I think...
...that if you can keep the high-frequency noise under control then a switcher is a better proposition for a big amp than a conventional supply on the grounds of weight, size and especially voltage regulation. Efficiency is probably similar. Making a one-off transformer type supply from scratch is just a bit easier though. If the noise levels are similar then I don't buy the assertion that conventional supplies sound "better".
As far as reliability is concerned, I have looked at all kinds of things but in practice I have seen nothing better than a simple one fet forward converter that can put out 400 watts dc with an ETD49 size tranny. PCB weight is about 1.5 kg.
Having said all that though, I have never actually used a switcher in an amp. It is just too easy to hook up a tranny and diodes and caps and away you go. But if the junk box didn't have what I needed then I would think seriously about a switcher.
...that if you can keep the high-frequency noise under control then a switcher is a better proposition for a big amp than a conventional supply on the grounds of weight, size and especially voltage regulation. Efficiency is probably similar. Making a one-off transformer type supply from scratch is just a bit easier though. If the noise levels are similar then I don't buy the assertion that conventional supplies sound "better".
As far as reliability is concerned, I have looked at all kinds of things but in practice I have seen nothing better than a simple one fet forward converter that can put out 400 watts dc with an ETD49 size tranny. PCB weight is about 1.5 kg.
Having said all that though, I have never actually used a switcher in an amp. It is just too easy to hook up a tranny and diodes and caps and away you go. But if the junk box didn't have what I needed then I would think seriously about a switcher.
electricity isn't expensive enough
the power conversion of a switcher can approach 90% -- and at relatively small size -- the motivation comes in applications where a lot of heat equals a lot of money.
I am fascinated by the advances in battery management, power management, size and economy.
the power conversion of a switcher can approach 90% -- and at relatively small size -- the motivation comes in applications where a lot of heat equals a lot of money.
I am fascinated by the advances in battery management, power management, size and economy.
Could someone explain what a SMPS does and how it works as opposed to a standard power supply and why it is more suited to audio applications in cars.
I read here
http://www.smpstech.com/tutorial/t00con.htm
but it doesn't really make anything clear to me about what makes an SMPS different.
I read here
http://www.smpstech.com/tutorial/t00con.htm
but it doesn't really make anything clear to me about what makes an SMPS different.
In a nutshell, 240vac (or whatever is your local AC voltage) goes in and gets rectified to 339 vdc and smoothed by a big capacitor. Then that 339 vdc gets switched on and off by a mosfet at typically 100kHz or so and this chopped voltage is applied to the primary winding of a relatively small and light ferrite cored transformer that doesn't have many turns. (I have seen 700 watts going through a tranny the size of a matchbox, another with a one-turn secondary producing 60 volts!) The transformer secondary feeds a rectifier diode(s) and then a filter choke with a powdered iron core and not many turns of wire, and then to a filter capacitor. A feedback cct measures the dc output voltage and sends a signal back to the control cct that varies the on-to-off time percentage of the switching mosfet. This has the effect of varying the output voltage as necessary.
This is for an offline psu running from the AC mains. It can work just as well though with a low voltage dc going in, as from a car battery, and a higher or lower voltage coming out the other end. For a car amplifier smps you might have + / - 40 vdc coming out.
The good -> they don't waste a lot of power as heat like a normal regulated power supply that dumps the excess voltage as heat in a big power transistor on a big heatsink.
The bad -> a poorly engineered one can generate a lot of electrical noise and interference that can wipe out things like your TV reception when you want to watch something important like The Three Stooges.
The ugly -> they can be tricky, scary and downright dangerous to work on unless you know *exactly* what you are doing. There only appears to be black magic involved but it is only plain old physics and engineering but with less room for gross error.
This is for an offline psu running from the AC mains. It can work just as well though with a low voltage dc going in, as from a car battery, and a higher or lower voltage coming out the other end. For a car amplifier smps you might have + / - 40 vdc coming out.
The good -> they don't waste a lot of power as heat like a normal regulated power supply that dumps the excess voltage as heat in a big power transistor on a big heatsink.
The bad -> a poorly engineered one can generate a lot of electrical noise and interference that can wipe out things like your TV reception when you want to watch something important like The Three Stooges.
The ugly -> they can be tricky, scary and downright dangerous to work on unless you know *exactly* what you are doing. There only appears to be black magic involved but it is only plain old physics and engineering but with less room for gross error.
yes but
a lot of the mystery has dropped away as companies like National, On-Semi, Infineon, Linear, TI, Maxim etc provide you with the software for component selection and have the actual PCB layouts right on their websites.
