Overclock.net › Forums › Components › Power Supplies › Discussion on SMPSU theory
New Posts  All Forums:Forum Nav:

# Discussion on SMPSU theory

Had this conversation with another member earlier today, just thought I'd share for anyone else who wants to read.

Warning: Spoiler! (Click to show)
[1/23/2011 1:20:22 PM] phaedrus2401: ok
[1/23/2011 1:20:45 PM] phaedrus2401: So do you want to start with SMPS theory?
[1/23/2011 1:21:00 PM] **********: sure. ill take whatever you will give me
[1/23/2011 1:21:04 PM] phaedrus2401: ok
[1/23/2011 1:21:11 PM] phaedrus2401: You know the difference between AC and DC right?
[1/23/2011 1:21:21 PM] **********: alternating current and direct current
[1/23/2011 1:21:24 PM] **********: yes
[1/23/2011 1:22:05 PM] phaedrus2401: Ok, the purpose of the PSU is to convert AC to DC, since it's difficult to design complex circuitry that uses AC
[1/23/2011 1:22:21 PM] phaedrus2401: And to reduce the voltage to levels that won't fry components
[1/23/2011 1:22:24 PM] phaedrus2401: so
[1/23/2011 1:22:32 PM] phaedrus2401: The simplest power supply is the linear power supply
[1/23/2011 1:22:48 PM] phaedrus2401: Say you want to turn 120VAC into 12VDC
[1/23/2011 1:22:53 PM] phaedrus2401: All you need is three components
[1/23/2011 1:23:14 PM] phaedrus2401: You take a transformer, which converts AC to AC of another voltage
[1/23/2011 1:23:15 PM] phaedrus2401: via black magic
[1/23/2011 1:23:21 PM] **********: yeah
[1/23/2011 1:23:32 PM] phaedrus2401: And you have that transformer drop voltage from 120VAC to say 20VAC
[1/23/2011 1:24:19 PM] phaedrus2401: Then you have a rectifier, which is basically a different name for a diode, which turns the AC into DC, with a voltage drop
[1/23/2011 1:25:00 PM] **********: ok
[1/23/2011 1:25:12 PM] phaedrus2401: Then a capacitor filters the resulting output
[1/23/2011 1:25:37 PM] **********: ok. so the capacitors are what help with ripple suppression?
[1/23/2011 1:25:43 PM] phaedrus2401: Capacitors and inductors
[1/23/2011 1:25:50 PM] **********: ok
[1/23/2011 1:26:18 PM] phaedrus2401: When you have a combination of capacitors and inductors designed to filter or amplify certain signal frequencies, that's called a pi filter
[1/23/2011 1:26:36 PM] phaedrus2401: (ripple is just an unwanted, random signal--with a power source you want no signal at all)
[1/23/2011 1:26:52 PM] phaedrus2401: But anyway
[1/23/2011 1:26:55 PM] phaedrus2401: Linear PSUs are very simple
[1/23/2011 1:27:03 PM] phaedrus2401: Three to five components minimum, easy peasy
[1/23/2011 1:27:11 PM] phaedrus2401: But they have problems
[1/23/2011 1:27:26 PM] **********: what kind of things/systems use linear PSUs? a toster?
[1/23/2011 1:27:33 PM] phaedrus2401: Not a toaster, too much load
[1/23/2011 1:27:36 PM] phaedrus2401: Nothing nowadays
[1/23/2011 1:27:39 PM] **********: haha.
[1/23/2011 1:27:45 PM] **********: ok. just curious
[1/23/2011 1:27:48 PM] phaedrus2401: But old stuff, like old amplifiers and speakers
[1/23/2011 1:27:58 PM] phaedrus2401: I've seen an old calculator with a linear PSU
[1/23/2011 1:28:02 PM] phaedrus2401: But linear PSUs have problems
[1/23/2011 1:28:12 PM] phaedrus2401: 1. Inefficient, generally 50-65% efficient
[1/23/2011 1:28:27 PM] phaedrus2401: 2. Suffer a lot of voltage drop with increased load
[1/23/2011 1:28:43 PM] phaedrus2401: 3. Require massive transformers for anything but a miniscule load
[1/23/2011 1:28:43 PM] **********: ooo...thats bad
[1/23/2011 1:28:51 PM] phaedrus2401: 3 is the biggy
[1/23/2011 1:28:56 PM] **********: yeah
[1/23/2011 1:29:04 PM] **********: that makes them impractical for most uses
[1/23/2011 1:29:05 PM] phaedrus2401: A 300W linear PSU would require a transformer the size of a small dog
[1/23/2011 1:29:14 PM] phaedrus2401: And would generate a 60Hz hum
[1/23/2011 1:29:26 PM] phaedrus2401: That's the important bit there
[1/23/2011 1:29:32 PM] **********: yeah
[1/23/2011 1:29:34 PM] phaedrus2401: 60Hz
[1/23/2011 1:29:39 PM] phaedrus2401: The frequency of US mains
[1/23/2011 1:29:44 PM] phaedrus2401: In the EU it's 50Hz
[1/23/2011 1:29:51 PM] phaedrus2401: The hum itself isn't important, but that gives you a clue
[1/23/2011 1:29:52 PM] phaedrus2401: You see
[1/23/2011 1:30:13 PM] phaedrus2401: The required size of a transformer is inversely proportional to the frequency of the AC signal it is modifying
[1/23/2011 1:30:37 PM] phaedrus2401: So if that AC signal were, say, 40,000Hz instead of 60Hz
[1/23/2011 1:30:57 PM] phaedrus2401: The same transformer could handle 10,000W instead of 300W
[1/23/2011 1:31:08 PM] phaedrus2401: (assuming that it doesn't melt from increased current draw)
[1/23/2011 1:31:21 PM] phaedrus2401: You follow?
