Most Efficient Way of charging the Capacitor from Boost Circuit?

[AngelWing]

Hi All,
This may not look very complete now cause I am still collecting data. I like to read up everything before I take out the lab kits.
So here is the thing: I am trying to make my own Portable Battery powered camera Strobe.
For this purpose I have till now decided to use 5 Lithium ion cells to give me 18.5 Volts DC. This will be stepped up to 300 - 330 Vdc via a simple fly-back inverter transformer. Which will in turn be controlled via an Arduino(I am still finalizing it) and MOSFETS.
Now I have with me 360V 4400uF capacitor for xenon tube (they have very low ESR being photo flash capacitor and all).
Which I need to charge.
I want my charge time to be in the vicinity of 2 -4 Sec, In fact it can be a little bit more.
Now Capacitor in Flash system are charged from 110v approx to 330v every time a flash cycles. I want to minimize as many losses as possible. Efficiency is the target here.
I understand that charging a capacitor through inductor is more efficient as it reduces the inrush current.
So question 1: Will the Secondary of My inverter Transformer Work as Current controlled source and be efficient enough to charge capacitor with minimal losses and not tax the Lithium ion cells with huge inrush currents? ??

I also realized since I am planning to control the boost circuit via arduino I can change duty cycles to gradually increase voltage of secondary during each flash charge cycle. Say from : 110v to 140 to 180.. till 330 etc .
This should also reduce the inrush current. And is easily doable with little bit of feedback and coding.
So Question 2: Is the above mentioned hack a good way to continue??
My primary focus is to reduce all the losses so that i can extract as many flashes from battery as possible.

Now time for declaration: I understand the risks involved with such projects. I also have quite less hands on with SMPS and high frequency boost circuit so will be extra bit careful. Energy levels in this project are beyond lethal hence I am gathering all inof before starting. So please do not waste your time reminding me the same.

Note 2: I am writing it at the end of the day after a full day reading stuff. I still do not have circuit diagrams ready but idea is forming. I will upload the same as and when I have finalized stuff.
Thanks.

[Buriedcode]

For just charging a cap, rather than trying to create a regulated output voltage, then a constant on-time boost converter will work just fine.

1. Yes, it is essentially a current source, dumping energy into the cap during every primary 'off' time.
2. Using constant on-time for the transformer (technically, coupled inductor) means what controls the input current is time between 'on' periods. 

If the output cap starts with 0v then it will take a long time (compared to the on time) for the current in the secondary to drop to zero, but as the caps voltage increases, this time gets shorter and shorter.  This is why self-oscillating circuits such as those used in camera flashes increase in pitch as they charge. Those converters are quasi resonant in that the primary gets switched just as the secondary's current drops to zero. 

If you have a constant off-time then the current in the inductor can increase every cycle until saturation, so you may want to either use a feedback winding to sense when the magnetic current in the transformer/flyback/coupled inductor is zero, or wait a fixed period.  Waiting a fixed period would have to be long enough top prevent saturation, but then later on in the charging cycle it will just be 'dead time' that necessarily increases charge time.  A compromise is to have a fixed off time at the start of charge, then when the output is say, 60V, reduce this off-time to something more sensible.

If you have full control over the on/off time of the switch, then you can change the on time (duty) and period to trade input current for charge time. 

I have never liked using micro's for switched mode power supplies, mostly because if that is all that is controlling the switch, you can make things go pop but leaving the switch on too long and also because it makes a very slow control loop for regulation.  But for cap charging the most important factors are the transformers inductance, saturation current, max Vds of switch, and Ron of switch.  The diodes reverse breakdown voltage and recovery time also play a part.  Once you know the primary inductance and saturation current, then you can calculate a Ton(max) for your controller, and vary the off time.

I think it would be better to have the micro 'monitor' the boost, but not actually be part of it. That is to say, it monitors the output voltage, and can control the peak current limit (to reduce input current, whilst increasing charge times) but not directly drive the switch.  It can be done but if you're switching at 100kHz (not uncommon) then few micro's have PWM hardware fast enough to give reasonable resolution. Where-as having the micro 'set' certain parameters gives flexibility without limiting yourself to lower frequency (= higher inductance = larger transformer).

You seem to know what you're doing, so sorry if you've heard this before. OTOH if that was 'too much' I'll explain further.

Just a quick calculation of the power.. 4400uF @ 360 = 285J.  To charge in 4 seconds, thats 285J/4s = ~71.3W.  If your converter is roughly say 75% (linear tech claims their chargers can be up to 80%+) then the average input power whilst charging is 71.3W/0.75 = 95W.   with 5 lithium poly cells at their average voltage, that does indeed give 18.5V.  95/18.5 = ~5A average current draw.
As that is 'average' current draw, the peak current i the inductor will be much more than this, so you'll need a rather large flyback, especially if you're trying to keep the switching frequency down. Even without mentioning the fact the cap will be charged enough to kill you a few times over, a 100W boost is quite a serious converter.

