H-Bridge Circuit Design

[Fish4Fun]

I would like to build a 4 channel H-Bridge suitable for driving an inductive load (predominantly three-phase "BLDC" and steppers, but also for testing the reactive components of an SMPS) Rather than being a dedicated driver for any particular motor, I would like to build a "prototyping" driver for testing various motors/circuits at voltages from 20V to 100V and currents upwards of 20A/phase. Obviously 20A at the upper end of the voltage range is 2kW, a non-trivial amount of power to switch on and off into an inductive load at any frequency. I may well end up having to design/build two devices, one for higher current/lower voltage the other for higher voltage but with lower current, but I would prefer to build a single device even if it cost 4x more than 2 narrower range devices. Switching Frequency considerations are 0Hz to 100kHz.

Design priorities are as follows:
1) Robustness
2) Flexibility (wide range of voltages/currents/switching frequencies)
3) Simplicity from a Connectivity Point-Of-View
4) Efficiency ==> Note this is a LOW priority
5) Cost **Note: I include this only as a last consideration, ultimately this is a "one-off" designed specifically to save money/time/effort in testing motors and prototyping other circuits, and I really don't care if it cost $50 or $500 to build**

The first choices to make are:

Mosfets vs IGBTs for the output driver.
Complementary vs High/Low Side Drivers for N-Channel Devices.

or

a Module like:
FSBB30CH60
or
FSAM50SM60A

(These are three channel devices, so to achieve 4 channels would either require 2 modules (six channels) or 1 module + a secondary circuit.)

I would love some thoughts/input on which direction to focus on. This will be my first high power h-bridge design so any experienced input would be welcome; including an "off the shelf solution" if there is one (for instance I have several "ESCs" designed for the RC hobby industry, and while not ideal for what I want, are certainly functional for testing BLDCs I also have numerous "Stepper Drivers", again functional for Testing/Driving Steppers, but not versatile enough to suit me, but there maybe an "off the shelf" solution that would meet the project design goals, actually there is one that is close: the "SnapAmp" by Dynomotion [ http://dynomotion.com/ ], and I have not completely ruled it out if I get in "over my head" on this project. Any other "off the shelf" recommendations would be great! ).

If there is general interest in this topic I will keep this thread updated with my progress. (Realistic time frame for project completion would be Spring of 2014.)

Thanks in advance!

Fish

 

[Fish4Fun]

Well I had some time to play the past few days, and I got tired of reading about all of the caveats and peculiarities involved with the various HS drivers and H-Bridge modules, so I thought to myself, "how hard can it be to build a floating supply?" So I sketched up the attached schematic, connected it to an AVR via an optocoupler (not shown in the schematic) broke out a virgin breadboard and fired up my Lecroy DSO....hehe after a bit of fiddling with things (no smoke, but some pretty serious heat ;-) ) I have a very stable floating Vhs + 16V PS with galvanic isolation from the low voltage supply. With a few tweaks I should be able to supply plenty of HS switching power to whatever drivers I finally settle on, LOL!

If any are interested the components are as follows:

Q1 IRLZ34
D1 Generic 15V 1W Zener
D2-4 STTH2R06RL 600V 2A trr = 5nS (Way over kill, but I had them on hand)
R1 120ohm 1/4W
C2 4.7uF 250V Film
C3 0.1uF 250V Film

T1 was made from a 5700 series Toroid inductor core with 23 turns 18ga Magnet wire primary 46 Turns 30ga Magnet wire secondary. The primary measured 33uH and the secondary measured 106uH (As a side note, reversing the primary and the secondary in the circuit actually reduced the ripple in the first filter and showed no discernible difference in the regulated stage. I will investigate this further when I get another chance to play.)

The PWM signal is currently ~580hz with a 2% Duty Cycle (1730uS off, 35uS on). I plan to add a current sense circuit to the Zener to adjust the duty cycle as needed, As it sits now the supply can manage a constant 13mA with a drop of 4V (Vhs + 12V); I have not even started on the Math to calculate how much drive current I will actually need, but I suspect even with minor tweaking I will have plenty (Simply increasing the PWM frequency to 735hz increased the current to 15mA with a drop of only 1.8V; Increasing the PWM Frequency to 1.1kHz increased the output to 16mA with no voltage drop, but the zener was got hot pretty quickly Lots to investigate and test, but I am tentatively going to call this a win.)

