Flyback and Boost converters may latch-up at start-up when power supply impedance is too high, this happens because the error amplifier is asking for a too high peak current level that cannot be reached or for a too high duty cycle that doesn't allow the output voltage to rise.
As you have found out, the problem is reduced to a much lesser extent by adding a soft-start feature, that may be as simple as clamping the output of the error amplifier to a slowly rising voltage ramp. Further protection is achieved if a undervoltage lockout feature is added in such a way that when the supply voltage gets too low the soft-start voltage ramp is restarted from zero, thus resetting the PSU.
Concerning your design, I think that you should put the transformer, the switches, the current sense resistors and the input and output filter capacitors in a thightly laid-out PCB with short tracks, because with such long wires and such wide current-loop areas, your point-to-point prototype is likely to radiate EMI like hell. Don't you have an oscilloscope to check waveforms and ringing issues?
As you have found out, the problem is reduced to a much lesser extent by adding a soft-start feature, that may be as simple as clamping the output of the error amplifier to a slowly rising voltage ramp. Further protection is achieved if a undervoltage lockout feature is added in such a way that when the supply voltage gets too low the soft-start voltage ramp is restarted from zero, thus resetting the PSU.
Concerning your design, I think that you should put the transformer, the switches, the current sense resistors and the input and output filter capacitors in a thightly laid-out PCB with short tracks, because with such long wires and such wide current-loop areas, your point-to-point prototype is likely to radiate EMI like hell. Don't you have an oscilloscope to check waveforms and ringing issues?
No i don't have a scope
I think emi is kept to a minimum as i no longer disturb tv and the audio out from the amp is quiet with no signal, juse faint hiss but thats normal for a smp like this and the puters soundcard out is not the best either.
When i box the thing i gonna use thick wire on the input part, and shortest possible wires to minimize excess inductance.
Now i have made a circuit that drives the controller from the output so if the smps gets overloaded it just shuts down and requires a manual restart.
I think emi is kept to a minimum as i no longer disturb tv and the audio out from the amp is quiet with no signal, juse faint hiss but thats normal for a smp like this and the puters soundcard out is not the best either.When i box the thing i gonna use thick wire on the input part, and shortest possible wires to minimize excess inductance.
Now i have made a circuit that drives the controller from the output so if the smps gets overloaded it just shuts down and requires a manual restart.
Audio signals are quite hard to distort in an audible way with EMI, whose most usual symptom is just an increased noise floor. Also, TV and FM radio are somewhat hard to distort since they employ high frequencies well above 50Mhz. However, AM radio is easily distorted because it operates well below 50Mhz, in the frequency range where most SMPS radiate, so I would recommend that kind of testing.
I am going to mount the thing in a metal box as well as shorten all wires as mutch as possible.
I suppose that the performanve could only be better from where i stand now if i shorten wires and add thickness to the ones in the primary part.
Which box material would be best, a steel case from a computer psu or a aluminum box ?
I suppose that the performanve could only be better from where i stand now if i shorten wires and add thickness to the ones in the primary part.
Which box material would be best, a steel case from a computer psu or a aluminum box ?
Ok. But as i meant this to be used in car, shielding would be contained by the car chassis.
I think that a real car amp emits more EMI if run like my diy version tho. Most car amps i have seen inside doesent even have a input filter, only some small coils on the output on the amp rails so it wont hiss from the smps switching.
I think that a real car amp emits more EMI if run like my diy version tho. Most car amps i have seen inside doesent even have a input filter, only some small coils on the output on the amp rails so it wont hiss from the smps switching.
Remember that car-audio amplifiers employ push-pull power supplies that generate much less EMI than a flyback to start with. They require little filtering and little attention to stray magnetic fields due to the toroid core. Manufacturers use that approach beacuase they know its simplicity and advantages quite well, even at the expense of bigger size.
Car Amp SPMS
Hi Tekko,
Al always, Eva is right. And the most common PWM chip you will fing in almost every car amp employing a DC-DC Converter is the TL494. Same is also true for almost any AT or ATX PSU box.
While the '494 is, by far, the most widely used chip, there are much better ones, like the SG3525 (voltage-mode operation) or the UC1846/2846/3846 (current-mode operation), because these two chips have true totem-pole outputs, enabling them to drive almost any MOSFET or power BJT (with speed-up turn-off caps for the BJTs) directly, without the need for a separate driver circuit.
