After third time winding transformer I feel I need an advice
. I made DC-DC PSU with push-pull topology (12->+-35V, tl494, 4xIRF540). When applying an 1.5 Amps load, the voltage dropped to 26v. Is it normal ? I don't think so... Where should I search the problem? Most likely it's transformer. I found a huge collection of formulas, calculations e.t.c. , but I can't find just HOW-TO wind it! Whether I should overwind wires?.. Whether all surface should be occupied?.. and so on.. I shall be grateful for any information covering this question!
p.s. Maybe somebody can make photos of their own tranformers, so I'd take a look!
. I made DC-DC PSU with push-pull topology (12->+-35V, tl494, 4xIRF540). When applying an 1.5 Amps load, the voltage dropped to 26v. Is it normal ? I don't think so... Where should I search the problem? Most likely it's transformer. I found a huge collection of formulas, calculations e.t.c. , but I can't find just HOW-TO wind it! Whether I should overwind wires?.. Whether all surface should be occupied?.. and so on.. I shall be grateful for any information covering this question!p.s. Maybe somebody can make photos of their own tranformers, so I'd take a look!
Voltage Drops
Couple o' questions:
1) What type of core material/geometry are you using? PWM IC switching frequency?
2) What reference material(s) is(are) you using to wind it?
I have a great source for you to use. In another thread I regularly post to, I recommended the following book: "High Frequency Switching Power Supplies" by George Chryssis (c) 1989 edition. In this book, you will learn everything you need to know to design, fabricate and construct a prototype of a switching power supply. Each chapter deals with a specific component or part of the power supply: switch, output recifier(s), transformer, control ckt, etc. Chapter 4, which deals entirely with the transformer, contains a 10-14 step equation-oriented development process that is very cut-n-dried. The theory is kept to a minimum, but theoretical stuff is explained where necessary.
I have used this process to design several DC-DC Converters with great success. Hope this helps, and Best of Luck!
'73,
de N8XO
Couple o' questions:
1) What type of core material/geometry are you using? PWM IC switching frequency?
2) What reference material(s) is(are) you using to wind it?
I have a great source for you to use. In another thread I regularly post to, I recommended the following book: "High Frequency Switching Power Supplies" by George Chryssis (c) 1989 edition. In this book, you will learn everything you need to know to design, fabricate and construct a prototype of a switching power supply. Each chapter deals with a specific component or part of the power supply: switch, output recifier(s), transformer, control ckt, etc. Chapter 4, which deals entirely with the transformer, contains a 10-14 step equation-oriented development process that is very cut-n-dried. The theory is kept to a minimum, but theoretical stuff is explained where necessary.
I have used this process to design several DC-DC Converters with great success. Hope this helps, and Best of Luck!
'73,
de N8XO
great voltage drop question
Before I can answer or give advice, I'd need to know the input and output voltages (+/- 35V I presume) and currents (1.5A full load on both?). What is your switching frequency, control is voltage mode, I presume, vs. current mode, continuous conduction mode, or possibly discontinuous, etc.? Also, how did you design the transformer? From what you've stated so far, it could be a number of things, with transformer leakage inductance coming to mind first. If the TL494 is voltage mode control, the transformer could be saturating due to flux imbalance. For push pull topology, current mode control should be employed. Is the TL494 internally compensated? If not, what type of external frequency compensation are you using? Best regards.
Before I can answer or give advice, I'd need to know the input and output voltages (+/- 35V I presume) and currents (1.5A full load on both?). What is your switching frequency, control is voltage mode, I presume, vs. current mode, continuous conduction mode, or possibly discontinuous, etc.? Also, how did you design the transformer? From what you've stated so far, it could be a number of things, with transformer leakage inductance coming to mind first. If the TL494 is voltage mode control, the transformer could be saturating due to flux imbalance. For push pull topology, current mode control should be employed. Is the TL494 internally compensated? If not, what type of external frequency compensation are you using? Best regards.
I use two toroidal 42x25x11 cores, pulled together with isolation tape(0.3 mm width).material initial magnetic permeability is 2000(this is "the best russian material for SMPS" as I heard from clever men). All reference materials I use are unfortunately in Russian (.Approximate schematic may be found at : http://www.bluesmobil.com/shikhman/arts/jenspow.gif
Thanks for Book recomendations, but here in Russia, we have some difficulties with buying such books
. I got some not so bad book (can't remember the name and the author for now). It covers almost all SMPS creation aspects, except HOW-TO wind transformer!
And now full data about my PSU:
1. output voltage +-35. Load (~1.5A) was applied to 70v and Voltage has dropped to ~52 V (26V on each winding).
