Help regarding Mosfet driver

[Svasaya]

hi, all
I have build by simulation a circuit of mosfet driver along with protection scheme.
source V1 is pulses with period 20m and 9m On time.
I need help regarding the protection scheme comprising R8,D2,D3,D4,C1,R9,Q3,D1
whether it should be used?
since the pulses will be turned off to the Mosfets in case of
both Mosfets failing. If one fails the pulses will be fed. If both are
failed whether will it make any difference if we keep on feeding the pulses
since Mosfets are not going to respond any way.
Is the protection scheme of any practical meaning? or should it be changed
to something else.
Any Sugesstions are welcome.
Thanks Regards.
Samir.

[newboy]

Can you elaborate more what's you going to do with this circuit and what's the problem ?

Also, why dont you try simulate in proteus and send it over here for any comments !

If I can help, I will help you but you must submit the simulation in proteus.

bye

[Svasaya]

I am making an dc 12v to 230V ac converter using push pull type transformer.
I am working on the driver section of z44 mosfets used in the design.
I was looking into the aspect of design to shut down the converter on failure of
any mosfet. Since the converter will will be wired in the house wiring any household load going
problematic or any surge load should not cause damage.
Also I will get proteus and give simulation in proteus
will take time.
Thanks.
samir

[Kwaigo]

Hi Samir,
I already have built circut on this principle, using micro and built-in comparator.
I think is very good idea, but, protection circut require flip-flop.
There suitable D-type flip-flop with reset input. Put "1" to the "D" input, input pulses to the "CLK" and output of failure sensor to "RESET". (invert signals, if you need)
Output of flip-flop - AND-ed with an input signal -  to the gate driver.
It`s make your protection circut very effective.

Its works this way:
1) Incoming pulse write "1" to the flip-flop.
2) While flip-flop is set, input signal come to the gate driver as is.
3) If fault detected, flip-flop comes "0", and stay "0" untill next pulse come. So mosfets closed, until next pulse. Its prevent oscillations and high power dissipation on the mosfets.

Best regards,
Kwaigo

[Svasaya]

Kwaigo.
Kindly can you give some more information about failure sensor
Also can you post a circuit about D flip flop. I couldn't get the
concept clearly.
Thanks
Samir.

[newboy]

One more thing ... if mosfet failure, is it in open circuit or short circuit ?

From my experience (not much experience actually), the failure mosfet act short circuit.

So, that's why your circuit is important prior any advice on the sensor that's going to be attached to.

 

[Svasaya]

I further found by study that the circuit can be effective for preventing over current situation.
Also if mosfet fails then it's gate and drain gets shorted.
It is important not to feed that voltage to gates of other mosfets.

[Walkura]

From my experience almost always the drain and gate gets fused together .
Only drasticly exploding mosfets might break all the connections .
When i look at your schema i have a few questions about it .
Since your gate probably gets fused in case of problems your drainvoltage oft destroys your drivers c.q. mcu .
Ending gatepulses is a nice idea but its surely not your biggest worry .
Why not use a standard totempoledriver behind your emittercoupled schmitt trigger input instead of this  combi of rc timer & lots of diodes ?
(which in this case you do need to properly close the mosfet)
That will greatly reduce your switching losses (if you lower your gateresistors anyway) .
Also if you actualy want to know if your mosfets work why not measure your drain voltages possibly controling the enable/shutdown pin of a dedicated gatedriver .
Without the Vbatt&Vsupply its not possible for me to calculate your switching speed .
But when i calculate as example 45 Volt & 75 nQ total gatecharger the switchingspeed was ~4 uS instead of the 50 nS's i usualy choose myself .

Also since you use 2 mosfets it is clear you intend power .
With powers like that' your bigger concern is how to stop it from burning down in case of anything instead of your gatepulse .
I mean your gatepulse would be shorted but your Vbatt would also be shorted through your load ,in case of an inductive load meaning shorted all together very very fast .

For proper assesment i would need more data like what is your intention and what voltages & currents are at play.

Good luck.

