Re: Re: Re: 2kW Switcher
I received the design manuals that Magnetics inc. publishes. They have all the info on how to select a core. Most ferrites will operate at 50Khz, it's just a matter of winding the transformer correctly.
Yes, I have taken physics, and I have tried filter chokes on the input to automotive amplifiers. I have done so with a scope on the input to the amplifier. If there is not enough input capacitance in the amplifier, with the high frequency currents the added inductance in the choke will see the transient currents from the switching transformer, therefore, generate transient voltages.
The noise on the amplifier is probably not getting through the power leads to the amplifier, it's most likely a ground loop issue between battery, head unit, and amplifier. Star grounding as suggested by one of the other members is the best solution.
-Dan
BenY said:
I received the design manuals that Magnetics inc. publishes. They have all the info on how to select a core. Most ferrites will operate at 50Khz, it's just a matter of winding the transformer correctly.
jackinnj said:
Yes, I have taken physics, and I have tried filter chokes on the input to automotive amplifiers. I have done so with a scope on the input to the amplifier. If there is not enough input capacitance in the amplifier, with the high frequency currents the added inductance in the choke will see the transient currents from the switching transformer, therefore, generate transient voltages.
The noise on the amplifier is probably not getting through the power leads to the amplifier, it's most likely a ground loop issue between battery, head unit, and amplifier. Star grounding as suggested by one of the other members is the best solution.
-Dan
Ground Loops & Toroids
Ground Loops:
I have the worst time with alternator whine, and I can tell from experience that a good grounding shceme will take care of almost any anternator-induced whine. Yes indeedy, star-grounding is the best way to go, but it is a good idea to keep input grounds (on the primary side of the PSU) separate from the output grounds (the secondary, or audio side of the PSU) to further remove the audio circuitry from the noise source. then run a 10-ohm resistor from the RCA input jacks of the amp to vehicle (primary-side) ground to tie it all together. Bypass with a 0.01uF cap is there's still a little noise.
Toroids:
Amidon Associates of CA suggests that you use their ferrite material #77mix with a permeability of u=2000 @ 20kHz. They have a graph to extrapolate what permeability # to use at higher frequencies (like 45-50 kHz), and it is all pretty straightforward.
Best of luck!
Ground Loops:
I have the worst time with alternator whine, and I can tell from experience that a good grounding shceme will take care of almost any anternator-induced whine. Yes indeedy, star-grounding is the best way to go, but it is a good idea to keep input grounds (on the primary side of the PSU) separate from the output grounds (the secondary, or audio side of the PSU) to further remove the audio circuitry from the noise source. then run a 10-ohm resistor from the RCA input jacks of the amp to vehicle (primary-side) ground to tie it all together. Bypass with a 0.01uF cap is there's still a little noise.
Toroids:
Amidon Associates of CA suggests that you use their ferrite material #77mix with a permeability of u=2000 @ 20kHz. They have a graph to extrapolate what permeability # to use at higher frequencies (like 45-50 kHz), and it is all pretty straightforward.
Best of luck!
Re: Re: Re: Re: Re: 2kW Switcher
Also meant the power inputs. 12 volts in to the amp. The choke used was several paralled #12 coppers on a core...
Now, a low pass filter (choke in series with a good cap bank.. ..then you may be onto something! But, even then, I still think grounding is a more important issue.
I think we're on the same page
I hope so, anyway
-Dan
jackinnj said:
Also meant the power inputs. 12 volts in to the amp. The choke used was several paralled #12 coppers on a core...
Now, a low pass filter (choke in series with a good cap bank.. ..then you may be onto something! But, even then, I still think grounding is a more important issue.
I think we're on the same page
I hope so, anyway -Dan
Grounding Inputs & Outputs
I think we're all singing from the same sheet of music
, and the coil/cap thingy is a great idea. Nuttin' like brute force where necessary!
What I mean wrt input/outputs is in the supply: 12V-side is the inpit, and the +/-V is the output side. The RCA inputs to the amp should be tied to the power supply's output ground, then to the psu's input (12V) ground via a 10 ohm resistor and a 0.01uf cap in parallel.
N8XO
I think we're all singing from the same sheet of music
, and the coil/cap thingy is a great idea. Nuttin' like brute force where necessary! What I mean wrt input/outputs is in the supply: 12V-side is the inpit, and the +/-V is the output side. The RCA inputs to the amp should be tied to the power supply's output ground, then to the psu's input (12V) ground via a 10 ohm resistor and a 0.01uf cap in parallel.
