[Mariss Freimanis]

Wow! That comes as a shock to me; IR2104s not suitable for driving IRF540s.

Here we've gone and built nearly 40,000 drives this year, used up 320,000 IRF540s and 160,000 IR2104s just to find out now it doesn't work.:-)

Yellow trace: IR2104 driven IRF540 gate to source voltage.
Blue trace: IRF540 drain to source voltage.

Test conditions: 62VDC supply, 600uH winding at 7A.

Darn!

Mariss

[Mariss Freimanis]

Not cool. It is a pirate copy of our G201 REV3.

Mariss

So the concusion is that the A3986 can drive an IRF540 as well as the IR2104 (roughly)

See http://www.protobyte.com/images/P2000/P2000-03.jpg

I've used this drive before and it runs very cool even at 6 amps.

[kreutz]

Mariss;

I had never seen any of your boards before. I am sorry for the comment!
What switching times are you getting with the IRF540 and IR2104?

Thanks.

[kreutz]

Mariss;

[QUOTE][Wow! That comes as a shock to me; IR2104s not suitable for driving IRF540s.
/QUOTE]

That is not the conclusion, read Riko's post

No worse

--------------------------------------------------------------------------------

So the conclusion is that the A3986 can drive an IRF540 as well as the IR2104 (roughly)

On one of my prior post I stated:

The IR2104 is not a good choice for driving the IRF540. Short circuit current specifications are for Vo= 0 volt with Vcc= 15 volt (for Output high) and Vo= 15 V for Output Low short circuit current (< 10 microseconds). No test circuit details are given from International Rectifiers.

The IR2104(S) is not recommended for new designs, and nowadays there are many drivers on the market that can offer faster turn on and off times for the IRF540 than that part, in the same, or lower price range (for small quantities).

Thanks for the picture!

[Mariss Freimanis]

kreutz,

I meant my reply to be humorous and certainly no offense was taken.

First off I have a typo; we use International Rectifier IRF540N MOSFETs only. The intrinsic diode reverse recovery behavior is important. The IR parts have a "soft" instead of a "snap" recovery.

I like to limit Vds dv/dt to no faster than 3V/ns; MOSFETs get into trouble above 50V/ns. This is accomplished by limiting Vgs drive current and the IR2104's source/sink of 130mA/260mA meshes nicely with this requirement.

The switching scope trace is shown in message #161. It's darn hard to properly instrument the scope probes for accurate readings at these speeds. The traces were a "quickie" using Tektronix P3010 probes with 6" (150mm) grounding clips. I was too lazy to get out my really good probes.

Mariss

[kreutz]

Mariss;

The switching scope trace is shown in message #161. It's darn hard to properly instrument the scope probes for accurate readings at these speeds. The traces were a "quickie" using Tektronix P3010 probes with 6" (150mm) grounding clips.

I know it from experience. You've got a very clean trace!, it doesn't even look that you were using long grounding clips at the time!

When I first read your post I had not logged yet so I did not see the picture before I asked for the switching times.

[Mariss Freimanis]

"The IR2104 is not a good choice for driving the IRF540. Short circuit current specifications are for Vo= 0 volt with Vcc= 15 volt (for Output high) and Vo= 15 V for Output Low short circuit current (< 10 microseconds). No test circuit details are given from International Rectifiers."

Not quite sure what to make of the above. What I take the IR specs to mean is the CMOS outputs have a p-channel Rds of about 100 Ohms and an n-channel Rds of about 50 Ohms. The 100uS spec is probably the secondary breakdown SOA limit. No circuit details? Then build your own evaluation circuit. I have tested and confirmed these specs by working into a resistive 10 Ohm load.

100uS certainly not a problem; a MOSFET gate is all done switching in 500nS.

Altogether we have used over 300,000 IR2104 half-bridge drivers, almost 160,000 this year. In that time we have had only 3 confirmed IR2104 failures in our database.

Yes, we build a lot of drives. We ship about 200 a day now in a large part owing to the reliability and ruggedness of the IR bridge drivers and MOSFETs.

Mariss

[Mariss Freimanis]

"I know it from experience. You've got a very clean trace!"

Thanks! The yellow trace doesn't show the Vds plateau at 6V wich most assuredly is there. It's being washed out by the trr current spike common mode voltage which puts a false inflection point in the waveform. Note the 20nS wide low -dv/dt at the 200nS mark on the Vds blue trace. That marks the trr current spike.

Mariss

[Mariss Freimanis]

"Information needed to make the Spice model should be provided by the manufacturer."

