currently, i have 4x470uf capacitors on the +60v rail, and 4x470uf on the negative rail. as you say, perhaps this is too much?
seeing as it is important that i finally add these inductors on the rectifier stage, i have lots of 22uh 25a toroid inductors, maybe these could be effective?
if so, i will try again with feedback, but with an inductor between the capacitors and diode on each rail side
when i had that feedback issue, i did notice really loud buzzing from the transformer when the current would peak an waveform was messy.
seeing as it is important that i finally add these inductors on the rectifier stage, i have lots of 22uh 25a toroid inductors, maybe these could be effective?
if so, i will try again with feedback, but with an inductor between the capacitors and diode on each rail side
when i had that feedback issue, i did notice really loud buzzing from the transformer when the current would peak an waveform was messy.
No ,i think such capacity should be ok for such load as yours ,especially if there's no feedback .Had problem with feedback with adjustable flyback converter ,for 11Ampere max output ,to reduce ripple ,used 2x3300 uf .That made feedback to oscillate a bit ,had to do more tricks to stabilise ,but problem with constant current mode remains . Current limiting is working ,but transformer gives additional 1khz sound ,This psu i have at work ,so don't care much about that .But if i need to charge Li-on cell with bms integrated ,i have current and voltage kick ,when connecting battery , and that turns on battery protection. Then i have to increase voltage slowly with load connected ,to prevent battery not charging .
If you really would like to implement output voltage regulation ,i will try to explain how that works and whats needs to be changed .
The big inductor at output is not for ripple reduction ,but for voltage reduction ,energy absorbtion and returning it to load . Try to google " forward converter current doubler" . To have +-60V output after inductor , you need to have more than twice voltage from transformer secondary , lets say 2x150V . As example ,atx psu have such inductor ,and transformer outputs are 2x29V for +12v rail . Part of energy is absorbed by inductor, and returned to output when there's a deadtime period ,when no primary mosfets conducting .Output current doubles .This is the same stepdown buck dc/dc converter ,just without mosfet switch, pulses already coming from transformer .Inductance is critical ,and load current dependent .Have recently modded one psu , original toroid inductor heated at half load current ,a few watts of loss ,ant that was not wire hot ,the core itself . Good result were reached only when tested EI core from old flyback psu , massive , and with 1mm air gap (a few layers of insulation tape helped ), and with 2,5mm2 wire wounded .Tried also small good toroid ,with high current rating ,similar inductance ,no go .Also one benefit ,what current doubler does - reduce diode loss . When you have let's say 10 Ampere output as you have now ,no inductor, all 10ampere goes through diodes ,one by one .With transformer voltage higher twice and inductor present after diodes ,current is 5A , because inductor current rises slow .Lower power loss on diodes .
Output capacitors and inductor will cause additional delay in output response ,and then feedback oscillates . For 494 ic there's resistor in series with capacitor ,between pins 3 and 2 as i remember ,to slow down feedback , but depending on load , inductance ,output capacitors , needs testing of values . Also , if additional filter is added ,feedback must be connected before that filter . Or feedback resistor split into two resistors in series , and midpoint connected through 1-10uf to first output capacitor , end resistor to output after all filters . That trick allows feedback to sense voltage after all filters ,compensate voltage drop on inductors DC resistance ,but slowly .Better slowly ,than oscillate .Some of these things can be seen in simulation ,but harder to notice . Feedback oscillation shows as not constant pulse length , and that variation is slower than pulse length itself ,this is what we hear as buzzing .
If you really would like to implement output voltage regulation ,i will try to explain how that works and whats needs to be changed .
The big inductor at output is not for ripple reduction ,but for voltage reduction ,energy absorbtion and returning it to load . Try to google " forward converter current doubler" . To have +-60V output after inductor , you need to have more than twice voltage from transformer secondary , lets say 2x150V . As example ,atx psu have such inductor ,and transformer outputs are 2x29V for +12v rail . Part of energy is absorbed by inductor, and returned to output when there's a deadtime period ,when no primary mosfets conducting .Output current doubles .This is the same stepdown buck dc/dc converter ,just without mosfet switch, pulses already coming from transformer .Inductance is critical ,and load current dependent .Have recently modded one psu , original toroid inductor heated at half load current ,a few watts of loss ,ant that was not wire hot ,the core itself . Good result were reached only when tested EI core from old flyback psu , massive , and with 1mm air gap (a few layers of insulation tape helped ), and with 2,5mm2 wire wounded .Tried also small good toroid ,with high current rating ,similar inductance ,no go .Also one benefit ,what current doubler does - reduce diode loss . When you have let's say 10 Ampere output as you have now ,no inductor, all 10ampere goes through diodes ,one by one .With transformer voltage higher twice and inductor present after diodes ,current is 5A , because inductor current rises slow .Lower power loss on diodes .
