Battery Operated Inverted Project

[eebaker]

Gentlemen,

I am new to this forum. To introduce myself, I am a retired engineer with circuit design experience in broadcast communications and fiber optics. However, I have not had much experience
with SMPS or magnetic circuits other than some cookbook boost converters. So I am catching up on the magnetics and reading your posts, learning a lot.

I am hoping to build a battery operated 230 VAC inverter. There are several different battery scenarios that cover a possible 13 vdc, 24 VDC and 48 VDC. My inverted will be different than most in that it will provide a center tap.

To get some experience, and understanding, I have taken apart a 1200 watt 12 VDC to +/- 50 VDC output converter ( push/pull) available on eBay for $70 that looks very much like the IRAUDPS1 reference design. I have tried winding my own transformers and dropped them into their existing design and made a few observations:

While the "factory" circuit has a waveform at the drain that is picture perfect ie no ringing, the hand wound transformer has dramatic ringing. In addition, the circuit draws more idle current with the hand made transformer. How much ringing is acceptable? I played with some snubbers but it was trial and error. In all cases, the snubber resistors got warm.

Is it possible to saturate the core with small asymmetries in the drive of a push pull ( due to imperfect hand winding??). Would I be better off with a full bridge topology to avoid magnetic creep?

Given I am in a battery operated environment, efficiency is important to me. Is the full bridge topology that much more efficient?

Since I am not driven by accountants on this, what components are worth paying a premium for to improve efficiency? I know that ExcellentIT 7300 will make recommendations, but is there any room for improvement for a few dollars more?

I notice that torroids are often used in the output inductors. But in literature, I read gapped inductors are better in cases where there is a DC component to the current. What am I missing?

And if I do chose a full bridge drive, I notice they are AC coupled to the transformer to avoid saturating the core. Intuitively, this seems like a tall order for a cap. If the switching current is hypothetically 75 amps, what rating on the cap is going to tell me it is not a source of failure?

And finally, once I have got something working, what is the possibility of transitioning the design to one that can use planar magnetics? I see many modules( 1/4 , 1/8 bricks) capable of 100's of watts, isolated and FETS with little or no heat sinking. Seems almost like magic !

Thanks for taking the time and I look forward to getting to know you.
Regards,

eebaker

 

[KX36]

Your basic battery to mains inverter design will usually be an isolated DC-DC step up converter followed by an unisolated DC to low frequency inverter, that way you use a small ferrite transformer instead of a big line frequency transformer. There are still a few different ways to do it. You can have a single mains DC bus and an H bridge inverter or + and - DC buses and a half bridge inverter for example. Single bus and H bridge is probably easier to kep DC offset out of the mains. Either way you need to drive a high side switch or 2 in the inverter at a low frequency which isn't as easy as driving them at high frequency. Each high side switch driver will probably need a dedicated floating power supply and then you need to level shift the signal to it. Then you have a choice of square wave, modified sine or pure sine output. Modified sine is most often used.

In terms of the isolated DC-DC converter. Push pull would probably be OK as the input voltage is low. For both push-pull and full bridge you can avoid transformer saturation by using current mode control. Series capacitor is only necessary in voltage mode control or if no control and just a free running oscillator is used. That's not to say you can't do it like that - Current mode control is usually used in regulated supplies and since mains has wide tolerances, regulation isn't strictly necessary. I'd still do it regulated but I have some experience in the tricky task of closing feedback loops across the isolation boundary. You could do primary side regulation with current mode control, that'd probably be quite easy and the regulation is probably still going to be better than the full range of mains tolerance.

Output inductors should be gapped if they are ferrite. Toroids will generally be powdered iron which have a "distributed" gap.

Planar magnetics are just a different shape of transformer core that have a larger surface area to volume ratio to dissipate the heat and a window that suits windings made of PCB layers or stacked layers of copper sheet or tape (such as a single U shaped winding per layer) put in series/parallel for the desired turns ratio and current density. The modules seeming to produce a large amount of power with minimal heatsinking must be very efficient so they are most likely resonant topologies such as phase shifted full bridge or LLC, which include zero voltage switching and/or zero current switching on certain semiconductors at certain parts of the cycle.

I've discussed snubber design elsewhere on this forum a few times, but that's one of the last things you do. You need to design your transformer as well as possible, minimising parasitics like leakage inductance. That will minimise the loss in the snubber, but you have to tailor the snubber to the parasitic inductances and capacitances in your circuit.

EDIT: Just noticed you said you want a centre tap. not 100% sure what you mean by that, you mean like on building site equipment for example where in UK you get a 240 to 115V isolation transformer with a centre tapped earth on the secondary?

 

[KX36]

To do a centre tapped output, I'd probably think of that as a 2 phase output with both outputs 180 deg out of phase and at half the nominal output voltage relative to the centre tap. e.g. for 240V CTE, you might have DC bus of positive and negative 170-200V (very approx) and a half bridge type switch pair for each phase. DC offset on the centre tap may become a slight problem.

