actually, i am using 12 turns on the primary (6 turns, centre tap, 6 turns) and 24 turns on the secondary(12 turns, centre tap, 12 turns). this produces about +/-45v from my measurements, however i havent been able to test the voltage drop on load, as the converter is drawing 5 amps at 9v without any load at all, and my power supply maxes out at 5A.
Hi. If you have high current drawn from psu without load, probably ttansformer core is not suitable for this job. Try to increase operating frequency first, maybe install potentiometet for adjusting, and change frequency while checking current. Also , maybe deadtime is too small for mosfets chosen . Deadtime is short pause, when both mosfets are off. Adjusted by voltage of dtc pin of 494, by simple resistor divider from gnd to vref. Maybe output diodes are too low voltage?
Also, if you are using 50percent duty cycle, no feedback, you can move 220 uh inductors after output capacitor, output voltage would increase and match transformer's primary to secondary winding ratio. With inductor right after diodes and no capacitor, inductor acts like current doubler - absorbs some current, and during pause returns it to load, same as in buck stepdown comverter. Also, so high 220 uh inductance is easy to saturate, must be physically big size and with air gap preferably. Yellow cores , used in atx , are crappy, they are heating by itselft, not the wire resistance heats them.
Also, if you are using 50percent duty cycle, no feedback, you can move 220 uh inductors after output capacitor, output voltage would increase and match transformer's primary to secondary winding ratio. With inductor right after diodes and no capacitor, inductor acts like current doubler - absorbs some current, and during pause returns it to load, same as in buck stepdown comverter. Also, so high 220 uh inductance is easy to saturate, must be physically big size and with air gap preferably. Yellow cores , used in atx , are crappy, they are heating by itselft, not the wire resistance heats them.
thank you for the great reply, i like the sound of adding the inductors, would it implemented look something like this:
also, adding dead time control is a great idea, im pretty sure that with the tl494 dead time pin, 3.3v is 0% duty cycle and 0v(ground) is 50% duty cycle, so i could create a voltage divider that varies from 0v-3.3v. The tl494 is in push pull mode, so does varying the dead time apply to both outputs?
being able to change the frequency sounds great, at the moment i have measured the tl494s output frequency which is at around 35khz.
To vary the frequency, would i vary the voltage going to tl494s pin 1 (1in+)? i could maybe use a 10k variable resistor between ground, pin1, and vref.
edit: even with the bridge rectifier completely removed from my circuit and the secondary outputs of my transformer connected to nothing, the current draw is still just as high. diodes i am using on the rectifier side is GBU808 diode matrix chip
also, adding dead time control is a great idea, im pretty sure that with the tl494 dead time pin, 3.3v is 0% duty cycle and 0v(ground) is 50% duty cycle, so i could create a voltage divider that varies from 0v-3.3v. The tl494 is in push pull mode, so does varying the dead time apply to both outputs?
being able to change the frequency sounds great, at the moment i have measured the tl494s output frequency which is at around 35khz.
To vary the frequency, would i vary the voltage going to tl494s pin 1 (1in+)? i could maybe use a 10k variable resistor between ground, pin1, and vref.
edit: even with the bridge rectifier completely removed from my circuit and the secondary outputs of my transformer connected to nothing, the current draw is still just as high. diodes i am using on the rectifier side is GBU808 diode matrix chip
I just saw - you using regular diode bridge, which is usable only at mains frequency, not designed for switchmode supplies and would make higher losses. You need to choose four separate diodes, fast ones. If you want +-60v output, and in the future plan to add voktage regulation, keep the inductors , but then increase secondary winding turns. Without inductors, diodes must have at least 200 volts reverse voltage . With inductors even more, maybe 400, volts.
For adjusting frequency capacitor at Ct pin must be adjustable, or resistor at Rt pin. Capacitor would be problematic to adjust, so replace fixed value resistor with potentiometer and some fixed resistor in series , to limit maximum frequency.
Pins 1-2 typically used for feedback, 15-16 for current limiting ,. Changing frequency should be used just for choosing proper frequency, at which all components run fine - transformer and mosfets, this would affect output voltage a little .
LLC topology is another thing, where adjustable frequency is used to regulate output voltage.
For adjusting frequency capacitor at Ct pin must be adjustable, or resistor at Rt pin. Capacitor would be problematic to adjust, so replace fixed value resistor with potentiometer and some fixed resistor in series , to limit maximum frequency.
Pins 1-2 typically used for feedback, 15-16 for current limiting ,. Changing frequency should be used just for choosing proper frequency, at which all components run fine - transformer and mosfets, this would affect output voltage a little .
LLC topology is another thing, where adjustable frequency is used to regulate output voltage.
