB on page 132 and added my tansformers. I have the parts on the way from Digi-key the two 2uf voltage divider caps are indicated as electrolytic but I am sure they are polyester film capacitors so I ordered ECG 250VAC caps What do you think ? I would have rather had Sprague 430P's but Digi-key does not stock these. I have one other question before I finish laying out my board the grounding, I ground the input power and the output is grounded using the CT of transformer and the analog is grounded in a star simular to audio with a copper flood. The layout is single sided. In the schematic it has a GND1 which is connected to the power ground and earth ground
connected to output which when boxed would invite ground loops I think, also I am going to regulate from the postive rail and use a zener to lower the voltage to the tl431, what do you think?
again thanks for input
Chas
feedback
I have seen versions of this in many designs mainly SMPS for car amps where both rails are regulated but I think by using this it may require using the internal error amp for compensation and I feel more confortable using a stand alone opamp which the tl431 is, however I plan on breadboarding both and since I am going to use the MC33025 I am going to try compensation using its internal amp. I am not trying to reinvent the wheel but I see no reason not to use the internal amp except isolation.
chas
I have seen versions of this in many designs mainly SMPS for car amps where both rails are regulated but I think by using this it may require using the internal error amp for compensation and I feel more confortable using a stand alone opamp which the tl431 is, however I plan on breadboarding both and since I am going to use the MC33025 I am going to try compensation using its internal amp. I am not trying to reinvent the wheel but I see no reason not to use the internal amp except isolation.
chas