[1/23/2011 1:31:23 PM] **********: ok
[1/23/2011 1:31:30 PM] **********: yea.
[1/23/2011 1:31:59 PM] phaedrus2401: Switch mode PSUs adress that issue, and efficiency and voltage drop, but that's more complicated as to why
[1/23/2011 1:32:07 PM] phaedrus2401: So a switch mode power supply
[1/23/2011 1:32:30 PM] phaedrus2401: Does something a little complicated
[1/23/2011 1:32:59 PM] phaedrus2401: A switch mode PSU (SMPS) takes the 120VAC 60Hz signal from the mains, and rectifies it using a special type of rectifier called a bridge rectifier
[1/23/2011 1:33:56 PM] phaedrus2401: So you've turned a 120VAC signal into a ~115VDC signal
[1/23/2011 1:34:17 PM] phaedrus2401: Then, you have two transistors (switches) that switch on and off very, very quickly
[1/23/2011 1:34:41 PM] phaedrus2401: When they're on, current flows forward. When they're off, current flows backward.
[1/23/2011 1:34:51 PM] **********: ok
[1/23/2011 1:34:56 PM] phaedrus2401: That makes it an AC signal
[1/23/2011 1:35:08 PM] **********: becasue the current changes flow
[1/23/2011 1:35:13 PM] phaedrus2401: Exactly
[1/23/2011 1:35:17 PM] **********: ok
[1/23/2011 1:35:32 PM] phaedrus2401: Except they're switching on and off >20,000Hz
[1/23/2011 1:35:34 PM] phaedrus2401: Instead of 60hz
[1/23/2011 1:35:42 PM] phaedrus2401: 20,000-30,000 generally
[1/23/2011 1:35:48 PM] phaedrus2401: High enough that you can't hear the hum
[1/23/2011 1:35:50 PM] **********: ok
[1/23/2011 1:36:04 PM] phaedrus2401: Then you have a transformer, and a rectifier, and a pi filter
[1/23/2011 1:36:05 PM] **********: outside of the human range of sound
[1/23/2011 1:36:08 PM] phaedrus2401: yeah
[1/23/2011 1:36:38 PM] phaedrus2401: You basically have an AC-AC converter whose main job is to increase the frequency of the AC signal, then a basically linear PSU
[1/23/2011 1:36:52 PM] phaedrus2401: That's the ultra simple, bare bones, simple as it gets explanation
[1/23/2011 1:37:18 PM] **********: ok. so it increased the amperage as it creates the AC?
[1/23/2011 1:37:35 PM] **********: is it amperage? i still get them mixed up
[1/23/2011 1:37:47 PM] phaedrus2401: no no, frequency
[1/23/2011 1:37:54 PM] **********: ok.
[1/23/2011 1:37:57 PM] phaedrus2401: Voltage and current you understand, right?
[1/23/2011 1:37:58 PM] **********: duh...hz
[1/23/2011 1:38:06 PM] **********: yeah. I do.
