So, what would _you_ use if you needed to have moderately fast (< 1ms settling time) linear pass devices that needed to deal with several tens of volts and several hundreds of amps?
Linear MOSFET? Bipolar? Tell the boss that a switcher really is appropriate, even if the speed is a challenge?
Something else*?
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
More info needed...
Is the load inductive?
Is "moderately fast", the turn-on time, or the turn-off time, or both?
Also depends on what the definition of "several" is ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Meaning your pass element(s) might be disspiating something near 1 kW? Sounds like the big problem will be getting rid of the heat and not what device you choose.
Can you do something to switch the supply voltage? I'm thinking of the relays on different transformer taps used in linear power supplies. Though you get glitches when it swtiches. Maybe some sort of switching power supply to coarsely set the supply voltage and then a linear pass element.
George H.
Perhaps that is on the right track...
Switching supply generates VDD into a big-ass capacitor
Then "pass device" is simply an ON/OFF switch? ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Using a linear mode for this purpose would be ridiculous. Few hundred amps is no problem; a dozen or so of good MOSFETs in parallel will handle that in the switched mode.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
Class-D style output with a transformer?
-- Failure does not prove something is impossible, failure simply indicates you are not using the right tools... nico@nctdevpuntnl (punt=.) --------------------------------------------------------------
I _did_ just sign an NDA, so I'm trying to filter things.
Take several as being 'around five' and go from there.
Noninductive load, or at least only parasitic inductance from whatever wiring it takes to get from the honkin' big array of honkin' big transistors to the load.
Turn-on and turn-off both need to be about the same speed.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Oops -- it's pulsed, for several (there's that word again) ms at a time. So the transistors themselves need to be fairly stout, but the average heat dissipation is low.
Well, lower.
(several tens of amps)(several hundreds of volts) = tens of kW.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
That's where I'm going, actually. I'm just working on arguments (blandishments?) to gently lead the customer in that direction. If you could get the switching times up into several* tens of kHz, or even
100's of kHz, then I think you could get the response that you needed, possibly at the cost of a really fancy controller.In fact, I'm thinking that even if the controller had to be a DSP and some fast ADC resources, you'd still be ahead of the game compared to linear pass devices.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Been there, done that:
Catching the flyback transients is a big problem at >100A of current. Multiple-leg switching is the way to go.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
That'd be my "switcher" option, yes.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Feedback loop cutoff frequency ~ (PWM_frequency*number_of_legs)/8
The right way to do that would be a multiple leg switcher controlled by something like TMS 28xx. The output capacitor should be able to handle the ripple current of ~ output_current/number_of_legs; that could take some thinking.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
Any physics reason for that, or just a rule of thumb that you've worked up or found?
Assuming a fast enough processor, and knowledge of the load, it seems you'd be able to make the controller nearly deadbeat. With work.
(Not that it matters in this case -- I think your above rule of thumb would provide _plenty_ of bandwidth).
That makes way lots of sense.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
If pulsed and at low duty cycle that would be my route. Simple, cheap, no EMC issues. But watch the SOA ratings and stay awya from the limits.
Nah, too much base current needed.
If the duty cycle and thus dissipation is too high then a switcher is the ticket.
Depends on the application. If this is some sort of charge/load balancer or selector look at charge controllers from TI and similar companies.
The filament current alone would make a big transformer emit a TUNGGG sound when turned on :-)
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
Tens of volts at large amps? sounds like a metal heater or some kind?
I'd use a phase firing system at high freq through a toroidal transformer with a low ratio output and large cross section wire on the secondary, like copper tape wire etc..
If that output needs to be DC, then use some fast rectifiers and low ESR caps.
Both with feed back regulation of course.
Jamie
A linear approach requires current to be equally shared between the devices. That complicates the things somewhat.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
In this case it's not so bad because it doesn't have to be fast. Shunt resistors and opamps are cheap. In fact, the resistors could just be meander traces because only the ratios matter, the absolute value doesn't.
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
Hello Tim,
With the current info, it can be both. When it is just for: current limiting (followed by shut down), inrush current limitation (where normally the voltage drop across the pass device is zero) and other transient dissipation things, I would go for the linear solution. I made such circuits for hot-swap things, but at low power levels.
When there is prolonged dissipation in the pass device, I would go for some first order hysteresis like switching scheme (based on current).
Can there be a very large capacitor between your device and the load? If so, your circuit may not "experience" current transients caused by the load. Or is current drawn from a capacitor seen as current that has to be limited?
In case of tens of volts, I would go for mosfet rather than bipolar (just because of drive requirements)
With kind regards,
Wim PA3DJS
ms
ns
s esomething like this:
-Lasse
1 ms isn't fast for a switcher. People build class D switchmode audio amps and fast servos. A switcher will sure save weight/size/cost/cooling.
John
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.