A agree up to a point. I use Spice for the first "rough cut" of a circuit during the initial to middle stage of a design. It serves mostly to catch conceptual oversights or as a means to iteratively navigate thru a difficult and an intially partially understood problem.

At some later point empiricism replaces the analytical; you have to build a prototype to test an idea. Prototypes are expensive in time and money so this stage is best delayed until there is a better than even chance it will work, sort of, on the first cut.

Then old fashioned 'scope troubleshooting, Exacto-knife and soldering-iron skills come into play. The prototype is basically bludgeoned into working.

The final iteration takes the result of these gentle skills and integrates them into the original analytically-derived design. The result is something fine that works.

No one tool is complete by itself to do a major task.

Mariss

[misterstarshine]

That was a serious discussion about the mosfet driverstage of a stepper driver. I can't really claim I catched up on the details but I guess it's all about if the system can switch fast enough at the desired voltage and current rating so that the driver performs well without breaking.

There is certainly less discussion about the anti-stall feature though. :|

/Tom

[riko]

I am seeing +-16V for abs max rating for Vgs

See http://www.irf.com/product-info/data...ta/irl540n.pdf

Where are you seeing that +- 10v spec.

[riko]

Would the G201 run better or at an even yet higher reliabilty (had to phrase that just right) if it used the other part:

http://www.irf.com/technical-info/appnotes/an-1104.pdf

Or is it unnecessary?

What about that miller effect referenced in commnet " 2 on page 3. Could that old miller problem effect the MTBF? Would it improve the drive's MTBF?

MOSFET devices have considerable "Miller capacitance" between their gate and drain terminals. In low voltage applications this gate-drain capacitance is rarely a concern because the Miller effect is dependent of the VDS.

A potential problem occurs when the drain voltage of the bottom device rises very quickly due to turn on of the top MOSFET. This high rate of rise of voltage couples capacitively to the gate of the MOSFET via the Miller capacitance. This can cause the gate voltage of the bottom MOSFET to rise resulting in turn on of this device as well. A shoot-through condition exists and MOSFET failure is certain if not immediate.

[misterstarshine]

This stuff is ridiciously complicated. I really wish International Rectifier could propose a fresh new solution for us guys who are not proffessors in solid state physics.

Still there is a terribly low intrest in the anti-stall feature.

[kreutz]

Riko;

I was looking at his one.

[riko]

I think it is relatively easy.

Just forget about putting the finer points on all of this and look at:

1. The current drive at the top of the IR2104 spec sheet reads 120 ma source/270 sink. These are the min values listed for the short circuit. Typical is 210/360
2. The same spec values for the A3986 are min 140/160 & typical 110/200.
3. The G201 uses the IR2104 and it works.

So just concentrate on these values and make your guess at how well the A3986 will work (keeping in mind that the G201 has auto current reduction at idle)

These type of specs are usually termed "figure of merit" and
"rule of thumb" and evidently there is some amount of 'specsmanship" in the writing of them.

Time to put the soldering Iron to the chip and put all of the previously mention considerations & caveats aside until a working model is looked at.

[riko]

Kreutz,

I guess you have to get the avalanche rated part which also has the lower RDson.

All of this discussion is great as we get further along.

It seems everyone has something partly right and partly wrong about this.

I know I've learned a lot so far.

Thanks to all, especially the original post about "the little fellow" which got the ball rolling. Nothing like a little controversey and a lively give and take discussion.

And it is important to do the math.

I will be studying those spec sheets anew.

Do the curves show us anything?

[kreutz]

What about that miller effect referenced in comment " 2 on page 3. Could that old miller problem effect the MTBF? Would it improve the drive's MTBF?

Part of the protection against that kind of failure is the ability of the mosfet driver to keep the Low Mosfet's gate below the trigger voltage when it is in the off state (here the pull down resistance and the VoL of the driver stage play a role), the other part of that protection is to keep dV /dt to a level where that extreme effect will not show. Mariss keeps his dV/dt low, and in particular, his choice of the IR2104S made his life easier controlling that.

[kreutz]

Do the curves show us anything?

You get most of the information for your particular design from the curves since the given figures on the data-sheet are only good for the manufacturer's test circuit configuration and voltages.

[riko]

There are curves on the IR2104 but not the A3986.


I think there is an upcoming application paper on the A3986.

[kreutz]

There is certainly less discussion about the anti-stall feature though. :|

In order to provide this protection you will need a supervision micro-controller added to that A3986 board design. Look at the patents on the subject. Adding that feature will involve a little more effort and testing (R&D) and some potential legal side-effects..