Current Fed or voltage fed Dual Active bridge topology?

Hello People!
I have to build a dc -dc converter with following specifications:

1. Galvanic isolation between input side and output side
2. input voltage 28-42V
   
3. Output voltage 350V
4. Output power 5KW
5. efficiency >97%
6. High power density

First I went with "Voltage Fed Full Bridge Phase Shifted Converter"
Later on, I found out that it has following demerits:

1. It is an inherent buck converter so only way of boosting voltage was high transformer turns ratio. Hence more size and more loss.
   
2. Involtage fed phase shift pwm full bridge input current has to go from negative peak to positive peak in very short duration(L{\frac{di}{dt}}) , hence leakage inductance of transformer had to be neglegible which is very hard to achieve.
   
3. There is inherent loss of duty cycle in this type of converter which is not good for higher efficiency.

Because of these reasons I shifted to current fed converters. But in meantime my specifications changed and one more attribute was added in the list. ie
Bidirectional power transfer.
Thus I moved my focus to Dual Active Bridge. And there is only slight difference between single active bridge circuits and dual active bridge and that is, on output (HV) side, instead of diodes, mosfets/igbts are placed. So I natuarlly thought that current fed topologies will be better option. But it is complex and people have built high boosting circuits with voltage fed. I am unable to find advantages of current fed DAB over voltage fed? Only odvantage is since no inductor output is there in current fed, multiport outputs are convenient but I need only one output.
Also transformer saturation ican be avoided in CF (i don't know how)
Please suggest me which topology should I use?? I know it is a long post but I am really in trouble and I need to get this done anyhow. Thanks for taking your time..

Unfortunately what you are playing with and the efficiency levels you want nothing is going to be simple or easy.

The only advice I can give on the active bridge circuits and related components it ot sit down and do some realistic numbers comparisons for min, max and average power level losses for the different designs and components you can come up with and see which circuit shows the most favorable numbers for the most realistic conditions it will be operating in.

Beyond that, it may be worth looking at realistic cost of build numbers for various designs as will being when you get into the high 90's% efficiency ranges cost and complexity to build tends to go up extremely fast for tiny gains.

So much so that what it may cost to build just to get 1 - 2% more efficiency may greatly eclipse the value of the energy saved in any realistic real world - as used in service - working lifetime, which for a highly complex design might not be all that great in life expectancy to begin with. (Rather like a high fuel efficiency vehicle that cost $20K more than a normal one but only saves the owner $10K in fuel in its lifetime while also costing him $10k in service costs to keep it running. )

As for bidirectional power transfer, I am assuming you are referring to the ability to work on both DC - DC step up and DC - DC step down modes? It can be done but it adds a whole new layer of complexity and cost that might not be worth it if it's not a normal active function that will be in constant use.

Shadow_warrior said:

Bidirectional power transfer.

Click to expand...

Please define.

The requirement for galvanic isolation means that a transformer is unavoidable.

Uspide: That also takes care of the high input - output voltage ratio.
Downside: It also drastically increases complexity.

As I said in answer to a previous post, start with low power designs using cheap throw-away parts; don't start with high power stuff until you have a basic design working flawlessly at eg. 12V <> 48V and 1A or so.

Your 5000W design needs a minimum input current rating of 200 Amps; in real world designs every part needs to be able to withstand at least a 50% overload for some time without harm, to allow for transient variations.

At those power and current levels, you are looking at possibly a thousand dollars just for all the fuses needed to protect the semiconductors...
(you need special ultra-rapid fuses to protect high power transistors or thyristors). That's before you get to the costs of any active components.
https://uk.rs-online.com/web/p/centred-tag-fuses/4483500/
https://uk.rs-online.com/web/p/centred-tag-fuses/4483370/

I've been designing electronics for over 40 years and routinely work with high power drives and converters - but there is no way on earth I would consider trying to design what you describe from scratch unless a customer put something like £100,000 in as a guaranteed minimum budget...

[Ron - by using bidirectional he means regenerative, so if the load side voltage increases above some point, power is returned to the input side.
Think of it like an electric vehicle drive, normally battery > motor but when using dynamic braking, returning power from the motor to the battery.]

ronsimpson said:

Please define.

Click to expand...

From a dc source to a 350 volt battery during charging and from battery to dc source during discharge

[QUOTE="
[Ron - by using bidirectional he means regenerative, so if the load side voltage increases above some point, power is returned to the input side.
Think of it like an electric vehicle drive, normally battery > motor but when using dynamic braking, returning power from the motor to the battery.][/QUOTE]
Yes

Shadow_warrior said:

From a dc source to a 350 volt battery during charging and from battery to dc source during discharge

Click to expand...

If the power flow is not reversing literally cycle-by-cycle, two separate one-way units plus a voltage monitoring controller to switch "enable" signals between them would likely be simpler and more reliable.

If the charging source is separate from the load, it's even easier - everything goes in-line and just operates.

rjenkinsgb said:

If the power flow is not reversing literally cycle-by-cycle.

Click to expand...

No, power is not reversing cycle by cycle. The mode change is basically on the discretion of the user/client

That makes it simpler still.

You need a custom transformer regardless. Imagine each side having two sets of voltage taps (or winding end and a tap).
eg. 24 & 48V --- 320 & 380V (depending on the voltage range of the batteries).

The "high" voltage taps / ends at each side just feed rectifiers, so the transformer output goes to the respective load.
The lower taps can be fed by power transistors or IGBTs etc. at one side or the other, depending which direction power transfer is needed.

