This is a branch off a previous thread.

I have a need to rectify 120VAC to DC for a project I'm working on. In order to smooth out the ripple/pulsing, I naively thought I'd just add a large can-size capacitor after the rectifier. But I know these are dangerous, and fortunately came here and got some good advice on how to keep my limbs attached and eyes in their sockets.

One of the obvious flaws that was pointed out was that this large capacitor was going to have a huge inrush current- probably tripping the circuit breaker and/or frying the switch among other havoc. I also had under-sized the capacitor voltage but I've changed that already.

(Note: the capacitor I'm looking at is this one:
CDE 250v 26000uf Computer Grade Bus Capacitor

It was suggested that I put an inductor between the rectifier and the capacitor to limit the inrush current. My circuit will normally draw about 6 or 7 amps with a primarily resistive load- though could occasionally require as much as 20 or 30 amps.

I've been searching around and haven't been able to find the right formulas, or figure out how to apply them at least to determine what inductor I need.

Any suggestions on what size of inductor I'll need for this, or how to figure it out? Any special considerations for the on/off switch? I assume I'll want a flyback diode across the inductor correct?

Thanks in advance,
Eric

 

Hi Eric.
The choke doesn't just limit inrush current, it also spreads the working current over a greater time reducing the ripple current in the capacitor and the "peaky" current waveform on the mains. The downside is a lower output voltage but in your other thread you said you were aiming for 120V dc output anyway.

Choke input power supplies were commonly used on valve equipment (due mainly to the rectifier properties and the limited availiablity of large high voltage capacitors at the time). I would suggest an internet search for background information on this.

If you switch after the choke a flyback diode would be needed but I would strongly reccomend switching happens on the ac side before the rectifer.

You need a choke of sufficient inductance to provide the smoothing you need, but also sufficient current handling capacity so it doesn't saturate. 7 amps isn't too hard, 30 might be rather difficult to accomodate. Do you know how much ripple you can tollerate? That's your starting point for calculating your choke.

I would suggest modelling this in LTSpice or similar. A sinusoidal voltage source (note in Spice these are specified by peak voltage), a suitable diode bridge, an inductor, your 26000uF capacitor, and a resistive load. For something like this I would just select a similar-looking diode from their list rather than bothering to create a model for the diodes you're planning on using.

Vary the inductance and take a look at the waveforms to see what output voltage and ripple you get. Don't forget that any practical inductor will also have some (a few ohms or tens of ohms) resistance as well. When you know how much L you need you can then look at buying or winding a suitable choke.

 

Here's a tutorial on inductor input power supplies. Might get your head in the right place.


Wretched website wouldn't let me edit the entry.

 

Here's a tutorial on inductor input power supplies. Might get your head in the right place.

Oops. Can't upload it.

 

You may want to consider an Inrush Current Limiter.

 

Inrush current limiters would help with the switch-one surge but wouldn't smooth out the pulsating input current in normal use. It may be worth having one in addition to the input inductor.

#12 - I've be interested in that tutorial. If you can't upload it here I've got somewhere I could host it and post a link.

 

The page I wanted to display is somewhere in here:

http://www.qsl.net/i0jx/index.html

here it is, 4th one down.

http://www.qsl.net/i0jx/supply.html

One of the interesting points is how to find the right inductance and how it operates with respect to current. The inductor isn't just an impedance in series with a capacitor. It needs to have the current continuous and in the same direction at all times.

 

Note that you can get a large resonant over-voltage (theoretically up to twice the steady-state voltage) from an LC filter when the input voltage is suddenly applied near the peak of the input sinewave (as could occur with random switching on of the input voltage). This can be readily seen in simulation. Thus you may still need something like a negative-coefficient thermistor inrush limiter with an LC filter.

 

Note that you can get a large resonant over-voltage....

This is true, although I'm not sure how it would behave with the load connected. Not only is there a lot of loss to de-Q the LC but it's non-linear as well. So far as I'm aware electrolysis cells appear resistive initially but then tend towards constant voltage once there's enough potential to split the O-H bonds.

I would include the negative-coefficient thermistor as crutschow says but if anyone has a good circuit model of an electrolysis cell I'd be interested to see what it does to the LC filter.

