Author Topic: IRF H-Bridge Design - shorting the supply at switch time  (Read 3866 times)

0 Members and 1 Guest are viewing this topic.

Offline guruTopic starter

  • Full Member
  • ***
  • Posts: 59
  • Helpful? 1
  • Robotics, CNC, Electronics, Programming
IRF H-Bridge Design - shorting the supply at switch time
« on: June 04, 2009, 12:22:18 PM »
I am looking at the complimentary pair IRF530 and IRF9530 N/P channel mosfets for an h-bridge motor controller. The N channel has a total rise time of 9+22ns, the P channel has a fall time of 45+46ns. So it looks like I will short the supply for about 60ns during switching. I am thinking I can add some delay in the CPLD controlling the switching by passing through some buffers? 60ns is a lot, it would take 6 buffers to delay if I use a 10ns CPLD. Probably still cheaper than sequencing with a clock. Any other ideas?

btw: When I say 9+22ns, I mean 9ms delay time, 22ns rise time as per the datasheets.

Thanks,
Colin

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #1 on: June 04, 2009, 04:21:24 PM »
Hi,

Switching times are determined by how much charge you move how fast, just plucking some ideal or maximum values from a datasheet won't ensure a proper deadband.

It may be easier for you to feedback the actual conditions to the CPLD and then disallowing a short alltogether - this way you would get the minimum deadband needed at all times.

Switching MOSFETs off (as well as on) is faster when you move the charge faster - think at least 3A..5A short term to get a decent shift. IOW, use drivers between the CPLD and the gates of the transistors.
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline guruTopic starter

  • Full Member
  • ***
  • Posts: 59
  • Helpful? 1
  • Robotics, CNC, Electronics, Programming
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #2 on: June 05, 2009, 08:50:56 PM »
Thanks for the heads up. I designed the original h-bridge for the superservo, but that was low power RC servo type motors. I knew high voltage/current was different. I've been reading a lot about the high power mosfets and using them in h-bridges so I just finished reading about the charge rate on the gate and keeping the mosfet out of "resistor mode" to sum it all up. The IRF530/9530 requires over 8 volts to fully switch on, so I think the 2n3904/3906 should suffice to drive the mosfets. The CPLD certainly isnt capable on its own. The mosfets gate have a lot of inherent capacitance too I have to watch out for. I really like your advice about feeding the signal back into the CPLD, I never though of that, but it's perfect; since any time that the h-bridge is fully off is a loss in power at the motor!

You talk about amp, 3a to 5a, but from what I've read the mosfet's gate does not see current, it sees voltage, the impedance of the gate is so high that current isnt a factor. Thus, as we often see a floating gate on a mosfet will often keep it's last known state....due mostly to it's inherent capacitance of 1000pf or thereabouts.


Hi,

Switching times are determined by how much charge you move how fast, just plucking some ideal or maximum values from a datasheet won't ensure a proper deadband.

It may be easier for you to feedback the actual conditions to the CPLD and then disallowing a short alltogether - this way you would get the minimum deadband needed at all times.

Switching MOSFETs off (as well as on) is faster when you move the charge faster - think at least 3A..5A short term to get a decent shift. IOW, use drivers between the CPLD and the gates of the transistors.


Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #3 on: June 05, 2009, 10:00:07 PM »
Hi,

You talk about amp, 3a to 5a, but from what I've read the mosfet's gate does not see current, it sees voltage, the impedance of the gate is so high that current isnt a factor. Thus, as we often see a floating gate on a mosfet will often keep it's last known state....due mostly to it's inherent capacitance of 1000pf or thereabouts.
MOSFETs are very misunderstood creatures unfortunately - lots of people actually think they're just electronic switches and that the resistive part is just a good-for-nothing snake in their digital paradise!  :-/

How do you change the charge of a capacitor without current? (You don't!)
Charge is Q, equivalent to As (Ampere seconds), measured in Coulomb
Capacity is Q/V, measured in Farad

A charged gate certainly is Hi-Z (That's what the FE in MOSFET is about), but you have to charge the gate capacitance and drive it through the miller plateau to get the magic happening - The MOSFET cannot switch any faster than you charge up this capacity.

You need to look at Qg (total gate charge).
A datasheet for the IRF530 from ST says, that Qg is 32 nC typ. and 45 nC max.
If you want to switch with a drive current of say 50mA, your fastest switch will be:  (32nC/50mA) = 640ns typ. (900ns max.) assuming no losses anywhere (i.e. achievable only in Utopia and its close surroundings), so add a spoonfull of extra oomph for good measure.

