What is the best heat transfer method

[kinarfi]

I redid the heat sink and the boards holding it together, put a new board on it for the FAN7842s and later remove the fan7842 and I am putting totems on to do the job of FET driver. I'm still working on the cct board, got some cuts to do on it.

 

[ronv]

I like just the push pull for drivers. Here is one. You might be able to find faster transistors with fairly high current. The problem I see with this is that your comparators have a pretty slow rise time (about 5 usec. I think) so the shoot thru current is going to be high. I made this circuit initially for a fast fet driver. 20, 30 ns. You can change the rise time in the sim and see what happens to the shoot thru current. So maybe just a low side driver would work best. I found some at mouser.

 

[kinarfi]

Like I said, my spice doesn't like totems, I had to change 2N2907 to 2N4403, and I labelled every wire, but look what i get.
The totem I'm using in push pull using 2n222 & 2n4403, each base has 3K to signal, emitters tied together.

 

[ronv]

Maybe skip the initial operating point in the place where you set how long the sim runs. If not post your esc file & I'll look and see if I can figure it out.

 

[kinarfi]

I need a new model for 2N44032, I did a pick new transistor and picked 2N2905 and it simmed beautifully, now to go find a good model, Here's a sneak peak at what I got , need to tweek a few pots and put some input and output terminals on and I think I'm ready to go. I didn't totem the PFETs because they turn on and off while the NFETs do the PWM.

 

[ronv]

OK, good. So If you are really using that comparator have a look at the current thru the .1 ohm resistor when the rise time is set to 5 or 6 usec.

 

[kinarfi]

These are more correct, I have found spice likes using Linear's component and if you use imported models and subs, sometimes it slows the sim down to a crawl. My models for LM358 and LM393 don't run and I haven't found any that do, I used Linear's components, I'm gonna get shot for this, because for the most part, an opamp is an opamp and a comp is a comp.

 

[ChrisP58]

There is a problem with driving an N and a P mosfet from the same driver. There is a period of time when the common gate voltage has both mosfets fully on. The time may be very short, but it's not zero. This will cause a large shoot through current through the two mosfets.

As I mentioned before, allowing both mosfets in a vertical pair to be on at the same time is a Very Bad Thing.

I suspect that insufficient dead time during transitions may be the root cause of some of your meltdowns.

 

[ronv]

Take a look at this one. I think I understand your comparator set up. One will always be low ( 0 volts) and the other is PWM. Is that correct?
So this driver pretty much eliminates the shoot thru Chris is talking about. To turn on the FET the gate voltage has to go thru the gate resistor. Because the FETs have gate capacitance it slows down the turn on time. During turn off the gate current goes thru the diode making it turn off quick. It is kind of a compromise between slow turn on time and shoot thru. Both waste some power.
But I think Chris is right you can use a little heat sink if you get it right.

 

[kinarfi]

Take a look at this one. I think I understand your comparator set up. One will always be low ( 0 volts) and the other is PWM. Is that correct?
So this driver pretty much eliminates the shoot thru Chris is talking about. To turn on the FET the gate voltage has to go thru the gate resistor. Because the FETs have gate capacitance it slows down the turn on time. During turn off the gate current goes thru the diode making it turn off quick. It is kind of a compromise between slow turn on time and shoot thru. Both waste some power.
But I think Chris is right you can use a little heat sink if you get it right.

Can you please upload the .asc file for this?

 

[ronv]

Sure, but we need to get your models working because it depends on the actual fets and the comparator.

 

[kinarfi]

Take a look at this one. I think I understand your comparator set up. One will always be low ( 0 volts) and the other is PWM. Is that correct?

Yes

So this driver pretty much eliminates the shoot thru Chris is talking about. To turn on the FET the gate voltage has to go thru the gate resistor. Because the FETs have gate capacitance it slows down the turn on time. During turn off the gate current goes thru the diode making it turn off quick. It is kind of a compromise between slow turn on time and shoot thru. Both waste some power.
But I think Chris is right you can use a little heat sink if you get it right.

The controlling Hall device is tied to the steering wheel of my OHV, so the mechanical speed is super slow compared to the electronics, OR SO I THINK!
The voltage from the Hall goes up and down from 5.51 which is changed to 5.0 by the first op amp, 1a that is sent to the Pfet drivers , oops- not labelled, LM393 driving P1&P2 and is compared the voltage divided by 2 1Ks and it has a dead band created by the pot. Is that not enough of a dead time to allow the fet to turn off before the other turns on, or am I not understanding properly.
the 5.0 is also mixed with the triangle for the pwm using the same voltage dividing method with the dead band in the middle.
Thanks for the .asc

 

[kinarfi]

Sure, but we need to get your models working because it depends on the actual fets and the comparator.

comparators are ST LM393 8 dip PFET is IRF4905, NFET is IRF3205, opamps are LM358, 8 dip
I did reload the scad3.exe, it's been updated lately, and that's where I got my functioning 2N4403, still cleaning up some other files.

 

[ronv]

Lets try this to get you going.

In spice create 2 folders named new. One in the Lib/sub file and one in the lib/sym file.

Unzip these files and save them to their respective sub and sym folder as new. First one is the sub... second the symbols.
There may be a couple in there that don't work, but most do. I haven't cleaned house in a while.

Now when you click on the add component file you will see a file called new (kind of like misc.) Click on it and you can see what parts are in there..

 

[kinarfi]

can you send me your LM393. sub file? I think the ones I've downloaded were Pspice you beat me downloaded and looking at them

 

[ronv]

The controlling Hall device is tied to the steering wheel of my OHV, so the mechanical speed is super slow compared to the electronics, OR SO I THINK!
The voltage from the Hall goes up and down from 5.51 which is changed to 5.0 by the first op amp, 1a that is sent to the Pfet drivers , oops- not labelled, LM393 driving P1&P2 and is compared the voltage divided by 2 1Ks and it has a dead band created by the pot. Is that not enough of a dead time to allow the fet to turn off before the other turns on, or am I not understanding properly.
the 5.0 is also mixed with the triangle for the pwm using the same voltage dividing method with the dead band in the middle.
Thanks for the .asc

I think one of the comparators will be minus while the other is doing pwm. So the PFET on one side will be on all the time and the NFET below it in the H always off. But on the other side when the PWM is running the PFET will be on when the NFET is off, then the PFET needs to turn off and the NFET on. It is on this side that they are both on for a short period of time and there is nothing to limit the current except the resistance of your wires and the FETs. It's not a long time, probably less than a microsecond but the current is really high.

 

[kinarfi]

run this .asc and look only at the P1, P2, N1 &N2 and you'll see how I have it figured, then run the hall, MXD and the inputs to the comparators
It is even more clear if you delete the load

 

[kinarfi]

Another thing to remember is that the TIME is actually of no concern, the hall is steady at 10 to 11 ms and as I turn right, hall volt move to the right of center and vise versa left. I need to redo the hall so really show this,

 

[kinarfi]

With the previous explanations, do you still see a problem with my set up? , please & thanks
Jeff

 

[ronv]

I'm having a little trouble getting the sim to run right. Where does FB come from? Something on the steering? How big is it?