SMPS, control loop?

Hello Tim,

Thou shalt not provide too small of a load side capacitor. So don't play uncle Scrooge with that cap ;-)

In some constellations (small cap, >50% duty) this can result in stability issues. You can read up on that under the topic "right half plane zero" or "RHP".

Current mode control prevents the inductor current to reach "kablouie" levels. I usually employ current mode in my (often discrete) designs.

Regards, Joerg

Presumably what gives is that the power supply is designed to be cheap and adequate, even with the load-side ripple that may happen when you change load current. I suspect that if you had to charge $5 more for an AT supply with current control you wouldn't make your money back, unless you could somehow convince the earring-in-the-navel set that having 'current mode' power supplies could somehow get them laid.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

As usual, Joerg is spot on. Jerald Foults' website (spelling?) has a whole bunch more useful stuff. (P)CMC is the way to go. If only the UC384x was developed before the poxy SG3524 or TL494....

NB: real men dont use smps chips ;)

As long as the peak energy stored in the output cap(s) is >> that in the choke, the overshoot is small. If electrolytic caps are used and ESR is minimised, capacitance tends to be very high (multiple large cans are often required for low ESR, resulting in huge capacitance), maximising capacitive energy storage. Likewise if inductor size is minimised (eg

10-20% ripple) then so is inductor energy. Overall, overshoot is minimised.

Conversely, a high inductance, low capacitance design has the potential for significant overshoot, which must be dealt with somehow. and lots of inductance means lots of weight, cost and volume.

PC supplies are cheap, so inductive components are kept as small as possible, helping reduce the severity of this problem. The output inductors are coupled, so the high-voltage capacitance has a proportionately larger effect than the low-voltage capacitance.

That being said, I have not measured the step recovery performance of a cheap PC supply. I have measured some cheap OEM supplies, and found responses ranging from OK to unacceptable.

IMO if the smps cant run continuously into a dead short, its not good enough. A brutal test is to attach a pair of metal rods to an output, and intermittently short them with a large, coarse file. If the smps can handle that without evil overshoot (the filter inductor current ramps up to the current limit value, even with (P)CMC) and/or explosions, then its a good design.

thats actually pretty easy with a single output PCMC smps, but multiple-output supplies PCMC tend to have problems, as whichever output is shorted tends to get all the power...

Cheers Terry

Mmmm, discrete.

Meh, I'm short on transistors and don't care to use that much breadboard space, so I'd just as well use an LM393, 339, or better yet, a 3524 which has everything rolled together for a free-running PWM bridge generator.

Either way, you need a PWM generator, and a 3524 has two op-amps inside it for control, not inconvienient (though the 7500 is more universal as the amps are identical).

???

Personally I don't see an UC384x working for half bridge...

Yeah, Ec might be 8 * E_L, but it's still an overshoot of about a volt, bare minimum, and far more if the circuit works theoretically (i.e., doesn't explode or saturate from the excessive current) with delay dominated by the filter.

Yeah, I want to make a bench supply for electrolysis, think 4 to 6V and

50-100A max. Short proof operation would definetly be a nice addition. Regulation isn't critical as far as transients, but I can't have this thing exploding every time the load changes quickly, either.

Tim

--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

"Barton has a lengthy section on rectifier-capacitor analysis, replete with a plethora of evil equations. Hell, a paper even appeared in IEEE Industry applications just last year."

and:

"Michael Gaspari wrote a great paper ( "IEEEXplore only returns 12 papers with a Gaspari as an author, and only one of the was by an M. Gaspari; that paper was "An open framework for cooperative problem solving". I don't think that's the one you mean. Can you give a more detailed reference? I looked through the 2004 Industry Applications and didn't see a paper about rectifier-capacitor analysis; can you give a more detailed reference for that one as well?"

But you vanished from the group for a while, and apparently didn't see my request. Can you provide the references? Thanks.

Hello Tim,

ROFL! There is a grain of truth in that. Oh, and real man drive stick shift cars and never use anything other than charcoal or, better, firewood for the barbie.

Seriously, I'd make my life easy, swallow the pride and buy a 6V golf cart battery. They are cheap. You just have to make sure that it's gases can go somewhere safe. If you have to lop off a little you could rig a ladder of monster diodes on heat sinks.

Regards, Joerg

What gives is that you can't analyze a closed-loop control system by qualitative thinking. The usual line of reasoning, going around the loop one element at a time, generally predicts that any closed-loop system will oscillate.

You see weird closed-loop explanations, and wildly unlikely extrapolations, in the press all the time.

John

Unless you have to drive one in rush hour. I used to be stick-shift-only, but haven't had one since 1987.

I now prefer 6 passenger tanks ;-)

Of course my oldest daughter's Mazda MX-5 is in my garage today, because I put her on a plane to Ciudad Obregon yesterday, for a political conference.