Sources like the American Radio Relay League (www.arrl.org) in their annual handbooks have given step by step instructions for constructing the transformers, etc.
a lot of the mystery has dropped away as companies like National, On-Semi, Infineon, Linear, TI, Maxim etc provide you with the software for component selection and have the actual PCB layouts right on their websites.
Sources like the American Radio Relay League (www.arrl.org) in their annual handbooks have given step by step instructions for constructing the transformers, etc.
Okay I think I got a better understanding now. Basically the switching mode name comes fom the switching between AC and DC voltages? Looking at this http://sound.westhost.com/project89.htm it goes fom DC to AC then through a transformer to bump up the voltage then back to DC via the rectification bridge. I understand (well it appears that) DC voltages can't be run through transformers to get voltage transformation.
So some questions... what is the reason for such a high frequency? Is that so the transformers can be smaller than what they'd be if it was 50-60hz signal? Does the frequency have to be as high as what Rod Elliot uses? (35kHz, 54kHz) or can it be lower? Does lowering it make the circuit any more simple? Lower costs? As for simplicity, there is this circuit. http://www.i4at.org/lib2/inverter.htm which is really simple. I forget the name of this circuit, some multivibrator kinda name. That surely could be rectified and smoothed with some caps after the transformer. Would that work? If so, work well, or what would be wrong with it?
Sorry for the million and one questions
-Mike
So some questions... what is the reason for such a high frequency? Is that so the transformers can be smaller than what they'd be if it was 50-60hz signal? Does the frequency have to be as high as what Rod Elliot uses? (35kHz, 54kHz) or can it be lower? Does lowering it make the circuit any more simple? Lower costs? As for simplicity, there is this circuit. http://www.i4at.org/lib2/inverter.htm which is really simple. I forget the name of this circuit, some multivibrator kinda name. That surely could be rectified and smoothed with some caps after the transformer. Would that work? If so, work well, or what would be wrong with it?
Sorry for the million and one questions
-Mike
PS: jackinnj I looked up National's website, the LM3478MM chip seems interesting... but I am not too knowledgeable on what all the specs mean. The TI website was too hard for me to figure out but the specs all did seem rather low (2A maximum output)
the higher frequency means that you can use fewer turns on the transformer primary -- since impedance is a linear function of frequency -- further the storage capacitors can have smaller value -- but they need to be designed for switching supplies.
there is a very good tutorial on the National Semi website -- goes over the physics of what happens in a bot linear and switch mode supplies.
you try to get away with as much current handling capacity as possible with a single-chip design, but the peak currents are much higher than the average -- some chips will handle 3A or 5A peak.
while not my preferred switching chip, the LM3524 (SG3524 in Euroland) is a good one to start out on as National has a PCB diagram right on their website. this chip is cheap enough that you can practice with it.
there is a very good tutorial on the National Semi website -- goes over the physics of what happens in a bot linear and switch mode supplies.
you try to get away with as much current handling capacity as possible with a single-chip design, but the peak currents are much higher than the average -- some chips will handle 3A or 5A peak.
while not my preferred switching chip, the LM3524 (SG3524 in Euroland) is a good one to start out on as National has a PCB diagram right on their website. this chip is cheap enough that you can practice with it.
Basic switching power supply notes:
OnSemi/Motorola Switchmode Power Supply Reference Manual
OnSemi/Motorola - Linear and Switching Voltage Regulator Handbook
National - Introduction to Power Supplies
National - Linear and Switching Voltage Regulator Fundamentals (Part 1)
National - Linear and Switching Voltage Regulator Fundamentals (Part 2)
ePanorama.net - Switching Power Supply Links
National and Maxim both have lines of easy to use switching regulators for medium power applications - as far as car applications, they are particularly useful for powering computers and other electronics requiring stable voltages. I recently built a car PC power supply with the MAX787, MAX788, and LM2587 - very straightforward application. The few components needed can be recycled from old PC power supplies, including magnetics.
The switching power supply on Rod Elliot's site is fairly simple when you break down the individual sections:
High switching frequencies allow for smaller magnetics, reducing weight - this is the industry's driving force in switching power supply development, along with increasing efficiency. Switching frequencies can range up to 1MHz, requiring careful design layout to minimize noise radiation. However, lowering the switching frequency usually will not simplify the number of components in a given topology aside from switching noise considerations. For what it's worth, I'll be testing out the above design this weekend for power output at 100kHz....