N8XO
Push-Pull vs. Full Bridge topology issue
ai" :
"Problem with H-bridge is that you have two pairs of switching devices in series with each other, on each side of the transformer. Therefore conduction losses double, so it's not as efficient. H-Bridge is great at high input voltages, where currents are small, and conduction losses are minimal."
Be careful when making direct comparisons. Although the push-pull has only one FET conducting at a time, vs. two FETs for the full bridge, the FETs in the push pull circuit must withstand **twice** the supply voltage when off, as opposed to the full bridge topology, whose FETs only need withstand the supply voltage. Therefore, the push-pull topology requires FETs with twice the drain-source breakdown voltage rating vs. full bridge. Higher voltage FETs always have higher Rdson for the same die size. The result is increased conduction losses. Also, switching losses are higher for push-pull due to the double voltage present at drain to source. It is well known to all skilled in the art of SMPS design, that full bridge can handle higher power requirements better than push-pull. Although I feel that push-pull is a very good choice in this application, it is incorrect to assert that full bridge incurs greater losses. To say that full bridge is "not as efficient" is dead wrong. Why then, does every power supply OEM and designer insist on full bridge for the highest power requirements? As far as transformer isolated converters go, the pecking order has always been from lowest to highest power as follows. Flyback, forward, push-pull, half bridge, full bridge. I just thought I should set the record straight. Best regards.
ai" :
"Problem with H-bridge is that you have two pairs of switching devices in series with each other, on each side of the transformer. Therefore conduction losses double, so it's not as efficient. H-Bridge is great at high input voltages, where currents are small, and conduction losses are minimal."
Be careful when making direct comparisons. Although the push-pull has only one FET conducting at a time, vs. two FETs for the full bridge, the FETs in the push pull circuit must withstand **twice** the supply voltage when off, as opposed to the full bridge topology, whose FETs only need withstand the supply voltage. Therefore, the push-pull topology requires FETs with twice the drain-source breakdown voltage rating vs. full bridge. Higher voltage FETs always have higher Rdson for the same die size. The result is increased conduction losses. Also, switching losses are higher for push-pull due to the double voltage present at drain to source. It is well known to all skilled in the art of SMPS design, that full bridge can handle higher power requirements better than push-pull. Although I feel that push-pull is a very good choice in this application, it is incorrect to assert that full bridge incurs greater losses. To say that full bridge is "not as efficient" is dead wrong. Why then, does every power supply OEM and designer insist on full bridge for the highest power requirements? As far as transformer isolated converters go, the pecking order has always been from lowest to highest power as follows. Flyback, forward, push-pull, half bridge, full bridge. I just thought I should set the record straight. Best regards.
Re: Push-Pull vs. Full Bridge topology issue
It was not a general comparison, It was a comparison at a specific input voltage. 12V (ie Low) input voltage. The pecking order is not same when dealing with 12 volts or lower input voltages.
I've yet to see any step up 12 volt supply over a few watts that doesn't use Push-Pull. If you can give me an example of one, I'd love to see it!
Yes, the fets need to withstand twice the input voltage. For 12volts in, you need to withstand 24 volts. Most suitable mosfets are in the 50 to 60 volt range, anyway. 50 volts should even be more than the probable transients (even then avalanche energy ratings apply).
As far as the thread goes. If someone here is looking for information on building a 12 power supply to power audio amplifiers in a car, Push-Pull is the best topology. It has the least part count, is the easiest to build, and is the only common type for the application.
Wouldn't you agree?
-Dan
Claude Abraham said:
It was not a general comparison, It was a comparison at a specific input voltage. 12V (ie Low) input voltage. The pecking order is not same when dealing with 12 volts or lower input voltages.
I've yet to see any step up 12 volt supply over a few watts that doesn't use Push-Pull. If you can give me an example of one, I'd love to see it!
Yes, the fets need to withstand twice the input voltage. For 12volts in, you need to withstand 24 volts. Most suitable mosfets are in the 50 to 60 volt range, anyway. 50 volts should even be more than the probable transients (even then avalanche energy ratings apply).
As far as the thread goes. If someone here is looking for information on building a 12 power supply to power audio amplifiers in a car, Push-Pull is the best topology. It has the least part count, is the easiest to build, and is the only common type for the application.
Wouldn't you agree?
