The current draw from a device I am using increases over time. So at time=zero I expect the output to be DC (or close to it) and over time the pulse width to narrow to maintain a similar RMS current. I am thinking of using something like a LM2524 driving a power FET. I have some questions:

1.From the data sheet I noticed a duty cycle of approx 0-50% - on startup I would require about 100% - presumably I would have to invert the output ? – any suggestions?
2.The current sense appears to be about 200mv and they state the duty cycle drops to about 25% - so at what sense voltage would it start to reduce the current i.e. what is the range between no limiting to maximum.
3.If I take the approach of inverting the output to give me close to DC at time zero - then that will mess up any current limiting because the duty cycle of the device would be close to zero ?

Any advice would be appreciated - even the suggestion of a more suitable component.

 

If the current increases over time, the startup current should be somewhat lower - implying a low duty cycle drive from the regulator.

Inverting what output?

With no values given, anything from us will be no better than guessing. Please post up a schematic if you can.

 

Thanks for the response, I don’t have a schematic yet – as I am just looking at the specifications for PWM with current limit capabilities.
My reference to inversion was prompted by the specification stating a 0 – 50% duty cycle, so with the duty cycle set to 0 the power transistor would need to be on – as the unit warms up the current demand increases along with the duty cycle to maintain a constant RMS current.
If I used a PWM without any modification at start up I would have to run at 50% duty cycle and it would reduce when the current went over the required limit, according to the spec sheet the output duty cycle drops to about t 25% which is fine. My concern is starting with a duty cycle of 50% - it basically means the current would have to be double to start with – taking the startup up current from 10-15Ams up to 20-30Amps – that is the problem I am facing. I hope I have explained the problem clearly enough for you, I have not built any circuits for about 20 years – so excuse me for being a bit slow!

Data Sheet
http://pdf1.alldatasheet.com/datasheet-pdf/view/8735/NSC/LM2524.html

 

I envision your load as a 250 W fluid heater ,2.4 ohm negative temperature coef., base current of 10 A. @ 24 V.dc. A power FET With a .02 ohm current sensing resistor Source to gnd. Re vol 6, ch.6/9.PWM power controller 555ramp generator. Remove one op amp&connect FET source to - input. connect slider of 1 k pot to + input., supply pot with 1 V.,groundbottom side. Disconnect top of 10 k pot & connect to op amp output. At power turn on adjust 1 k pot for 10 A. First maybe replace the diodes with a small resistor[ 4.7 k] to gate of FET. Sense V will be .2 & pot willbe about .205 v _+ . Gate pulses will be wide [70-80 % duty ] As fluid heats up current goes up sense v goes up pulse width comes down. with a long time constant maybe it will be stable. Lots of luck as ckt. not tested.

 

Thank you for the input - I am currently looking at a few options. The circuit I am currently using does the job the PWM V2 from http://alt-nrg.org/

 

Simply using PWM on a hydrolizer will limit your current used, but only by shutting it off for a portion of the cycle. Unless you have an inductor in the loop, your current flow will stop, and so will your gas production while the current is off. When the current is ON, the excess current will result in your electrolyte solution being heated instead of being converted to gas.

If you have 6 or fewer cells, in series, what you really need is a "buck" DC-DC converter that will limit the current, yet keep it fairly constant.

What you need to know is at what current do you get maximum production with minimal heating. That will depend upon a number of things, one of them being electrolyte (type, concentration) another being plate area.

National Semiconductor has tools available from their main page for figuring out various power management designs. I plugged in somewhat arbitrary numbers for you:

Input voltage from 11.0 to 16.0, Vout=10, Iout=20.
The results are on this page:
http://webench.national.com/ss1/ss?...1V=10&O1I=20&op_TA=30&submit.x=45&submit.y=17

But, if you're running just a single cell, plates all stacked up and connected together, you'll probably get best gas production with somewhere around 1.5v to 2.5v across the cell. After a certain point, more current just makes more heat.

It's completely up to you to find that point.

[eta]
PWM V2: http://alt-nrg.org/pwm-v2.html
...uses the LM324 quad operational amplifier, which was an amazing device when it was first introduced around 34 years ago, when personal computers were still a thing of the future. But, that was 34 years ago, and time has certainly moved beyond that now seemingly stone-age device.

The PWM V2 won't work at all much above 2.5kHz without frying a MOSFET. The link I provided shows a couple of PWM ICs that run at 250kHz. Come up a couple of orders of magnitude in performance, will you? Besides, their efficiency can exceed 95%. I'll guarantee you're not getting that much efficiency from PWM V2.

 

Finding that point of resonance could be a bit tedious, too. We have seen anecodtal reports of it happening from 6 KHz up to just over 143 KHz. Some authors insist that multiple waveforms are required.

We would be very interested in hearing positive results.

 

OK, cell resonance is one thing, simple current limiting via a "buck" converter is something else.

I've heard various claims of increased gas production vs current consumption when the resonant frequency was achieved. The cell provides the "C", and an external inductor (wound toroid, etc.) provides the "L" for a resonant LC tank circuit. One big problem is keeping up with the changing "C" of the cell; as the dielectric constant of water changes with temperature (positive coefficient), and gas has a somewhat different dielectric constant than water. Gas production begins, the dielectric constant changes.

When the dielectric constant changes, the capacitance changes, as does the LC frequency product. In order to compensate for that change, you either need a large external variable capacitor, or a variable inductor, or a method to track the changing C portion and compensate by changing the frequency.

So, if you want to experiment with finding resonant frequencies, you'll probably need to focus on the lower frequency ranges - the most common numbers I hear being bandied about are from 10kHz to 45kHz.

But if you're looking for an efficient way to limit current, the higher frequencies are your ticket. Take a look at an old-style linear power supply. They have big transformers that weigh several pounds; even for low current units. I have an old Lambda 160W variable linear supply who's transformer is 4"x4"x3.5". Sitting next to it is a 250W ATX-form-factor switching power supply that's been converted into a bench supply. Instead of the big chunk of iron wrapped in copper, the switching supply has a few much smaller toroids in it that are driven at high frequencies. It's the high frequencies that allow the significantly higher power to be output from a significantly smaller inductor.