[1/23/2011 1:38:07 PM] phaedrus2401: ok
[1/23/2011 1:38:19 PM] phaedrus2401: Now there's one other important thing to mention
[1/23/2011 1:38:23 PM] **********: ok
[1/23/2011 1:38:32 PM] phaedrus2401: With SMPS
[1/23/2011 1:38:38 PM] phaedrus2401: The AC-AC converter (the primary side)
[1/23/2011 1:38:44 PM] phaedrus2401: Doesn't just increase frequency
[1/23/2011 1:38:58 PM] phaedrus2401: It also regulates voltage on the output
[1/23/2011 1:39:14 PM] phaedrus2401: Because the switching transistors are turned on and off by a chip called the PWM controller
[1/23/2011 1:39:35 PM] phaedrus2401: PWM is pulse width modulation
[1/23/2011 1:39:47 PM] phaedrus2401: You probably know that it's used to control fan speed, for instance
[1/23/2011 1:39:58 PM] **********: ahhh. ok. yeah
[1/23/2011 1:40:16 PM] phaedrus2401: Basically, it controls how long the transistors are switched on, vs. off
[1/23/2011 1:40:20 PM] **********: ive been wondering what separates the primary and secondary in a PSU
[1/23/2011 1:40:35 PM] phaedrus2401: this is an important bit, read closely
[1/23/2011 1:40:41 PM] **********: gotcha
[1/23/2011 1:41:29 PM] phaedrus2401: The PWM controller (often combined with the PFC controller... later) controls how long the transistors are on or off, and is hooked up to the secondary controller, which monitors voltages on the outputs (among other things)
[1/23/2011 1:41:48 PM] **********: ok
[1/23/2011 1:41:49 PM] phaedrus2401: If more current is demanded, the voltage on the outputs will drop a bit
[1/23/2011 1:42:10 PM] phaedrus2401: So the PWM controller increases the length of time the transistors are on
[1/23/2011 1:42:22 PM] phaedrus2401: Which allows more current through
[1/23/2011 1:42:26 PM] **********: tomcompensate
[1/23/2011 1:42:30 PM] phaedrus2401: yeah
[1/23/2011 1:42:31 PM] **********: to compensate
[1/23/2011 1:42:37 PM] **********: blah. crappy keyboard
[1/23/2011 1:43:06 PM] phaedrus2401: 90% of voltage regulation is done by the PWM chip modifying the duty cycle of the transistors
[1/23/2011 1:43:46 PM] **********: ok
[1/23/2011 1:43:51 PM] phaedrus2401: Any questions?
[1/23/2011 1:43:53 PM] phaedrus2401:
[1/23/2011 1:45:15 PM] **********: not so far. im going to save this conversationa nd re read it a few times to make sure that I understand it all. But as of right now i understand. my dad gave me a basic education in electronics, components, thier function and the like.
[1/23/2011 1:45:24 PM] phaedrus2401: good
[1/23/2011 1:45:31 PM] **********: so ive got a foundation at least
[1/23/2011 1:45:43 PM] phaedrus2401: Now, I will say
[1/23/2011 1:45:58 PM] phaedrus2401: Wait, one more word on the primary
[1/23/2011 1:46:03 PM] **********: no problem
[1/23/2011 1:47:12 PM] phaedrus2401: You also have a capacitor on the primary (or multiple in parallel), which act as a resevoir when the transistors are off, and then when the transistors are switched on they supply all their charge at once. Otherwise the throughput of the transistors would always be the same and current couldn't be adjusted
[1/23/2011 1:47:15 PM] phaedrus2401: That's an important bit I forgot
[1/23/2011 1:47:42 PM] **********: ok. hence why capacitors are so important
[1/23/2011 1:47:49 PM] phaedrus2401: well, those are on the primary
[1/23/2011 1:47:58 PM] **********: well...one of many reasons. they take a HUGE amount of load
[1/23/2011 1:48:07 PM] phaedrus2401: Those ones, yeah
[1/23/2011 1:48:26 PM] **********: that reminds me. which is more important. a good quality primary or a good quality secondary?
[1/23/2011 1:48:34 PM] phaedrus2401: Both
[1/23/2011 1:48:36 PM] **********: ideally both but
[1/23/2011 1:48:39 PM] **********: ok
[1/23/2011 1:48:43 PM] phaedrus2401: But generally the secondary caps will fail first
[1/23/2011 1:48:46 PM] phaedrus2401: Even in quality units
[1/23/2011 1:48:53 PM] phaedrus2401: Since they have to be a smaller package
[1/23/2011 1:48:53 PM] **********: ok
[1/23/2011 1:49:11 PM] phaedrus2401: Also
[1/23/2011 1:49:15 PM] phaedrus2401: Not going to go into PFC yet
[1/23/2011 1:49:29 PM] phaedrus2401: But in the two transistor forward topology used in 99% of decent PSUs
[1/23/2011 1:49:38 PM] phaedrus2401: PFC is implemented by a DC-DC converter between the bridge rectifier and the primary caps
[1/23/2011 1:49:45 PM] phaedrus2401: That boosts the input voltage to 400VDC
[1/23/2011 1:50:09 PM] phaedrus2401: So
[1/23/2011 1:50:49 PM] phaedrus2401: 120VAC enters the PSU, goes through the transient filter (which protects the PSU's components, and protects your mains from major backwash of current), hits the bridge rectifier and is converted to about 110VDC
[1/23/2011 1:50:54 PM] phaedrus2401: Then the PFC transistors boost it to 400VDC
[1/23/2011 1:51:20 PM] **********: ok
[1/23/2011 1:51:27 PM] phaedrus2401: The switching transistors and primary capacitor convert it to 400VAC 20kHz
[1/23/2011 1:51:47 PM] **********: ok
[1/23/2011 1:51:47 PM] phaedrus2401: Then the transformer drops it to ~20-40VAC (depending on design)
[1/23/2011 1:51:53 PM] phaedrus2401: Then things get complicated
[1/23/2011 1:52:09 PM] phaedrus2401: In modern units, the +12V is the first concern, so we'll start there
[1/23/2011 1:52:56 PM] phaedrus2401: The 20VAC hits the +12V schottky (or sometimes MOSFETs, in which case the secondary controller plays a role in regulating those) and is rectified to about 18VDC
[1/23/2011 1:53:02 PM] phaedrus2401: Or
[1/23/2011 1:53:03 PM] phaedrus2401: hang on
[1/23/2011 1:53:04 PM] phaedrus2401: um
[1/23/2011 1:53:32 PM] phaedrus2401: Whoops, think I skipped a step
[1/23/2011 1:53:41 PM] phaedrus2401: **** it
[1/23/2011 1:53:46 PM] **********: no problem. im learning exactly what I wanted to.