 

Thanks for all the great advice. I've been studying that and scouring the web tonight, learning a bit maybe even. It looks like the PTC thermistor combined with the inductor is the best bet. Oh and thanks for the advice on LTSPice - I've actually never used it before and finally did. There are many times it would have saved me frustration.

Anyway, from playing with ltspice, it looks like anything coming close to a reasonable inductor/choke has little impact on the circuit. It looks like I'll need something pretty massive. Does that sound about right? Either that or live with a much lower capacitance and significantly more ripple. I'll have to figure out where I can get an inductor like that.

Another problem is that as I add the inductor and PTC and get the ripple under control - my voltage starts to drop too low. Now I could scale back my HHO cell and that wouldn't be a problem, but that's no fun. One idea I'm playing with is to use both legs of my home 240v single phase current. This will give me more voltage than I need, but then I could turn down the input with a high power dimmer, something like this:
http://www.ebay.com/itm/New-4000W-A...445?pt=LH_DefaultDomain_0&hash=item3cd04414bd

Dunno if that would work at all though. I'm thinking of just going out and buying a canister of hydrogen, but after everything I've put into this I kind of want to figure it out.

Its pretty clear the guy that sold me the HHO cell just runs it straight off the rectifier, so worst case is I could end up doing that.

Thanks again,
Eric

 

What size inductor are you thinking of as "reasonable"? I'm expecting you to need a Henry or so. It'd be a big iron-cored wirewound device that looks like a power transformer. You still haven't said what your ripple target is?

So far as I'm aware most people just run them without smoothing capacitors and it just works. I personally would be inclined to do that.

Be aware that those cheap dimmers don't really reduce the voltage in the way you'd expect, look up "phase angle control". Definately don't try to feed a capacitor input power supply with one as the high dV/dt (rapid voltage rise) will kill your capacitor but I suspect you could feed an inductor input supply with one. With a little ingenuity you can model the voltage output of a phase-angle dimmer in spice and see what effect it has. Hint: use a behavioural voltage source to logical AND a sine source and a square 1V pulse.

 

I found my inductor

http://www.hammondmfg.com/195.htm

Part # 195T10 at the bottom.

Does this look about right for the task?

It weighs 35 pounds and costs $300. This will keep the inrush current from blowing a fuse and keep the ripple really low. Now I just gotta decide if I really want to do this that bad. lol

 

What size inductor are you thinking of as "reasonable"? I'm expecting you to need a Henry or so. It'd be a big iron-cored wirewound device that looks like a power transformer. You still haven't said what your ripple target is?

So far as I'm aware most people just run them without smoothing capacitors and it just works. I personally would be inclined to do that.

Be aware that those cheap dimmers don't really reduce the voltage in the way you'd expect, look up "phase angle control". Definately don't try to feed a capacitor input power supply with one as the high dV/dt (rapid voltage rise) will kill your capacitor but I suspect you could feed an inductor input supply with one. With a little ingenuity you can model the voltage output of a phase-angle dimmer in spice and see what effect it has. Hint: use a behavioural voltage source to logical AND a sine source and a square 1V pulse.

Yeah I was wondering about the dimmer.

Well the ripple I was going for is just "low" haha. But after looking at ltspice I think I understand the reason people have trouble with ripple. Below a certain voltage the HHO splitting stops, above a certain voltage you just get heat. The amount of ripple on a 120hz dc pulse spends a lot of time in both regions. So I think if I keep my voltage right around 120-125V with about 5-10V ripple I'd be pumping out HHO about as efficiently as it can (and for the cell I got, about 6.5A should be the current). If I've modeled things correctly in ltspice that 100mH inductor, an NTC thermistor and about a 2200uf capacitor look like a decent solution.

Like I mentioned earlier though, I need 240vac in order to have enough voltage due to the drop in the inductor. This looks like a pretty big drop across the inductor but it seems to be well within the ratings. Of course, I could be missing something I really only know enough to be dangerous here haha.

 

The page I wanted to display is somewhere in here:

http://www.qsl.net/i0jx/index.html

here it is, 4th one down.

http://www.qsl.net/i0jx/supply.html

One of the interesting points is how to find the right inductance and how it operates with respect to current. The inductor isn't just an impedance in series with a capacitor. It needs to have the current continuous and in the same direction at all times.