-Dan
Re: 2 kW
Hey Steve,
Serously beefed up traces... ...3/8 inch by 1/16 inch copper soldered on anything carrying currents. Very custom design... ...brute force overkill. However, from what I'm learning, the transformer could have been done a little differently.
As for fets, IRF only! Digikey lists more than 7000 of the N version. I've found the 1010's can be hard to use in push-pull. They switch very fast compared to other commonly used in car audio.
So, you built a PS? Have any pics of it? Schematics?
I don't have any schematics for mine. Basic push pull topology, doesn't really need a schematic...
73,
Dan
P.S. Actually, was thinking of gearing it towards a kit, but that'll be some years down the road. Many improvements to be made yet!
N-Channel said:Since you're not putting it into mass-production, don't worry too much about down-sizing. Probably put it on a PC board with seriously beefed-up traces where appropriate, but don't downsize the MOSFETs. Those 100A FETs are just the ticket! Too bad ON doesn't make them anymore!
'73,
Steve
Hey Steve,
Serously beefed up traces... ...3/8 inch by 1/16 inch copper soldered on anything carrying currents. Very custom design... ...brute force overkill. However, from what I'm learning, the transformer could have been done a little differently.
As for fets, IRF only! Digikey lists more than 7000 of the N version. I've found the 1010's can be hard to use in push-pull. They switch very fast compared to other commonly used in car audio.
So, you built a PS? Have any pics of it? Schematics?
I don't have any schematics for mine. Basic push pull topology, doesn't really need a schematic...
73,
Dan
P.S. Actually, was thinking of gearing it towards a kit, but that'll be some years down the road. Many improvements to be made yet!
FETs
I thought you'd used the old Motorola FETS before Mot spun off into ON semi. I went and looked up the IRF1010N. Pretty impressive. Try the MTB75N06 from ON. Drain current is 75A and 10 milliohms Rds(on). I agree with you on the IR quality. On is mainly concenteating on 250V and lower (bye bye 400- & 500V MOSFETS), but IS paying particular attention to the developemnt of their P-channel FETs (more current, less Rds(on)).
Yes, we finally have the camera. Will take pics of switcher tomorrow and post. That one used all Motorola semi's (PWM chip, MOSFETS and output Schottkys- 100V devices). Next one will use the UC1846 (MILSpec- It's nice working for DoD)) PWM IC, and some BIG BEEFY RFG70N06s from Harris. Output diodes are MBR20200CTs- 20A, 200V Schottkys. I like the lower forward voltage of the schottkys, and the few extra percent points in efficiency.
I thought you'd used the old Motorola FETS before Mot spun off into ON semi. I went and looked up the IRF1010N. Pretty impressive. Try the MTB75N06 from ON. Drain current is 75A and 10 milliohms Rds(on). I agree with you on the IR quality. On is mainly concenteating on 250V and lower (bye bye 400- & 500V MOSFETS), but IS paying particular attention to the developemnt of their P-channel FETs (more current, less Rds(on)).
Yes, we finally have the camera. Will take pics of switcher tomorrow and post. That one used all Motorola semi's (PWM chip, MOSFETS and output Schottkys- 100V devices). Next one will use the UC1846 (MILSpec- It's nice working for DoD)) PWM IC, and some BIG BEEFY RFG70N06s from Harris. Output diodes are MBR20200CTs- 20A, 200V Schottkys. I like the lower forward voltage of the schottkys, and the few extra percent points in efficiency.
Wouldn't a second chip be a waste of space?
The ESP design only uses positive rail feedback and current draw would be relatively similar for both rails so the voltage drop would be round about the same.
If they were in sync it would be basically the same as having all the fets driven from the same chip but with feedback for only one rail?
push-pull vs. full bridge
Posted by "dkemppai":
"It was not a general comparison, It was a comparison at a specific input voltage. 12V (ie Low) input voltage. The pecking order is not same when dealing with 12 volts or lower input voltages.
I've yet to see any step up 12 volt supply over a few watts that doesn't use Push-Pull. If you can give me an example of one, I'd love to see it!
Yes, the fets need to withstand twice the input voltage. For 12volts in, you need to withstand 24 volts. Most suitable mosfets are in the 50 to 60 volt range, anyway. 50 volts should even be more than the probable transients (even then avalanche energy ratings apply).
As far as the thread goes. If someone here is looking for information on building a 12 power supply to power audio amplifiers in a car, Push-Pull is the best topology. It has the least part count, is the easiest to build, and is the only common type for the application.