Only the input side (for whichever direction) is active, with feedback from the opposite side for current limiting and voltage regulation.

Each side is then in principle just a (relatively) simple push-pull forward converter - eg. something like this:
**broken link removed**


For info, quite a few commercial inverters use multiple small transformers, each with their own drive transistors. That keeps peak currents down.
If you drive them all at the same frequency but with a phase offset from each pair to the next, you can reduce ripple current on both the input and output sides, simplifying smoothing.

One of the DSPIC33...MC devices with multiple PWM outputs may be suitable to control a setup like that.


Edit - Doing a mental simulation of the tapped winding config, I realise it's too simple and cannot work; whichever side is driven, the other end (rectifier) on the same winding will be trying to exceed its own DC rail voltage. It would need a bit more control or stick to simpler, separate units in each direction.

Shadow_warrior said:

No, power is not reversing cycle by cycle. The mode change is basically on the discretion of the user/client

Click to expand...

Do you want to use the transformer in direction 1 and direction 2?

Shadow_warrior said:

From a dc source to a 350 volt battery during charging and from battery to dc source during discharge

Click to expand...

This seems like a poor design to have a 350 volt battery bank for a 24 or 48 volt DC system.

Why use it to begin with rather than a battery that's compatible with the far lower working voltage?

ronsimpson said:

Do you want to use the transformer in direction 1 and direction 2?

Click to expand...

I didnt get you sir.

tcmtech said:

This seems like a poor design to have a 350 volt battery bank for a 24 or 48 volt DC system.

Why use it to begin with rather than a battery that's compatible with the far lower working voltage?

Click to expand...

I don't know but this was the only specification given to me. I know nothing about how 350Volt battery will they use or something else.

Shadow_warrior said:

I don't know but this was the only specification given to me. I know nothing about how 350Volt battery will they use or something else.

Click to expand...

Unfortunately you are likely doomed on this whole project if you don't have a full understanding of what is being used and why on both ends of the system and that both have very solid reasons for being what they are. Especially if a highly complex and expensive to design and build device is required just make two simple and common, but mismatched, things or systems work together.

Oddly it's one of the most overlooked primary aspects, and points of concern and contention (especially on forums for some reason.), in custom designing anything despite how obvious it should be that items A and B, that need to work seamlessly with each other, shouldn't need any complicated mid points between them to function as they were made to.

For me such a odd project would need considerably more information and justifications as to why there is such a wide mismatch in things before I would take it serious to any level beyond being an academic thought exercise in conceptual designs.

Which I do have to admit is interesting being I have played with such designs in the past and in normal efficiency ranges it's not terribly difficult to do but the way it can be done, as I have done it, comes across as a bit counter intuitive when first looked at but at the bare minimal design level does work like a real non regulating bidirectional DC - DC transformer.

tcmtech said:

Unfortunately you are likely doomed on this whole project if you don't have a full understanding of what is being used and why on both ends of the system and that both have very solid reasons for being what they are. Especially if a highly complex and expensive to design and build device is required just make two simple and common, but mismatched, things or systems work together.

Oddly it's one of the most overlooked primary aspects, and points of concern and contention (especially on forums for some reason.), in custom designing anything despite how obvious it should be that items A and B, that need to work seamlessly with each other, shouldn't need any complicated mid points between them to function as they were made to.

For me such a odd project would need considerably more information and justifications as to why there is such a wide mismatch in things before I would take it serious to any level beyond being an academic thought exercise in conceptual designs.

Which I do have to admit is interesting being I have played with such designs in the past and in normal efficiency ranges it's not terribly difficult to do but the way it can be done, as I have done it, comes across as a bit counter intuitive when first looked at but at the bare minimal design level does work like a real non regulating bidirectional DC - DC transformer.

Click to expand...

We have a meeting fixed with them next week. I will surely ask them. Btw client is a cto of a space startup an I think the product will be used for ion thruster. I am not sure though

Shadow_warrior said:

We have a meeting fixed with them next week. I will surely ask them. Btw client is a cto of a space startup an I think the product will be used for ion thruster. I am not sure though

Click to expand...

Seems off that something potentially being launched into space would use directly incompatible power sources plus add unnecessary payloads to compensate for it.

Is elon musk planning to launch one of his semi practical and semi funtional powerwall systems into space now?

Shadow_warrior said:

space startup

Click to expand...

Now you have problems. In this area there are several space companies. One I did designs for one, in another division.
When there is no air in the box the heat can not escape. The parts jest get hotter. I potted everything and now the cabinet gets hot. I could not get rid of the heat.

ronsimpson said:

Now you have problems. In this area there are several space companies. One I did designs for one, in another division.
When there is no air in the box the heat can not escape. The parts jest get hotter. I potted everything and now the cabinet gets hot. I could not get rid of the heat.

Click to expand...

Now you have my curiosity as to how they do deal with dissipating heat in a vacuum. What's the trick or tricks to dumping heat when solar side temps can be +200 - +400F and shadow side is at the far end bottom and all the electronics are buried in a box in the middle?

tcmtech said:

how they do deal with dissipating heat in a vacuum

Click to expand...

Not well.

I live at 5280 feet (+/- a little) 1500m You must increase the heat sink to 1.1.
I have built things for 20,000feet, 6000m. You increase the heat sink to 1.5. (if there is a fan then 2x on the heat sink)

I think a vacuum is a very good insulator. Probably a heat sink is worthless in a vacuum.