Forums

Need the topology for a current to current SMPS, not voltage to voltage SMPS

Started by RobertMacy October 16, 2013
Jim Thompson wrote:
> On Thu, 17 Oct 2013 08:04:25 -0700, RobertMacy > <robert.a.macy@gmail.com> wrote: > >> On Thu, 17 Oct 2013 07:55:22 -0700, Jim Thompson >> <To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote: >> >>>> ...snip... >>> I'm not quite understanding your requirement. You can send me details >>> in private if you like. >>> >>> ...Jim Thompson >> Thank you for your kind offer. I am being vague for two reasons: >> 1. purposely obfuscating the actual application/techniques under cloak of >> NDA. >> 2. inadvertantly floundering around, albeit embarrassingly, but in my >> defence, this is NEW territory here. > > BTDT ;-) MANY of the projects I take on are new territory... that's > what makes them fun. >
As the president of a client said: If, immediately after accepting an assignment, you get a serious knot in your stomach it promises to be a really fulfilling kind of work. Or as John Wayne sad it, "Courage is being scared to death - and saddling up anyway". -- Regards, Joerg http://www.analogconsultants.com/
RobertMacy wrote:
> On Wed, 16 Oct 2013 11:53:49 -0700, George Herold > <gherold@teachspin.com> wrote: > > .....snip... >> I scribbled this at lunch... which I think works as a current buck. >> https://www.dropbox.com/s/h2a8dyz0vzoovev/DSCF0031.JPG >> >> (Fun question) >> George H. > > Thanks for the schematic, I see the philosophy in it. > > I neglected to mention the MOST important aspect, the input is AC, not > DC. There is a potential full compliance could 'pop' to over 1kV, which > is obviously destructive. But, with a dead short voltage is not much, > because current is not much. Thus, my question here.
1kV doesn't have to be destructive, they make FETs for well above that. AC presents a challenge but not an impossible one. Got to bridge-rectify that and your circuit must withstand a "brown-out" every 8.3msec. -- Regards, Joerg http://www.analogconsultants.com/
RobertMacy wrote:
> On Thu, 17 Oct 2013 06:01:01 -0700, George Herold > <gherold@teachspin.com> wrote: > >>> ...snip... >> OK... We haven't done many AC circuit yet.. but I'm guessing the same >> sort of thing can be done. >> For completeness here's the way a current source buck-boost should look. >> >> https://www.dropbox.com/s/eey79dkci1sqpp0/DSCF0034.JPG >> (I figured out in the shower that the capacitor was in wrong.) >> >> George H. > > Thanks again for the images, surprisingly this time dropbox let me view > them - usually doesn't. > > IN: AC with source current maybe 30-50mA, compliance to 1kV > OUT: approx 5Vdc, with variable current, adjusted with time envelope > from 0.8A down to 50uA. > > with source current that low, can't just feed it straight thru. > > assuming NO losses... > with input compliance at 80V and using all 50mA, supply the 5V at 0.8A > You see why I first said current to current conversion? > > After major floundering around [and rather embarrassingly] in front of > all of you; I am starting to perceive the final topology as a > multi-step: AC to DC at power required, then DC to DC at current required. > But even 90% at each step yields a terrible 80% over all, so I was > trying for a single step: AC to DC at current required. >
That's what they invented flybacks for. Huge range, single step. But you need to employ a recuperating winding to get above your 90%. The magnetics design also promises to be "interesting".
> I can always flip the diode to make it a positive supply. That does look > like a 'single' step. > > Tim's right about paying attention to the RFI/EMI kicking back into that > source. Any parasitics and !!! It's going to be fun trying to filter out > that 50V ON/OFF appearing across the MOSFET terminals let out into the > world through that inductor at 1MHz.
If you need to go to 1kV compliance I would not operate at 1MHz. -- Regards, Joerg http://www.analogconsultants.com/
George Herold wrote:
> On Thursday, October 17, 2013 10:58:24 AM UTC-4, Robert Macy wrote: >> On Wed, 16 Oct 2013 09:43:03 -0700, Jim Thompson >> >> <To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote: >> >> >> >>>> ...snip... >>> So SMPS SHUNT regulator ?:-} ...Jim Thompson >> Alas, the compliance current is around 50mA with a compliance >> voltage as high as 1kV, which is destructive. A 'straight' shunt >> regulator either shorts out the incoming 50 mA or let's it pass >> thru, but in this case I sometimes need 800mA at a small 5Vdc. So, >> need some 'multiplication' of the current - a capacitor. There is >> sufficient power if the system operates with the compliance at >> around 80V at 50mA just have to somehow convert that to 5Vdc at >> 800mA. But, in this case it will be super necessary to monitor >> compliance voltage so it won't EVER go above 100V in order to NOT >> kill parts. In this supply that protection circuit, or 'voltage >> limit', is the corollary of the standard 'current limit' in a >> voltage supply! >> >> So far it appears a variation of George Herold's approach may get >> me there. > > Whoa! Hey, a big word of caution Robert. I've not built a spms in > my life. ...
But now, if it don't work right, Robert could claim "But George Harrold said it would!" :-) I'm taking this web course "Fundamentals of power
> electronics" by Robert Erickson. (Colorado U.) I've done two weeks > of lectures and HW. So your question sparked the theoretical part of > my brain. Which is great for me. But for a practical circuit, I'd > feel a lot better if one of the smps guru's signed off on it. > Putting in your numbers, you're hoping for a current gain of ~16 at > greater than 80% efficiency. The few curves I've looked at for > voltage smps show that losses start to really bite into the > efficiency with higher gains. So you'll need to put in some real > numbers. (We've only started serious loss calculations this week.. so > I'll know more after the weekend :^)* Maybe a combo of transformer > followed by current smps would work? > > Anyway good luck... and not to worry about floundering around on SED. > That's what's great about it. Throw out some crazy idea and see what > sticks. >
Or get into a hot discussion about Obamacare :-) -- Regards, Joerg http://www.analogconsultants.com/
On Thu, 17 Oct 2013 09:51:55 -0700, Joerg <invalid@invalid.invalid>
wrote:

