it's not that simple, but it's also not that complex
to calculate the transformer parameters take a look at this spreadsheet link on my website (for push-pull supplies), it's cobbled from Linear Technology's equations LT1683:
http://www.tech-diy.com/smps_xfmr.xls
the transformer core you use will depend upon, among other things, the peak and average current drawn, switching frequency , how much temperature swing you can accomodate, etc. then you have to figure out how to get the windings through the hole of the donut.
if you pick the wrong core, the transformer will go into saturation, the impedance will drop to zero and you will blow the switching transistors.
there are much better switching chips than the SG3524 and SG3525 for audio applications, but these are cheap and pretty much ubiquitous now.
to calculate the transformer parameters take a look at this spreadsheet link on my website (for push-pull supplies), it's cobbled from Linear Technology's equations LT1683:
http://www.tech-diy.com/smps_xfmr.xls
the transformer core you use will depend upon, among other things, the peak and average current drawn, switching frequency , how much temperature swing you can accomodate, etc. then you have to figure out how to get the windings through the hole of the donut.
if you pick the wrong core, the transformer will go into saturation, the impedance will drop to zero and you will blow the switching transistors.
there are much better switching chips than the SG3524 and SG3525 for audio applications, but these are cheap and pretty much ubiquitous now.
Has anyone actually got a decent PCB layout for a car SMPS? I made one but its pretty poor and doesnt work well at all. It was based on a SG3525, but i'm willing to try a TL494 if anyone has a layout for that?
Thanks
Ben
Thanks
Ben
DC-DC Converter
Man, I am ticked off. Last night I wrote this long-winded diatribe on how things in the SMPS world should be, and when I went back a couple of screens to see who I was making my points to, the text I typed (all 12 paragraphs of it) was gone. UGH!
Anyway, here is a synopsis as best as I can remember it:
fr0st: I’ll do your comments first and try to answer your questions at the same time. You don’t want to use a P-channel with the LM2677, because to drive a P-channel, you will use “negative” logic (pull the gate low), where an active “low”output at the PWM chip turns the MOSFET on. The LM2677, like all monolithic Simple Switchers, uses “positive” logic, where active “low” output means the transistor is off. The TL494 is indeed a good chip for this. However, I need for you to answer a few questions for my help to be really effective, so here goes: 1) What is (are) your input and out put voltage(s)?, 2) is output higher or lower than output?, and 3) does your amp need a single voltage, or split (+/-) voltages, and high/low voltages, too? (like +/-35V(amp) and +/-15V(crossover)?😕
shyfx: I wholeheartedly agree with you in that it would greatly help to have a separate power supply forum. Even the best-designed amps from this website are no good without a good supply to back them up. In answer to your question, I agree with jackinnj: there ARE better chips than the 3525 or the 3524 (basically a TL494 with a current-sense amp in place of the 2nd error amp). Motorola, before they spun off their Semiconductor Products Sector (SPS) into ON Semiconductor, came out with the “next generation” SG3525: The MC33025. Major differences include an improved pin-out scheme (error amp’s output pin is right next to the input pins), high-speed oscillator up to 2MHz (1MHZ switching speed and 50nS propagation delay), and the ability to operate either voltage- OR current-mode, reducing the risk of converter damage to MOSFET imbalance. Go to ON’s webpages and download the ‘33025’s datasheet and application notes. Very helpful. Also, fore the N-Channel MOSFETs in the push-pull DC-DC converter, use ON’s MTB75N06s- 75A, 60V devices, in parallel. These have an Rds(on) of 0.01 Ohms.
RobM: I, too am a fan id primary-side control. To get a good solid feedback from the output, yo9u will need to use an optocoupler, like the MOC8102 from ON semi. Use the optocoupler to maintain DC-isolation across the transformer galvanic boundary. Ok, this is getting long-winded again, and I will pick up from here later, but these pointers should get you started in the right direction. I will be getting a digital camera in a few days, so I will post a few pics of the low- and high-powered DC-DC converters I’ve completed.