2. Switching Frequency is ~40 kHz(Maybe I should try to higher it?)
3. PSU have no output control. I am planning to do Voltage control, but i'm interested in current control mode too (can't you explain how to realize it?). Pulse width in my PSY is maximum.
4. Unfortunately I don't know what is "Frequency compensation"
few more questions:
How to detect "transformer leakage inductance" and why happens Flux imbalance.
What will happen if copper ways leading to transformers windings are too thin?
Is it too bad or not too bad that I have a difference of 0.5 V on the secondary windings?
p.s. One of last days I tried new PSU and got a new result:
when applying a small load of 0.2 A scheme works for 2-3 sec (a time while output capacitors holding energy i suppose), then voltage drops, (I disconnect load), and I view that Impulses on the gate of all of the output transistors become very short- 0.1% of their initial lenght. then I wait 30-60 sec and impulses begin to restore their lenght( At first it is going slow ,then very fast! )
I can understand that transformer leakage or flux imbalance is difficult thing, but new result really drives me mad!!
Beforehand is grateful for the answers!
Thanks for Book recomendations, but here in Russia, we have some difficulties with buying such books
. I got some not so bad book (can't remember the name and the author for now). It covers almost all SMPS creation aspects, except HOW-TO wind transformer!And now full data about my PSU:
1. output voltage +-35. Load (~1.5A) was applied to 70v and Voltage has dropped to ~52 V (26V on each winding).
2. Switching Frequency is ~40 kHz(Maybe I should try to higher it?)
3. PSU have no output control. I am planning to do Voltage control, but i'm interested in current control mode too (can't you explain how to realize it?). Pulse width in my PSY is maximum.
4. Unfortunately I don't know what is "Frequency compensation"
few more questions:
How to detect "transformer leakage inductance" and why happens Flux imbalance.
What will happen if copper ways leading to transformers windings are too thin?
Is it too bad or not too bad that I have a difference of 0.5 V on the secondary windings?
p.s. One of last days I tried new PSU and got a new result:
when applying a small load of 0.2 A scheme works for 2-3 sec (a time while output capacitors holding energy i suppose), then voltage drops, (I disconnect load), and I view that Impulses on the gate of all of the output transistors become very short- 0.1% of their initial lenght. then I wait 30-60 sec and impulses begin to restore their lenght( At first it is going slow ,then very fast! )
I can understand that transformer leakage or flux imbalance is difficult thing, but new result really drives me mad!!
Beforehand is grateful for the answers!
Switcher Schematic
drob:
OK, I've looked at the schematic. Basically, it looks like a good start for the design. Since my Cyrillic is a little rusty, I'll do my best.
1) TL494: This is a good chip to use- if you don't want or need current-mode control. Since you've indicated a desire to use current-mode, pick another chip. Two chips come to mind right away: the UC3846 (Unitrode division of Texas Instruments) and the MC33025 (ON Semiconductor, formerly Motorola). The '3846 is a good chip, but I would stick with the '33025, because it is a newer chip, and it will continue to be supported by ON for quite some time. Go to www.onsemi.com, and punch in MC33025 until you get to its page. Download the spec sheet (it's about 12-20 pages), but it's worth it.
2) 40kHz Switching frequency: Was this your SWITCHING frequency or your CLOCK frequency? For push-pulls, the switching freq is half the clock freq. If this is the switching frequency, then this is fine.
3) Frequency Compensation: This is VERY important. It determines how fast (or slow) your circuit responds to changing loads and changing supply voltages. There are different compensation networks for different topologies and control methods. I would check the books you have available and read the chapter(s) on Compensation. Very important.
4) Voltage control: Otherwise known as feedback, you will need to use the (+) input of the error amp to sense output voltage. Using an optocoupler will help in isolating the secondary side ground from the primary side ground and decrease noise on the (+) and (-) outputs.
I'm running out of steam for now, So I will leave it here. But I do wish there were a good source for figuring out number of turns for the transformer. Before I go, you do not need to isolate the two cores from each other. Just tape them together. Their Flux numbers (u=2000) is a good number. Just remember, this number is probably chosen at 20 kHz, and goes down as frequency goes up, and not linearly, either. Also, since you're using two cores, the cross-sectional area is DOUBLE, so your turns will be half. If you can get a data sheet on the cores, that would help, too. Best of luck, and stay tuned for part II......
'73, de N8XO
drob:
OK, I've looked at the schematic. Basically, it looks like a good start for the design. Since my Cyrillic is a little rusty, I'll do my best.
1) TL494: This is a good chip to use- if you don't want or need current-mode control. Since you've indicated a desire to use current-mode, pick another chip. Two chips come to mind right away: the UC3846 (Unitrode division of Texas Instruments) and the MC33025 (ON Semiconductor, formerly Motorola). The '3846 is a good chip, but I would stick with the '33025, because it is a newer chip, and it will continue to be supported by ON for quite some time. Go to www.onsemi.com, and punch in MC33025 until you get to its page. Download the spec sheet (it's about 12-20 pages), but it's worth it.
