Problem with Hbridge circuit?

one of the them current running in reverse recovery for the body diodes I know there was added discrete diodes but unless they can be garanteed to have lower Vf than the fet body diodes I'm not sure they do much good.

Since the supply 12V a nice gate drive voltage, it might be an idea to change the upper fets to P-channel and change or modify the driver a to drive them instead.

I'm guessing that changing upper fets to P-channel, connecting BST to Vcc, Cap from BST to ground, HS connected to ground, and inverting the drive signal for upper fets might work.

-Lasse

There are plenty of other ways that are just as good, but I'm not aware of any that are simpler or easier; it's a very common scheme.

Yes.

They are, just not at really low frequencies (unless you start using a really big caps). There's the additional presumption that, if you leave the bridge off for awhile, it's OK if it takes a few switching cycles before it's fully "working" again when you resume operation.

Even at 60Hz they can work fine, just not at DC.

---Joel

Yep. That's why I asked if the added diodes were fast-recovery types.

But driving a 30nS edge directly into a motor is just generally a bad idea.

Cheers, James Arthur

No, I'm not a god and I'm not perfect, You just were not listening and I can understand why. You were more interested in hearing answers to resolve your problem instead of listening to why it wasn't working.

And as I said, depending on what type of load you have on that circuit you can fix it. Yes, those drivers are made for a half bridge circuit for which they work just fine using all N channels fets.

change the code in the pic so that both low side fets come on during the off cycle or, if you want to keep the driving low side on just turn on the low side on the non driving side in the off cycle of the PWM.

This in effect will also cause braking if you were to have something like a DC motor on the output otherwise, it'll just discharge what ever else is there during the off cycle. If you see this as a problem, then increase the carrier freq and use series inductors/chokes in the load circuit to help with this.

You may find it to work fine simply by turning on the low side fet that isn't supplying the low side drive to the load just to keep the boost cap in check!

if they are fast recovery really doesn't matter unless they turn on and they won't unless they have a lower Vf than the (hot) fets

-Lasse

No, I wanted to know why it wasn't working. That is the reason I asked. It does me no good to magically fix my problem without understanding it... except that sometimes it does end up becoming a waste of trime trying to figure out why it isn' t work.

The problem is that a lot of things you say seem to contradict my understanding of how fundamentally the thing works... such as your next statement

Why does the load matter? Obviously it works with inductors because they show all there schematics with them... and since I'm mainly using it with inductors it should be ok... but a purely resistive load should work fine just as long as its within spec. (in fact an inductive load is much worse because of the inductive kickback... of course it's current is frequency dependent but that is also not an issue)

Yes, which is why I thought you were nuts when you kept harping on P channels. In fact it would be much easier to use P-ch and also cheaper... But when I was researching different types for purchasing I could not find any good p-ch and they tended to be more expensive(I was trying to find larger current and low Rds_on and Nch were the best). I then read on several sites that N-ch's were more popular because they were cheaper and more efficient at the same task. (of course you then need a driver).

Since you need to drive both low and high sides with enough current I need a driver anyways... so getting a driver that drives Nch high side isn't really any worse than getting one that drives a p-ch high side.

I was thinking about that but when I read about it I remember someone saying that its not efficient. (essientially you end up running the motor as a generator on the "off" cycle and it then brakes itself instead of freewheeling)

This seems like it might help with the problem of the boost caps though... but there might be some cross-conduction involved causing the fets to heat up more(probably not really but possible).

Yeah, that was the problem I was talking about. It seems much worse than leaving the fet off... of course this was before I found out that that the chip didn't work as I thought ;/

yes, probably so but also might cause other problems.

I'm wondering if there is simply a way to trick the boost part into charging/discharging but I cant' see it without making the circuit more complicated.

I think its probably best to just try and find a better driver that doesn't use that trick or to use some discrete fet's (dual n and p ch in cmos config) and p ch high side.

I think the cost will be about the same and it will be easier to troubleshoot and route.. I was just trying to minimize the heat dissipation of the fets as much as possible. (IIRC the p-ch had almost 2x the on resistance of equivalent n-ch and about 1.3x the cost if not more)

Still seems that the boost cap will not be discharged on the off cycle of the pwm.

I probably can trick it though by hooking the boost cap to a small fet that will ground out on the off cycle and go high on the on cycle? (of course this seems like a waste and I'll probably not do it)

I have a pc scope but since I brought a new computer without a parallel port I haven't been able to use it ;/ Since my MB is in about 5gal of vegtable oil I haven't added an expansion card yet ;/ (just not looking forward to adding the card because it will get messy ;/)

well, do normal h-bridge circuits toggle the non-conducting path's low side fet so that the boost can work?

i.e., basically in my configuration for running the motor, I was planing on only PWM one high side fet and let the opposite low side be in conduction while the other two fets were always off(preventing cross conduction). Of course when I wanted reverse I would flip this configuration.

I know I can ~PWM the low side of the same side as the high side that is being PWM and that seems like it will make the thing work but I didn't do this when I designed it(although its not big deal to implement as I just have to output a ~PWM signal from my pic) because I wanted to avoid cross conduction... just, say, incase the ~PWM transitions overlap with the PWM transitions causing both to be on for too long(say if my transitions are not very fast for some reason).

I guess though that I probably don't need to change anything on my circuit and just increase the PWM frequency and include the ~PWM to the proper low side. When I get some new IC's in I'll go ahead and do that and hopefully it will work.

I imagined that they couldn't hurt anything (since either the fet's diode or the discrete diode will work by themselfs).

You got me.

I'm Slaughtered, and you're kill-filed.

RL

That makes a lot of sense.

Just change the code in the pic to turn on the low side FET that needs to charge the cap and turn off the other low side fet that is acting like the low side load path. DO this in the off cycle and you'll get what you're looking for with out the problem of regen drain in your circuit. Normally, full bridges fully turn off during the off cycle accept if they wish to have regen braking to help slow down a motor or for a fast stop in which case, both low side fets can be locked on.

P.S. You may want to have a small cap or a bi-directional TVS diode across the output of the bridge so that when it goes into H-Z mode you'll won't have a problem of wheeling voltages at the bridge.

Thats what I'm going to try...