’73,
Steve
Man, I am ticked off. Last night I wrote this long-winded diatribe on how things in the SMPS world should be, and when I went back a couple of screens to see who I was making my points to, the text I typed (all 12 paragraphs of it) was gone. UGH!
Anyway, here is a synopsis as best as I can remember it:
fr0st: I’ll do your comments first and try to answer your questions at the same time. You don’t want to use a P-channel with the LM2677, because to drive a P-channel, you will use “negative” logic (pull the gate low), where an active “low”output at the PWM chip turns the MOSFET on. The LM2677, like all monolithic Simple Switchers, uses “positive” logic, where active “low” output means the transistor is off. The TL494 is indeed a good chip for this. However, I need for you to answer a few questions for my help to be really effective, so here goes: 1) What is (are) your input and out put voltage(s)?, 2) is output higher or lower than output?, and 3) does your amp need a single voltage, or split (+/-) voltages, and high/low voltages, too? (like +/-35V(amp) and +/-15V(crossover)?😕
shyfx: I wholeheartedly agree with you in that it would greatly help to have a separate power supply forum. Even the best-designed amps from this website are no good without a good supply to back them up. In answer to your question, I agree with jackinnj: there ARE better chips than the 3525 or the 3524 (basically a TL494 with a current-sense amp in place of the 2nd error amp). Motorola, before they spun off their Semiconductor Products Sector (SPS) into ON Semiconductor, came out with the “next generation” SG3525: The MC33025. Major differences include an improved pin-out scheme (error amp’s output pin is right next to the input pins), high-speed oscillator up to 2MHz (1MHZ switching speed and 50nS propagation delay), and the ability to operate either voltage- OR current-mode, reducing the risk of converter damage to MOSFET imbalance. Go to ON’s webpages and download the ‘33025’s datasheet and application notes. Very helpful. Also, fore the N-Channel MOSFETs in the push-pull DC-DC converter, use ON’s MTB75N06s- 75A, 60V devices, in parallel. These have an Rds(on) of 0.01 Ohms.
RobM: I, too am a fan id primary-side control. To get a good solid feedback from the output, yo9u will need to use an optocoupler, like the MOC8102 from ON semi. Use the optocoupler to maintain DC-isolation across the transformer galvanic boundary. Ok, this is getting long-winded again, and I will pick up from here later, but these pointers should get you started in the right direction. I will be getting a digital camera in a few days, so I will post a few pics of the low- and high-powered DC-DC converters I’ve completed.
’73,
Steve
Thanks for clarififying the P-fets... Basicly i'm trying to get out of using a linear regulator so I don't spend heaps on heatsinks and transformers.
Since I'm not to keen on rectifying mains a Full smps (basicly half/full bridge) would be out. But if I use transformers and a buck regulator I wouldn't loose much efficientcy at the cost of some load regulation which would be wasted anyway powering a car smps.
The problem is I havn't found a chip capable of driving external FET's/transisters. I still have 5 lm2677's which I could probly pull 20 amps or so off in sync but I'd rather more amps to make the PSU more usefull for testing (my amps keep getting larger 🙂 ).
Is the feedback the same in a buck reg as any other switching PSU? As far as I've found if the output voltage drops duty cycle is increased and the opposite if its too high.
Thanks for your help
SMPS Feedback schemes
Yes and No. It is the same in that it does sense positive voltage, and a higher voltage will result in narrower pulses to the switching transistors, but since you're only sensing a positive voltage (not a +/- bipolar output), you can do the same. If you WERE using +/- outputs (say +/-35 to 40V), you would also need to use the (-) input of the PWM chip's error amp. However, this geta complicated, so I would use the optocoupler I mentioned, then it would equally sense the + and - outputs, maintaining balance about ground (0V).
When I get aroung to learning how to use the scanner attached to my CPU, I will scribble something together, scan it and attach it to a reply in this thread.
Regarding the paralelled 2677s, I also tried to parallel 5 LM2677S-ADJs to get a theoretical 25 Amps out. No (bleep)ing way did it work. Here's why: If you're gonna wind the toroids for each section (around 15-33 uH each), you MUST MUST MUST wind them EXACTLY the same, or one section will dominate, supplying all, or most of the current, and cause an imbalance between sections.