Thanks, yes i have lots of 10a10 rectifier diodes that might be better suited for the output rectifier.
I played around with varying the voltage to dead time pin, and got some better results. At around 10% duty cycle, the rectifier output capacitors charged from 0v to 45v slowly, but the current draw was only around 200ma and the mosfets stayed cool. Increasing the duty cycle made the capacitors charge up much faster from 0-55v, drawing about 2A.
For current limiting as you mentioned with pin 16, i am going to use this circuit, the “50v” arrow goes to the rectifier side 50v output:
I played around with varying the voltage to dead time pin, and got some better results. At around 10% duty cycle, the rectifier output capacitors charged from 0v to 45v slowly, but the current draw was only around 200ma and the mosfets stayed cool. Increasing the duty cycle made the capacitors charge up much faster from 0-55v, drawing about 2A.
For current limiting as you mentioned with pin 16, i am going to use this circuit, the “50v” arrow goes to the rectifier side 50v output:
431 connection is incorrect in this schematic. Anode must be grounded, voltage divider connected to ref pin, cathode to optocoupler LED cathode, led anode through resistor to some small voltage,not larger than 36v. At best to connect divider directly to 494 pin1 , and another divider from 494 reference to pin2, omitting optocoupler. Also there must be compensation components. Google as example atx psu schematic with tl494 and study examples .
If with low duty cycle circuit is operating, probably primary inductance is too low.
If with low duty cycle circuit is operating, probably primary inductance is too low.
hello again, did some further research and i have managed to improve everything a lot by winding a better transformer, at idle its only drawing 200ma, and it holds its voltage on loads up to 250w. Turned out the wire thickness I was using for winding my transformers, the core type and winding method was wrong and I'm surprised that I even got a voltage output on the secondary. I'm still using a similar circuit with irf3205s at 50khz, but with the added inductors that were recommended.
Just asking if anyone knows of a good diode model to use on the rectifier side and if I should be using a full wave rectifier or a different type.
also, another issue I don't understand yet is that if there is no load on the transformer secondaries, then i put a load on the secondaries very suddenly, there is a huge voltage spike on the primary side that blows out the mosfets. If I take the same load and put it on the secondaries, and slowly increase the converters input voltage, it handles the load fine.
Just asking if anyone knows of a good diode model to use on the rectifier side and if I should be using a full wave rectifier or a different type.
also, another issue I don't understand yet is that if there is no load on the transformer secondaries, then i put a load on the secondaries very suddenly, there is a huge voltage spike on the primary side that blows out the mosfets. If I take the same load and put it on the secondaries, and slowly increase the converters input voltage, it handles the load fine.
Probably when you add load ,transformer core saturates and acts like short circuit,and MOSFETs can't handle that. Very short time , when load is connected, should not cause problem, because capacitors supply the current to load. But if load connected longer, then transformer has to handle load and charge capacitors at same time, which can be expected as very high load. Soft start would ensure, that duty cycle increases slower and is handled without core overload. Also some current limiting at primary side would be helpful. Some time ago when playing with sg3525 , i've implemented high side power limit , by using small value shunt on minus wire of mosfet, and comparator, also one npn , which discharged soft start capacitor of sg3525, thus preventing current increase. You may try something similar, you can set limit to level, at which nothing blows.
added in the voltage divider and ommited the tl431, with divider going strait to pin 1 and 2.5v on pin 2, so i think im correct in saying that if the voltage on pin 1 goes above 2.5v, it decreases output voltage? everything is working really nicely, mosfets stay cool and voltage is really well regulated, attached a neat little 200w amp and that also played nice until unfortunately the converters rectifier diodes exploded!
The diodes are 10a10 and started to make a squealing noise as they broke down.
Im now using irfz44n mosfets that are working far better than the irf3205s
Im not sure what to do with diodes
The diodes are 10a10 and started to make a squealing noise as they broke down.
Im now using irfz44n mosfets that are working far better than the irf3205s
Im not sure what to do with diodes
Also just curious, would it be possible to use 4x common anode diode chips to make the rectifier section, rather than using a common anode chip and a common cathode chip (i say chip as in to-220 diodes)
There seems to be a severe deficiency in any common anode and common cathode diode pairs in the uk.
For example, 4x mur1560 wired as full wave rectifier.
There seems to be a severe deficiency in any common anode and common cathode diode pairs in the uk.
For example, 4x mur1560 wired as full wave rectifier.