Another thing to take into consideration is that you are looking for +/-35v @ 2.5A = 175W. Assuming a 65% efficiency rating, make this something like 270W. OK, so at low line of 11V, 270W will result in the primary MOSFET drain current of just under 25A. Remember for a flyback, the PEAK drain current for the primary switching MOSFET is something like 3.2X the max rated drain current. EVA, correct me on this if I'm wrong.
So for a 25A drain current, you will be peaking out at just shy of 80A at full power. There are few MOSFETs that can handle that kind of drain current and still keep it together, even with excellent heatsinking.
Using the push-pull topology, like EVA suggests, with voltage-mode, or perhaps even current-mode control GREATLY reduces your PEAK drain currents, cuts your EMI by at least half, and just works out nicely. You could use the TL494, but if you can get an SG3525, or a UC1846, or even an MC33025, you would have some great beginnings for a really good supply. Try Rod Elliot's pages or search some of the Power Supply threads here for some great examples.
Hope this helps!
Regards,
Steve
Hi Tekko,
Al always, Eva is right. And the most common PWM chip you will fing in almost every car amp employing a DC-DC Converter is the TL494. Same is also true for almost any AT or ATX PSU box.
While the '494 is, by far, the most widely used chip, there are much better ones, like the SG3525 (voltage-mode operation) or the UC1846/2846/3846 (current-mode operation), because these two chips have true totem-pole outputs, enabling them to drive almost any MOSFET or power BJT (with speed-up turn-off caps for the BJTs) directly, without the need for a separate driver circuit.
Another thing to take into consideration is that you are looking for +/-35v @ 2.5A = 175W. Assuming a 65% efficiency rating, make this something like 270W. OK, so at low line of 11V, 270W will result in the primary MOSFET drain current of just under 25A. Remember for a flyback, the PEAK drain current for the primary switching MOSFET is something like 3.2X the max rated drain current. EVA, correct me on this if I'm wrong.
So for a 25A drain current, you will be peaking out at just shy of 80A at full power. There are few MOSFETs that can handle that kind of drain current and still keep it together, even with excellent heatsinking.
Using the push-pull topology, like EVA suggests, with voltage-mode, or perhaps even current-mode control GREATLY reduces your PEAK drain currents, cuts your EMI by at least half, and just works out nicely. You could use the TL494, but if you can get an SG3525, or a UC1846, or even an MC33025, you would have some great beginnings for a really good supply. Try Rod Elliot's pages or search some of the Power Supply threads here for some great examples.
Hope this helps!
Regards,
Steve
The primary peak current at full load in a flyback converter depends on a lot of things:
- Turn ratios.
- Primary inductance
- Continuous or discontinuous mode.
These parameters may be tailored to reduce the primary-side peak current, that is inherently high, at the expense of increasing the secondary side peak current, that is inherently low.
It's quite simple to make a working flyback, but it's quite hard to make it optimum.
Tekko: Have you done the corresponding math in order to choose optimum primary inductance, turn ratios and operating frequency? I have written a small program that helps me to calculate flybacks...
- Turn ratios.
- Primary inductance
- Continuous or discontinuous mode.
These parameters may be tailored to reduce the primary-side peak current, that is inherently high, at the expense of increasing the secondary side peak current, that is inherently low.
It's quite simple to make a working flyback, but it's quite hard to make it optimum.
Tekko: Have you done the corresponding math in order to choose optimum primary inductance, turn ratios and operating frequency? I have written a small program that helps me to calculate flybacks...
I've got a nice continuous-mode approach :
- Primary inductance: 20uH (>30A sat.)
- Turn ratios: 2+2:1 (sec+secri)
- Operating frequency: 50Khz
Predicted steady state full load conditions (175W):
- Primary current sweeps from 21A to 28A during 11.86uS on-time.
- Secondary currents sweep from 7A to 5.25A during 8.13uS off-time (each).
I've also tried critical-conduction-mode :
- Primary inductance: 5uH (>50A sat.)
- Turn ratios 2+2:1
- Minimum frequency at full load: 28Khz (depends only on inductance).
Predicted steady state full load conditions (175W):
- Primary current sweeps from 0A to 49A during 20.9uS on-time.
- Secondary currents sweep from 12.3A to 0A during 14.3uS off-time (each).
- Primary inductance: 20uH (>30A sat.)
- Turn ratios: 2+2:1 (sec+sec
ri)- Operating frequency: 50Khz
Predicted steady state full load conditions (175W):
- Primary current sweeps from 21A to 28A during 11.86uS on-time.
- Secondary currents sweep from 7A to 5.25A during 8.13uS off-time (each).
