EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: David on April 30, 2016, 02:22:09 pm

Title: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: David on April 30, 2016, 02:22:09 pm
Hi all,

It's currently raining so what better to do than reverse engineer a 2.5kW 24Vdc to 110Vac Inverter...

I am currently looking at the DC-DC stage and a little confused with things. Attached is the topology I seem to have come up with from the unit but I can't say I've seen this kind of stacking before? Some background info:

* I am fairly certain it's a push-pull topology.
* There are 4 blocks of circuitry (with essentially identical layouts). - The secondaries of each transformer are wired in series to the next circuit block.
* Each block has it's own transformer and two N-channel FETs.
* There are no floating supplies, bootstrap circuits etc (hence one of the reasons I believe it must be push-pull).
* Each N-channel FET shares the same gate drive signal as the FET in the same position in the other blocks.
* There are no centre taps on any of the transformer outputs.

If this is the correct topology, are the primaries effectively working in parallel to spread the heat across all 4 transformers? :-//


Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: amspire on April 30, 2016, 02:31:16 pm
They have obviously decided that about 30A is as much current a single switching circuit can pull from 24V, and so they have duplicated the circuit 4 times. When you are designing low voltage high current switchers, you usually reach a point where to increase the current further starts to dramatically increase costs.

Probably all the switchers work in unison.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: diyaudio on April 30, 2016, 02:48:34 pm
Hi all,

It's currently raining so what better to do than reverse engineer a 2.5kW 24Vdc to 110Vac Inverter...

I am currently looking at the DC-DC stage and a little confused with things. Attached is the topology I seem to have come up with from the unit but I can't say I've seen this kind of stacking before? Some background info:

* I am fairly certain it's a push-pull topology.
* There are 4 blocks of circuitry (with essentially identical layouts). - The secondaries of each transformer are wired in series to the next circuit block.
* Each block has it's own transformer and two N-channel FETs.
* There are no floating supplies, bootstrap circuits etc (hence one of the reasons I believe it must be push-pull).
* Each N-channel FET shares the same gate drive signal as the FET in the same position in the other blocks.
* There are no centre taps on any of the transformer outputs.

If this is the correct topology, are the primaries effectively working in parallel to spread the heat across all 4 transformers? :-//

If your drawing is error free, it looks like a typical (single phase) current sharing topology , each switch will charge each primary at the same time,  forcing a current share path per fet switch (reducing looses),  this topology is similar to the power supply section of high power car amplifiers "SMPS DC to DC" sections for high power audio.
   


why are there DC diodes on the outputs ? for a 24Vdc(input) to 110Vac converter. ?
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: David on April 30, 2016, 03:01:51 pm

why are there DC diodes on the outputs ? for a 24Vdc(input) to 110Vac converter. ?

This stage provides the DC link which is then fed into a H-bridge to produce the desired AC output (@50Hz).
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: NiHaoMike on April 30, 2016, 03:12:42 pm
Is it an isolated inverter? Since if it isn't, they could do a shortcut by connecting the output in series with the input, giving about 14% of the output power "for free". Or indeed, add some diodes from each drain to an "intermediate" rail, using the load as part of the regenerative snubber.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: David on April 30, 2016, 03:27:51 pm
Is it an isolated inverter? Since if it isn't, they could do a shortcut by connecting the output in series with the input, giving about 14% of the output power "for free". Or indeed, add some diodes from each drain to an "intermediate" rail, using the load as part of the regenerative snubber.

The transformers all provide galvanically isolated secondaries. There are some diodes and a cap near the FETs but I'm fairly certain there just there for snubbing.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: nctnico on April 30, 2016, 03:53:02 pm
It looks like a forward converter with multiples transformers to me. I have build similar circuits.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: retrolefty on April 30, 2016, 04:04:56 pm
Boy, be user to test for and label the proper phasing for all those windings, both parallel primary and series secondary. In phase adds, out of phase subtracts.



Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: David on April 30, 2016, 04:38:42 pm
Boy, be user to test for and label the proper phasing for all those windings, both parallel primary and series secondary. In phase adds, out of phase subtracts.

The FET configuration wouldn't be right for a forward...
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: chris_leyson on April 30, 2016, 05:01:21 pm
You're right it's a push-pull forward converter with parallel series combination of transformers.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: nctnico on April 30, 2016, 05:38:34 pm
Boy, be user to test for and label the proper phasing for all those windings, both parallel primary and series secondary. In phase adds, out of phase subtracts.
The FET configuration wouldn't be right for a forward...
It is but you don't need a half-bridge configuration to drive it. Instead you have an extra winding on the transformer which is used only half of the time. As a bonus the diode in the non-conducting MOSFET feeds the reset current back into the DC supply.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: David on April 30, 2016, 10:47:30 pm
Hmm I need to read that a few times...it's not sinking in yet! |O
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: David on April 30, 2016, 11:23:47 pm
Boy, be user to test for and label the proper phasing for all those windings, both parallel primary and series secondary. In phase adds, out of phase subtracts.
The FET configuration wouldn't be right for a forward...
It is but you don't need a half-bridge configuration to drive it. Instead you have an extra winding on the transformer which is used only half of the time. As a bonus the diode in the non-conducting MOSFET feeds the reset current back into the DC supply.

OK, I've re-read your post and think I was just confused with the terms. From what I think your saying, it's four 'standard' push-pull converters (I thought you were talking about a two switch forward converter) with all the secondary windings wired in series.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: Circlotron on May 01, 2016, 12:09:51 am
The designer might have found that one big fat primary winding was hard to wind and terminate, as well as having bad skin effect at high frequencies so they settled for a quadfilar primary. From there it's a tiny step to add extra mosfets, one to each termination pin.
Title: Re: Reverse Engineering a 2.5kW Inverter - Topology Question
Post by: T3sl4co1l on May 01, 2016, 04:48:28 am
Likely, it's a PP forward converter, but no inductor.

They run those things wide open (full duty cycle). The only reason they don't explode the instant you turn them on?  Generous soft start time constant.

Accordingly, absolutely no regulation or current limiting, either.

Tim