[Kwaigo]

Hmm, I`m think about short circut/current overload protection. Not a mosfet break.
In one of these circuits MOSFET channel used as Isense resistor. I think, MOSFET is sufficient reliable device, so its never fault, when all protections is ok.
Anyway, here is two ideas:
(quick draw, may have some mistakes)

[MAXPAYNE]

Hmm, I`m think about short circut/current overload protection. Not a mosfet break.
Anyway, here is two ideas:
http://rapidshare.com/files/176510985/Sheet1.pdf.html
http://rapidshare.com/files/176510986/Sheet2.pdf.html
(quick draw, may have some mistakes)
In one of these circuits MOSFET channel used as Isense resistor. I think, MOSFET is sufficient reliable device, so its never fault, when all protections is ok.

Why u r using rapidshare for such a small file. u could have attach the files here.......

[Jehan]

I think that it will not matter if you give pulses or not after the mosfet has failed, even if one mosfet is failed, the other will not work because mosfets generally fail as a short circuit,   a fuse is reguired in the power supply for protection.

[sam_des]

Hi,

A fuse in power-supply rail will not protect the control circuits since ususally fuses(even quick-blow) takes hundreds of milliseconds to blow. I've seen even a optocoupler breached before fuse.
It is better to use a zener diode, usually of 12V, from gate-to-source very near to FETs. Now whenever FET blows to short, supply rail will be shorted to gnd through zener breakdown. Heavy currents will blow the zener, but will protect driver & all other circuits. You will only have to change FETs & zener.

For over-current/short circuit you can do that sensing mosfet drain-source voltage, only when MOSFET is being driven. You can calculate the max safe Vds, by using your max current & effective Rds(on) of your selected mosfet(s). Use comparator to turn off gate pulses whenever this Vds exceeds safe limits. You can google for further details-this is known as MOSFET Desaturation Sense.

Hope that is useful. I'll be glad to help further.
regards,
sam_des

[Svasaya]

The Frequency of converter is 50Hz. 12Vdc to 230V ac.
Also the load to Mosfets is the inductance of push pull transformer.
the reason for using discrete s is to have a low cost design. also power handling will be around 200 - 250W.
Here is proteus simulation file. I could not get a 12-0-12 to 230v Transformer insted i have used 12-0-12 to 12v transformer. If any can suggest how to I would be grateful to.
This ckt will act as an over current limit.
If load is reduced from 0.3 Ohms to 0.2Ohms the current limit in action can be seen
as the gate pulses are not passed fully.
Is the design ok?
Samir.

[sohel]

Hello,

here some design(bmp) for ur requerment. its 24vdc to 230ac,

please dont asked for sorce code

[sam_des]

Hello Svasaya,

Unable to open your dsn file with my ISIS 7.2SP2. Nevertheless I attaching my idea of sc/ol sense & driver for push-pull inverter. Check it out.


Hello Sohel,
Are those 4x5W resistors for current sensing Are they wire-wound You should avoid them since wire-wound resistors have too much parasitic inductance which hampers the mosfets very badly. You can have low-inductance ww resistors, but they are much costly.

I think it'll be better to use fuse in DC rail itself for current sensing. These fusese have very low resistance- around 0.0022 ohms & inductance. They are used in auto-industry & have current-rating of 25,30,40 Amps.

Also if possible these sensing must be done in FET drain rather than in Source.

Hope that's useful.

regards,
sam_des

[sohel]

r u talking about shunt? if yes then u should use an 12no copper wire arround 2 inch with replacing 0.01R resistor. and useing fuse u may use.

[sam_des]

Hello Sohel,

I didn't get your question clearly...
AUTO-FUSEes I am talikng about are simple battery-fuses used in vehicals. They have very small size 8mm x 12mm, colored(green/red/yellow/orange) plastic body with rectanguar inseration leads costing around 1.5-2 Indian Rs. I will post the image if I could. I have measured their resistance on a R-merter from local ww resistor manufacturer, it is on avarage 0.0022 ohms, inductance is low, can withstand heavy currents for quite a few times before breaking open.

On the other hand shunt resistors are specifically made for current sensing,they are much costly as the current rating increases. But they are also more accurate than auto-fuses.
Since nowadays, demands on dc-ac inverter's cost are extremely high(atleast here in India), that shunts are no longer a suitable option.