I've also tried critical-conduction-mode :
- Primary inductance: 5uH (>50A sat.)
- Turn ratios 2+2:1
- Minimum frequency at full load: 28Khz (depends only on inductance).
Predicted steady state full load conditions (175W):
- Primary current sweeps from 0A to 49A during 20.9uS on-time.
- Secondary currents sweep from 12.3A to 0A during 14.3uS off-time (each).
According to my multimeter i have had over 9A out from the thing with only 8A in
Overtunity or busted multimeter ?? i had it between the rails and set on the high current 10A mode.
The fets i have used can do around 38A continous together and over 100A pulsed.
The transformer has the turns ratio of 3-4 turns primary and 9 turns secondary.
Eva, show me the proggy.
Overtunity or busted multimeter ?? i had it between the rails and set on the high current 10A mode.The fets i have used can do around 38A continous together and over 100A pulsed.
The transformer has the turns ratio of 3-4 turns primary and 9 turns secondary.
Eva, show me the proggy.
Where is the schematics? 
They disappeared or what ? 
I'm VERY INTERESTED in this topic, I NEED see the schematics and build this thing
uahuahuahua..
Look !
http://www.diyaudio.com/forums/showthread.php?s=&threadid=69613
I dind't find it before open the topic
, sorry !
Well, I'm learnig about theese supplies, and thank you all !
Sooo, can you guys send me the schematics? It appeared to have been removed by the admin or happend some error in the forum...
Here's my email, fuinhox@gmail.com I apreciate very much if you send me the prototypes
Thanks !!!
They disappeared or what ? I'm VERY INTERESTED in this topic, I NEED see the schematics and build this thing
uahuahuahua.. Look !
http://www.diyaudio.com/forums/showthread.php?s=&threadid=69613
I dind't find it before open the topic
, sorry !Well, I'm learnig about theese supplies, and thank you all !
Sooo, can you guys send me the schematics? It appeared to have been removed by the admin or happend some error in the forum...
Here's my email, fuinhox@gmail.com I apreciate very much if you send me the prototypes
Thanks !!!
Have they failed? why?
By the way, this is my flyback calculation helper program. Yes, I know that it's GWBASIC so don't blame me... Lines 10 and 20 contain input parameters :
VP=primary voltage
TP=primary turns
VS=secondary voltage
TS=secondary turns
PIN=power level
F=operating frequency (zero for critical conduction mode)
LT=core Al value in Henries per turn^2
10 VP=12: TP=12: VS=35: TS=24: PIN=175
20 F= 50000!: LT=1.4E-07: REM F=0 for critical conduction mode
30 LP=LT*TP^2: LS=LT*TS^2
40 VTR=VS/TS: VRP=TP*VTR: DC=VRP/(VP+VRP)
45 IF F=0 THEN F=.5*LP*VP^2*DC^2/(LP^2*PIN)RINT "f=";F
50 TON=DC/F: IRP=VP*TON/LP
60 EP=PIN/F: IF (.5*LP*IRP^2)>EP THEN TON=SQR(EP*LP^2/(.5*LP*VP^2))
70 TOFF=TON*(1-DC)/DC: IRP=VP*TON/LP:IRS=VS*TOFF/LS
80 I0P=(PIN/(.5*LP*F)-IRP^2)/(2*IRP)
90 I0S=(PIN/(.5*LS*F)-IRS^2)/(2*IRS)
95 IRMSP=SQR((LP/VP)*(((IRP+I0P)^3-I0P^3)/3)*F)
96 IRMSS=SQR((LS/VS)*(((IRS+I0S)^3-I0S^3)/3)*F)
100 PRINT "p:";INT(TON*1E+09)/1000!;"us",INT(I0P*1000!)/1000!;"A to",INT((I0P+IRP)*1000!)/1000!;"A",INT(IRMSP*1000!)/1000!;"A_rms"
110 PRINT "s:";INT(TOFF*1E+09)/1000!;"us",INT((I0S+IRS)*1000!)/1000!;"A to",INT(I0S*1000!)/1000!;"A",INT(IRMSS*1000!)/1000!;"A_rms"
120 PRINT INT(.5*LP*IRP^2*F*1000!)/1000!;"W discont.",INT(TON*F*1E+05)/1000!;"% duty"
130 PRINT INT(LP*1E+09)/1000!;"uH L pri"
And this is the output that it should produce with these parameters:
p: 11.864 us 21.052 A to 28.114 A 18.999 A_rms
s: 8.135 us 14.057 A to 10.526 A 7.866 A_rms
25.136 W discont. 59.322 % duty
20.16 uH L pri
The 'W discont.' rating is the power level at which the discontinuous to continuous mode transition happens.