2) 40kHz Switching frequency: Was this your SWITCHING frequency or your CLOCK frequency? For push-pulls, the switching freq is half the clock freq. If this is the switching frequency, then this is fine.
3) Frequency Compensation: This is VERY important. It determines how fast (or slow) your circuit responds to changing loads and changing supply voltages. There are different compensation networks for different topologies and control methods. I would check the books you have available and read the chapter(s) on Compensation. Very important.
4) Voltage control: Otherwise known as feedback, you will need to use the (+) input of the error amp to sense output voltage. Using an optocoupler will help in isolating the secondary side ground from the primary side ground and decrease noise on the (+) and (-) outputs.
I'm running out of steam for now, So I will leave it here. But I do wish there were a good source for figuring out number of turns for the transformer. Before I go, you do not need to isolate the two cores from each other. Just tape them together. Their Flux numbers (u=2000) is a good number. Just remember, this number is probably chosen at 20 kHz, and goes down as frequency goes up, and not linearly, either. Also, since you're using two cores, the cross-sectional area is DOUBLE, so your turns will be half. If you can get a data sheet on the cores, that would help, too. Best of luck, and stay tuned for part II......
'73, de N8XO
voltage drop : recommendations
With plus and minus 35 volts, 1.5 amp each, the total output power is 105 watts. At this power level there are other topology choices besides push-pull, such as forward converter and flyback converter. With a forward or push-pull, you need inductors after the transformers to store the energy on a per cycle basis. You are using inductors after the transformer, aren't you? If not, it could very well be the caps discharging, as you said, since there would be no inductors to store the energy.
With any transformer-isolated topology except flyback, the transformer merely couples and transfers energy from the primary side to the secondary side. It does not store any significant amount of energy. That is why inductors are needed on the secondary side, one inductor for each output. When the power transistor is turned on, current flows from the input battery, or supply, into the transformer primary winding, and current also flows in the secondary winding through the forward biased rectifier on the secondary side, through the output filter inductor, and into the output filter capacitor and load. When the power transistor is turned off, the current in the output filter inductor continues through the catch rectifier, into the output filter cap and load. During the on time, energy is transferred from the input to the load. During the off time, the inductor transfers the energy it stored during the on time, into the output cap and load. The inductor must be there.
If you aren't using an output filter inductor, that's the trouble. Let me know, and we'll go from there. Best regards.
With plus and minus 35 volts, 1.5 amp each, the total output power is 105 watts. At this power level there are other topology choices besides push-pull, such as forward converter and flyback converter. With a forward or push-pull, you need inductors after the transformers to store the energy on a per cycle basis. You are using inductors after the transformer, aren't you? If not, it could very well be the caps discharging, as you said, since there would be no inductors to store the energy.
With any transformer-isolated topology except flyback, the transformer merely couples and transfers energy from the primary side to the secondary side. It does not store any significant amount of energy. That is why inductors are needed on the secondary side, one inductor for each output. When the power transistor is turned on, current flows from the input battery, or supply, into the transformer primary winding, and current also flows in the secondary winding through the forward biased rectifier on the secondary side, through the output filter inductor, and into the output filter capacitor and load. When the power transistor is turned off, the current in the output filter inductor continues through the catch rectifier, into the output filter cap and load. During the on time, energy is transferred from the input to the load. During the off time, the inductor transfers the energy it stored during the on time, into the output cap and load. The inductor must be there.
If you aren't using an output filter inductor, that's the trouble. Let me know, and we'll go from there. Best regards.
Thanks for advices!
Last days I had troubles with my diagnostic equipment : My 220->12v supply and my oscillograph were broken simultaneously!
Thats why I don't have new shocking results!
Today I got Car battery for 12V supply and new 2 channel oscillograph. Will make new experiments!.
I am using output inductors and capacitors. capacitors are 4700 uF x 35V. Inductors are: 10 coils of 1.8 mm wire on unknown ferrite core (cylinder 10mm diameter, 30mm length ripped from some Chinese radio). What minimal Capacitance and Inductance will you recommend me (wished SMPS power is at least 200-300w)?
Thanks for the new chips recomendations, but there are none of them in stock of our shop. Maybe I can order them, but it is very difficult. I'll do that only when I'll be in the total Dead End!.
the clock frequency is 40 kHz.
Frequency Compensation is interesting thing, but I don't undersstand it completely. Maybe you could show me some examples or some articles in Net. Better with reference to TL494 (if type of controller is important).