> > >1kV doesn't have to be destructive, they make FETs for well above that. >AC presents a challenge but not an impossible one. Got to bridge-rectify >that and your circuit must withstand a "brown-out" every 8.3msec. > >-- >Regards, Joerg > >http://www.analogconsultants.com/
Did he say 60Hz? I could imagine RF here. The diodes only have to withstand the SMPS input voltage + one diode drop if the bridge rectifier output is clamped.
Spehro Pefhany wrote:
> On Thu, 17 Oct 2013 09:51:55 -0700, Joerg <invalid@invalid.invalid> > wrote: > >> >> 1kV doesn't have to be destructive, they make FETs for well above that. >> AC presents a challenge but not an impossible one. Got to bridge-rectify >> that and your circuit must withstand a "brown-out" every 8.3msec. >> >> -- >> Regards, Joerg >> >> http://www.analogconsultants.com/ > > Did he say 60Hz? I could imagine RF here. >
Well, that's the difficulty here, Robert can't divulge too much so all we can do is take potshots. [...] -- Regards, Joerg http://www.analogconsultants.com/
On Thu, 17 Oct 2013 09:03:20 -0700, Spehro Pefhany  
<speffSNIP@interlogdotyou.knowwhat> wrote:

>> ...snip.... > Your monitor could just be a 5W zener diode across the input (or a > small zener + BJT). > >> >> So far it appears a variation of George Herold's approach may get me >> there. the current charges a cap, which then supplies large amount of >> power. Aain, corollary, using cap instead of inductor for the energy >> storage/conversion. > > What the heck source are you working with, if you can talk about it? >
No, a simple shunt will NOT work. Remember, 50 mA will NOT provide the 800 mA. Only get that if the voltage coming in is allowed tor rise a bit. Sorry, can't talk too much.
On Thu, 17 Oct 2013 09:26:47 -0700, George Herold <gherold@teachspin.com>  
wrote:

>> ...snip... > Whoa! Hey, a big word of caution Robert. I've not built a spms in my > life. I'm taking this web course "Fundamentals of power electronics" by > Robert Erickson. (Colorado U.) I've done two weeks of lectures and HW. > So your question sparked the theoretical part of my brain. Which is > great for me. But for a practical circuit, I'd feel a lot better if one > of the smps guru's signed off on it. Putting in your numbers, you're > hoping for a current gain of ~16 at greater than 80% efficiency. The > few curves I've looked at for voltage smps show that losses start to > really bite into the efficiency with higher gains. So you'll need to > put in some real numbers. > (We've only started serious loss calculations this week.. so I'll know > more after the weekend :^)* > Maybe a combo of transformer followed by current smps would work?
Doesn't matter about the 'source' of concepts, what matters is that you 'triggered' inside my brain an approach that almost in a single step does what I want. ....with a few changes.
On Thu, 17 Oct 2013 10:38:28 -0700, Joerg <invalid@invalid.invalid> wrote:

>>> ....snip... > Or get into a hot discussion about Obamacare :-) >
grrrrr.
On Thu, 17 Oct 2013 11:08:49 -0700, Joerg <invalid@invalid.invalid> wrote:

> Spehro Pefhany wrote: >> ..snip... >> Did he say 60Hz? I could imagine RF here. >> > > Well, that's the difficulty here, Robert can't divulge too much so all > we can do is take potshots. >
Actually, go ahead and try at 60Hz.