Optocoupled H bridge gate driver circuit??

[shazan]

Hi
I am looking for a way to drive an isolated MOSFET half bridge - later maybe a full bridge also - using optocouplers and discrete components. My idea is to work at 20KHz or so max.
I found this ckt from the IRF website.

http://www.irf.com/technical-info/designtp/dt94-12.pdf

I have some questions, though
1. I see there is an isolated supply formed with the zener etc, so if I duplicate this circuit taking the reference from the bottom of the bridge, can I use this ckt(original plus duplicated) for a half bridge gate drive?
Please advise.

Thank you all for so many enlightening hours spent in this valuable forum.

Please pardon any mistakes I may have made. I am a relative newbie in this field.

 

[karesz]

Hali,
I think a "half bridge" means; it has two FETs between switched voltage & GND, and your load is to the middle of FETs connected...
In that case I cannot see that its your solution :-(.
K.

 

[FvM]

The circuit can work for a half bridge, but I think, it's not safe. It's not protected against loss of control circuit supply
(causes a bridge short) and will power up in an unclear state, possibly driving the transistors to linear mode with high losses.

 

[shazan]

Hi
Thak you for your responses.
I mean to use it in a half bridge circuit, with the top switch switching ON when the bottom is OFF and vice versa.
Thank you FvM, for the heads up on the power sequencing. This circuit is going to be sequenced ON after the logic/control system is ON.
I will also make sure that I keep some dead band so as to prevent any shoot-through.

 

[FvM]

In addition, you should consider the rather low gate drive capability, 30-50 mA for switch on, at maximum 100-200 mA for switch off.
It's only suitable for driving small FETs respectively slow switching, I think. Typical driver ICs (e.g. from IRF/Fairchildsemi) provide
1 or 2 A gate current.

 

[shazan]

Thanks again, FvM. I was aiming to use it with a drain voltage of 300-340VDC, so is there any way, if I scale the components values (wattage, amps, etc) I will be able to use this circuit? Some guidance in how I should approach my need here would be most appreciated.
I have been looking at IR2110 as an alternative, but am a bit chary about its a. Cost and b.Can it be isolated from logic completely?
Are there any other discrete opto isolated circuits that I can use which resolve all the issues I now face? 340VDC, logic input, isolated GND, 1-2A gate drive......

 

[karesz]

Hi,
You can drive an IR2110 exactly similar as these FET driving-even with an optocoppler...
In my opinion is a an "ADUMxxxx Digital Isolator" i.e. from Analog Dev. is the better way...
K.

 

[shazan]

Hi Karesz
You're suggesting I use an opto with an IR2110? Maybe you're right, I should stop breaking my head and go with the easiest way out.
I have looked at the ADUM chip but found it a bit costly, almost the same as an IR2110
I really would like to build a discrete driver, though.

On another topic, I dont see any "helped" button? or is there a lower limit to number of posts before I am able to virtually thank you guys?
regards
S

 

[FvM]

The main disadvantage of IR2110 comapred to the presented is that it needs a low voltage supply of 12-15V, common ground
with the DC bus negative node. But I doubt, that you can easily "harvest" the supply for an 1A to 2A gate driver through series
resistors. The input of IR2110 could be in fact driven by logic optocoupler.

 

[karesz]

@shazan,
you can see the "helped button" only if you are the iniciant of the topic...
Greetings!
K
P.S.:
ADUM has newer generations too, & dont forget pls, it has multi channel, and from other vendors are similar functions to find, can be some cheaper ...

 

[shazan]

@Karesz:
Thank you. I am the starter of this topic, but I dont see a "helped" button anywhere?????
I will look for similar ICs, like you said, maybe some are more reasonably priced.
@FvM:
Thanks again. I'll look for some chips like Karesz said, before deciding on one. I found this TLP250/350 from Toshiba, thats being touted by some as an optocoupler with 400V and 2.5A capability. I understand it will need a 2.5A supply derived from the DC bus to function. I could replace the entire irf circuit with that, is my thinking going on the right lines?

 

[FvM]

I don't see a voltage quantity in the order of 400V in the TLP250 datasheet. It's an optical isolated gate driver, that needs n low voltage
supply. If you want to derive it from the bus voltage, you most likely need a DC/DC converter, in some cases a resistor/Z-diode
supply may be possible (with rather high resistor power dissipation). In any case, you should determine the expected dynamic current
consumption from gate capacitance, gate voltage and switching frequency. With moderate switching frequencies and gate capacitances,
the dynamic current is probably low, so you mainly have to provide the static 11 mA supply current of the device. The 1.5A peak gate
current will be sourced from bypass capacitors-

 

[shazan]

Hi FvM
I agree, but this is a page I found, where the 400V is mentioned
**broken link removed**
I will most probably use the TLP350, lets see.
Thank you - and Karesz - for your very perceptive advice and suggestions.
bye for now.

 

[FvM]

I guess, the vendor wanted to suggest 400V working voltage (mains) for the inverter, which sounds reasonable.

 

[shazan]

Hi again
@FvM: This is just an idea, Please point out any mistakes I make in this:
Suppose I replace the supply from the IRF circuit (presently using a zener with a big series Resistor) with a series pass transistor regulator (with BJT of appropriate VCE limit), and scale up the components and the bypass cap to be able to charge the Ciss fast, won't the circuit be almost the same as the TLP250/350, except where the photodiode is being used so speed up the TLP, but with my freq being 20-25kHz, should not matter too much since I have 40-50uS to charge the bypass?

 

[FvM]

I have listed the disadvantages of the IRF cicrcuit compared to a state-of-the-art driver (as e.g. TLPx50) in my above posts.
Just to repeat it in a brief:
- no fail-safe (output on with control current off)
- no low-voltage lockout
- low drive current
Not yet addressed, but also a possible issue is the rather large delay and delay skew caused by the slow optocoupler. It's particularly
important in bridge (push-pull) operation.

Some points can be of course improved by changing the design. The main advantage (low quiscent current) would be lost if
using a fast logic optocoupler to reduce the switching delay.

 

[shazan]

Thank you FvM
I will stick with TLPx50 for my design, this was just me stretching a lil