S.m.p.s

Status
Not open for further replies.
Jump to Latest
#21
If anyone is interested in trying a SMPS design I can give them a few pointers and suggestions. I am a power design engineer. Feel free to drop an email. For the meantime I can suggest looking at Abraham Pressman's "Switching Power Supply Design" and Keith Billing's "Switchmode Power Supply Handbook" for starters. There are also a lot of good handbooks, papers, app-notes, and manuals on TI's analog section.

BeanZ

BeanZAudio@columbus.rr.com
 
#24
anyone else think this calls for a communal SMPS design?... I'm sure they'd be lots of people with the knowlage required to get it on its feet and you get feedback from everyones experiences with them to make a better design.
who's interested?
 
#26
Couple o' things: fr)st makes a good point about the ESP design- if you need more power, simply stack the cores. For these power levels, you will use the push-pull topology because either a single-ended forward- or a flyback converter will have too high of drain currents for the MOSFETs. This is easier if you're doing toroids, but toroids can be harder to wind. :bawling:

Remember to get the phasing right, or the converter will let the magic smoke out. :dead::flame:

Having said those two things, let's assume for a moment, that the auto industry is NOT :stop: going to a 42V bus voltage. (This has been talked about by the auto industry for some time, but all the OEMS are now backing away from it. Lack of business case)

For a 12-14V system, the MTB75N06E's (now NTB75N06E) from ON Semi are a good starting for the MOSFETs, use IRF540s for a 24-28V system since they have a Vbdss of 100V. Don't bother with NPN bipolars- they're just too complicated to drive. :whazzat: Depending on the controller you're using, N-Channel MOSFETs require a single series-gate resistor for the drive cktry.

The SG3525 is the first controller I've worked with, and with good results, EXCEPT when there is transformer imbalance, then the aforementioned MOSFET destruction will occur. This is a Voltage-Mode controller that can be made into current-mode by making a current-sense amp (made out of an op-amp across a series-sense resistor coming off the MOSFET sources to ground) and plugging its output into the SHUTDOWN pin (pin 10). This is a crude way of giving it pulse-by-pulse current-limiting.

Or, you can use the MC33025, a great current-mode controller, having a similar pinout to the 3525, with a much faster oscillator (2MHz, I think), and better gain-bandwidth on the error amp. This lends it to having better (read: tighter) control in the feedback loop. ON has a very good tutorial for this (and other) chips.

Speaking of feedback, I don't remember if I saw an opto-isolator in the ESP supply, but for DC isolation, this would be a good idea.:idea:

I know I'm running on and on, but someday, If I figure out how to scan one of my schematics into the computer, I will post one of them. Breakfast is on the table, so I have to go for now. Until next time........
 
#28
N-channel OT - Dayton

it's ex-WA8PZL, one year I forgot to renew. I have the complete Halli-Scratchers ready to go when I re-up my General.

my son and I are going to fly out to Dayton -- seems that this is the last year and it will be in Columbus or Indianapolis next year.
 
#29
2 Bad for Hara if it does go. :bawling: I run an Icom 706MKII-G thru an SGC-237 to a long wire. QRZ=good.

Presently I am prototyping three SMPSs:

The first is for a 120W micro-ATX box for my Book PC BKi-810 computer to run off 10-32VDC, using the MC33025 current-mode controller in a push-pull topology.

The second is for a multi-channel home theater amp. The power supply uses the MC33025 in an off-line, half-bridge topology. The amp circuit is based on Nelson Pass' and Norm Thagard's A75 (Audio Amateur 1-94), biased into high-bias AB 50W, nominal, into 8 ohms (100W into 4 ohms), with both RCA and XLR inputs for each channel.