It only supports one secondary winding, it's quite hard to calculate output RMS currents for multi-output flybacks since not all diodes conduct during all the off-time and complex load-sharing phenomena may be observed. However, two symmetrical outputs with balanced loading may be modelled as a single output carrying double the current, or providing double the voltage. Also, for calculation purposes, low power outputs may be neglected, or the highest power output may be assumed to deliver all the output power...
Enjoy
By the way, this is my flyback calculation helper program. Yes, I know that it's GWBASIC so don't blame me... Lines 10 and 20 contain input parameters :
VP=primary voltage
TP=primary turns
VS=secondary voltage
TS=secondary turns
PIN=power level
F=operating frequency (zero for critical conduction mode)
LT=core Al value in Henries per turn^2
10 VP=12: TP=12: VS=35: TS=24: PIN=175
20 F= 50000!: LT=1.4E-07: REM F=0 for critical conduction mode
30 LP=LT*TP^2: LS=LT*TS^2
40 VTR=VS/TS: VRP=TP*VTR: DC=VRP/(VP+VRP)
45 IF F=0 THEN F=.5*LP*VP^2*DC^2/(LP^2*PIN)
RINT "f=";F50 TON=DC/F: IRP=VP*TON/LP
60 EP=PIN/F: IF (.5*LP*IRP^2)>EP THEN TON=SQR(EP*LP^2/(.5*LP*VP^2))
70 TOFF=TON*(1-DC)/DC: IRP=VP*TON/LP:IRS=VS*TOFF/LS
80 I0P=(PIN/(.5*LP*F)-IRP^2)/(2*IRP)
90 I0S=(PIN/(.5*LS*F)-IRS^2)/(2*IRS)
95 IRMSP=SQR((LP/VP)*(((IRP+I0P)^3-I0P^3)/3)*F)
96 IRMSS=SQR((LS/VS)*(((IRS+I0S)^3-I0S^3)/3)*F)
100 PRINT "p:";INT(TON*1E+09)/1000!;"us",INT(I0P*1000!)/1000!;"A to",INT((I0P+IRP)*1000!)/1000!;"A",INT(IRMSP*1000!)/1000!;"A_rms"
110 PRINT "s:";INT(TOFF*1E+09)/1000!;"us",INT((I0S+IRS)*1000!)/1000!;"A to",INT(I0S*1000!)/1000!;"A",INT(IRMSS*1000!)/1000!;"A_rms"
120 PRINT INT(.5*LP*IRP^2*F*1000!)/1000!;"W discont.",INT(TON*F*1E+05)/1000!;"% duty"
130 PRINT INT(LP*1E+09)/1000!;"uH L pri"
And this is the output that it should produce with these parameters:
p: 11.864 us 21.052 A to 28.114 A 18.999 A_rms
s: 8.135 us 14.057 A to 10.526 A 7.866 A_rms
25.136 W discont. 59.322 % duty
20.16 uH L pri
The 'W discont.' rating is the power level at which the discontinuous to continuous mode transition happens.
It only supports one secondary winding, it's quite hard to calculate output RMS currents for multi-output flybacks since not all diodes conduct during all the off-time and complex load-sharing phenomena may be observed. However, two symmetrical outputs with balanced loading may be modelled as a single output carrying double the current, or providing double the voltage. Also, for calculation purposes, low power outputs may be neglected, or the highest power output may be assumed to deliver all the output power...
Enjoy
SMPS ideas again
Once again im planning a audio smps for 12v in, this time a PP one with SG2524 for controller.
Output criteria is split supply around 35v unloaded and 25-30v @ 2A per output rail under max load.
I plan on using two gate drive IC´s for driving 2-4 IRFP460 per side till i get hold of better fets.
This all i plan to use a hand drawn boar for where i solder the controller and gate drive ic´s SMD style as i dont have a drill.
The thing will prolly be something similar to Rod Elliot´s smps.
Once again im planning a audio smps for 12v in, this time a PP one with SG2524 for controller.
Output criteria is split supply around 35v unloaded and 25-30v @ 2A per output rail under max load.
I plan on using two gate drive IC´s for driving 2-4 IRFP460 per side till i get hold of better fets.
This all i plan to use a hand drawn boar for where i solder the controller and gate drive ic´s SMD style as i dont have a drill.
The thing will prolly be something similar to Rod Elliot´s smps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.