C_iss is the main capacity, but C_oss, C_rss and T_rr (150 ns), Q_rr (800 nC) of the source-drain diode are some of the reasons you have to add to the current drive.

If this sounds too strange, either ask or take my word for the substantial (albeit short*) current needed for fast switching.

(* the supply doesn't need to be up to this, as long as you use adequately sized (low-ESR) capacitors to decouple it).
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline Webbot

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 2,166
  • Helpful? 111
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #4 on: June 06, 2009, 07:06:24 PM »
Hi Soeren,

You've obviously got some good knowledge about this MOSFET stuff. Hows about a tutorial ?

Webbot Home: http://webbot.org.uk/
WebbotLib online docs: http://webbot.org.uk/WebbotLibDocs
If your in the neighbourhood: http://www.hovinghamspa.co.uk

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #5 on: June 08, 2009, 11:59:46 AM »
Hi,

I've made a mental note about it, for when i get some other planned stuff aside, but I'm a bit busy presently, travelling too much for comfort, which is taken time from my cyborg patients and piling up paperwork :(

Will probably be an eBook on one of my own sites though, but that shouldn't detract from the assumed usefullness :)
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline Webbot

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 2,166
  • Helpful? 111
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #6 on: June 08, 2009, 02:38:47 PM »
Will probably be an eBook on one of my own sites though, but that shouldn't detract from the assumed usefullness :)

Or you could post on SoR and ref to it from your own site? Would probably find a wider audience?
Webbot Home: http://webbot.org.uk/
WebbotLib online docs: http://webbot.org.uk/WebbotLibDocs
If your in the neighbourhood: http://www.hovinghamspa.co.uk

Offline guruTopic starter

  • Full Member
  • ***
  • Posts: 59
  • Helpful? 1
  • Robotics, CNC, Electronics, Programming
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #7 on: June 29, 2009, 01:14:38 AM »
I breadboarded the design here, except I used the IRF9530 and IRF530 mosfets instead.
http://www.cadvision.com/blanchas/hexfet/np-s.htm

This doesnt work because the lower N channel MOSFETs are not turning off completely causing one of the N channels to get very hot. The gate voltage on the P channel is good, on the N channel it ranges from 4v to 12v .... not good! For now, I removed the B-side to make a half-bridge so I don't fry a mosfet.

Of course, I can have the high side P fet turn on the low side N fet. That works fine. I want to introduce the CPLD later and control the lower fets directly, I am just breadboarding it right now and testing with thumb switches. The thumb switches have a 10k to gnd, and a 120ohm to the gate of 3904 NPN on 1 pin, and direct to 12v on the other pin. 

Maybe the circuit just doesnt work with IRF530's, but even looking at the circuit it doesnt look like it would be very good. The resistors would cause the switching of the mosfets to be *slow* as Soeren was talking about earlier.

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #8 on: June 29, 2009, 05:13:32 AM »
Hi,

I breadboarded the design here, except I used the IRF9530 and IRF530 mosfets instead.
http://www.cadvision.com/blanchas/hexfet/np-s.htm

This doesnt work because the lower N channel MOSFETs are not turning off completely causing one of the N channels to get very hot.

Change R6 and R12 to something between 330 Ohm and 470 Ohm (and while you're at it, do the same for R2 and R8).
That will give you a faster switching and a better OFF.
I think your immediate concern is something else though, as Q3 and Q7 needs pull up resistors on their bases, as they obviously doesn't close properly.
Try that and tell us how it went.


Now back to my holiday, sipping Scotch&Coke on the sunny porch in our summer cabin, a stone throw from one of the finest beaches in Denmark... Cheers  ;D ;D ;D
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline guruTopic starter

  • Full Member
  • ***
  • Posts: 59
  • Helpful? 1
  • Robotics, CNC, Electronics, Programming
Re: IRF H-Bridge Design - shorting the supply at switch time
« Reply #9 on: June 29, 2009, 09:37:00 AM »
Hi Soeren,

Thanks for the help. The PNPs are responding better but I am still not getting very good voltage ranges on the bases. I am not happy with the driver method using resistor dividers. I don't think this circuit would perform well in a PWM driven mode. I am thinking I should totem pole the drivers. That way I would get the fastest on/off times. Using 4 more general transistors wouldnt kill me. What do you think?

Sounds like a fun time at the cabin. I need a vacation, [sigh], work has been rough lately...and this h-bridge is not work but play. Oh well, at least I can drink!

Here's a recipe I recently rediscovered from my college days.

Singapore Sling:
Shot or 2 of Lemon Gin (sometimes referred to as "panty remover")
half sprite and half orange juice
dash of grenadine

C

 


Get Your Ad Here

data_list