High-Side N-Channel Driver Voltage Problem

[NotANewbie]

I've got a real tough and complicated problem here that I am having trouble solving without some help.
I'm building a battery charger using a DC-DC buck converter. 
I am using a N-Channel mosfet as the high-side switch.  The driver for the mosfet is an IR2110, which has both a high-side and low-side mosfet driver in one chip.   

Now I have built a BLDC motor controller before using n-channel high-side switches.  I used LT1161 chips in those, which work great, but they are limited to 48vdc, and I need to charge a 48v battery with a 60-75v input.   
I have also built higher voltage BLDC motor controllers using the IR2110 (or equivalent, IR2130),
and that works good too.
So no problem, right?  Test charging a 3v battery from a 24v power supply all works great. 
Then I go to charge a 21v battery and it doesn't work. 
What is the problem?  It is the high-side charge-pump boost voltage.   
This works for a motor controller or other load, because when the mosfet is turned on, the source terminal has a path to ground,
and can charge up the boost capacitor.  But for a battery charger, there is no path to ground!!  When the high-side is switched on, the source terminal is now connected to the + side of the battery, which at 3v there is still headroom for the cap to charge above 10v, but at 21v the boost cap can only be charged to 3v from a 24v charge source.   Dooooooh.   Now what???

After hours of google searches for a solution, there were none to be found.
Well of course I can use a P-channel mosfet instead and the problem goes away, but that is just not an acceptable answer,
we need the efficiency of the N-channel mosfet even at the cost of more parts for the driver circuit.

OK, I figure we just need a dc boost converter that will boost the voltage of the battery being charged to 10-15v higher,
and we can feed that into the IR2011 VH input, and sure enough this works!   Umm, but no it doesn't.
A simple cheap LM2577 type of dc-dc boost converter will not work with over 30v input or generate more than 48v. output,
so while it works with the 21v battery, it will explode with the 48v battery. 
And yet here is another major problem, that battery voltage GOES UP as it charges, so if we have a DC-DC converter
that takes 21v and produces 36v (all good!)  then the battery charges up to 27v, and now 36v is not enough!   (all BAD)

So what we need is a floating input to the DC-DC, it only needs say battery+ and battery+(-5v) as input, and can then
generate battery+(+15v) as output.  As long as the DC-DC floats, it won't be limited by the battery voltage (could be 200v, no matter).

So this works also!   Yeah, I am a GENIUS...NOT.   Because how do we get a floating 5v input?  Well I used a -5v regulator,
and that WORKS!   But now I just shifted the problem of 70v inputs to the voltage regulator, and no -5v regulator can handle more than 40v.   

So I turn to EEVBlog for help.   HELP! 

[NotANewbie]

Thanks for the quick reply.  I'm not getting how this solves the problem(s)?

What you need is an isolated DC-DC converter.  You can buy these, but you can also make them pretty easily with an LT8301 from Linear Technology.  It is a flyback type converter that provides an isolated output [or outputs], and it does not require the normal opto-isolator type feedback circuit.  Check it out here:

http://www.linear.com/product/LT8301

-Ken

[NotANewbie]

Right, I have used a separate power supply, the problem is the input and output voltage range are out of bounds for the battery voltage.
This chip is limited to 42v input, and I can have 48v.   A transformer might help to boost the output voltage.  I still have the problem of regulating the boost level to only 10-20v above the battery voltage.   I also noticed these transformers are not cheap.

What would really solve my problem is a -5v regulator that can handle 100v.
What I do now is use a LM7905 to take 24v input and produce 19v output, which gives me a floating (-)19v (+24v) 5v input
to the dc-dc and generates +20v output, which is +15v from the +24v battery high.
This works great, but a LM7905 will not work at higher battery voltages. 
I can use a zener diode on the ground connection of the lm7905 so it doesn't see more than 35v, but now I'm really locked into
a restricted voltage range of the battery I can charge.

[ajb]

Why do you need a -5V regulator do drive a high side N FET?  Are you talking about generating a point 5V below the battery positive for bootstrapping the high side gate driver?  You could use an isolated d DC-DC converter to directly provide a positive supply for your highside gate drive.  You'd connect the (isolated) Out- to your gate driver's common and and the (isolated) Out+ would provide the gate driver's positive drive supply.

[Zero999]

Not only do you need an isolated floating PSU for the MOSFET but you also need to isolate the gate drive with an opto-coupler.

[NotANewbie]

Why do you need a -5V regulator do drive a high side N FET?  Are you talking about generating a point 5V below the battery positive for bootstrapping the high side gate driver? 

Exactly, so the dc-dc floats and generates the 15v above the bat+

You could use an isolated d DC-DC converter to directly provide a positive supply for your highside gate drive.  You'd connect the (isolated) Out- to your gate driver's common and and the (isolated) Out+ would provide the gate driver's positive drive supply.

Ahh, ok, so yeah because it is isolated, I just need a straight 15v output, not bat+15v,  to feed the IR2011.
And could then use something like this:
http://www.digikey.com/product-search/en?KeyWords=NMA1215SC&WT.z_header=search_go
which at $8, isn't too expensive.  I don't really want to fab one up myself.
Or this one is even less expensive, and I have a solid 5v supply, my 12v power is not neccessarily within the 10-13 range.
http://www.digikey.com/product-detail/en/CME0512S3C/811-2892-5-ND/4693720
Thanks for the help!   