I don't want to implement Voltage/Current Control for now, as I believe it is important for the "Quality" of the output voltage, not for the Output Power. As I heard many Industrial Car Audio Amplifiers use unregulated PSU and are feeling good!
Best regards!
Waiting for THE part II
Last days I had troubles with my diagnostic equipment : My 220->12v supply and my oscillograph were broken simultaneously!
Thats why I don't have new shocking
results!Today I got Car battery for 12V supply and new 2 channel oscillograph. Will make new experiments!.
I am using output inductors and capacitors. capacitors are 4700 uF x 35V. Inductors are: 10 coils of 1.8 mm wire on unknown ferrite core (cylinder 10mm diameter, 30mm length ripped from some Chinese radio). What minimal Capacitance and Inductance will you recommend me (wished SMPS power is at least 200-300w)?
Thanks for the new chips recomendations, but there are none of them in stock of our shop
. Maybe I can order them, but it is very difficult. I'll do that only when I'll be in the total Dead End!.the clock frequency is 40 kHz.
Frequency Compensation is interesting thing, but I don't undersstand it completely. Maybe you could show me some examples or some articles in Net. Better with reference to TL494 (if type of controller is important).
I don't want to implement Voltage/Current Control for now, as I believe it is important for the "Quality" of the output voltage, not for the Output Power. As I heard many Industrial Car Audio Amplifiers use unregulated PSU and are feeling good!
Best regards!
Waiting for THE part II
I can't remember for now the exact output voltage, but I remember that I saw interesting phenomenon:
Voltage was dropped strongly(1.5-2.5 V as I remember) when applying not so big load (in my case 100-200 mA).
As I remember when applied greater load voltage drop was less.
I mean that voltage drop was not linear with increase of a load!
Today I got new precision digital multimeter and I'll try to measure exact voltage drop with different loads.
Best regards!
Hello drob! I'm your neighbour - Lithuania
About schematic you are doing. I done many car amplifiers and DC - DC converters too and this one you are doing now. I want to share some information about this schematic. My SMPS is done with TL494, driver transistors are BD140 (also it can be kt814, bd910, kt818) the main thing for this transistors - I > 1A, F > 3MHz. Diodes may be russian type or analogue - no difference - bove are good Mosfet transistors are IRFZ44. They have 0,028mO RDS and one pair of this transistors can supply power for 100W amplifier very nice I had used 6 (3 pairs) transistors for 300W (RMS offcourse) amplifier and have no problems ever Also there are many others transistors, but i think irf540 is better for amplifier than SMPS (i think so). I use one toroidal K45x28x8 core, 2000u (russian type) for transformer. First winding contain 2*4 winds, second 2*15 (for +-45V supply). Diodes KD2999A - in my case they handle about 7A (U = +-45 V) without cooling. Switching frequency is 50 kHz (clock 100kHz). No voltage or frequency compensation. One more thing - i use 10 000uF capasitor on +12V rail right beside transformer - it helps very much not to drop voltage in secondary (hold about 7 volts at 200W load). Other things like in schematic. Now i don't remember any precise measurements, but with load of about 200W i have 5 volts drop on secondary. One more thing - capasitor of 2200uF i put near chip - it helps to hold voltage on the output (TL494) or mosfet input (gate voltage). I think that you use enough thickness of windings and there is no comments about it.
If there any questions - ask
Good luck!
About schematic you are doing. I done many car amplifiers and DC - DC converters too and this one you are doing now. I want to share some information about this schematic. My SMPS is done with TL494, driver transistors are BD140 (also it can be kt814, bd910, kt818) the main thing for this transistors - I > 1A, F > 3MHz. Diodes may be russian type or analogue - no difference - bove are good
Mosfet transistors are IRFZ44. They have 0,028mO RDS and one pair of this transistors can supply power for 100W amplifier very nice I had used 6 (3 pairs) transistors for 300W (RMS offcourse) amplifier and have no problems ever Also there are many others transistors, but i think irf540 is better for amplifier than SMPS (i think so). I use one toroidal K45x28x8 core, 2000u (russian type) for transformer. First winding contain 2*4 winds, second 2*15 (for +-45V supply). Diodes KD2999A - in my case they handle about 7A (U = +-45 V) without cooling. Switching frequency is 50 kHz (clock 100kHz). No voltage or frequency compensation. One more thing - i use 10 000uF capasitor on +12V rail right beside transformer - it helps very much not to drop voltage in secondary (hold about 7 volts at 200W load). Other things like in schematic. Now i don't remember any precise measurements, but with load of about 200W i have 5 volts drop on secondary. One more thing - capasitor of 2200uF i put near chip - it helps to hold voltage on the output (TL494) or mosfet input (gate voltage). I think that you use enough thickness of windings and there is no comments about it.If there any questions - ask
Good luck!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.