I have decided I am going to prototype this project in "modules". The high-side driver being what I thought was going to be the biggest hurdle is seeming like it might actually be through the first phase (bread boarding, lol). I still need to tweak the transformer; I likely need to replace the IRLZ34 with a faster switch; I need to play a bit with the diodes, I need to add the voltage feedback and I might even try adding an LM7812 in place of the 16V zener....ok, so I am a long way from done with the HS Supply, but I did make a lot more progress than I thought I would, LOL!

I will update this again when I make more progress.

Fish

 

[KrisBlueNZ]

1. The MOSFET source/drain are reversed. The source should be grounded.
2. I don't see any need for D3 and D4.

If you're generating a constant supply voltage, you should use a fixed duty cycle drive signal to the MOSFET.

It would probably be simpler to use an isolated DC-DC converter such as http://www.digikey.com/product-detail/en/UEI15-150-Q12P-C/811-1870-5-ND/1980328

That one is rated at 15W output (i.e. 1A). There are smaller, cheaper ones available if you don't need that much current.

 

[Fish4Fun]

Kris,

Thanks for the response!

1) Mosfet: My dyslexia does not ALWAYS migrate to actual circuits. (!me)

2) D4 is likely not important in any case, but the filtering provided by D3, in the actual circuit, removes a lot of transients and switching artifacts; likely more careful attention to component selection/layout would achieve the same.

3) While I think using a "plug-n-play" DCC converter might well be a viable option, it neither needs to step up or down; 12-15V DC will already be available, I just need to "float it" on top of Vhs for high-side switching, preferably with galvanic isolation.

Again, THANKS for the response!

Fish

 

[KrisBlueNZ]

1) Mosfet: My dyslexia does not ALWAYS migrate to actual circuits. (!me)

LOL

2) D4 is likely not important in any case, but the filtering provided by D3, in the actual circuit, removes a lot of transients and switching artifacts; likely more careful attention to component selection/layout would achieve the same.

A small inductor would be better, or just a resistor if you want to save space.

3) While I think using a "plug-n-play" DC-DC converter might well be a viable option, it neither needs to step up or down; 12-15V DC will already be available, I just need to "float it" on top of Vhs for high-side switching, preferably with galvanic isolation.

The point of using a DC-DC converter is not necessarily to step up or down; it's to provide the floating supply (galvanic isolation) in a compact, easy-to-use package. They're not cheap though, and you might be able to save money by using discrete circuitry. Personally, for a one-off project, I would save the development time and uncertainty, and buy a converter module.

 

[Fish4Fun]

Kris,

A little light came on in my head last night; even with the dense fog, it was pretty clear: I think I have WAY over thought the floating power supply! When I first started looking @ modern H-Bridge designs I kept running into various tricks to provide a floating high-side voltage to enable the use of N-Channel devices on the high side. I obviously failed to think this all the way through, or realize what the "obvious" alternative was.

Please correct me here if I am wrong: I could very simply take an old line-powered 12V transformer and a bridge rectifier and tie it's ground to Vhs and have exactly what I was attempting to do above (have Vhs + 12V regardless of Vhs)? In short the charge pump circuits in all of the various "high-side driver chips" are simply an effort to reduce the component cost in production units? Forest // Trees LOL. As I stated, I have never designed//built anything like this, and for whatever reason I have been making this part of it A LOT more complicated than it needs to be.

Heck, for my initial testing purposes I could simply take a 12V battery pack and tie its ground to Vhs w/o any regard to isolation at all! Wow. I am truly an idiot.

Thank you for patiently guiding me through to the painfully obvious.

Fish

 

[KrisBlueNZ]

Well, you're right, but the usual solution to providing a "floating" supply for the gate drive for the top N-channel MOSFET is even simpler than that. You just connect a capacitor to the top MOSFET's source, and feed the top end of it from the supply voltage via a diode.

When the bottom MOSFET is ON, the capacitor charges up though the diode. When the top MOSFET is ON, the capacitor's stored charge gives you a positive supply voltage relative to the top MOSFET's source, from which you can drive the top MOSFET's gate.

MOSFET driver ICs use this bootstrapping method to generate the "floating" supply voltage for the top MOSFET's gate. It's standard practice.