Wouldn't you agree?
-Dan"
Yes, I'll agree that a push-pull is more desirable due to its lower parts count and simplicity. Full bridge should only be used at very high (> 500 W) power levels. Since you now stipulate low power levels, I'd agree with you. For this application, if I could choose between those two only, I'd choose the push-pull hands down. Push-pull is generally recommended around 200 to 500 watts. From roughly 75 to around 200 W, forward is the preferred topology. Below 75 watts, the flyback is simplest of all, more so than push-pull. Also, flybacks can have multiple secondaries, and offer better cross regulation of multiple output voltages than do other topologies. Seeing that the power is a mere 1.5 watts, I still don't see what is gained by using a transformer isolated topology.
For 115/230 V ac offline supplies, isolation is required per UL, VDE, etc. Or, if the input & output voltages greatly differ, non-isolated converters must run at near 0% or near 100% duty cycle, which increases noise and ripple current. A transformer allows the turns ratio to be properly scaled, optimizing the duty factor. In this case, no isolation is needed, and the in/out voltages are nearly equal, which results in around 50% duty cycle for a buck-boost topolgy, which is optimum. Oh well. A push-pull would no doubt work great. But, a custom transformer would have to be wound. An inductor stock from a supplier sure sounds attractive, wouldn't you think?
As far as an example of a converter over a few watts not using push-pull, I've designed many over the last 12 years which are in consumer OEM equipment you've used numerous times in your daily life. This is a 1.5 W op-amp supply. If it was a 200 or 300 watt power amp supply, I wouldn't hesitate to use a push-pull. Do we agree?
Best regards.
Posted by "dkemppai":
"It was not a general comparison, It was a comparison at a specific input voltage. 12V (ie Low) input voltage. The pecking order is not same when dealing with 12 volts or lower input voltages.
I've yet to see any step up 12 volt supply over a few watts that doesn't use Push-Pull. If you can give me an example of one, I'd love to see it!
Yes, the fets need to withstand twice the input voltage. For 12volts in, you need to withstand 24 volts. Most suitable mosfets are in the 50 to 60 volt range, anyway. 50 volts should even be more than the probable transients (even then avalanche energy ratings apply).
As far as the thread goes. If someone here is looking for information on building a 12 power supply to power audio amplifiers in a car, Push-Pull is the best topology. It has the least part count, is the easiest to build, and is the only common type for the application.
Wouldn't you agree?
-Dan"
Yes, I'll agree that a push-pull is more desirable due to its lower parts count and simplicity. Full bridge should only be used at very high (> 500 W) power levels. Since you now stipulate low power levels, I'd agree with you. For this application, if I could choose between those two only, I'd choose the push-pull hands down. Push-pull is generally recommended around 200 to 500 watts. From roughly 75 to around 200 W, forward is the preferred topology. Below 75 watts, the flyback is simplest of all, more so than push-pull. Also, flybacks can have multiple secondaries, and offer better cross regulation of multiple output voltages than do other topologies. Seeing that the power is a mere 1.5 watts, I still don't see what is gained by using a transformer isolated topology.
For 115/230 V ac offline supplies, isolation is required per UL, VDE, etc. Or, if the input & output voltages greatly differ, non-isolated converters must run at near 0% or near 100% duty cycle, which increases noise and ripple current. A transformer allows the turns ratio to be properly scaled, optimizing the duty factor. In this case, no isolation is needed, and the in/out voltages are nearly equal, which results in around 50% duty cycle for a buck-boost topolgy, which is optimum. Oh well. A push-pull would no doubt work great. But, a custom transformer would have to be wound. An inductor stock from a supplier sure sounds attractive, wouldn't you think?
As far as an example of a converter over a few watts not using push-pull, I've designed many over the last 12 years which are in consumer OEM equipment you've used numerous times in your daily life. This is a 1.5 W op-amp supply. If it was a 200 or 300 watt power amp supply, I wouldn't hesitate to use a push-pull. Do we agree?
Best regards.
Dualing Chips
Now that I think of it, just use 1 PWM chip, 1 set of MOSFETs and 1 transformer, wtth 1 set of 4 diodes to make a traditional +/- bi-polar output configuration. Most +/- bipolar SMPSs for commercial Amps use this configuration. And, yes, you should sense from both the (+) and (-) rails, since their ground will be isolated from the primary ground.