Awesome, very helpful THANKS!

 

$300 seems a lot but if your happy with that and your models suggest it'll do what you want then that's fine. If you wanted to there are ways of making one (I'm thinking of using the primary of a microwave oven transformer and putting an air gap in the core). That would be a lot more work though.

I would expect that with 120 Vrms in you'd see maybe 100V dc output, and 200V dc output with 240Vrms in - does that fit with what your models suggest? Try throwing phase angle control into your model and see that you can get the voltage you want off a 240V supply. An alternative is to use a transformer.

If you determined that you really need 160V rms (for example) to get 120V d.c. out you can use a 120V to 40V transformer and connect its secondary so it adds to the 120V incoming giving you 160V ac. The transformer would only pass 1/4 of the power because 3/4 of the voltage is delivered direct from the 120V mains. That means you'd need a 250VA transromer which might cost maybe $70 - more expensive than the dimmer off 240V but giving you sine wave input and probably greater reliability.

 

Another approach you guys could consider is an active power factor controller widely used in modern equipment to make the rectifier input current a sine wave.

PowerFactorCorrectionHandbook

It would be much cheaper but more complex especially at the required current levels.

Timescope

 

Now I don't know about a HHO at the voltage level your talking about, but a lot of the battery level (12V) ones use PWM. So how can ripple even be a factor?

 

Now I don't know about a HHO at the voltage level your talking about, but a lot of the battery level (12V) ones use PWM. So how can ripple even be a factor?

That's a good question, and maybe it isn't. I might just be overthinking/over-engineering the problem. Wouldn't be the first time haha. But I think the shape of the pulses is the problem. If I can believe the internet, HHO production starts when you have 1.8V across each plate, and above 2V (or 2.2V maybe) the extra energy goes only to heat. Then I want a certain amount of current, maximizing HHO while avoiding excessive degradation of the steel plates (and production of toxic byproducts). Reportedly, about 0.5A per square inch of one plate is the sweet spot. So on a 12V system with PWM you're dialing this in. On 120V I may be able to achieve the same with the dimmer, but if that doesn't work I'll have to manage current by experimenting with electrolyte concentration.

At this point, its become not so much a practical project (though I still do need the HHO torch) but rather an interesting learning project. So I do appreciate all the comments and suggestions. Just learning about ltspice has already made it worth everything.

Cheers,
Eric

 

I believe that by now you see that your beginning premise was completely wrong. Trying to start with 170 volts to get to 2 or 3 volts is the hard way, ripple doesn't matter, and reversing polarity keeps the wear on the plates even. (You can do that if you don't need your hydrogen and oxygen kept separate from each other.)

Think: What would happen if I just used a Variac to run a microwave oven transformer backwards and applied low voltage, high current AC to the water? It suddenly becomes very simple. Need separate containers? Add a few diodes.

"We don't need no stinking inductors to split water."
Tony Montana

 

Here's the actual HHO cell I bought:

http://www.ebay.com/itm/HHO-DRY-CELL-316L-62-PLATES-HYDROGEN-GENERATOR-3-7-LPM-120-VOLT-AC-/251266000336?pt=Motors_Car_Truck_Parts_Accessories&hash=item3a809ee5d0

Here's the same exact cell wired for 12V (multiple smaller cells in parallel):
http://www.ebay.com/itm/HHO-Dry-Cel...r_Truck_Parts_Accessories&hash=item3a80c8b0c0

Notice that the 12V has a higher gas output (according to the manuf)? I assume that is because the portion of the rectified wave that can't generate gas.

Regarding running AC on the cell that sounds reasonable. But people out there caution strongly against it. Now these are not EE's but mostly hacks trying to prove that you can achieve above unity energy efficiency. But they seem to have figured out from experimentation not to do AC.

But you're definitely right I should have used the 12V cell and a decent PWM. It would be up and running already. It was actually the manuf of the cell who suggested the 120V. But then again, he's just putting the rectifier straight across it and just accepting the lower efficiency. That looks to be the easiest route for me but I have to admit I'm curious to put this filter together I've modeled.