IC identification needed for 2500w "Mobile Power" Inverter

[cornwallav8r]

55pilot, your transformer statement...There is no transformer per se, this unit is 2 staged...it uses 8 banks of step up converters(in 2 banks also apparently driven by 3525 ICs) to build up the high voltage, and then that high voltage is supplied to the output NMOSes. Wouldn't want to get my fingers on that line while debugging. Likely lethal.

 

[cornwallav8r]

Let's use some logic here...why would the MOSFETS be on at the same time if I use the wrong one? One chip drives ALL the output MOSFETS. Just wait half a 60Hz cycle, and the outputs will be the same as the other version, right? There is nothing else on the board that this thing syncs with. The only external control lines are on pin 8 (soft start) and pin 2, I believe to be a logic ENABLE line.
The alternate inverter type is shown as a 1527, which in the series would equate to an SG3527 , not a 3525, so by my logic this is a win.

It really does make a difference. You need to use the version with the "On = High" output stage. The data sheet is a little confusing, but it is the "SG1525A" drawing on page 2 of the datasheet. If you use the other one, you will blow the MOSFETs because they will be on at the same time.

The circuit drawn by cornwallav8r is s little suspicious, but it could be just poor drawing. I can see pins 11 and 14 driving 3 MOSFETs each with the resistor/diode combination. The driver is terribly under-powered, but it will work.

I do not understand what the transistor is doing, but it could just be poorly drawn. The only thing I can think of is that the transistor is pulling the other set of MOSFETs' gate low. So when pin 11 is high, it drives its MOSFETs on, while the transistor pulls the gates of pin 14's MOSFETs low. This will supplement pin 14's output drive to keep those MOSFETs off even as their drains are whipped high by pin 11's MOSFETs turning on. We need a better drawing to know for sure what the circuit is doing, but that woudl be my guess.

Much more importantly, if the driver is dead, it may be because something else in the circuit killed it. Just replacing it is not going to help until its killer is identified and fixed.

---55p

 

[cornwallav8r]

Steve, the inverter transistors turn off the mosfets when their base goes high. Note that the same logic output pin feeds 2 separate quarters of the h-bridge. The other set fed from the same pin go high while the transistor's base is hig, bringing its collector to ground. A clever and cheap way to effect an h-bridge IMO.

Good find. But check out the datasheet. There are 2 different versions. They have outputs that are logically inverted. I don't think you'd want to put the wrong one into the circuit.

On second thoughts it may not matter, since there appear to be a set of mosfets driven by the logically inverted output signal.

Is it just me, or does anyone else think that the three mosfets turned on (presumably) by the transistors need something to turn them off again.

 

[55pilot]

Let's use some logic here...why would the MOSFETS be on at the same time if I use the wrong one?

You are right, it is a good idea to use some logic rather than just opining without understanding.

Look at the data sheet. The sequence is: A-on, Both Off, B-On, Both Off, A-On....

On one chip, when the output is active, it is 1. So the sequence is A-high/B-low, both low, A-low/B-high, both low.....

On the other chip, the output is inverted, so when an output is active, it is low. This means the sequence is A-low/B-high, both high, A-high/B-low, both high...

Still think the ICs are interchangeable? You are using low-side MOSFETs which are on when the input is high, so using the wrong IC will have both MOSFETs on at the same time.

---55p

 

[cornwallav8r]

Hmm, good point. And this is why we debate these things
I looked at 2 data sheets, I see reference to adjustable dead time, but don't see a logic diagram or more info, where am I missing that info? Different datasheet?

 

[55pilot]

Hmm, good point. And this is why we debate these things
I looked at 2 data sheets, I see reference to adjustable dead time, but don't see a logic diagram or more info, where am I missing that info? Different datasheet?

Here's the datasheet that Steve linked in his post. You can also find it directly from ST's website. Bottom half of page 2 explains the output structure and gives sample output waveforms.

---55p

 

[cornwallav8r]

Hmm, I think you might be missing something I didn't supply. Each of the pins 11 and 14, output pins supplies one half the bridge. Let's say the h-bridge is made up of 2 halves split vertically, each side (half) comprised of 2 sets of NMOSes in series with high voltage and ground. In this configuration, because of the hard wired inverter, the NMOses can never be on at the same time, shorting the high voltage to ground.

The Ac 120v output however does see the change. My brain doesn't wanna stretch that far, but it doesn't look like it will kill NMOS parts directly does it? Something yet I am not seeing....

 

[cornwallav8r]

55pilot, I believe you are incorrect here. Explain to me how, when the left half is hard wired with a logic inverting transistor such that the Left top is on, the left bottom MUST be off?
By design, regardless of what the chip does, all 4 transistors CAN NEVER all be on.
What DOES happen, is that the AC load, wired in the middle between left and right, sees a different off time, which I believe, now is an extended on time instead. Right?

P.S. You keep stating that the circuit schematic is suspect. It is not, I just confirmed the wiring is as I drew it, and the transistors are configured as I drew it.

 

[55pilot]

First, I have a hard time believing your circuit description. What you are describing is a really bad way to do a voltage inverter. But since I have seen some really bad stuff coming out of cheap Chinese companies, it is not out of the realm of possibilities.

The "normal" way to do an inverter is to have a center tapped transformer. The center tap stays at the DC voltage and each of the primary legs is pulled low alternately to generate the AC voltage. That does not require a H-bridge. H-bridges are not that easy to control and if there is a way to do something without a H-bridge, that is generally the better way, especially if that way requires only low side MOSFETs.