Last edited:
Added in 4x mur1560 diodes for the rectifier section, and the converter works nice and stable, rectifier is no longer exploding or getting hot. hooking the +60, 0v and -60v up to my amp, i can play music quite loud. Now, i just need to work on improving how much power the converter can handle, as at high current draw(high volume on the amp) the legit irfz44n mosfets i am using break. currently i have 2 of the mosfets switching one side of the transformer primary, and 2 mosfets switching the other side. So far, i havent been using a gate driver, instead i have been using a totem pole with a npn and pnp transistor(one totem pole for each tl494 channel).
to try and combat the mosfets blowing up, im thinking of adding more so that there is 4 irfz44ns driving each side of the dual primary. considering that my totem poles going to each mosfet side use one a1015(pnp) and one c1815(npn), would adding more mosfets be risky without a gate driver?
each mosfet gate has a 10 ohm resistor between it and the totem pole.
any other ideas for more power are really appreciated, unfortunately i dont have any inductors for rectifier filtering.
Last edited:
If you have fixed duty cycle, you don't need filtering inductors that much .If you have unused atx psu , they have vertical inductors 1uH ,with ferrite and good wire ,at 3,3 and 5V output , you can try using them .If you would try to regulate voltage by pwm duty cycle ,then another connection of inductors are needed and inductors itself must be different .
I see some problems with driver wiring . As i understand ,you have placed totem pole driver and 494 ic on separate small board .What resistance you are using for 494 output load(totem pole input) ? Actually , not more than 1k to ground should be there ,if frequency is like 70-100Khz . Try using scope on mosfets gate pins ,check for waveform,if its clean .If mosfets blowing at high power, there may be few possibilities. 1.Due to pnp/npn b-e voltage loss , mosfets at lowest can get +0,7V and at highest VCC - vbe - 494 loss (~1,2V approx). So not fully using mosfet possibilities and maybe with ringing they not full closing. 2. Bad waveform , Gate-source voltage squarewave have not enough sharp edges, and this causes too long transition time ,from on-off ,but less radiated emi noise .But mosfets will heat more . 3. Transformer core comes to saturation at some load level,and then inductance decreases ,current increases ,and so If better mosfets fitted ,they will blow too ,but at little higher power . 4. Decoupling problems ,long trances ,also i see no bypass capacitors close to mosfets and transformer . Traces must be as short as possible and wide ,additionally wire can be soldered on tracks , especially ground . 5. Totem pole driver is not close to mosfets , and due to inductances and long traces mosfets getting noisy waveform.I remember in car amplifier converters mosfets are placed at opposite sides of case/heatsink , and at each side have its own totem pole driver . Also ,IRFZ44 are 3x in parralel ,in some amplifiers were place for 4th estimated ,and upgrade was possible. Maybe 4th mosfet was installed in more powerful model . For increasing number of mosfets , totem pole transistors probably needs to be upgraded too .I would recommend to try 1-2Ampere rated . As example SS8050 / SS8550 pair , also there are some from BC series ,but now can't say exact numbers . The ground point ,where gate driver's ground is connected ,is critical .At best , each mosfet group should have its own driver . Something like one driver for 2x mosfets max .Gate driver should improve things ,but then wiring must be done according to datasheet .I would not recommend IR2110 there ,as very critical to ground, and ringing on ground causes both mosfets turn-on at same time and blowing everything .Also IR2110 have level shifting ,which is not needed there ,and misbehaves due to ringing .Maybe discrete design would work more reliably .
I see some problems with driver wiring . As i understand ,you have placed totem pole driver and 494 ic on separate small board .What resistance you are using for 494 output load(totem pole input) ? Actually , not more than 1k to ground should be there ,if frequency is like 70-100Khz . Try using scope on mosfets gate pins ,check for waveform,if its clean .If mosfets blowing at high power, there may be few possibilities. 1.Due to pnp/npn b-e voltage loss , mosfets at lowest can get +0,7V and at highest VCC - vbe - 494 loss (~1,2V approx). So not fully using mosfet possibilities and maybe with ringing they not full closing. 2. Bad waveform , Gate-source voltage squarewave have not enough sharp edges, and this causes too long transition time ,from on-off ,but less radiated emi noise .But mosfets will heat more . 3. Transformer core comes to saturation at some load level,and then inductance decreases ,current increases ,and so If better mosfets fitted ,they will blow too ,but at little higher power . 4. Decoupling problems ,long trances ,also i see no bypass capacitors close to mosfets and transformer . Traces must be as short as possible and wide ,additionally wire can be soldered on tracks , especially ground . 5. Totem pole driver is not close to mosfets , and due to inductances and long traces mosfets getting noisy waveform.I remember in car amplifier converters mosfets are placed at opposite sides of case/heatsink , and at each side have its own totem pole driver . Also ,IRFZ44 are 3x in parralel ,in some amplifiers were place for 4th estimated ,and upgrade was possible. Maybe 4th mosfet was installed in more powerful model . For increasing number of mosfets , totem pole transistors probably needs to be upgraded too .I would recommend to try 1-2Ampere rated . As example SS8050 / SS8550 pair , also there are some from BC series ,but now can't say exact numbers . The ground point ,where gate driver's ground is connected ,is critical .At best , each mosfet group should have its own driver . Something like one driver for 2x mosfets max .Gate driver should improve things ,but then wiring must be done according to datasheet .I would not recommend IR2110 there ,as very critical to ground, and ringing on ground causes both mosfets turn-on at same time and blowing everything .Also IR2110 have level shifting ,which is not needed there ,and misbehaves due to ringing .Maybe discrete design would work more reliably .