-Nikhil
OnSemi/Motorola Switchmode Power Supply Reference Manual
OnSemi/Motorola - Linear and Switching Voltage Regulator Handbook
National - Introduction to Power Supplies
National - Linear and Switching Voltage Regulator Fundamentals (Part 1)
National - Linear and Switching Voltage Regulator Fundamentals (Part 2)
ePanorama.net - Switching Power Supply Links
National and Maxim both have lines of easy to use switching regulators for medium power applications - as far as car applications, they are particularly useful for powering computers and other electronics requiring stable voltages. I recently built a car PC power supply with the MAX787, MAX788, and LM2587 - very straightforward application. The few components needed can be recycled from old PC power supplies, including magnetics.
The switching power supply on Rod Elliot's site is fairly simple when you break down the individual sections:
- DC --> AC square wave (controller, MOSFETs)
- voltage step-up (ferrite transformer)
- AC -> DC (fast recovery diodes, low esr capacitors)
High switching frequencies allow for smaller magnetics, reducing weight - this is the industry's driving force in switching power supply development, along with increasing efficiency. Switching frequencies can range up to 1MHz, requiring careful design layout to minimize noise radiation. However, lowering the switching frequency usually will not simplify the number of components in a given topology aside from switching noise considerations. For what it's worth, I'll be testing out the above design this weekend for power output at 100kHz....
-Nikhil
Switching PSU
I recently build a SMPS. I used the UC3525 controller IC and mosfets from Fairchild. I'm able to get about 300W of power from the power supply at +- 25V. It's quite small and the efficiency is around 85%. And the output is isolated from the input.
At the beginning (about 3 years ago) I was a dump in switching PSU, but after many research and many questions, I learned how it's working. I think the hardest part in a smps, is building the transformer, choosing the right core size and the correct number of turns. If there's not enought turn, the core will saturate and the mosfets just go up in smoke. Another tricky things about smps is noise. A mosfet gate is very sensitive to noise, so it's easy to loss efficiency just because of a noisy board.
Now I'm pretty satisfied and I'm on the way the make 4 amps that will work with the power supply. I hope to have 4 x 50W rms in 4 ohms.
Ask for more info if you wish to.
I recently build a SMPS. I used the UC3525 controller IC and mosfets from Fairchild. I'm able to get about 300W of power from the power supply at +- 25V. It's quite small and the efficiency is around 85%. And the output is isolated from the input.
At the beginning (about 3 years ago) I was a dump in switching PSU, but after many research and many questions, I learned how it's working. I think the hardest part in a smps, is building the transformer, choosing the right core size and the correct number of turns. If there's not enought turn, the core will saturate and the mosfets just go up in smoke. Another tricky things about smps is noise. A mosfet gate is very sensitive to noise, so it's easy to loss efficiency just because of a noisy board.
Now I'm pretty satisfied and I'm on the way the make 4 amps that will work with the power supply. I hope to have 4 x 50W rms in 4 ohms.
Ask for more info if you wish to.
Re: Switching PSU
I put a spreadsheet with the math for determining the transformer specs on my website at:
http://www.tech-diy.com/smps_xfmr.xls
Unless you have a pile of LM3525 chips (UC3525 or SG3525) you might want to try the current mode controllers -- these measure current on each pulse and limit the pwm such that the core can't go into saturation. Curent mode controller chips also allow you to use BJT switching transistors.
RedFloyd said:
I put a spreadsheet with the math for determining the transformer specs on my website at:
http://www.tech-diy.com/smps_xfmr.xls
Unless you have a pile of LM3525 chips (UC3525 or SG3525) you might want to try the current mode controllers -- these measure current on each pulse and limit the pwm such that the core can't go into saturation. Curent mode controller chips also allow you to use BJT switching transistors.
Now that everything is working good I dont want to change that controller. But remember that the peak current in a smps for a car amp can have peak current over 75A, so it's quite hard to mesure those currents without disturbing the circuit.
For the transformer, is used app notes from Magnetics. They provide pretty much all the information the correctly design the transformer.
For the transformer, is used app notes from Magnetics. They provide pretty much all the information the correctly design the transformer.
I built the ESP smps just recently and I'm impressed with how well it works. There is some audible switching noise when it starts up but thats gone after a second (soft start is my guess).
The only thing he didn't go into much detail on was the transformer core.
People here help me with that and I got mine sorted and it now does +-35v rails off 12v.
I'm still to make a stacked transformer so i'm limited to about 400w before the core saturates but with dual p3a's its not going to need alot more than that at full power anyway.
Mines runs @ 50khz (not measured but its on my to do list). As far as his design goes would increasing the frequency up to 100khz let me get away with a smaller core?
My diodes are up to the task and all the fets aswell... It might save my time/money with adding an extra core on.