I would urge you to design and build a much lower power version first so you can optimize it and have practice in winding transformers.

[AngelWing]

Hi Buriedcode
Thanks for the inputs.
I have few doubts will ask them later, when I have acquired a little bit more knowledge.
I have always noticed that scifi noise strobes make when they charge. But if its audible doesn't it means they are subjected to resonant frequencies in upper 20Khz range? Which is quite low.
Anyway. Here is what I am doing now: 
1. Went to a wedding photographer yesterday. He gave me a busted flash with broken hotshoe for free, I am cannibalizing it to study the circuit.
2. Have ordered a 200W Chinese car inverter just to again study it.

So I will hookup the flash transformer to oscilloscope an will study on times and duty cycles etc.

My main concern is to make it as efficient as possible in a DIY.
I understand that the charge cycles will be huge when starting from 0V. But flash capacitor do not really discharge to 0V, Plasma in Flash tubes stops emitting light at 110V. Mordern flash use IGBT to quench the flash when sufficient light has been delivered but that is separate problem all together and will be discussed once I can reliably charge capacitors without blowing Mosfets or diodes (One step at a Time).

[Buriedcode]

It is true that xenon lamps will stop conducting below a certain voltage, meaning the voltage on the capacitor does not drop back to 0V - so it will be recharging from 90-150V back up to your 330/360V target.
Also true, as you mentioned that an IGBT can be used to extinguish the xenon arc to give a fixed flash period or to control peak current.

An automotive 12-220/110V inverter is a different beast to a boost converter.  These are usually forward converters where the transformer transfers energy to the secondary during the primary 'on' time, without storing any appreciable amount of energy.  The boost converters (or flybacks as we're using a transformer/coupled inductor) are designed to store energy, so the transformer stores 'packets' of energy during the primary on time, then transfers this to the secondary on the primary off times - this is why they often use gapped cores to increase the amount of energy stored.

The 'audible whine' you hear from most camera flashes is indeed in the audio range - but mostly at the start when the charging caps voltage is relatively low.  Also note those powered by a single AA or two AA cells operate at 1.5-3V.  This low voltage means the on-time for a given inductance is longer, reducing frequency, and so typical camera flashes do operate in the upper audio range - but only at the start. They can quickly get to 100kHz+.

Another factor that influences efficiency is the turns ratio.  A pretty ballpark figure for 12V to 330V would be 1:10.  As its a flyback (not a forward converter) it isn't as simple as Vout/Vin.  A 1:10 ratio would mean when the output is charging to 360V, the primary voltage will 'fly back' to 360/10 = 36V.  This voltage determines the max Vds of the MOSFET you'll use, which is generally inversely proportional to Rds - higher voltage MOSFETS that can cope with voltage spikes of 100V will have a much higher on-resistance, lowering efficiency.  A 60V MOSFET could be a good compromise.

I too ahve taken apart 'horseshoe' flashes, and they are almost always simple self-oscillating flybacks, using a single transistor (bipolar, NPN) as the switch, and a feedback winding in the flyback.  The only relatively complex parts are the trigger circuits.  These chargers can have quite good efficiency, but the charge time, current draw and power supply voltage dependency is at the mercy of the transformer specifications - they're just designed to charge the cap within a certain time, and current draw goes up significantly with power supply voltage, but as they are designed to use batteries anyway there is no provision made to run off higher voltages.

Some helpful links:
http://www.dos4ever.com/flyback/flyback.html - excellent page
https://www.diodes.com/assets/App-Note-Files/dn25.pdf - zetex used to have great app notes.

And attached is another zetex app note, for a higher power strobe. Nothing like the power you're attempting, but nevertheless and all analogue approach which is still common.

[AngelWing]

Hi Buried code, thanks for advice:

Here are my findings as of now:
1. Car Inverter used an ETD29 transformer in center tapped winding. It was wired in Push Pull configuration with frequency of 55Khz. The circuit used KA7500BS ic for gate drive control etc. It looked kind of neat.
I have yet to determine its Efficiency.
So I went ahead and looked for details of push pull converter, and found that its nothing but kind of 2 Forward inverters. And looks to be quite energy efficient.

So again with the Questions:
1. Will the design implemented by this car inverter work for charging flash ? As I believe it may not provide the flexibility of a resonant flyback used in Flashes where switching frequency climbs with capacitor charging.

2. I also did not note any Inductor at the output of the Push Pull converter, Which is kind of off, as every article/forum-post etc mentions that Inductor being used to store energy. There was though a 400v 10uF capacitor after bridge rectifiers. So is that Inductor necessary if such config is to be used for DC/DC conversion ? 

One thing I could not find as of now is a definite formula for Push Pull converter. Also the entire working explained somewhere.
I do have formula for Forward inverters: Vout = Vin x TurnRatio x PwmDutyCycle. Hopefully the push pull will follow same formula albeit for each primary winding separately.

