AC-fed SMPS topologies or the lack thereof

On a sunny day (Sun, 2 Sep 2018 09:48:02 +0200) it happened Piotr Wyderski wrote in :

Rectified 3 phase AC gives some percents ripple, then one does not need the storage capacitor, but just RF decoupling on the AC side.

>

For why?

If you want to make a switching buck equivalent of a VARIAC, you can. You need two bidirectional switches, driven with isolated gate drivers, in synchronous buck mode, plus filtering. The filtering is tricky because it will consume VARs at line frequency; you can't have a zero-current off state like you can at DC.

In that case, the output is AC, in phase with the input.

A single phase PFC isn't supplied with continuous power, but delivers DC. The input dips to zero periodically. A capacitor is required. That's obvious enough, but it was a point of confusion among a lot of participants in the Google Little Box challenge (which, it turned out, was about half inverter and half ripple management).

To do anything else, you need multiphase, at minimum two phase (90 degrees), of course three phase (120 degrees) is more traditional these days. To get DC output from that, you need four quadrant converters, with each converter's output rotating in phase with the mains, to form the difference frequency (0Hz = DC). Just as a DC-to-AC inverter (VFD, say) must deliver a rotating output vector. It's doable, but that doesn't mean you want to do it...

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/ 

"Piotr Wyderski"  wrote in message  
news:pmg4jk$hnf$1@node2.news.atman.pl... 
> Why there seem to be no AC-fed SMPS topologies? Google is surprisingly 
> silent about that. There are some attempts to bulid a bridgeless PFC, 
> but they still have the high-voltage electrolytic capacitor in the middle  
> and in most cases they do not address the inrush current issues 
> at all, as the PFC is a more or less typical buck, i.e. a series  
> connection of L->D->C. The lack of this storage capacitor together 
> with a reasonable phase factor would either require another storage 
> technology (a bulky inductor) or high output ripple tolerance. But 
> we were used to the latter, any iron core trafo-based PSU had to 
> contain a big output capacitor bank and sometimes a choke. 
> 
> I understand all this, if the directly AC-fed topologies were a niche, 
> I wouldn't ask this question. But they are not even a fringe and I find 
> this mysterious. So what's wrong with basically chopped sines? 
> 
> Best regards, Piotr 
> 
>

There are if you accept a BR on the input. Otherwise you'd need to duplicat e the control circuitry, and have it tolerate reverse voltage. A BR is chea per.

le

u t

ate

ts

Actually that's not the case. I've encountered numerous lighting SMPSUs tha t use no reservoir cap. Some output a hf modulated sine wave, some output d c with bits chopped out at each crossover. Good for batery chargers too. In principle one could even run a micro like that for some cost cutting apps, with it going to sleep during each chopped-out bit of power.

NT

The latest generation of "pure sine" stage lighting dimmers are basically just that: a synchronous buck convertor where both switch and "catch diode" are implemented by series back to back FETs. Effectively a Variac, input and output at 50/60Hz but internally at 40kHz+

piglet

^ ^ ^ ^ ^ ^

Those don't deliver (continuous, stable) DC...

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/

Robustness + higher efficiency as a side effect. The bridge itself can waste considerable amount of power, 2-5W in the mid power range. You don't have the inrush current/charging spikes related problems either. It is also cos(phi)=~1 at no cost. So the input stage is far superior to a traditional switcher. The "only" problem is what to do with the chopped sine afterwards.

Complexity comparable to, say, a 2T synchronous isolated forward. And now you have GaN parts without the body diode, which, in fact, is the reason I asked this question.

Or rectified AC, and then you basically end up with a high tech version of a normal transformer. IMHO useful enough not to be *that* rare. But the VARIAC analogy is a good one, thanks.

Sure.

But then it's cheating, as it is extremely easy to have continuous power source in the multiphase case. You can even use the wye/star trick to effectively double the already high ripple frequency. BTW, do they get rid of the big input capacitors in that case?

Best regards, Piotr

no, but it is dc.

NT

You can do an active FWB just as well, which is probably simpler to support with off the shelf controllers.

Direct conversion PFC is well known, I think the usual case is two transistors and two diodes. You can change the diodes to sync FETs quite easily. You'll still get inrush due to body diodes. A flyback or SEPIC configuration could be used to avoid inrush, or a buck-boost, or whatever else you like.