[1/23/2011 1:53:48 PM] **********: thank you
[1/23/2011 1:54:49 PM] phaedrus2401: It's rectified to ~12VDC, and an inductor (big coil) acts as a magnetic amplifier to regulate it to precisely 12VDC
[1/23/2011 1:55:42 PM] phaedrus2401: Then you have the pi filter (with that coil doing double duty as part of that as well), which filters unwanted ripple and noise
[1/23/2011 1:56:03 PM] phaedrus2401: Then complicated crap involving the controller and protections sets, blah blah
[1/23/2011 1:56:40 PM] **********: ok. so that big coul is what makes sure its exavtly 12v (ideally)?
[1/23/2011 1:56:46 PM] phaedrus2401: yeah, hang on
[1/23/2011 1:57:59 PM] **********: no worries
[1/23/2011 1:58:15 PM] **********: you are doing me a favor. so im at your mercy.
[1/23/2011 1:58:31 PM] phaedrus2401: lol
[1/23/2011 1:58:33 PM] phaedrus2401: ok
[1/23/2011 1:58:37 PM] phaedrus2401: So the coil
[1/23/2011 1:58:43 PM] **********: da
[1/23/2011 1:59:12 PM] phaedrus2401: An inductor has the property of always wanting the same voltage to be going across it, you tune it right so that it wants to output +12VDC
[1/23/2011 1:59:24 PM] phaedrus2401: You just have to regulate its input so that it's close enough that it can regulate it
[1/23/2011 1:59:34 PM] phaedrus2401: The mag amp does the last 10% of the voltage regulation
[1/23/2011 1:59:36 PM] **********: ok
[1/23/2011 1:59:45 PM] phaedrus2401: But the PWM controller does 90% of it, making sure that its input is close enough
[1/23/2011 2:00:08 PM] phaedrus2401: As for ripple
[1/23/2011 2:00:22 PM] phaedrus2401: What a rectifier does, basically
[1/23/2011 2:00:29 PM] phaedrus2401: Is turn a sin wave into a flat line'
[1/23/2011 2:00:33 PM] phaedrus2401: But not perfectly
[1/23/2011 2:00:49 PM] **********: thats the lin eyou see on an oscilliscope?
[1/23/2011 2:01:01 PM] phaedrus2401: Yeah
[1/23/2011 2:01:05 PM] phaedrus2401: That's the graph of voltage over time
[1/23/2011 2:01:11 PM] **********: ok. just making sure
[1/23/2011 2:01:45 PM] phaedrus2401: All a rectifier is really doing is transforming that sine wave by flipping the negative parts to positive (like doing an absolute value function) and moving the center line to its own output voltage
[1/23/2011 2:01:47 PM] phaedrus2401: Put VEEEERRRRRRRYYYYYY simply anyway
[1/23/2011 2:02:03 PM] phaedrus2401: But you still have all the ups and downs of the sine wave
[1/23/2011 2:02:11 PM] phaedrus2401: Just none of them go negative anymore
[1/23/2011 2:02:28 PM] **********: yeah. the idea is to make it as flat as possible
[1/23/2011 2:02:29 PM] **********: ok
[1/23/2011 2:02:31 PM] phaedrus2401: Yes
[1/23/2011 2:02:50 PM] phaedrus2401: So you put capacitors and inductors in parallel on the line
[1/23/2011 2:03:13 PM] **********: so components are shielded from a slight amount of over voltage and thats why it trys to only go positive?