Another thing to lower the noise is to put a coupled-inductor after the rectifying diodes, but before the output caps to form an L-C filter. Separate inductors for each rail, followed by a second set of caps should be a good output filter.
Now that I think of it, just use 1 PWM chip, 1 set of MOSFETs and 1 transformer, wtth 1 set of 4 diodes to make a traditional +/- bi-polar output configuration. Most +/- bipolar SMPSs for commercial Amps use this configuration. And, yes, you should sense from both the (+) and (-) rails, since their ground will be isolated from the primary ground.
Another thing to lower the noise is to put a coupled-inductor after the rectifying diodes, but before the output caps to form an L-C filter. Separate inductors for each rail, followed by a second set of caps should be a good output filter.
Re: Dualing Chips
I posted my problem on another thread here re: getting more out of the little cores I have available.
The cores I've got are only tiny in comparison (37mm OD 17?mm high) and definately won't do the power outputI need. If I stack 3 cores that should give me about 1.2kw. If I do a transformer per rail that should get the power rating I want. The problem is getting them both on there.
Instead of having both primaries on a single transformer can I have one per transformer?
Can anyone recommend a solution (considering the little cores I have)?
N-Channel said:
I posted my problem on another thread here re: getting more out of the little cores I have available.
The cores I've got are only tiny in comparison (37mm OD 17?mm high) and definately won't do the power outputI need. If I stack 3 cores that should give me about 1.2kw. If I do a transformer per rail that should get the power rating I want. The problem is getting them both on there.
Instead of having both primaries on a single transformer can I have one per transformer?
Can anyone recommend a solution (considering the little cores I have)?
Re: push-pull vs. full bridge
Please go back an re-read my post. I'm specifying low input VOLTAGE, HIGH POWER. Push-pull is the only way to go!
Show me a single SMPS with a 12volt INPUT used for car audio that uses a topology other than push-pull. Please, I'd actually like to see one. Even just one.
At 100 amps input, and a RDS on of .01Ohms, that's 100Watts conduction losses. You would double that with any full bridge topology! That kills efficientcy by around 10% (that's roughly figuring 12volts in, 1000 watts output, at 90% max dutycycle for the power switches with an Rds on of .01 Ohms. )
-Dan
Claude Abraham said:
Please go back an re-read my post. I'm specifying low input VOLTAGE, HIGH POWER. Push-pull is the only way to go!
Show me a single SMPS with a 12volt INPUT used for car audio that uses a topology other than push-pull. Please, I'd actually like to see one. Even just one.
At 100 amps input, and a RDS on of .01Ohms, that's 100Watts conduction losses. You would double that with any full bridge topology! That kills efficientcy by around 10% (that's roughly figuring 12volts in, 1000 watts output, at 90% max dutycycle for the power switches with an Rds on of .01 Ohms. )
-Dan
Re: Re: Dualing Chips
I'll look up some rough info on your core... ...it'll give you a feel for the power you can get.
If you're having trouble finding a core, I several here that will do 1.2kW or better. Email me if can't find anything else, and we can talk about a trade or something similar.
-Dan
fr0st said:
I'll look up some rough info on your core... ...it'll give you a feel for the power you can get.
If you're having trouble finding a core, I several here that will do 1.2kW or better. Email me if can't find anything else, and we can talk about a trade or something similar.
-Dan
This is the cores I have
On my last SMPS at 50khz (not measured but it'll probly be close) I had 6 primary turn on a single core which worked well if thats any help.
I just kept guessing till I got it right
The only info I was able to get on it was the info on the site - MATERIAL L15 SIZE A35 X B21 X C13
Basicly any extra power I can get can be soaked up by my amp so anythings good (preferable ~2.5kw typ.).
I've got plenty of chunky fets (yay for samples) so upping the switching frequency a bit wouldn't be a large drop in effientcy. I'm fairly famililar with high frequency so that won't be much of a problem.
Thank for your help
On my last SMPS at 50khz (not measured but it'll probly be close) I had 6 primary turn on a single core which worked well if thats any help.
I just kept guessing till I got it right
The only info I was able to get on it was the info on the site - MATERIAL L15 SIZE A35 X B21 X C13
Basicly any extra power I can get can be soaked up by my amp so anythings good (preferable ~2.5kw typ.).
I've got plenty of chunky fets (yay for samples
) so upping the switching frequency a bit wouldn't be a large drop in effientcy. I'm fairly famililar with high frequency so that won't be much of a problem.Thank for your help
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.