The only thing he didn't go into much detail on was the transformer core.
People here help me with that and I got mine sorted and it now does +-35v rails off 12v.
I'm still to make a stacked transformer so i'm limited to about 400w before the core saturates but with dual p3a's its not going to need alot more than that at full power anyway.
Mines runs @ 50khz (not measured but its on my to do list). As far as his design goes would increasing the frequency up to 100khz let me get away with a smaller core?
My diodes are up to the task and all the fets aswell... It might save my time/money with adding an extra core on.
fr0st said:
There is an economic and engineering tradeoff between switching frequency and efficiency. You save a little on the transformer but you give it up with the ESR of the filter/storage capacitors and such.
One of the important thing to remember about switching frequency is the rapid transition of current flows -- every millimeter adds nano-henries of inductance (well, maybe femto-henries)
In DIY we concern ourselves less with efficiency than accomplishing better results on our own. For this reason I keep suggesting that folks try out the Linear Tech LT1533 slew rate controlled switching chip. It is expensive -- about 20X the price of a SG3524 -- but the noise is comparble or less than linear regulators. (Linear also has a slew controlled gate driver.)
Hey guys,
All this talk of SMPS keeps questioning the electrical noise generated by the PS. What if you mounted the SMPS in a seperate case, and then filtered the output of the supply. And then on the amp you could filter it again. This would also allow a few different amps to use the same power supply. Seems like it would be a neat thing for a car audio project.
I myself want to try the ESP SMPS for a car audio system. I'm sure I could build a few amps and a power supply shared for all of them. I'm also sure I can beat commercial quality amps, at least the lower cost ones.
However, how well would the noise escape through an aluminum heatsink/case. If you use torroids the noise would go out the middle of them, but I think the transistors generate most of the noise in SMPS anyways don't they?
Pete
All this talk of SMPS keeps questioning the electrical noise generated by the PS. What if you mounted the SMPS in a seperate case, and then filtered the output of the supply. And then on the amp you could filter it again. This would also allow a few different amps to use the same power supply. Seems like it would be a neat thing for a car audio project.
I myself want to try the ESP SMPS for a car audio system. I'm sure I could build a few amps and a power supply shared for all of them. I'm also sure I can beat commercial quality amps, at least the lower cost ones.
However, how well would the noise escape through an aluminum heatsink/case. If you use torroids the noise would go out the middle of them, but I think the transistors generate most of the noise in SMPS anyways don't they?
Pete
if you look at the output of a your computer's SMPS you will see a repetitive pulse -- a pulse mind you -- not the random kind of noise you see on an LM317 regulator -- the pulse is of very short duration and is difficult to get out, even with inductors, ferrites etc.
one trick which seems to work -- partly -- is to use a post-regulator -- i.e. a LM317 or a pass transistor and error amplifier, but this adds expense both in dollars and in lost efficiency.
one trick which seems to work -- partly -- is to use a post-regulator -- i.e. a LM317 or a pass transistor and error amplifier, but this adds expense both in dollars and in lost efficiency.
May I reccommend a few good books? In amother DIY thread, I recommended George Chryssis' book "High Frequency Switching Power Supplies" because it was THE book that helped me to unravel the mysteries of switcher design and theory. This book is broken into chapters like the sections of a typical switcher. Chapter 4, which is devoted entirely to transformers, gives all the info you will need in choosing a transformer.
It explains all the design trade-offs (yes, there ARE going to be some), materials available and their magnetic properties, and a very cut-n-dried design process for both flyback and forward (both single- and double-ended) transformers. Using these design techniques, I have been able to reduce the design process down to 9 or 10 equations, with great success. This book is accompaniedf by several others, invcluding one be Marty Brown, a design engineer at Motorols (probably now ON semiconductor). Two good double-ended current-mode controllers worth looking at are the UC1846 (Unitrode) and the MC33025 (On semiconductor).
Please let me know how these work for you. Best of Luck!
'73 de N8XO
It explains all the design trade-offs (yes, there ARE going to be some), materials available and their magnetic properties, and a very cut-n-dried design process for both flyback and forward (both single- and double-ended) transformers. Using these design techniques, I have been able to reduce the design process down to 9 or 10 equations, with great success. This book is accompaniedf by several others, invcluding one be Marty Brown, a design engineer at Motorols (probably now ON semiconductor). Two good double-ended current-mode controllers worth looking at are the UC1846 (Unitrode) and the MC33025 (On semiconductor).
Please let me know how these work for you. Best of Luck!
'73 de N8XO
- Status
- Not open for further replies.