This is the schematic for the little tl494 board that you recognised. From the tl494 outputs to the totem pole inputs, i use 4.7k pull down resistors to ground(r4, r5). Is that value too large, as you mentioned with your point about pull downs having to be under 1k.
Unfortunately i cant have a totem pole for each Mosfet, my pcbs have to be fairly small as my milling machine does not have much of a large milling area.
For increasing the current handling of my totem poles, could i use multiple a1015/c1815 transistors in series rather than buying higher power transistors?
tl494 frequency set to 98khz, and the waveform on the mosfet gate looks pretty clean
From what i understand, a set voltage is applied to the - pin, e.g 2.5v, and when the voltage on the + pin goes above the - pin voltage (2.5v), the tl494 shuts off its outputs.
I did a bit of experimenting with feedback, adding it in to my circuit. Using tl494 vref, i set the voltage on - to 2.5v and had + going through voltage divider to the output. Voltage divider was set so that if the output went above 60v, the voltage on pin + would be above 2.5v.
What i found, was that the tl494 would try ti regulate the output, by shutting off when output was above 60v, and turning on when below 60v.
However, when the tl494 outputs would turn back on, they would not immediately put out a nice square output. This caused messy output caused extreme current draw that would occur each time the tl494 kicker back in to try and increase output voltage. With a load, the combination of the load and the messy output would cause all the mosfets to blow.
Overall i have found that without feedback, even at the highest load i can go to before i have the other mentioned issue of Mosfets blowing up, the output voltage still stays above 60v, so i have avoided the feedback.
I have found the tl494 comparators are much more usefull for over current protection.
Perhaps a better method of feedback would be to have the dead time varied based on the output, rather than closing or opening the tl494 outputs in a binary way.
I did a bit of experimenting with feedback, adding it in to my circuit. Using tl494 vref, i set the voltage on - to 2.5v and had + going through voltage divider to the output. Voltage divider was set so that if the output went above 60v, the voltage on pin + would be above 2.5v.
What i found, was that the tl494 would try ti regulate the output, by shutting off when output was above 60v, and turning on when below 60v.
However, when the tl494 outputs would turn back on, they would not immediately put out a nice square output. This caused messy output caused extreme current draw that would occur each time the tl494 kicker back in to try and increase output voltage. With a load, the combination of the load and the messy output would cause all the mosfets to blow.
Overall i have found that without feedback, even at the highest load i can go to before i have the other mentioned issue of Mosfets blowing up, the output voltage still stays above 60v, so i have avoided the feedback.
I have found the tl494 comparators are much more usefull for over current protection.
Perhaps a better method of feedback would be to have the dead time varied based on the output, rather than closing or opening the tl494 outputs in a binary way.
494 have two amplifiers, for feedback, for liear control of pwm duty cycle, and not comparators. On-pff operation is caused by slow output response, as example too much uf output capacitors, and then all the feedback system oscillating at low frequency, with buzzing sound. To prevent that, a compensation is needed. Additional inductors must be inswrted , after diodes output and before capacitrs. The return stored energy to losd diring pwm pause. The ssme asin dc/dc converter. But those inductors need to be large. So probably not for this design.
Yes, you can use multiple totem pole drivers
If you using now 4,7k pulldown resistor, try adding another stage of totem poles in setirs, waveform will improve, but another problem my appear, that now mosfets will get 1,2v at minimum gate voltage , twice vbe drop. Also, you can try to keep existing totem pole in board and add each one for two mosfets, if more mosfets is used in parralel.
Yes, you can use multiple totem pole drivers
If you using now 4,7k pulldown resistor, try adding another stage of totem poles in setirs, waveform will improve, but another problem my appear, that now mosfets will get 1,2v at minimum gate voltage , twice vbe drop. Also, you can try to keep existing totem pole in board and add each one for two mosfets, if more mosfets is used in parralel.