I don't see that any of these are "no cost" though...

Yeah, but that's a lie, sort of. They might not have a body diode as such, but they de facto have the same behavior. If Vgs(min) were opposite to Vds(max) you could use it like a JFET, but with the extremely stringent limit (0-6V usually) you can't get more than about -2V D-S. So it acts like a higher Vf schottky, or really, not far from an SiC schottky with SFA capacitance. So you don't have any choice but to use them back-to-back for true AC applications.

Don't know what "they" do, it's probably a matter of hold-up or ride-through time. The induction heaters I made, didn't bother, just a few 10s of uF to deal with switching ripple.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/

If you mean AC to DC, the now-classic PFC converter (boost switch into a big cap, than an isolated dc-dc switcher) is it. It has no inrush issue, and the line current is a sine. Any ac/dc converter needs energy storage, and a high-voltage electrolytic cap is the best way to do that.'

--

John Larkin   Highland Technology, Inc   trk 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

The ordinary three phase six pulse rectifier handles the ripple _voltage_ issue quite well, but look at the phase _currents_, they are quite ugly. Sooner or later, you are going to need some three phase PFC.

Going to 12 pulse (or even 18 pulse) rectifier reducing the ripple voltage even further and also cleans up the phase currents.

Unfortunately wye/delta feeds need heavy iron transformers at 50/60 Hz. If you are talking about significant power and start with a medium voltage (MV) feed, you are going to need a MV/LV transformer (say

20/0.4 kV) anyway, so get two MV/LV transformers, one connected in wye and the other in delta.

I've seen circuits that use 2 active chopper devices as the incoming diodes, so there are no separate diodes. I don't have a schematic in my collection though and haven't played with them. Energy efficiency is the goal of course.

NT

It does. If the reservoir cap is empty, the diode current is only limited by the small inductance of the choke.

Unfortunately.

Best regards, Piotr

triac lamp dimmer?

--
     ?

On 09/02/2018 11:06 AM, John Larkin wrote:

Look ma! No capaci-taters! And almost 50% efficient

Version 4 SHEET 1 2996 908 WIRE 800 -144 576 -144 WIRE 992 -144 800 -144 WIRE 1216 -144 1056 -144 WIRE 1312 -144 1216 -144 WIRE 1504 -144 1312 -144 WIRE 1744 -144 1600 -144 WIRE 2304 -144 2016 -144 WIRE 2592 -144 2304 -144 WIRE 2928 -144 2688 -144 WIRE 576 -64 576 -144 WIRE 1312 -64 1312 -144 WIRE 2016 0 2016 -144 WIRE 2384 0 2016 0 WIRE 2640 0 2640 -80 WIRE 2640 0 2464 0 WIRE 2304 16 2304 -144 WIRE 576 112 576 16 WIRE 992 112 576 112 WIRE 1216 112 1216 -144 WIRE 1216 112 1056 112 WIRE 1312 112 1312 16 WIRE 1552 112 1552 -80 WIRE 1552 112 1312 112 WIRE 1856 112 1552 112 WIRE 2192 112 1920 112 WIRE 2192 176 2192 112 WIRE 1744 224 1744 -144 WIRE 1840 224 1744 224 WIRE 2016 224 2016 0 WIRE 2016 224 1920 224 WIRE 2128 224 2016 224 WIRE 2640 240 2640 0 WIRE 2304 256 2304 80 WIRE 2928 272 2928 -144 WIRE 1840 304 1744 304 WIRE 2016 304 1920 304 WIRE 2240 304 2016 304 WIRE 576 384 576 112 WIRE 800 384 800 -144 WIRE 1552 432 1312 432 WIRE 1856 432 1552 432 WIRE 2304 432 2304 352 WIRE 2304 432 1920 432 WIRE 2192 544 2192 272 WIRE 1312 576 1312 432 WIRE 2016 640 2016 304 WIRE 2384 640 2016 640 WIRE 2640 640 2640 304 WIRE 2640 640 2464 640 WIRE 1552 688 1552 432 WIRE 2640 688 2640 640 WIRE 576 752 576 448 WIRE 800 752 800 448 WIRE 800 752 576 752 WIRE 1312 752 1312 656 WIRE 1312 752 800 752 WIRE 1504 752 1312 752 WIRE 1744 752 1744 304 WIRE 1744 752 1600 752 WIRE 2016 752 2016 640 WIRE 2192 752 2192 608 WIRE 2192 752 2016 752 WIRE 2592 752 2192 752 WIRE 2928 752 2928 352 WIRE 2928 752 2688 752 WIRE 800 816 800 752 FLAG 800 816 0 SYMBOL voltage 576 -80 R0 WINDOW 3 -231 53 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR Value SINE(0 170 60) SYMATTR InstName V1 SYMBOL diode 992 -128 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName D3 SYMATTR Value RRE07VS4S SYMBOL npn 1504 -80 R270 SYMATTR InstName Q1 SYMATTR Value 2SCR544P SYMBOL pnp 1504 688 M90 SYMATTR InstName Q2 SYMATTR Value 2SAR544P SYMBOL ind2 1936 208 R90 WINDOW 0 4 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName L1 SYMATTR Value 470mH SYMATTR SpiceLine Rser=5 SYMBOL ind2 1824 320 R270 WINDOW 0 32 54 VTop 2 WINDOW 3 4 56 VBottom 2 SYMATTR InstName L2 SYMATTR Value 470mH SYMATTR SpiceLine Rser=5 SYMBOL res 1296 560 R0 SYMATTR InstName R1 SYMATTR Value 470 SYMBOL res 1296 -80 R0 SYMATTR InstName R2 SYMATTR Value 470 SYMBOL zener 2320 80 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D5 SYMATTR Value 1N5375B SYMBOL zener 2208 608 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D6 SYMATTR Value 1N5375B SYMBOL res 2480 -16 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R4 SYMATTR Value 470 SYMBOL res 2480 624 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R5 SYMATTR Value 470 SYMBOL zener 2656 304 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D7 SYMATTR Value 1N5369B SYMBOL diode 992 128 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName D4 SYMATTR Value RRE07VS4S SYMBOL diode 816 448 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D1 SYMATTR Value RRE07VS4S SYMBOL diode 592 448 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D2 SYMATTR Value RRE07VS4S SYMBOL npn 2592 -80 R270 SYMATTR InstName Q3 SYMATTR Value 2SCR544P SYMBOL pnp 2592 688 M90 SYMATTR InstName Q4 SYMATTR Value 2SAR544P SYMBOL npn 2128 176 R0 SYMATTR InstName Q5 SYMATTR Value 2N3904 SYMBOL pnp 2240 352 M180 SYMATTR InstName Q6 SYMATTR Value 2N3906 SYMBOL schottky 1920 416 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName D11 SYMATTR Value PDS5100 SYMATTR Description Diode SYMATTR Type diode SYMBOL schottky 1856 128 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName D12 SYMATTR Value PDS5100 SYMATTR Description Diode SYMATTR Type diode SYMBOL res 2912 256 R0 SYMATTR InstName R3 SYMATTR Value 100 TEXT 344 24 Left 2 !.tran 1 TEXT 1584 264 Left 2 !K L1 L2 0.98

Yeah, let's see you

1. plot input current distortion
2. purchase that inductor :-)

Spamming inductors and zeners not a new idea:

Such deliberately strange approaches tend to be confined more to bespoke artisinal audiophoolery rather than anything properly commercial, of course.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/

It would be a weird fashion of coupled inductor that you probably couldn't buy off the shelf, but probably not large enough or expensive enough to impress an audiophool.

I'm definitely not cool enough to use 100H of inductance and 12 gas regulators :(

What if I wanted to mount an LED spotlight on a ceiling 65 feet high and didn't want to change it out that often/ever?

And its mutant cousin from Roswell, the cycloconverter. Not sure if they count as an SMPS, though.

Best regards, Piotr

They're not linear, so I'd count them as mains-synchronous switched-mode

--
     ?

Actually I don't see why a 1:1 power transformer connected as a differential choke wouldn't work. It's only about 20V RMS and a few tens of mA AC current across each inductor. would that plus the DC current be enough to saturate the ungapped core material? or would there be significant volt-second imbalance in a real circuit to walk it to saturation? don't know enough to know...