The third is for an 80W solar-photovoltaic array using 2 LM2677S-ADJs, sychronized in parallel to put out 14.2V at 5A to charge batteries and run radios. (Great for QRP). :yes:

The 'YL and I are house-hunting, so it might be a while before I actually complete them all. However, if this website is still in existence, I will post pics and results. 😎
 
#30
I'm *Attemping* to make a high current version of the LM2677 by putting adding fets in there off the output of it. I know theres porbly a better option but for simplicities sake I can't be stuffed 🙂
As far as I know if I raise the voltage 5-10v of the controller above the rail i'm switching it should drive fets.
Do i need a gate resistor for safety sakes? Also I got no idea of the value if anyone feels like enlightening me 😕

Also, rod SMPS doesn't have feedback at all. Thiers two plans on the site and rods modified one (the ones I built) although in the artivle he says he built one thats been regulated down by the switching fets. It still works fine, there just no voltage control.
 
#31
fr0st-

The ESP power supply DOES use feedback, although it is NOT dc-isolated. BTW, what is your raw power source, and what is your load? 😕 Some type of power amp? 😕 If you are powering it off the car's 12V system and want anything more than 25Watts and don't want to use a BTL chip like the uPC1230H2 (by NEC), you will definitely need a boost-type of DC-DC converter, with +/- bi-polar outputs. Also, DC-isolation is also a must if you wish to minimize noise :shhh: from the noisy 12V bus in the car. :smash:

I would use the ESP power supply as a good starting point for any high-powered DC-DC converter. Using an LM2677 to drive higher powered MOSFETs is really not advisable. :bawling: Beside, this chip will not work if you want to BOOST the car's 12V to something higher.
 
#32
The PSU i'm building is a mains powered one for car audio testing.
The rail is about 25v no load (18v transformer). My idea is to raise the rail powering the lm2677 (seperate from the one being switched from the added fets) to get enough voltage to turn on the external fets (The lm2677 just switches the voltage on its Vin with an internal fets so your just getting a switched output on its Vout) .
If you have a look at figure 9 on the ESP article thats the one I used and its go no feedback from the transformers output to the controller.
 
#33
S.M.P.S.

OK. I'm confused: 😕 Let me try and get this right: You're using the ESP power supply for a project, and this BIG 25V mains-powered DC supply for car audio testing, like servicing amps, EQs, headunits and the like, right?

If this is the case, then you don't even need to mess with the LM2677S, as it is an ultra-high efficiency chip (95%+), which is good if you're powering from batteries. A much better (and easier) 🙂 route is to go with a PWM chip like the TL494 (ON semi, TI, others). Tie its collector pins together and to the gate of a BIG P-channel MOSFET, via a 5K resistor. The '494's emitter pins should go to ground. A good MOSFET to use is Harris' RFG60P05E. 🙂 This is a 60AMP, 50Volt FET, with an Rds(on) of only 26milliohms! 😱 Tie a few of these in parallel to lower the Rds(on). From the gate(s) of the MOSFET(s), take a 10Kand run it to +Vcc. Now, you won't need a bootstrap overvoltage to drive the MOSFET. A good frequency is 50 kHz.

If you really want to use N-Channels, in between the PWM chip (if you don't want to use the '494) you will need an International Rectifier IR2117 High-side switch driver. Hope this clears things up a bit.
 
#35
since I am not familiar with the terms but I am following this threa could someone please tell me what these are:

- on/offline SMPS
- feedback: how is it used in an SMPS
-what is DC isolation and how do you achieve it?
- what is core stacking?
- push-pull topology/single-ended: how do these relate to an SMPS? is it the same as in an amplifier? except at the higher, set frequency of the PSU?-
- what is a flyback converter? http://henry.fbe.fh-darmstadt.de/smps_e/spw_hilfe_e.html
how does this make the drain currents too high? because it relies on one FET to make the waveform whereas P-P uses two to share the load?
- what is the difference between a current-mode controller and a voltage-moe controller. Obviously the current/voltage difference, but what does this mean in a circuit? you are supplying current/voltage to switch the transistors?

Sorry for all the questions, but I really am interested in this and find it a little hard to search for information online when I don't know what to look for.
 
