Ron, LTspice is notifying me that it can't find the IR21xx.sub file. I've checked, and couldn't find it in the zips you posted. And I haven't been able to locate it in Google either.

I bet it was because I named them both (sub & sym) .zip files.

 

Ok... after quite a bit of tweaking, and rewiring the schematic here and there (because I now realize that I didn't have a completely clear idea of what a mosfet driver does), I think I made it somehow work. Although I must confess I don't thoroughly yet understand what's happening.

The sim was running at a snail pace, so I removed the RC filter across the motor, and also one of the zeners. See if that would speed things a bit... .but it made no difference... the thing's still crawling at a speed that exceeds my patience.

Also, the sim kept crashing on me, and it wasn't until I placed R5 to R8 as gate pull-downs that it kind of worked. Although IR2301's datasheet does not show them as necessary.

I used Max's link to more or less calculate the value of C2 to C5, and it seems that it's within ballpark figures...






I'm going to add a few TVSs later here and there, to further protect the mosfets, but I'm not going to include them in the sim, which seems to be overwhelming LTSpice already.
When I lay my hands on the motor next week, I'm going to measure its resistance to see if the 1.5 ohms value that I used in the sim is acceptable. Is there any way to include the BEMF produced by the motor in the sim?

Questions:

• Does it matter if I feed the driver with 12V or 5V?
• If I were to power it with 12V, can I use 5V logic to switch its inputs?

Any changes, adjustments, suggestions, observations, and even nagging and scolding would be thoroughly appreciated... as long as they help to deliver me from producing magic smoke when the moment of truth arrives.

 

Does it matter if I feed the driver with 12V or 5V?

• If I were to power it with 12V, can I use 5V logic to switch its inputs?

.

They include level shifters so 5v logic should be OK, if available I would use 15v for VCC
Max.

 

They include level shifters so 5v logic should be OK, if available I would use 15v for VCC
Max.

That makes sense... in my (extremely) limited experience, nFets are best triggered with at least 10V, preferably 12V. Using 15V would allow for voltage drop within the ic before reaching the gate.

Many thanks!

 

Ok... after quite a bit of tweaking, and rewiring the schematic here and there (because I now realize that I didn't have a completely clear idea of what a mosfet driver does), I think I made it somehow work. Although I must confess I don't thoroughly yet understand what's happening.

The sim was running at a snail pace, so I removed the RC filter across the motor, and also one of the zeners. See if that would speed things a bit... .but it made no difference... the thing's still crawling at a speed that exceeds my patience.

Also, the sim kept crashing on me, and it wasn't until I placed R5 to R8 as gate pull-downs that it kind of worked. Although IR2301's datasheet does not show them as necessary.

I used Max's link to more or less calculate the value of C2 to C5, and it seems that it's within ballpark figures...

View attachment 105631

View attachment 105632


I'm going to add a few TVSs later here and there, to further protect the mosfets, but I'm not going to include them in the sim, which seems to be overwhelming LTSpice already.
When I lay my hands on the motor next week, I'm going to measure its resistance to see if the 1.5 ohms value that I used in the sim is acceptable. Is there any way to include the BEMF produced by the motor in the sim?

Questions:

• Does it matter if I feed the driver with 12V or 5V?
• If I were to power it with 12V, can I use 5V logic to switch its inputs?

Any changes, adjustments, suggestions, observations, and even nagging and scolding would be thoroughly appreciated... as long as they help to deliver me from producing magic smoke when the moment of truth arrives.

You need to PWM the top FET. So you can swap M1 & M3.
Usually you switch the top and bottom FETs on the same side, but I think you can do it the way you have it. You will notice it takes a while for the cap to charge up enough for the under voltage to turn off in the driver. If you switch them on the same side the cap will charge right away. But you have to put some leisure time between the drive signal so the top and bottom aren't on at the same time.

 

You need to PWM the top FET. So you can swap M1 & M3.
Usually you switch the top and bottom FETs on the same side, but I think you can do it the way you have it. You will notice it takes a while for the cap to charge up enough for the under voltage to turn off in the driver. If you switch them on the same side the cap will charge right away. But you have to put some leisure time between the drive signal so the top and bottom aren't on at the same time.