Also, the oscillator used to synchronize the chips must be clocked at at least 300kHz. A 555 might work here, but for better results, use a CMOS chip like a 4081 or 4011 or 4001.
I was using them in a 12V-powered ATX CPU supply (300W), but couldn't get it to work well, so I abandoned it. For now.......
Yes and No. It is the same in that it does sense positive voltage, and a higher voltage will result in narrower pulses to the switching transistors, but since you're only sensing a positive voltage (not a +/- bipolar output), you can do the same. If you WERE using +/- outputs (say +/-35 to 40V), you would also need to use the (-) input of the PWM chip's error amp. However, this geta complicated, so I would use the optocoupler I mentioned, then it would equally sense the + and - outputs, maintaining balance about ground (0V).
When I get aroung to learning how to use the scanner attached to my CPU, I will scribble something together, scan it and attach it to a reply in this thread.
Regarding the paralelled 2677s, I also tried to parallel 5 LM2677S-ADJs to get a theoretical 25 Amps out. No (bleep)ing way did it work. Here's why: If you're gonna wind the toroids for each section (around 15-33 uH each), you MUST MUST MUST wind them EXACTLY the same, or one section will dominate, supplying all, or most of the current, and cause an imbalance between sections.
Also, the oscillator used to synchronize the chips must be clocked at at least 300kHz. A 555 might work here, but for better results, use a CMOS chip like a 4081 or 4011 or 4001.
I was using them in a 12V-powered ATX CPU supply (300W), but couldn't get it to work well, so I abandoned it. For now.......
Little cores like that I just buy. so there within a few uH of the rated value. I'm hoping of getting a inductance DMM in the near future anyway.
Can you reccommend a way that I can boost the power of the lm2577 or another solution?
I started another thread here if u wanna reply in there to keep this thread a little more on topic 😉
Can you reccommend a way that I can boost the power of the lm2577 or another solution?
I started another thread here if u wanna reply in there to keep this thread a little more on topic 😉
are 40V mos-fet`s suitable for 12volt Switching SMPS ??
i found this NTB5404N: Power MOSFET 40 V, 136 A it has 4.5 mohm resistence nice...
i found this NTB5404N: Power MOSFET 40 V, 136 A it has 4.5 mohm resistence nice...
B.I.G said:
the low on resistance is good but this usually means a high gate capacitance. this makes it difficult to drive the fets fast enough for high switching speeds.
you could get over it by using totem pole drivers and low ohm gate resistors.
can anybody post a link to a website that sells big ferrite toroids for smps making?
(i need them to be able to ship it...)
(i need them to be able to ship it...)
TL494?
Has any use the TL494 ic, if so does any on have any schematics, not sure if I trust all the google searches.
Has any use the TL494 ic, if so does any on have any schematics, not sure if I trust all the google searches.
Re: TL494?
Yes. Try amidonassociates.com. You will want the FT-240-77 (Ferrite Toroid, 2.40" outer diameter, #77 ferrite material). At 50kHz, It will handle something like 1.2kW.
Yes again. Look at almost any car amplifier or AT/ATX pc supply. They almost always use the TL494, sometimes labelled KA7500, or KIA7500. Same Chip. Also, try searching the Power Supply Design forum for "TL494". Should get alot of hits.
Steve
B.I.G [/i][B] are 40V mos-fet`s suitable for 12volt Switching SMPS ?? [/B][/QUOTE] 40V is marginal for 12V apps. I would go with 50V units at a minimum. Reason is that your MOSFETS could see as much as ~2.6Vin(max). So for said:
Yes. Try amidonassociates.com. You will want the FT-240-77 (Ferrite Toroid, 2.40" outer diameter, #77 ferrite material). At 50kHz, It will handle something like 1.2kW.
shon35us said:
Yes again. Look at almost any car amplifier or AT/ATX pc supply. They almost always use the TL494, sometimes labelled KA7500, or KIA7500. Same Chip. Also, try searching the Power Supply Design forum for "TL494". Should get alot of hits.
Steve
- Status
- Not open for further replies.