Assuming there is a H-bridge...

For the H-bridge to work the way you are describing, there needs to be a voltage about 10V higher than the 12V to drive the top MOSFET. Where is that voltage coming from? It is not cheap to come up with that voltage, hence my doubts that a company that has thought things through will use that approach.

It is more likely that the top MOSFET is a P-Channel FET and when the transistor is on, it pulls the gate of the top MOSFET low and that turns it on. The transistor will drive the diagonal MOSFET so the bridge can work. If you were to turn on both pins 11 and 14 on at the same time, everything in the bridge will come on at the same time, resulting in a short.

---55p

 

[(*steve*)]

I think at least part of the problem is that you haven't drawn the mosfets in your diagram.

Without them we're all making assumptions about how they're connected. If there are 3 in parallel, just draw 1, it will make it clearer (but note which parts are duplicated).

 

[55pilot]

Explain to me how, when the left half is hard wired with a logic inverting transistor such that the Left top is on, the left bottom MUST be off?

Explain to me, how is the left top ever turned on? If it is a N channel MOSFET, it requires a voltage between 10 and 20V higher than the source voltage. Where is that voltage coming from? If it is a P channel MOSFET, then what you consider as "off" is actually on and what you consider left top is actually right top and vice versa.

---55p

 

[cornwallav8r]

Some clarifications. All the NMOSes are the exact same part. All NMOS type. I replaced them all, so I know for sure. The shown layout is exactly how it is wired.
Regarding transformer selection, versus H-bridge...all the little inverters do it this way in my experience. MUCH lighter and cheaper to implement versus a heavy transformer. But for some reason I see AC UPS units, 2 of which I just repaired, use big heavy transformers, as you describe. I guess they do it because they want line conditioning and the other benefits of a big heavy transformer design.

There is a lot of duplicity in this unit. (8) identical stages, each with its own flyback type transformer amplifier stage, all parallelled outputs, to supply the high voltage.
The NMOS high voltage output stage is similarly duplicated, with (3) parallel NMOses for more power. It's a big unit, which is why I wanted to spend the time to repair it, and learn something in the process. And your interest and comments are most appreciated.

 

[cornwallav8r]

Circuit is correct. All transistors are NMOS, identical. Driver configuration as drawn is correct, I checked it. What I didn't yet investigate is how the 35325 driver chip gets its power and how that relates to how it drives the NMOSes? I wonder whether the chip rides atop the high voltage like they do with light dimmers and such, so that 12v vcc plus the HV voltage is applied to the gates...a scary thought for troubleshooting.
Your comments about how the Chinese are doing this...I would surmise they know what they are doing and how to do it cheaply. A million of these are sold on Ebay... There are 2 quad op amps that I haven't deciphered (all that remains as far as IC chips), that handle the overtemp/overcurrent issues, etc. The 3525 chip is capable of overcurrent control and such, and is likely doing that job using the (2) separate 3525s for the first stage of HV step up...the latter switching stage I am repairing only has to switch the driver outputs all day at a 60Hz rate, as it is wired it has nothing to do with current or voltage that I can tell.

I will be happy to add clarifications later when I further investigate the circuit. A good learning experience.

First, I have a hard time believing your circuit description. What you are describing is a really bad way to do a voltage inverter. But since I have seen some really bad stuff coming out of cheap Chinese companies, it is not out of the realm of possibilities.

The "normal" way to do an inverter is to have a center tapped transformer. The center tap stays at the DC voltage and each of the primary legs is pulled low alternately to generate the AC voltage. That does not require a H-bridge. H-bridges are not that easy to control and if there is a way to do something without a H-bridge, that is generally the better way, especially if that way requires only low side MOSFETs.

Assuming there is a H-bridge...

For the H-bridge to work the way you are describing, there needs to be a voltage about 10V higher than the 12V to drive the top MOSFET. Where is that voltage coming from? It is not cheap to come up with that voltage, hence my doubts that a company that has thought things through will use that approach.

It is more likely that the top MOSFET is a P-Channel FET and when the transistor is on, it pulls the gate of the top MOSFET low and that turns it on. The transistor will drive the diagonal MOSFET so the bridge can work. If you were to turn on both pins 11 and 14 on at the same time, everything in the bridge will come on at the same time, resulting in a short.

---55p

 

[alertlink]

Let me explain. Attached is a typical chinese inverter output stage. A full hbridge stage. So hi side driver is needed. But instead using costy driver ics it can be done with discrete components like a diode ,capacitor transistor. Action is the same. Initialy when low side fets are on V2 ,12V charges capacitor c11 c12 alternativly, whith accumulate enough chage for next half cycle to supply needed 12V to hi side fets. G1,G2 are antiphase 60 Hz inputs from 3525. I is current sense output to 3525. V1 is 170V dc from dc to dc converter stage.So output is 120V ac from ac_l and ac_n terminals.

 

[alertlink]

For reference ,a 2500 watts inverter.

For reference ,a 2500 watts inverter schematics and pics.

 

[cornwallav8r]

Thanks alertlink for taking the time to post this, and filling in the gaps. My unit is repaired is was just the one remaining 3525 IC burnt. It is interesting how there are often inexpensive discrete ways to do things. And how versatile op amps are. My schematic has (2) LM324s doing lots of control stuff, so the board is a bit more complex but for the most part your schematic matches what I have. That boost cap is the missing link to the driver voltage question. Will save these schematics, thanks for posting!

 

[maxs]

what are the values of ZD1 and ZD2 in the control stage schematic above? the ics in power stage are tl494 i think...