Leisure time is what I need at this moment ... but I'll take your advice and switch the top FETs... see what happens.
I'm not quite following you... when you say that "If you switch them on the same side the cap will charge right away"...Do you mean that I should use, say, the left IR2301 to PWM M1, and the right IR2301 to PWM M2?

 

That makes sense... in my (extremely) limited experience, nFets are best triggered with at least 10V, preferably 12V. Using 15V would allow for voltage drop within the ic before reaching the gate.

Many thanks!

Here is a pretty good write up.
http://electronics.stackexchange.com/questions/58849/bootstrap-circuit-function

 

Leisure time is what I need at this moment ... but I'll take your advice and switch the top FETs... see what happens.
I'm not quite following you... when you say that "If you switch them on the same side the cap will charge right away"...Do you mean that I should use, say, the left IR2301 to PWM M1, and the right IR2301 to PWM M2?

Ooops they crossed.
Yes, usually to switch the top left on while turning the bottom left off. The bottom right can be on all the time.
So when the bottom left FET is on the cap charges thru the diode and the FET. The cap voltage is then "placed on top of the source voltage when the top FET turns on.

 

The way I have always done it in H bridge config. is to PWM M1 with M4 switched on, or M3 PWM with M2 switched on for reverse, otherwise the same side fires into each other.
Max.

 

The way I have always done it in H bridge config. is to PWM M1 with M4 switched on, or M3 PWM with M2 switched on for reverse, otherwise the same side fires into each other.
Max.

If you're referring to my schematic in post #22, then I'm not sure I get it. Did I number the fets right in that post? or is there a standard way for numbering them when working with H bridges?

 

Sorry I got the No.s from the OP.
Post #22 would be M1 -M3 & M4 -M2 pairs.
Max.

 

Sorry I got the No.s from the OP.
Post #22 would be M1 -M3 & M4 -M2 pairs.
Max.

So you would always PWM the high side fets then?

 

Yes for the charge pump to work.
Max.

 

Here's my latest sim, it took about 4 hours to complete a 0.030 second simulation!

​

Voltage at M1's gate looks weird (purple trace, V(n005))... I would have expected it to be pulsed! Also, current through the motor went up too high, almost 4 amps! Maybe it's because BEMF is not included in the sim, as you've already mentioned before, Max.

 

At its rated load of 1/6th HP it should be pulling 1.8amps, off load, the generated voltage will almost equal the applied so current would be decidedly less.
Max.

 

Aaaarggghhhh!!!!! I hate this stupid thing! I just can't seem to make it work!

Apparently, the value of the bootstrap capacitor is critical. It can't be too high or the driver won't have the time to charge it up between cycles, and it can't be too low or the nFet won't fire up.



What I did is run the sim at a much lower frequency (2.5 kHz), and tweaked the cap's value a bit, bringing it up from 480nF to 800nF.
Also, I changed the specs for the nFets to one in LTSpice's database that more closely resembles the FDA33N25. That is, FDB33N25.

But the driver stops pulsing the gate after the first half rectified cycle! What gives?

 

I don't use LTspice that much, but incidentally in practice IR/MOTO recommend high speed diodes for D11 & D12.
Max.

 

I'm almost giving up on this stupid simulation... it's behaving very strangely. It works fine for the first half-cycle, and then M1's gate stops being pulsed... Another thign that kind of bothers me is the very high value for the bootstrap cap that I set after hours of tweaking. It has to be between 1.5 and 2 µF for the sim to more or less work.

I guess what's next is to build the thing and see what happens.

 

Or... plan B would be to drive the fets using an isolated power supply, just like this one, and an optocopuler... I've done it before, based on a design posted by JDT. It's slower (although I'm not planning on running things at more than 20 kHz) and more expensive, but a lot easier to implement.

 

I have used the charge pump plug-in formula in the Silicone Labs link and it has worked for me.
Max.