Is this possible?

Of course it's possible. Everything is possible >:->

Source and sink, or just sink?

--James

You didn't specify what the 1 volt is, input? output? etc.. but here is something

that will output starting at .8 volts but needs supply voltage of

=A0 =A0 ...Jim Thompson

ets/ADP2105_2106_...

For a 1v output that sources and sinks all Jim needs to do is make a buck switcher with a synchronous rectifier, i.e. a totem-pole output. If the output gets too high, PWM the lower switch and you've got your

-- Cheers, James Arthur

something

or just put a 200ma load R on it at all times :)

something

Right. A synchronous switcher is like a pair of gears. It will move power in any direction, in any quadrants. A lossless 1-volt "output" sink will pump energy *into* the 5-volt supply.

John

Modern DRAMs often need termination. Since they have a lot of signals, it's less power overall if you terminate to the target voltage and that requires a regulator that can sink as well as source.

There should be lots of data sheets out there.

I assume you don't need to make a switcher that can pump the power back up to 5V rather than just turn it into heat.

Simple. Any 1V suply chip, and just load it with 5 ohm, so it can sink 200ma and source (Ispecified-200ma).

Wierd. A CC 200mA load across the supply? Maybe switched in or out depending on whether the +1 supply is on or off?

Please post how you solve it, if the NDA allows. Inquiring minds want to know! :-)

Ed

something

It has to have a straight voltage controller, though, and there has to be somewhere to store the energy on the 'input' side. These don't naturally occur and will complicate start-up and protection regimes.

Easier if a load profile and accumulated energy limits are specified.

Interesting to see how efficiency is going to be specified.....

RL

_sheets/ADP2105_2106_...

The 5v supply just has to absorb 40mA, worst case, for an unspecified period, possibly brief. Jim could plop a 1v shunt regulator across the output of an ordinary 1v synchronous buck, but a) he specified 'switcher' and b) waste 200mW? Horrors!--think of the planet, please! ;-)

-- Cheers, James Arthur

=A0 =A0 ...Jim Thompson

Use a buck converter turned upside-down so its power switch (e.g. an N- channel MOSFET) is grounded, sinking current to a freewheel diode and inductor/load that return to the 5-volt supply. Use a ground- referenced controller and voltage reference.

-- Joe

something

It works at a fixed 20% duty cycle, no "controller" needed. That makes a 5:1 gear ratio. With good fets and an low-resistance inductor, that might be good enough for many uses.

and there has to

Other +5 loads might absorb enough power.

John

something

Aha! Good idea! Just need to implement synchronous flyback and "flyforward" :-) ...Jim Thompson

something

That Marx-Lenin inductorless thingy might work, too.

I would take a flyback convertor IC, and invert the voltage feedback iso to stabilize the input rather then the output. The feedback parameters may have to be tweaked for stability.

If the current is sourced from 5V, there will be no problem with absorption the return.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

=A0 =A0 =A0 ...Jim Thompson

sheets/ADP2105_2106_...

yeh, maxim has a peltier driver that has the peltier between the outputs of two synchronous buck converters, one sinking, one sourcing depending on whether cooling or heating is needed.

seen the pumping mentioned as a potential problem for halfbridge class- d amplifiers, an asymmetric input could pump up the voltage on one side of a dual supply

-Lasse

That doesn't say whether this switcher will work in two quadrants (current source plus current sink).

"Source" is typical verbage for a power supply spec. So Jim's customer either used bad terminology in the spec or they really want something to sink current. I understand the confusion. I'd look over the spec for an additional 'source' specification. If none is found, I'd go back to the customer for clarification.

Customer has switching from +5V, 3 devices with different forward drops, but a common top connection :-), and wants a +1V relative to ground for a low forward element ;-) to limit dissipation. ...Jim Thompson

s:

Linear Tech has a number of DDR2/3 switchers for termination supplies that can sink and source...

-- Marc

something

I have a circuit that Phil calls the "Groucho Marx Generator". But it would take a lot of fets and caps to do 5:1, and the fats would be too big to integrate on-chip, I'm thinking.

A synchronous buck converter would work fine, regulating down to +1 volt. It should work fine operating backwards.

Right. Conservation of energy.

John