Hmm, I think my Spice model of the IR2110 is buggy. Re-running the sim without it, and using two independent gate drive voltage sources instead, gives the hoped-for turn-off of Q1 so that the freewheel current is wholly through the diodes and the supply. So, I now agree with Crutschow.
Anyone got a reliable Spice model of a FET driver IC equivalent to the one used in the OP?

 

Here's my sim using a de-bugged model of the IR2110 as the driver. The bootstrapping is now effective and heat in the FETS and diodes is minimal.

 

Here's my sim using a de-bugged model of the IR2110 as the driver. The bootstrapping is now effective and heat in the FETS and diodes is minimal.
View attachment 126200

Glad you got your simulation working.
Could you post your .asy and .lib files for the IR2110 model you used?
Thanks.

 

Sure. Like I said, it's a home-brew model. So no warranty .

 

Sorry, I wasn't around over the weekend.

I jmade the circuit on Friday and just tested it without D4 and getting noticable heating on D2. I am about to add D4 in and see what happens. It's just a nasty perfboard mockup, so it's ugly, but effective with a suitable degree of caution. Unfortunately, there was a problem on the pinouts for the IGBT I was using on Friday, so I lost some time to it.

 

I jmade the circuit on Friday

Which one?

 

I jmade the circuit on Friday and just tested it without D4 and getting noticable heating on D2.

Why would you do that?

 

Which one?

The first schematic in post #12

Why would you do that?

Prior to the activity this weekend, that was the one that we didn't think did anything useful.


I'm about to check over the circuit again, but it looks like Q1 was overheating after adding D4 in.

 

The first schematic in post #12

Did you connect pin Vs to the source of Q1 as is required?

 

Did you connect pin Vs to the source of Q1 as is required?

Are you sure you aren't actually here? You are damn good. I had it "connected," but the pin in the jumper wire between the perfboard and breadboard had a loose connection. The unit I am testing on is a mystery unit that has been in storage for years, so I need to tighten some of the springs before heavier testing, but all of the component temperatures appeared to be stable.

 

Since getting my sim running properly I'm confident that post #12 first circuit will do the business efficiently, provided Vs is connected to Q1 source as Crutschow said. However, if you're trying to get several Amps through the coil then a breadboard, even with the springs tightened, is likely to melt!

 

Still having some issues with this.

I was having an odd ringing / oscillation in my drive on Friday and burnt my mosfets not too shortly after. When I turned up the power, I was starting to get some heat issues on Q1 and D2. I'm not sure if the ringing was because my mosfets were damaged or if there was another issue. I had already added a 5k resistor and a 100nF capacitor snubber between Q1 and Q2 similar to Crutschow's simulation.

I should be getting some more (larger) perfboards to hook this back up on later today. Unfortunately, the original one did not leave enough space to mount heatsinks on Q1 or D2, but I should have enough space on the new perfboards.

Is there anything else I could add to reduce the heat on these components? Originally, both the high and low sides of the driver IC shared the same input signal. Should I separate these timings? IIRC, there are some similar circuits that use inverted timing on the low side with a small period that both sides are turned off.

I seem to also remember seeing Schottky diodes facing the Drain / Collector of Q1 and Q2 in similar circuits. I believe that these were used to improve the switching time since the body diodes in the mosefets / IGBTs are too slow to prevent current from flowing through them. Is this something that I should look into? Thanks.

 

Should I separate these timings?

No. Q1 and Q2 need to be on/off at the same time. The inverted timing and dead-time provision would be appropriate in an H-bridge; but that isn't the situation in this application.

 

What is the duty-cycle of the control signal going to the driver.
It should be no more than 50%.

 

I went ahead and just double-checked the output with the oscilloscope; The frequency is currently set to 60Hz and the duty cycle is capped at 35% (though, I can't really see a scenario outside drastic misuse that it would need to pass 25-30%).

 

I got my new perfboards in today and am planning to lay them out and solder them in the morning.

As mentioned in #52, I am planning to add two Schottky diodes as in the attached diagram. I'm thinking that the switch speed on Q3 isn't allowing the internal body diode to fully switch off before the current flows back through D5 and this may be part of why it is generating more heat (I had much less heat on Q4, but figure that it wouldn't hurt to add a matching Schottky diode for testing). I figure that D3 and D4 will block any reverse current through the body diodes and that I should use a fast response time diode for D5 and D6. Does this sound like it would work? Thanks.

 

If the circuit is working properly, the voltage across either MOSFET never goes significantly negative so there is never any reverse current through the substrate diodes.
So the added diodes won't do any harm but likely all they will do is just dissipate a little extra heat.