Output capacitors and inductor will cause additional delay in output response ,and then feedback oscillates . For 494 ic there's resistor in series with capacitor ,between pins 3 and 2 as i remember ,to slow down feedback , but depending on load , inductance ,output capacitors , needs testing of values . Also , if additional filter is added ,feedback must be connected before that filter . Or feedback resistor split into two resistors in series , and midpoint connected through 1-10uf to first output capacitor , end resistor to output after all filters . That trick allows feedback to sense voltage after all filters ,compensate voltage drop on inductors DC resistance ,but slowly .Better slowly ,than oscillate .Some of these things can be seen in simulation ,but harder to notice . Feedback oscillation shows as not constant pulse length , and that variation is slower than pulse length itself ,this is what we hear as buzzing .
Added in what inductors i could find, im not lucky enough to have any spare atx power supplies however i did find these two filter inductor toroids from an amplifer board - you can see them hanging off the sides of the board.
Added in further irfz44n mosfets, now with 4 mosfets for each side of transformer.
Ambient current draw is slightly higher with the extra mosfets. The upgrades with the added mosfets and inductors has definately improved the maximum output volume off the amp, before i have the issue where a single mosfet blows. Considering that only one mosfet goes when too much current is drawn makes me think that their might be a shoot-through issue? perhaps increasing dead time might help? open to ideas.
The board has optionable feedback, so far i havent tried with feedback enabled yet as i think some improvements to the tl494 pwm module are necessary, such as as you mentioned slowing down feedback, modifying dead time.
Just to clarify, should feedback from the tl494 connect between the diodes output and rectifier filter input?
Diode dc output - feedback connection - filter inductor - output caps?
Totem poles seem to be handling 4 mosfets pretty well, each with 10 ohm gate resistance.
Looking good . Question - when running at high power are mosfets hot ? If so , then problem may be too small toroid transformer core ,not able to output the amount of power you need . If you have another identical toroid core ,you can glue them together ,or just tightly place ,and wind same turns, and check what you have . Probably by trying to get maximum power ,you getting core too close to saturation ,and that makes current through mosfets very high ,then Rds ON resistance is very important ,if speed parameters are enough good . If a single mosfet blows , it can mean it has lowest RDs on and taking most of current ,another mosfet is just helping a little . Shoot-through is possible too ,you can try to add diodes , like UF4007 or at least 1N4148 ,in parralel to each 10ohms gate resistor ,anode to mosfets G ,cathode to driver ,to improve turn-off time .But with mosfet failure totem pole will be destroyed too . Without this additional diode ,it may survive . Also another reason of mosfets blowing can be too long rise and fall times ,and due that mosfets VGS opening voltage is very important , mosfets must be matched ,all whose are in parralel ,must have similar voltage . Remember when i measured mosfets IRF540N ,for amplifier ,it was almost impossible to find similars for pairing . One had 2,8V turn-on voltage, another had 4,1V and so on, IRF540 without N was good in this plane and i was able to find matching triplets . For parralel operation to have advantage , all devices must be open very very fast ,until transformer primary current not rised enough high .Otherwise who opens first ,will blow ,not withstanding power .You need to get very very sharp waveform at mosfet gate ,about 100ns ,from 1V to 10V .At oscilloscope screen fronts are almost invisible with good waveform .
You have added output inductor ,it will reduce the output voltage ,but will increase output current .It needs to have big core too , able to absorb energy from transformers secondary winding ,and not saturate .When saturation occurs , primary current increases ,and output voltage too .Again ,overload for mosfets .