Using a copper wire as sense R is very bad option if you are placing it in FET source lead. That will add inductance between FET source & system GND(i.e.Battery -ve). This inductance affects the FET switching, especially turn off. This may not be of much significance if switching frequency is low(& have large dead-time), but on high frequencies this must be avoided at all costs. Even the copper tracks between FET source & BAT -ve should be minimized & have as large as possible copper area, since they too, add to inductance.

regards,
sam_des

[sohel]

its good news that u r an indian. i have soft corner for indian, they are most helpfull.

i know u talk about 40 amp fuse but when u assemble in circuit u should use 2 pin base. most of delevered product will come to ur door for replace fuse. so what do u want ? if u r invited to come to u then u should use.

forget using fuse u will sleep.

[sam_des]

Hello Sohel,

Thanks for your advice. But I certainly know what I am talling about. I earn my bread as consultant & have designed inverters/ups from 10W/10000W, Battery Chargers(from 0.1A-100A), SMPS(1W-5000W).
My clients have delivered thousands of inverter/ups systems, not only in domestic but also in stringent, non-forgiving industrial market. After 10 years, I never had single complaint about those fuses.
It all depends on how the system is designed. If designer has done his job thoroughly then there are very few reasons for system to fail.

BTW, it is nothing personal, so don't get it wrong -- but I don't have any soft-corner for ...

reagrds,
sam_des

[sohel]

please sir. come to main topics.

[Svasaya]

Hello.
Sam_Des : Here is isis 7.0 file.
I have seen fuse ur talking about, for current sensing purpose can it give consistent res.
I have made an 18Swg wire of 110mm length. what do u think about it
I am gaining a lot from ur discussion and posts
But pls do not get personal.
Regards
Samir.

[tAhm1D]

Hi Svasaya,
Your effort is praiseworthy. As I have involved in designing and manufacturing with lot of Inverters/Power Supply products- though mostly in smps, not in Linear Type, I want to advice you one thing, instead of venturing in Mosfet Driver, you can use ready made Mosfet Driver ICs with little cost involved. Those ICs have inbuilt oscillator,driver, safety parameters and also feedback facilities. For Low Frequency, you can use SG3526 ic. This is a sincere advice from a Sonsivri friend and not an effort to distract you from your effort.

Hi sam_des,

For a Power Electronics Designer, using a copper wire as sense R and using inaccurate AUTO-FUSEes in place of shunt resistor are both not a proper option. But, if those are workable and can be used on ground reliably with less cost, I find nothing wrong and that is done efficiently by Sohel and you. Cost matters in this Sub-Continent.

Sohel can use the wire as he is using that in Low Frequency Linear Type Inverter and that design of Inverter is Indian Origin and I have seen that in Inverters in Kolkata and Delhi. As you have mentioned, yes, that cannot be used in high frequency application and Sohel  is not applying it in smps as his design is Linear Type.

One more thing. We are all Members of Sonsivri Family from different countries and please do not get personal as told by Svasaya.

[Kwaigo]

Hi, Samir.
I had tested your proteus model. I find, that protection schematic is not work properly.
That`s because gate driver too weak, and fronts too long.
You must:
a) Make gate driver some stronger.
b) Add circutry, that bloks MOSFET desaturation sensors when it`s switching.

P.S. Resistors R8 and R108 - too large. Must be approx. R7 range.

Best regards, Kwaigo

[sam_des]

Hello Samir,

Just seen your pdf. Havn't run on isis till. I suggest,

1) Add gate-source res to each mosfet of about 10k to prevent spurious turn on during power-up & down.
2) Purpose of D1 & D53 diodes are allowing fast turn-off by bypassing the gate res. There should be seperate diode(1n4007) for each mosfet, directly from gate to driver for fast sinking.
3) R7/107 should be from mosfet gate directly around 2k2.
4) Why don't you use simple CD4049UBE or CD4050B as gate driver. They are cheap & have enough source/sink capacity and are suitable upto few kHz.
5) Also LM358's dual op-amps can be used for both overload as well as short-ckt detect.
6) D6/D5 should both must very fast turn-off diodes with same forward drops. Should be 1n4148.


BTW, I had already said that it wasn't personal, but some experiences do affect one's emotions.
Any way, I did not intend to hurt anyone nor our forum's multi-national culture.

regards,
sam_des