A3986 DIP Module

[Matilda]

I did see those videos.
What FETs are you using?

[PMinMO]

Forget the dpack's you can't pull enough heat from them. Dpaks are best when they are stricktly hard On-Off, no reverse conduction, no linear region switching. If you notice in the data sheet they refer to up to 1.5W dissipation for the surface mount package.

2nd the logic level drive is also a bad match as logic level mosfets typically have higher gate charge at the Vgs levels the 3986 will supply.

I went kind of down the same road and learned the hard way, starting off with the small dpaks to252 also.

The 48V is to high considering back emf, and also the step innaccuracy is worse as you go up in supply voltage.

mosfets are STP16NF06L

[Matilda]

I looked up the spec of your Fet, good.
I have a whack of these IRF540N. RDSon is 0.044. That's low.
PD is 130W.
I should give them a try while I have them.

[Matilda]

Do you have a schematic you would like to share?
I'm curious what others are doing the LSS1 & LSS2 lines?
And the whole schematic as well.

[PMinMO]

At his point I don't want to disclose the schematic as there are some aspects of associated circuits that I have an agreement on. lss1 and lss2 are low current paths to the a3986 part itself as long as the physical layout accomidates the high current of the sense path appropriately.

[Matilda]

Sense path is good.
I have the LLS1&SENSE1 line running from the A3986 over to the sense resistor. Same goes with LSS2&SENSE2
Allegro says not to join them at the chip.
But I was wondering what the point was...and if that is all I do with those lines.

On the schematic Allegro gives, out they have caps across the sense resistors.
But on their demo boards, they yanked them.

Oh well?

[PMinMO]

lss is used for the low side driver, thus you just want the connection current for the source side of the fet, not the sense current in any way,

[Matilda]

I tried to upload my schematic, but pdf's are not allowed.
I'm sure were are talking about the same parts of the schematic, I was just gonna load my up for viewing.
On the LSS1-2 and SENSE1-2 lines, I basically copied what Allegro does/suggests.

[PMinMO]

You can now. I didn't realize it wasn't enabled.

[Matilda]

Here is my A3986 side of things

[Matilda]

I just finished a board using some IRF540N's.
The board powers up, but on the first step the A3986 blows.
With the original FET I had, IRLR024, the board would, and will work.

The difference between these two FETs is the Vdss. The IRF540N is 100V, and the other is 55V.
I assuming that having the 55v for the Vdss is whats saving the A3986?
My Vbb is 48Vdc.

This blows... literally

Any thoughts?

[PMinMO]

My recommendation is to stop wasting your time with the 3986.

[Matilda]

What do you recomend I start with.

[Matilda]

PMinMo,
When you did the laser video, did you do it at a 7 foot projection? I think thats what I'm reading.
Your laser does seem to have some sort of pattern to it.
I performed the same test. I placed a laser on my motor shaft, and projected to a wall 7 feet away. I piece of
paper on the wall with lines that were 0.1649" apart.
I noticed that some of my step pulses moved too much, and some too little. I also noticed that the motor did not move to some of the steps!
And when it would miss/loose/ignore these steps, then it would make up for them in one big jump.

The motor I'm using is: 2.2V, 7.0A/phase, 0.31 ohm/phase, 2mH, 1.8deg.

She is a hefty motor.

[PMinMO]

I don't recall the specific distance but it was in the neighborhood of 7'. I did the trigonometry, printed a target per step angle then measured. Your results are as expected and why I stopped any further development with the chip after the Allegro field engineer admitted the problem. They wouldn't admit there was a problem till that video, then they shared an Excel spreadsheet to compute the false triggers based on supply voltage, motor resistance and motor inductance to "HELP" find a suitable motor. After running the numbers with a number of COTS nema23 and nema34 motors, it didn't take long to realize none were even close to working. This wasn't a new spreadsheet, they knew the problem when I first contacted them, but didn't dividuldge it thus letting me spend time and money on several prototypes. Really ticked me off.