working on PSU, need to drop voltage but keep current same.

[Svart]

I am looking at many ways of doing this, but would like some suggestions on what other people would do in this situation, hopefully sparking new ideas for me.

I am attempting to use as few parts as possible, possibly even staying linear, but also trying to keep the heat down. As ohm's law constrains us, it's really looking pretty hard to do with what i have to work with..

I am attempting to drop 150VDC to 36vdc and feeding a simple switching PSU that I designed or dropping down to 24vdc for direct linear use.. both applications demand 1 amp of current with only around 1-2 amps available for the input of the PSU..

All suggestions welcome! I'll also be doing more sims with the ideas i have.

thanks!

 

[Svart]

hmm everything keeps coming back to using a trafo.. the very thing I am trying to avoid..

 

[bcarso]

Any reason why you wouldn't just do a buck converter from 150 to 24?

 

[Svart]

Bcarso, that's one of the various designs I am toying with right now. Parts count is one things I am trying to keep low and of course any odd magnetics are out of the question. I'm also looking hard at a version of Cuk converter too..

i was hoping that someone had a few opinions on a few of the topologies and maybe a few pointers and or suggestions of devices that folks like to use.

:thumb:

 

[PRR]

> drop voltage but keep current same.

Series.

> drop 150VDC to 36vdc and feeding a simple switching PSU that I designed or dropping down to 24vdc for direct linear use.. both applications demand 1 amp of current

Any linear trick will have to waste (150V-36V)*1A= 114 WATTS. This is a non-trivial heat load for a rack.

Since you have switcher skills, chop your 150V directly to 24V. At 75% efficiency, loss is only 8 Watts, a lot better than 114 Watts.

That does imply an odd choke. I'm no good at switcher values, but I suspect a very high inductance, higher than is generally stocked for low-volt work.

The standard PC power supply chops 320VDC (from 120/230VAC) down to 12V and 5V through a switcher transformer. What do you have to do to the primary to adapt it to 150V working? Reduce the turns? Maybe not even that if your load is much-less-than the original rating? Then you could probably take the windings for +12V and -12V and get your 24V. The -12V is only rated 0.5A, but if you are taking no load on the 5V and 3V windings, heat is not an issue. Drop would be high, but you will IAC move the voltage sensing from the 5V or 3V winding to the 12V line(s), and total load is light, so drop should not be large. You might melt a few parts before you get it figured out, but old (especially AT-type) power supplies are a glut, and new supplies are very cheap (try NewEgg and Jameco).

 

[Svart]

to clarify the type of buck converter I was looking at, I'm focused on a synchronous rectifier type, possibly with additional schottky's as flywheels. I'm not too concerned with efficiency, but the output must be clean as it will be supplying video equipment and digital circuitry.

"That does imply an odd choke. I'm no good at switcher values, but I suspect a very high inductance, higher than is generally stocked for low-volt work"

yeah I came to a similar conclusion.. :sad:

"Since you have switcher skills"

I wouldn't call them "skills".. rather, necessity to keep my job! but I don't think of myself as good by any respect.. just trying to take my time and get the opinions of those better than me and maybe learn a thing or two..

the heat is a problem as this will also be in the same box as a high voltage motor control and high current lamp contoller, both with the ability to give off a max of 750watts.. yup enough to heat a small room.. fans and heatsinks abound but those are worse case scenarios.. ambient in the box should sit around 150w under normal conditions.

The 150vdc is the output of the main motor control PSU. I am attempting to take what i need from here since this is a 670 VA toroid trafo, meaning HUGE, and not leaving much room for more trafos. but more importantly, there is no reason not to try to leave out more parts if i can do it.. :thumb:

thanks for the opinions! :guinness:

 

[PRR]

If you are not going into production, or into orbit, the sane way to get on with your life is to jam a 24VAC 50VA transformer in there somewhere.

A slightly less sane trick: you have a honker torroid. Get some #18 magnet wire. Wind it around the torroid. 100 to 200 turns should get you near 12VAC, which you can voltage-double to over 30VDC, and then regulate.

 

[PRR]

> I'm focused on a synchronous rectifier type

Why?

I thought the idea of a sync rect was to reduce the ~0.6V-1V drop in diodes. This is killer when you are going for 5V or 3V, but a non-issue at 24V.

 

[bcarso]

Yeah what he said...sync rectification is essential for super low voltage stuff, but one of those cool SiC schottkies, or just a UF1000ish part would be happy as a clam as a catch diode in a standard buck converter with high volts going in.

Cuk entails custom magnetics I believe, unless you got really lucky. It does have some attractive properties w.r.t. ripple but too much trouble imo for this app.

With the consdieration of video and audio in the vicinity, I would do something with a sync-able fixed frequency oscillator and vary the pulse width. Derive some sync info from the other circuitry so that the noise is stationary. I hope your current drain is at least X, where X is maybe 10% or so of max., o/w you tend to get into the right-half-plane zero territory and have to make the loop reallllly slow, or do some other trick like little burst-mode stuff, which tends to create its own interference (a TI regulator comtroller did this at about 8 kHz which produced a very audible whine in the audio of an early proto version of a portable product).

You could also try resonant mode at these voltages---more parts but quieter and less stress on the pass element. Unfortunately res mode and fixed freq don't mix easily, but maybe it would be quiet enough that sync would not be required.

 

[Svart]

excellent Bcarso and PRR, simply excellent.

I have only been looking at most of these designs and ruling them out one by one. This design is going to be in a production device, unfortunatly not for my DIY or audio. :?

I do understand that a decent catch diode will be sufficient but I am also planning to use this design again for 5v supplies for logic in the robotic system it's going in, so i am indeed looking for the most flexible design i can so that I can at least try to perfect it as time goes on! (if that's possible.. :shock: )

Yeah on further investigation the cuk won't be ideal even though some of it's attributes were nice. ah but there are more fish in the sea.. even though these taste like silicon..

I'm going to be doing some funky things trying to keep switching noise out of the system, I'll explain more later if i can get the things to work.. :wink:

I guess i should explain a little about the devices that I am designing for since that might spark some more opinions.. These are robotic cameras that simply go where people can't, like sewers, pipes, underground, nuclear powerplants.. etc. they are tethered by ~1k ft of multiconductor cable, video is over coax(yes i hear the groan and i share the feeling..). the camera has panning, rotating, focusing, zooming, iris and other options. the camera control signals are sent digitally over the coax during the retrace/blanking period.

:thumb:

 

[NewYorkDave]

Did you look into triax operation? That's how all those functions are handled in broadcast cameras nowadays. I don't miss the bulky old "multicore" at all!

Triax is thin (by comparison), flexible and inexpensive. We installed our triax cameras (Sony DXC-D35 with CCU-TX7 CCUs) a few years ago and I'm still impressed by the number of signals (in both directions) and degree of communication and control the system provides over one piece of cable.

The CCUs (camera control units) provide a single DC voltage (12V, I believe) over the triax and DC-DC converters within the cameras provide all the necessary internal operating voltages.

 

[CJ]

How about a 100 watt light-bulb? :idea:




That's high tech on my side of the tracks. :razz:

 

[Svart]

Hey NYD, the control is digitized through our tx/rx setup in a proprietary manner and then sent over the coax where it's picked back off the video, processed and sent to the camera via VISCA over a 232 port. the other conductors are for lighting and tractor(the part that actually carries the camera down the pipe) control. I'm in the process of replacing around 200 lbs and 15 rack spaces of control equipment with a nice little 2-3 rackspace unit capable of controlling the whole mess..

:guinness:

and CJ, that is so low tech it's simply GENIUS!