You have added output inductor ,it will reduce the output voltage ,but will increase output current .It needs to have big core too , able to absorb energy from transformers secondary winding ,and not saturate .When saturation occurs , primary current increases ,and output voltage too .Again ,overload for mosfets .
yes, the tranformer im using likely cant handle the current draw, it was taken from this cheap plate amplifier with fake ratings. it has worked fairly well but it is probably about time to upgrade to a toroid transformer that can handle more current, could you recomend a core type and wire gauge? i will buy them off amazon uk/ebay, then work on winding from there.
earlier in this thread i wound my own transformer, and i now know the reason that did not work was not only the winding method, but also because i was using a filter inductor toroid core rather than a power inductor toroid core, leading to the high current draws. i was probably also using incorrect coil wire type.
earlier in this thread i wound my own transformer, and i now know the reason that did not work was not only the winding method, but also because i was using a filter inductor toroid core rather than a power inductor toroid core, leading to the high current draws. i was probably also using incorrect coil wire type.
I would recommend to try use ready core first, with turn's, from dead or unused car amplfier, whose have 4 channels . It will already have primary winding, you just need to rewind secondary, calculate turns ratio by pulse voltage at output/ voltage at input. With inductors more turns needed. Most amplfiers have +-27V suplly voltage after rectifier, so without rewind no go. I have little experience with toroid cores, so can't recommend any . There are many manufacturers, many models, datasheets, and output power is rarely listed. Also power increase with frequency, but till some frequency, after that core heats up ( losses). There are transformer calculation apps available , allows choose core and play with parameters, frequency , output power, wire gauges etc, calculates if wire will fit.
Finally completed it, eventually bit the bullet and wound my own transformer.
Dc to dc converter Mosfets stay cool and dont really need the heatsinks, however the transformer needs a bit of air cooling as it does get quite hot at high load.
Measured the maximum output (of irs2092s amp module) at 350w into 4 ohms, ran this for 10 minuets and everything except the dc-dc
converter transformer stayed cool enough to touch 😀.
Currently the dc-dc converter is putting out +/-55v and In theory, 500w rms amp output could be achieved by once again using a larger dc-dc converter transformer core, with greater primary:secondary winding ratio to give +/-85v.
Files, if anyone feels interested:
Attachments
Nice work and progress ! About the transformer , maybe i would slightly repeat, but you can try to add another core of same size on top and wind turns through both cores ,so making one bigger core . Then decrease all turns count . Also , if you can get powerful atx psu (500 + Watts), you can try that core too .Those are not toroid ,but ER or EI , and depending on topology it had ,may have air gap , which may help to fight core saturation.
Also one thing i see from photo - you wound turns with big diameter wire , which by itself is not good for switching frequencies ,but it's ok for DC filters. Try to make litz wire ,by twisting lets say 100 wires by 0,1mm as example ,few meters each ,using electric drill or similar tool. Such 0,1mm wire can be found on 12v car relays coil ,mains frequency transformers primary winding and so .The smaller wire diameter , the higher frequency it allows to work with small loss . Other thing is heating of core itself .EI type ferrite core with air gap is less prone to that ,and by adding insulation tape layers ,you can increase saturation current ,by making I part a little distant from E part . Same apply to ER cores ,just they are different shape .Remember i've tested unknown massive EI ferrite core from old unknown power supply ,and output exceeded 500 Watts . Used halfbridge topology while testing and unregulated output,just diodes and capacitors .
ETD-59 is also massive core ,available to purchase (i think).
Also one thing i see from photo - you wound turns with big diameter wire , which by itself is not good for switching frequencies ,but it's ok for DC filters. Try to make litz wire ,by twisting lets say 100 wires by 0,1mm as example ,few meters each ,using electric drill or similar tool. Such 0,1mm wire can be found on 12v car relays coil ,mains frequency transformers primary winding and so .The smaller wire diameter , the higher frequency it allows to work with small loss . Other thing is heating of core itself .EI type ferrite core with air gap is less prone to that ,and by adding insulation tape layers ,you can increase saturation current ,by making I part a little distant from E part . Same apply to ER cores ,just they are different shape .Remember i've tested unknown massive EI ferrite core from old unknown power supply ,and output exceeded 500 Watts . Used halfbridge topology while testing and unregulated output,just diodes and capacitors .
ETD-59 is also massive core ,available to purchase (i think).
- Home
- Design & Build
- Electronic Design
- 12v to 60v, 0v and -60v circuits