 

[wally7856]

I did not understand the center tap part either, i thought he might of meant input.

 

[eebaker]

Dear KX36,

Thanks for your explanation of powdered iron being a distributed gap. It was an epiphany for me. I'm still brushing up on my magnetic circuits.

My plan is to break this into 2 parts, the DC-DC portion, ( 13VDC to +/-170 DC ) then the PWM modulator/ Class D output for the 230 VAC sine wave. I will start with smaller designs and work my way up to build understanding. No point letting the smoke out of expensive parts first. Baby steps. When complete, I hope to synchronize to the mains to create a seamless back up.

In the US, residential power comes to the house as 230 VAC with a center tap. The majority of appliances run on 115 VAC, with different circuits in the home getting split up evenly between the 2 (180 degree) phases. So My inverter should be able to energize each leg of the distribution panel. My thought was to design a "normal" Inverter with 230 VAC and attach the center to the midpoint of the DC supply

I have been tinkering with the TL494 and I feel I have a good understanding of how to use it. But am I limiting my self in any way with this choice? The 45% duty cycle limit seems very "safe" in terms of avoiding shoot through. But is that limiting my efficiency too much? Also, I am not familiar with the "current mode" control to avoid saturation the core with stray DC. Can the TL494 be used for this? Is this a "pulse by pulse" adjustment? Is CAP coupling an easy way out and what is the trade off?

For the moment, I will be using 12 VDC lead acid, charged with solar panels. However, I also have a Golf cart with a 48 VDC battery. This makes for a good portable power back up. Will the push-pull stress the FETs too much or does this mandate a full bridge approach?

 

[KX36]

TL494 is a very old general purpose PWM IC. It won't do current mode control or even proper over current protection without external circuits and I wouldn't use it for new designs. UCC38083 is my preferred IC for push pull.

You should be able to do a push pull on 10-55V in with the PWM adjusting the duty cycle to regulate the output. It's quite a wide range so may be difficult to optimise as for example the peak currents will be high on low input voltages, but probably still doable with 200V fets for example, or perhaps igbts. The primary current sense will probably need to be a current transformer, voltage feedback from a sense winding on the main transformer.

 

[eebaker]

I have taken a look at 38083 as you suggest. Thanks. I always like fewer pins In looking at their application on the TI data sheet, ( fig 5), I am struck by the use of gate drive transformers
vs using a Hi side/ Lo side driver. At first glance, it seems overly complex. But I am not the expert here so I assume I am missing the point on their use. Any reason an IR2110 would not do the same function?

I am trying to understand the function of the ramp current compensation scheme on the 38083.

I would not be attempting to use the same assembly for both 12 VDC and 48 VDC inputs. I expect I would simply have the same circuit/topology with the transformer wound differently to accommodate the different
input voltages. In one case, I would expect a 10.5-15 VDC input, and a 42 to 60 VDC input respectively. A 1 to 5 input range seems a stretch.

 

[KX36]

gate drive transformers Vs high side driver ICs is always a hot debate. There are minor pros and cons to each, but really both would work for most situations. Push pull has both switches on the low side so doesn't need either.

Fig 5 in the datasheet uses GDT to drive the secondary side synchronous rectifiers. You probably wouldn't use SR. they're more for low voltage high current secondaries where the loss in a diode rectifier is to high.


Slope compensation damps down subharmonic oscillation in peak current mode control with higher duty cycles (D>0.5 by the textbook but really it comes in to a lesser degree at lower duty cycles too)

 

[eebaker]

Thanks for that, KX36.

In looking closely at the data sheet of the 38083, I see the CS pin connected to the common sense resistor. Does this actually compensate for potential " flux walking" in the core?I donty see in the data sheet where it specifically says this, although it might be implied with a full understanding. Would it be appropriate to use this controller coupled with 2 IR2110s to create a Full Bridge topology, eliminating the need for the primary series cap?

Unrelated question to this. Is there any reason it would be wrong to run transformers backwards, assuming the Volts/turn are observed? For example, if a transformer was wound for an Offline to 12 VDC output, would the same transformer be appropriate for 12 VDC to 380 VDC, just run backwards?

 

[KX36]

Yes to both questions in the first paragraph. Peak current mode control gives a negative feedback to the duty cycle on a cycle by cycle basis. If one cycle has a higher peak current than the last it will turn off earlier, pulling it back from flux walking in topologies other than half bridge (in half bridge by doing this the capacitive divider mid point voltage goes towards one of the rails and worsens the imbalance). It also protects from transformer saturation which would be the end point of fluff walking as the FET will switch off if the current passes a set point.

CMC in brief: There is an outer voltage feedback loop and an inner current feedback loop. The inner loop, including the output inductor, looks like a voltage controlled current source which takes the LC complex conjugate double pole or of the voltage feedback loop making it significantly easier to stabilise than voltage mode control.

Assuming the winding layout is the same on primary and secondary, e.G. Both centre tapped, it's not a flyback "transformer" and the frequency is about the same as it was designed for so the volt seconds are about the same, it would probably work