[1/23/2011 2:03:26 PM] **********: just curious
[1/23/2011 2:03:28 PM] phaedrus2401: no
[1/23/2011 2:03:33 PM] **********: or is that just the narture of the system
[1/23/2011 2:03:34 PM] phaedrus2401: When voltage is negative, current is flowing backwards
[1/23/2011 2:03:41 PM] **********: ok. :F
[1/23/2011 2:03:43 PM] **********:
[1/23/2011 2:03:47 PM] phaedrus2401: A rectifier is a diode, and only allows current to pass in one direction
[1/23/2011 2:03:52 PM] **********: ok
[1/23/2011 2:03:54 PM] **********: cool
[1/23/2011 2:04:03 PM] phaedrus2401: So anything that goes through a rectifier *has* to be positive
[1/23/2011 2:04:11 PM] **********: ok. gotcha.
[1/23/2011 2:04:32 PM] phaedrus2401: So you have that ripple from the rectifiers, and some electromagnetic interference as well
[1/23/2011 2:04:42 PM] **********: ok
[1/23/2011 2:04:45 PM] phaedrus2401: From radio and magnetic waves interacting with the circuit
[1/23/2011 2:04:47 PM] phaedrus2401: So anyway
[1/23/2011 2:05:04 PM] phaedrus2401: Capacitors store a small charge, and inductors like to have a constant voltage going across them
[1/23/2011 2:05:35 PM] phaedrus2401: Capacitors get charged by the spikes in the output, then release that charge in the dips
[1/23/2011 2:06:11 PM] phaedrus2401: Inductors smooth the spikes and dips similarly, except they store charge in a magnetic field
[1/23/2011 2:06:33 PM] phaedrus2401: By manipulating the properties of the inductors and capacitors
[1/23/2011 2:06:50 PM] phaedrus2401: You can eliminate or enhance certain frequencies of signal
[1/23/2011 2:06:53 PM] phaedrus2401: A pi filter
[1/23/2011 2:07:02 PM] phaedrus2401: And in a PSU, they're tuned to eliminate as much signal as possible
[1/23/2011 2:07:34 PM] **********: cool. so THATS how ripple is regulated
[1/23/2011 2:07:37 PM] phaedrus2401: yeah
[1/23/2011 2:08:37 PM] phaedrus2401: So, any questions?
[1/23/2011 2:08:58 PM] **********: not yet
[1/23/2011 2:09:13 PM] phaedrus2401: What do you want to know about yet?
[1/23/2011 2:09:16 PM] phaedrus2401: *next
[1/23/2011 2:10:19 PM] **********: hrmmm...capacitor types? and any reading you can think of that I can do
[1/23/2011 2:10:35 PM] phaedrus2401: capacitor types
[1/23/2011 2:10:38 PM] phaedrus2401: Well, there are lots
[1/23/2011 2:10:44 PM] phaedrus2401: In a PSU generally there are five types
[1/23/2011 2:10:55 PM] phaedrus2401: X capacitors, Y capacitors, film capacitors, electrolytic capacitors, and polymer capacitors
[1/23/2011 2:11:11 PM] phaedrus2401: Not going to go into the details of how they all work
[1/23/2011 2:11:12 PM] phaedrus2401: But
[1/23/2011 2:11:21 PM] **********: ok
[1/23/2011 2:12:14 PM] phaedrus2401: X capacitors and Y capacitors are used in the transient filter to reduce incoming ripple. X capacitors can be used singly, and are made so that if they fail catastrophically they won't harm anything else. Y capacitors are used in pairs, and require a ground between them
[1/23/2011 2:12:27 PM] **********: ok
[1/23/2011 2:12:29 PM] phaedrus2401: A film capacitor is usually used right after the rectifying bridge to filter its ripple a bit
[1/23/2011 2:13:03 PM] phaedrus2401: Electrolytic capacitors are the main workhorse cap, they have the best capacity to size/cost ratio
[1/23/2011 2:13:27 PM] phaedrus2401: They're used as the main charge caps, for the pi filters, and for a few other roles
[1/23/2011 2:13:44 PM] phaedrus2401: However, electrolytic caps are sensitive to heat
[1/23/2011 2:13:53 PM] phaedrus2401: And due to a bad formula being spread around, a lot of cheap Chinese caps will fail prematurely
[1/23/2011 2:14:25 PM] phaedrus2401: Polymer capacitors are more expensive and can't get anywhere near the same capacity as electrolytics, but are much more reliable
[1/23/2011 2:14:39 PM] phaedrus2401: So they're often used in extremely critical areas, or where space is an issue
[1/23/2011 2:14:40 PM] phaedrus2401: However
[1/23/2011 2:14:56 PM] phaedrus2401: Sometimes a manufacturer will replace electrolytics with polymers for marketing
[1/23/2011 2:15:03 PM] phaedrus2401: Like Thermaltake did with its Toughpower XT units
[1/23/2011 2:15:28 PM] phaedrus2401: Problem is, polymer caps only go up to a certain capacitance, and have much less capacitance for a given size
[1/23/2011 2:15:45 PM] phaedrus2401: So Thermaltake replaced all the caps in the output pi filters with polymers
[1/23/2011 2:15:58 PM] phaedrus2401: But there was too little capacitance, and they didn't retune the filter right
[1/23/2011 2:16:02 PM] phaedrus2401: The result was epic fail ripple
[1/23/2011 2:16:16 PM] **********: so they marketed reliability over performance and sold it at a premium??