Now for Question 3:
Every one mentions various benefits in terms of EM noise and Output control etc for Push Pull inverter, But no body has given any quantitative comparison of efficiency of Push Pull vs Flyback etc. So do you have any idea which will be more efficient ?

Also, Thanks for providing those docs and links, I am going through them now. I agree I have lot to learn before I would be anywhere near designing of my boost circuit.
I have lots of experience with low power digital and analog circuit etc but have never gone into inductors and flyback as of now.
And am kind of apprehensive about it.. Hence want to carry on with established designs.
But the problem is most of the stuff available online is for very weak flash and I can't directly increase component sizes to get more power as its not that simple. Hence asking around.

[Buriedcode]

Whilst you can use a forward converter for charging a capacitor, it is designed to provide a fixed regulated output voltage.  This means the switching duty and possibly frequency must change as the output cap charges.  The output cap on forward converters is there purely for smoothing.  Although I am not overly experienced with forward converters, the turns ratio determines the output voltage.  You would either have to implement clever PWM techniques to control the output voltage, or limit the current.. or.. treat the forward converter as a fixed voltage source, and charge the cap via a resistor (which will waste 50% of the energy you use to charge the cap).

your questions.
1) Whilst I'm sure any converter can be used to charge a capacitor, by far the most efficient will be the flyback/boost.  The flyback is simply an isolated boost, and is much more suited to large Vin/Vout differences - such as 12V to 300V, or 315V to 5V. It does not necessarily have to be resonant, I mentioned that because the self-oscillating flybacks are, by their nature, quasi resonant.  And whilst flybacks are very often used to create a regulated voltage for <90W (generally) they are quite noisy, because you're building up a magnetic field in the transformer on the primary 'on' time, and letting it collapse in the off time. 

This makes them less suited to higher power DC-DC, but ideal for capacitor charging, where you dont' need a regulated output - the output capacitor determines the output voltage. The more 'packets' of energy you pump into the cap, the higher its voltage goes, until something gives (either the cap rated, the output diode, or the primary switch which will experience a spike of Vout/turns ratio).  You don't have to worry about voltage regulation, control loops or any of that jazz, all you have to do is switch the primary with an on-time that isn't long enough to saturate the core (= sudden increase in current draw, heat, efficiency takes a dive). Then once its turned off, wait for the current in the secondary to stop flowing, then repeat.

2. I cannot same I am an expert in push-pull converters, but often the output inductor is there as a filter, along with the output caps.  DC-AC covnerters (as in 12V DC to 220V AC) produce a 50/60Hz sine, or a pseudo sinewave output, which is done via PWM.  In order to attenuate or filter out the PWM switching frequency, to leave a relatively pure sine wave, it requires a good filter on the output so switching noise isn't injected into whatever its powering.  I agree that there probably should be an output inductor, and many automotive inverters are made pretty cheaply so it was probably just trying to cut costs.  Or perhaps it is some form of resonant converter that produces a rather clean output that doesn't require much filtering (unlikely, unless its very lower power).

3. As for efficiency of push-pull vs flyback, as always - it depends.  Again, I am no expert and I'm sure someone can explain much better.
Flybacks can be very efficient, especially at lower power because one can simply leave a longer time between switching the primary on, or switch it for a shorter time to give a lower peak current in the inductor/flyback transformer.  As you go up in power, because the flyback must *store* energy in the transformers core, the size of the transformer must increase in size to the point where you end up with a rather large transformer, that has to switch at very high frequency just to transfer the power.  Push-Pull and forward converters transfer energy *through* the transformer - current flows in both the primary and the secondary at the same time, so whilst the transformer size does increase with power, it does not have to be 'massive' since its not actually storing energy - its a true transformer rather than two magnetically coupled inductors.

Although of course you're after a way of saving some time by using an off-the shelf converter, and modifying it, for high voltage capacitor charging from low voltage, I don't think anything really beats a flyback.  I have seen other things like a 'royal' converter ( https://en.wikipedia.org/wiki/Royer_oscillator ) used in DIY strobe circuits, but I cannot comment on their efficiency.  They are ideal for driving CCFL tubes and produce relatively pure (read, low noise) AC high voltage, but given the fact an uncharged capacitor presents essentially a short circuit, I am unsure how these converters cope.

That 'flybacks for dummies' article is ideal.  Explains everything one needs to know, and the circuit he provides - whilst its for regulated DC output for nixie tubes, can be adapted to charge a capacitor fairly easily, although at low power.

Again: I keep going on about flybacks and boosts rather than other topologies - but I do not have much experience of other converters, so hopefully someone will chime in and either correct me, or provide other possible converters for your task.  Safe to say though, every xenon flash/strobe I have seen that runs off low voltage uses a flyback.  Perhaps the very high power ones (used for ruby lasers, or light pumped lasers in general) use a different type - I'll google.