[snip]

Naaah! LPG + ceramic briquettes. Just turn the knob and press the button ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

"Barton" refers to:

Thomas H. Barton, "Rectifiers, Cycloconverters, and AC Controllers," Oxford University Press, USA (June 1, 1994), ISBN: 0198561636

thats cos I remembered his last name wrong :(

I meant to "say" :

Michael L. Gasperi, "Life Prediction Modelling of Bus Capacitors in AC Variable-Frequency Drives," IEEE trans. Industry Applications Nov/Dec

2005 Vol. 41 No. 6 pp. 1430-1435

I was in overseas for 19 days :)

back now.

HTH

Cheers Terry

Peak Current-Mode Control

as opposed to, say Average CMC

works beautifully. and also has almost perfect line regulation on a pulse-by-pulse basis.

depends of course what voltage and how much capacitance. A capacitor on the +12V supply "looks" about 6x bigger when reflected through to the

+5V supply.

yes. in practice though its pretty easy to design the filter network to work well, which usually includes modelling inductor saturation and overload conditions.

adding an RC to critically damp the filter network can also help. Some applications may even require a clamp to limit overshoot. A lot depends on the FMECA - if a $1 clamp network can prevent snotting a $8,000 board, its probably a smart investment

you cant go wrong with current-mode control (apart from needing slope compensation for duty cycles > 50%). A good (P)CMC should be able to operate continuously with the transformer secondary shorted - hows that for brutal.

Many CMC chips now have a two-level current comparator, one for normal operation, and a higher threshold comparator for over-current tripping, which can shut trigger a shutdown-restart cycle.

So if say the transformer primary shorts out, the primary inductance plummets, so the rate of rise of primary current is much higher. During the inevitable delay from the current sense comparator to actually pulling Vg below Vth, primary current continues to rise, tripping the

2nd comparator.

you really cant do much better than control the peak current in your switch. Other than run a thermal model of junction temperature, and dynamically limit Tj and dTj/dt :)

Cheers Terry

I'd appreciate your comments on my question under the subject "switching power supplies in amplifiers", if you have time/interest.

Clifford Heath.

I've got one that is just a fat 60Hz transformer and a ~300A stud-mount diode.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Gee...

I must say Jim, I've just lost a bit of respect for you.

Tim (grills with oak logs, baby)

--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Yeah, but I can see that, if the phase shift isn't absurd, the gain will be corrective and, though it will oscillate, it will correct it in some time. Signals tend to go around the loop, so I don't see why such iterative reasoning wouldn't work, as long as you remember any interactions between elements.

I see absolutely no way it can correct something before it happens. Damped filters are currently a mystery. I can only see current mode working, which requires a current transformer from the primary, or some wildly low inductance resistor on the secondary.

And for regulation, you can't infer voltage from current, because the load isn't necessarily a constant resistor. Somehow, you'll have to adjust voltage, presumably faster than the filter delay, to avoid ugly transient response, but the only thing that can react momentarily is current... see what I mean?

I'll read more on your site later, Genome. It looks informative, and much more coherent than a dangerous number of your Usenet posts. ;-)

Tim

--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Without a time machine this is obviously the case. Although I use electricity for barbecueing I often use non-SMPS chips in my SMPS designs (typically I use a HCT123 and an opamp, somthing like half an LMC6482 or similar). Whether we will call the feedback voltage or current, under the score we are interested in the output voltage. My approach - which seems to be pretty unique as I have not seen it elsewhere - is to do voltage feedback, however a non-linear one; high gain for small changes and lower (sometimes much lower) for larger changes. A diode network in the feedback path of the opamp can do miracles... If combined with some filtering - so frequency response also gets amplitude dependent - there is usually really little left to wish. Oh yes, overcurrent protection etc. is included, if there is interest I might convert some of my schematics to .pdf (they are now available only as hpgl on my website).

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

Tim Williams wrote:

Hi

I would surely be interested to see the schematics :-)

Regards

Klaus

I began doing it, but I discovered some (especially those of interest) hpgl files were corrupted (may have been so for years...). I'll recover them within 2-3 days and convert to pdf, then I'll post a note.

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

Klaus Kragelund wrote:

JT got it right! Charcoal really sucks, got to wait too long for the oily smoke to burn off. Firewood is for smoking a piece of meat into jerky. Real men eat rare meat, cooked on a very hot fire for a short period of time. It's more about the drinking and socializing without hauling wood. Cheers, Harry

I have one of those "barrel" smokers, about six feet long. A real royal pain in the ass to keep a slow enough fire going to smoke without over-cooking.

I'm going to add a temperature-controlled LPG burner to heat the chips in a water bath. Then I can smoke over night without having to stay up and tend the fire ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.