[1/23/2011 2:16:21 PM] phaedrus2401: yup
[1/23/2011 2:16:29 PM] **********: ok
[1/23/2011 2:16:45 PM] phaedrus2401: You can rely on them to supply the same epic fail ripple until the whole thing falls apart, 'cause those polymer caps aren't going to fail.
[1/23/2011 2:16:53 PM] phaedrus2401: much anyway
[1/23/2011 2:17:21 PM] **********: OT Question. on a motherboard solid(Polymer?) caps are good because they last longer and dont need as much ripple suppression?
[1/23/2011 2:17:32 PM] phaedrus2401: It's the same principle
[1/23/2011 2:17:37 PM] **********: ok
[1/23/2011 2:17:42 PM] phaedrus2401: Your motherboard's VRMs are basically PSUs as well
[1/23/2011 2:17:57 PM] phaedrus2401: Though they don't need pi filters so much
[1/23/2011 2:18:19 PM] **********: so a low quality PSU would mean you would want a mobo with electrolytic caps to help protect it?
[1/23/2011 2:18:28 PM] phaedrus2401: Well
[1/23/2011 2:18:56 PM] phaedrus2401: Just any motherboard with more effective VRMs. Polymer caps aren't inherently bad, you just have to design around them from the ground up, you can't just stick them in the place of electrolytics and hope for the best
[1/23/2011 2:19:31 PM] **********: thats what I figured. it means a whole new design to compensate for the lack of capacitance
[1/23/2011 2:19:46 PM] phaedrus2401: Also, SeaSonic marketed the S12II units as using Polymer capacitors They put one polymer cap on the +12V pi filter
[1/23/2011 2:19:54 PM] **********: hasha
[1/23/2011 2:20:02 PM] phaedrus2401: But it was part of the design of the pi filter from the beginning
[1/23/2011 2:20:05 PM] **********: is that an example of a "meh" seasonic?
[1/23/2011 2:20:05 PM] phaedrus2401: So no harm was done
[1/23/2011 2:20:09 PM] **********: oh, nevermind
[1/23/2011 2:20:23 PM] phaedrus2401: So anyway
[1/23/2011 2:20:32 PM] **********: yeah
[1/23/2011 2:21:23 PM] phaedrus2401: SeaSonic is not the "best" OEM
[1/23/2011 2:21:41 PM] **********: i figured as much
[1/23/2011 2:21:51 PM] phaedrus2401: SeaSonic's build quality is actually pretty average for a high-end OEM
[1/23/2011 2:23:17 PM] phaedrus2401: SeaSonic, however, have always been able to produce some of the best *performing* PSUs. They were the first to make an 80+ unit, and every time they make a new flagship platform (S12, M12D, X) it sets a new standard
[1/23/2011 2:23:55 PM] phaedrus2401: But their units aren't any more reliable than, say, CWT or FSP (discounting FSP's low end, which is made on their older equipment)
[1/23/2011 2:24:07 PM] phaedrus2401: or Enhance or Seventeam
[1/23/2011 2:24:14 PM] phaedrus2401: It's just performance that they do best
[1/23/2011 2:24:47 PM] **********: ok
[1/23/2011 2:26:01 PM] **********: cool. thanks. thats everything i wanted to learn.
Great. Thanks you...
+rep
 Devil (17 items)
CPUMotherboardGraphicsRAM
3750k Asrock Z77 Extreme 4 MSI GTX 980 TI Kingstone Savage 2400 16 GB
Hard DriveHard DriveHard DriveCooling
Samsung EVO 850 500 GB Samsung EVO 850 250 GB WD Green 2 TB Prolimatech Megashadow
OSMonitorMonitorKeyboard
Windows 10 64 Dell U2415 LG W2600H Roccat Ryos MK Pro
Corsair HX 750 Coolermaster 690 III MX 1100 Razer Goliathus
Audio
XFI Gamer
 Devil (17 items)
CPUMotherboardGraphicsRAM
3750k Asrock Z77 Extreme 4 MSI GTX 980 TI Kingstone Savage 2400 16 GB
Hard DriveHard DriveHard DriveCooling
Samsung EVO 850 500 GB Samsung EVO 850 250 GB WD Green 2 TB Prolimatech Megashadow
OSMonitorMonitorKeyboard
Windows 10 64 Dell U2415 LG W2600H Roccat Ryos MK Pro
Corsair HX 750 Coolermaster 690 III MX 1100 Razer Goliathus
Audio
XFI Gamer
Just read the entire thing. I knew about 1/2 of it, but it's nice to know the rest, thanks!