#37
If this is the case, then you don't even need to mess with the LM2677S, as it is an ultra-high efficiency chip (95%+), which is good if you're powering from batteries. A much better (and easier) route is to go with a PWM chip like the TL494 (ON semi, TI, others). Tie its collector pins together and to the gate of a BIG P-channel MOSFET, via a 5K resistor. The '494's emitter pins should go to ground. A good MOSFET to use is Harris' RFG60P05E. This is a 60AMP, 50Volt FET, with an Rds(on) of only 26milliohms! Tie a few of these in parallel to lower the Rds(on). From the gate(s) of the MOSFET(s), take a 10Kand run it to +Vcc. Now, you won't need a bootstrap overvoltage to drive the MOSFET. A good frequency is 50 kHz.
Click to expand...

What I was going for was a PSU that I could change the voltage of and still retain alot of efficientcy. The lm2677 looked good for its simplicity, I just needed to beef up the inductor/fet and diode ratings to suit my needs.
With P channel MOSFETs to switch it on does the Gate voltage need to be less than the source for it to turn on?
So by tieing the gate to the V+ through a resistor your dropping the voltage to turn on the gate and then using the TL494 you short the gate to ground uning the internal transisters?
This is oposed to N channels which need a higher voltage to turn on so it requires a high side driver?
Can I run a P fet off the Vout of the lm2677 and use the internal voltage drop of the fet (plus a resistor if neccesary) to turn it on?
Would this require me to switch the drain and source around as to present the fet with a negative voltage?

Thanks for your persistance 🙂 its much appreciated
 
#38
with the transformer on the ESP SMPS, how would you get more than one output of different votlages?
Since I have never wound a transformers, I am guessing you would have to add a whole new winding? is that right, or do you just tap off from halfway through the secondary winding to give you 1/2 the voltage for example?

Also, how does one calculate the wattage/VA for a transformer like this? do you just add all the wattages/VA's for each tapping up to give you the overall?

The other question I had is, what if I wanted a constant voltage rather than afluctuating one? I would have to use a regulator right? Which dissapates the excess as heat? Now I also have come to the understanding that a SMPS in the car will fluctuate depending if the car is on or not, as the battery decreases etc. Therefore would I make sure the SMPS' lowest output voltage is the minimum required by my device so that I never go beneath my requirement?


Also maybe a suggestion for the message board people, about having a power supply forum. I have looked all over the net and haven't found anything where people discuss it. The only one is eio.com and they seem to have people just basically asking for others to design things for them.
 
#39
shyfx said:
with the transformer on the ESP SMPS, how would you get more than one output of different votlages?
Since I have never wound a transformers, I am guessing you would have to add a whole new winding? is that right, or do you just tap off from halfway through the secondary winding to give you 1/2 the voltage for example?

Also, how does one calculate the wattage/VA for a transformer like this? do you just add all the wattages/VA's for each tapping up to give you the overall?
Click to expand...

You could either "stack" the secondaries (that is to add taps into a secondary for extra voltages), or you could wind a separate winding. If you go the stacking/tap route then you'll want to make sure that the shared parts of the winding can handle the full current.

To calculate the VA you'd add up the VA of each winding. For example, if you had a transformer with two secondaries, one 5V @ 5A, and another with taps at +/- 10V @ 1A and +/- 15V @ 0.25A, the total VA would be 5*5 + 20*1 + 30*0.25 = 52.5VA. The +/- 10V part of the winding would have to handle 1.25A.
 
#40
If I was designing an SMPS for use in a car audio amplifier I would use the push-pull topology (simple and low parts count), the SG3525 (simple and well-suited for driving push-pull), low Ron power MOSFETs, and primary-side feedback (just to prevent overvoltage on the secondary side). I'm a big fan of primary-side feedback in car amplifier SMPS because it requires no extra parts for good isolation.
 
Status
Not open for further replies.
Top Bottom