[Seekonk]

I have found that many small electronic wall warts (12V) work fine at about 50V DC input, that is the power point voltage of my 36V solar array.  At that reduced voltage the power is about 1/10 the current rating.  That is more than enough to run my micro boards and drivers.   Bust them open and remove the resistors for EMI and the diodes used and they will work at even lower.  At reduced voltage ans accompanying higher current resistive losses are even greater.  Choose one rated only for 120V for best luck. I always grab these wall warts at a garage sale for a quarter, a cheap isolated supply.

This I almost hate to admit.  Instead of a capacitor bootstrap, I have a battery bootstrap in my buck converters.  I get free batteries from town recycle.  That insures the driver chip has full voltage when first starting.  It was a temporary setup to get the buck converter running.  It has been running that way a couple years.  The battery recharges in the same way as a capacitor does.

[NotANewbie]

So everything is peachy with the isolated DC-DC drive for the high-side.  Battery Charger works great.

But now I've been trying to make a simpler board using the IR2011 driver instead of the IR2110,
and it just keeps blowing up!  And because it blows up and turns on both the high and low side mosfets,
it blows them up also.   After weeks of this, and dozens of blown up parts, I once again turn for help!

I'm not really understanding why the 2110 works, and the 2011 crashes and burns.
The difference between the two driver chips is the 2011 has a separate power for the logic (5v),
and the low-side drive (15v), but the 2011 has just a single Vcc it uses for both.
To make it a little easier for you, here is the datasheet on the 2011 and 2110
http://www.irf.com/product-info/datasheets/data/ir2011.pdf
http://www.irf.com/product-info/datasheets/data/ir2110.pdf

I thought I'd licked the problem by adding another 1k pull-down resistor to the input pins, but that only
delayed the time when it blows up.  Sooner (like sometimes immediately) or later, both high and low are turned on
and everything blows up.   And the latest is without ANY inputs to the LIN and HIN inputs, only the 1k pulldowns,
and yet the chip is STILL blowing up.   My google searches did not find much, only one poor soul with the same kind
of problems and got no help and never came back to post he gave up or fixed the issues.

So basically what I'm I doing wrong?  something must be wrong.  The circuit is setup as the app notes go,
the same as the 2210 but no 5v supply.    One difference is the current setup is for 12v AC, and so I don't have a voltage regulator
to create 15vdc, I'm just using the rectified AC power (16vdc) to power the driver chip.   
Somehow I think it is getting over 20vdc that blows it up.  However, I added a 18v zener so it clamps anything over 18v, and had no effect, i.e. still blows up.
Any help appreciated, about as my wits end.

[moffy]

Biggest difference is in the voltage ratings, 200V vs 600V. If it is blowing up when you have no input then maybe it is oscillating?

[NotANewbie]

From what I've been reading, e.g. http://www.onsemi.com/pub_link/Collateral/TND379-D.PDF
I'm thinking that maybe the issue is the negative transient from the rectifier.
This document however, does not give me any clues on how to avoid these transients, what can I do?
My rectifier is four mosfets, using the intrinsic diodes in them to form a full bridge rectifier.
I don't see such transients on my scope, but my scope sucks, and it looks like I damaged my new one when
I attached it 1x to 120vac.

I note that I was originally inverting a solid 12vdc into 12vac using the driver chips and they blew up,
and I've since simplified to just one driver turnin on a high-side mosfet.

[diyaudio]

So everything is peachy with the isolated DC-DC drive for the high-side.  Battery Charger works great.

But now I've been trying to make a simpler board using the IR2011 driver instead of the IR2110,
and it just keeps blowing up!  And because it blows up and turns on both the high and low side mosfets,
it blows them up also.   After weeks of this, and dozens of blown up parts, I once again turn for help!

I'm not really understanding why the 2110 works, and the 2011 crashes and burns.
The difference between the two driver chips is the 2011 has a separate power for the logic (5v),
and the low-side drive (15v), but the 2011 has just a single Vcc it uses for both.
To make it a little easier for you, here is the datasheet on the 2011 and 2110
http://www.irf.com/product-info/datasheets/data/ir2011.pdf
http://www.irf.com/product-info/datasheets/data/ir2110.pdf

I thought I'd licked the problem by adding another 1k pull-down resistor to the input pins, but that only
delayed the time when it blows up.  Sooner (like sometimes immediately) or later, both high and low are turned on
and everything blows up.   And the latest is without ANY inputs to the LIN and HIN inputs, only the 1k pulldowns,
and yet the chip is STILL blowing up.   My google searches did not find much, only one poor soul with the same kind
of problems and got no help and never came back to post he gave up or fixed the issues.

So basically what I'm I doing wrong?  something must be wrong.  The circuit is setup as the app notes go,
the same as the 2210 but no 5v supply.    One difference is the current setup is for 12v AC, and so I don't have a voltage regulator
to create 15vdc, I'm just using the rectified AC power (16vdc) to power the driver chip.   
Somehow I think it is getting over 20vdc that blows it up.  However, I added a 18v zener so it clamps anything over 18v, and had no effect, i.e. still blows up.
Any help appreciated, about as my wits end.

IR2011  or  IR2110 will depend on supply voltage, operating frequency and total gate charge. Switching times are directly related to gate charge too.

The datasheet contains plots of chip temperature increase versus operating frequency for various reference MOSFET models.

Have you measured propagation delay?, probe and show us your and signals entering and leaving the chip. IR2110 has the lowest propagation delay, maybe you experiencing shot through, a DRC propagation delay maybe help you.