 Labor of Love (14 items)
CPUMotherboardGraphicsRAM
2600k @ 5.0 GHz/1.42v ASUS P8P67 Pro MSI GTX 670 OC 20 GB/1600/7-7-7-24
Hard DriveHard DriveOptical DriveOS
120GB Vertex SSD Western Digital 1TB None! Ubuntu 12.04/Win8/OSX 10.8.2
MonitorKeyboardPowerCase
27" 1440p X-Star OCN Mechanical - MX Browns Seasonic X-560 800D w/RX480 top mounted
Mouse
Logitech USB Generic
 Labor of Love (14 items)
CPUMotherboardGraphicsRAM
2600k @ 5.0 GHz/1.42v ASUS P8P67 Pro MSI GTX 670 OC 20 GB/1600/7-7-7-24
Hard DriveHard DriveOptical DriveOS
120GB Vertex SSD Western Digital 1TB None! Ubuntu 12.04/Win8/OSX 10.8.2
MonitorKeyboardPowerCase
27" 1440p X-Star OCN Mechanical - MX Browns Seasonic X-560 800D w/RX480 top mounted
Mouse
Logitech USB Generic
A few things:
1.) Linear PSU's do not have to have low efficiency. Nor are they limited to low output. Old Linear PSU's had that problem; but current PSU's are effectively linear PSU's with just the 12v being generated inside; and the DC-DC units making up your other rails (discounting the 5VSB transformer and inversion for -12v)

2.) Thermaltake's ToughpowerXTs didn't have fail ripple suppression?
Quote:
This Thermaltake was taken to the wall, and it clearly had no intention of giving in, because these scope shots are be-oo-tee-ful. 3.3V ripple under 25mV. 5V ripple under 20mV. 12V ripple under 40mV.
JonnyGuru.com's review by OklahomaWolf.

3.) Not all SMPS designs use a PWM Controller IC

I only skimmed it mostly; but there's some inherently bad statements in there.
 Test Bed (24 items) TTTbox (15 items)
CPUMotherboardGraphicsRAM
Core i5 3570K  Biostar TZ774XE Power Color AX5770 1GB 4x4GB 1866Mhz 9-10-9-27-2T
Hard DriveHard DriveHard DriveHard Drive
Samsung Spinpoint F3 1TB Samsung Spinpoint F3 1TB Western Digital Green 2TB Samsung Spinpoint M4 500GB
Hard DriveOptical DriveCoolingOS
Intel X-25M G2 80GB Lite-On BluRay Drive Arctic Cooling Freezer 13 Pro Windows 7 Ultimate
MonitorKeyboardPowerCase
Dell U2212HM Deck Legend Frost Tactile Silverstone Strider Plus 750w Custom Acrylic Test Bed
Logitech G500 Cooler Master Storm Tactics FPS Beyer Dynamics DT-990 600ohm Samson MediaOne 3A
AudioAudioOtherOther
FiiO E09k QOGIR FiiO E17 ALPEN Microsoft Lifecam Studio Blue Yeti Pro Microphone
CPUMotherboardGraphicsRAM
AMD A10-A5800K ASRock FM2A85X-ITX Radeon HD7770 1GB 2x8GB 1866mhz C9
Hard DriveHard DriveOptical DriveCooling
SanDisk Ultra Plus 128GB 2TB 5400 RPM  24x DVD Drive Noctua L9A
OSMonitorKeyboardPower
Windows 7 Home Premium 32" IPS TV Logitech K400 Seasonic G360
CaseMouseAudio
Cooler Master Elite 120 Advanced Logitech K400 Trackpad Custom 2.1 Home Theater Setup
 Test Bed (24 items) TTTbox (15 items)
CPUMotherboardGraphicsRAM
Core i5 3570K  Biostar TZ774XE Power Color AX5770 1GB 4x4GB 1866Mhz 9-10-9-27-2T
Hard DriveHard DriveHard DriveHard Drive
Samsung Spinpoint F3 1TB Samsung Spinpoint F3 1TB Western Digital Green 2TB Samsung Spinpoint M4 500GB
Hard DriveOptical DriveCoolingOS
Intel X-25M G2 80GB Lite-On BluRay Drive Arctic Cooling Freezer 13 Pro Windows 7 Ultimate
MonitorKeyboardPowerCase
Dell U2212HM Deck Legend Frost Tactile Silverstone Strider Plus 750w Custom Acrylic Test Bed
Logitech G500 Cooler Master Storm Tactics FPS Beyer Dynamics DT-990 600ohm Samson MediaOne 3A
AudioAudioOtherOther
FiiO E09k QOGIR FiiO E17 ALPEN Microsoft Lifecam Studio Blue Yeti Pro Microphone
CPUMotherboardGraphicsRAM
AMD A10-A5800K ASRock FM2A85X-ITX Radeon HD7770 1GB 2x8GB 1866mhz C9
Hard DriveHard DriveOptical DriveCooling
SanDisk Ultra Plus 128GB 2TB 5400 RPM  24x DVD Drive Noctua L9A
OSMonitorKeyboardPower
Windows 7 Home Premium 32" IPS TV Logitech K400 Seasonic G360
CaseMouseAudio
Cooler Master Elite 120 Advanced Logitech K400 Trackpad Custom 2.1 Home Theater Setup
Then what was the Toughpower that had such mediocre ripple?
Quote:
Originally Posted by Phaedrus2129;12123208
Then what was the Toughpower that had such mediocre ripple?
The FSP Built Toughpower 80+ Silver models. That was entirely do to the FSP design. Not cap choice or anything else.
 Test Bed (24 items) TTTbox (15 items)
CPUMotherboardGraphicsRAM
Core i5 3570K  Biostar TZ774XE Power Color AX5770 1GB 4x4GB 1866Mhz 9-10-9-27-2T
Hard DriveHard DriveHard DriveHard Drive
Samsung Spinpoint F3 1TB Samsung Spinpoint F3 1TB Western Digital Green 2TB Samsung Spinpoint M4 500GB
Hard DriveOptical DriveCoolingOS
Intel X-25M G2 80GB Lite-On BluRay Drive Arctic Cooling Freezer 13 Pro Windows 7 Ultimate
MonitorKeyboardPowerCase
Dell U2212HM Deck Legend Frost Tactile Silverstone Strider Plus 750w Custom Acrylic Test Bed
Logitech G500 Cooler Master Storm Tactics FPS Beyer Dynamics DT-990 600ohm Samson MediaOne 3A
AudioAudioOtherOther
FiiO E09k QOGIR FiiO E17 ALPEN Microsoft Lifecam Studio Blue Yeti Pro Microphone
CPUMotherboardGraphicsRAM
AMD A10-A5800K ASRock FM2A85X-ITX Radeon HD7770 1GB 2x8GB 1866mhz C9
Hard DriveHard DriveOptical DriveCooling
SanDisk Ultra Plus 128GB 2TB 5400 RPM  24x DVD Drive Noctua L9A
OSMonitorKeyboardPower
Windows 7 Home Premium 32" IPS TV Logitech K400 Seasonic G360
CaseMouseAudio
Cooler Master Elite 120 Advanced Logitech K400 Trackpad Custom 2.1 Home Theater Setup
 Test Bed (24 items) TTTbox (15 items)
CPUMotherboardGraphicsRAM
Core i5 3570K  Biostar TZ774XE Power Color AX5770 1GB 4x4GB 1866Mhz 9-10-9-27-2T
Hard DriveHard DriveHard DriveHard Drive
Samsung Spinpoint F3 1TB Samsung Spinpoint F3 1TB Western Digital Green 2TB Samsung Spinpoint M4 500GB
Hard DriveOptical DriveCoolingOS
Intel X-25M G2 80GB Lite-On BluRay Drive Arctic Cooling Freezer 13 Pro Windows 7 Ultimate
MonitorKeyboardPowerCase
Dell U2212HM Deck Legend Frost Tactile Silverstone Strider Plus 750w Custom Acrylic Test Bed
Logitech G500 Cooler Master Storm Tactics FPS Beyer Dynamics DT-990 600ohm Samson MediaOne 3A
AudioAudioOtherOther
FiiO E09k QOGIR FiiO E17 ALPEN Microsoft Lifecam Studio Blue Yeti Pro Microphone
CPUMotherboardGraphicsRAM
AMD A10-A5800K ASRock FM2A85X-ITX Radeon HD7770 1GB 2x8GB 1866mhz C9
Hard DriveHard DriveOptical DriveCooling
SanDisk Ultra Plus 128GB 2TB 5400 RPM  24x DVD Drive Noctua L9A
OSMonitorKeyboardPower
Windows 7 Home Premium 32" IPS TV Logitech K400 Seasonic G360
CaseMouseAudio
Cooler Master Elite 120 Advanced Logitech K400 Trackpad Custom 2.1 Home Theater Setup
New Posts  All Forums:Forum Nav:
Return Home
Back to Forum: Power Supplies
• Discussion on SMPSU theory
Overclock.net › Forums › Components › Power Supplies › Discussion on SMPSU theory