The only requirement is that the MOSFETs are switching continuously, with a duty cycle not too close to 0% or 100%, so there's enough time for the bootstrap capacitor to charge up, and its charge is being continuously topped up.

For an example of an IC that uses this technique, look at http://www.digikey.com/product-detail/en/LMD18200T/NOPB/LMD18200T/NOPB-ND/148219 and download the data sheet. It explains how the bootstrap capacitors are used to generate the gate driver supply for the high-side MOSFETs.

 

[Fish4Fun]

Kris,

Yes, it is exactly those data sheets that got me started down the path I was on; however, I really wanted to build a test circuit that allowed infinite 0 & 100% duty cycle....so I read an application note that showed using a 555 timer circuit to provide the HS floating voltage .... and that's how I ended up where i was.

Since component count/cost (and similarly overall size) are very low on my priority list for this project, and because it is absolutely a "1-Off", I would like to make the floating HS voltage circuit simple and "bullet proof", I just completely failed to realize how truly simple it could be. At the end of the day, I would be perfectly happy with something the size of a shoe box with terminal lugs for HS voltage in, 4 pairs of lugs for outputs , an RJ-45 for digital I/O and a power chord for the internal supplies, I will also likely add current sensors to each leg of each H-Bridge. This will allow me to quickly test steppers, BLDCs and Brushed DC motors on a single platform instead of wasting time custom configuring a test bed every time I need to test a motor.

I know it seems a bit off that someone might want to test such different types of motors on a single driver, but I love to "tinker", and I am into hobby oriented CNC among other things (hence the steppers). I am fascinated with the possibilities for the mass market RC hobby BLDCs. I am fairly convinced they could be driven in a semi-synchronous mode, if not completely synchronously. In my real-world job, several times a year I end up re-winding brushed DC motors (in fact I have one on my bench right now, lol) I will admit that a brushed DC motor does not require much effort to test (connect it to a power supply and it either goes, smokes or does nothing), but I would not mind replacing the 12V battery on my bench with a multi-purpose tool.

Anyway, hopefully at the end of this I will be in a better position to design drivers for specific purposes. I doubt building stepper drivers will ever be cost effective, but I think there is huge potential for BLDCs, and I am very interested in experimenting with them. My tinkering is an intellectual pursuit supported by my day job, I have no delusions about inventing or building something for profit, I will leave that to professionals and people much younger than me, lol.

Anyway, in the real world, I own a tackle shop in a Summer Tourist area, and this weekend is a busy one, so I won't be doing much except working till mid-week, but I would like to thank you again for your advice and input; I will continue to post my progress//problems and I hope you will continue to guide me away from stupidity and silly mistakes.

Fish

Fish

 

[Fish4Fun]

OK, there is nothing earth shattering here, this is a proof-of-concept schematic that represents what I currently have on a bread board for testing a 3-Phase BLDC motor from the RC Hobby industry. The Mosfets are FSL23N200, (200V, 24A 0.1 Ohm RDSon), absolutely NOT ideal for the purpose, but I have a ton on hand and they were dirt cheap...perfect for letting the smoke out of.

I am also concerned about the ability of the MCT61's to handle the MOSFET switching duties. The data sheet puts the switching time @ 2.4uS into a 100 ohm load ... we'll see about that, but assuming a 12 pole motor with a speed less than 18k RPM; this implies any given mosfet will have a minimum duty cycle of > 275uS. Initial testing will be a simple 0 to 3000 RPM ramp up in synchronous mode for a total of < 10 seconds run time.

The initial testing will be to test the theory that these motors can be run in synchronous mode with no feed-back (pole sensing). As existing sensorless BLDC motors "start" in synchronous mode, I expect this first phase to be successful (assuming I don't let the smoke out of the Mosfets, hehe).

Once I have proof-of-concept on the High-Side switching (and hopefully synchronous mode motor-run) the next step will be to implement PWM in the Vcc path via a buck or fly-back converter with current feed-back, rotor position sensing via "zero-crossing" feedback and a uController "learning phase" (and more suitable switches). I expect a lot of pitfalls and failures, but hope that all of these goals can be achieved.

For simplicity/safety I have decided to do initial testing using batteries. I have 3 14.8V LiPo 1.6AH battery packs for Vcc and am using an 8-cell AA battery pack for the Vcc +12V supply. The uController (not shown) will run from a standard 12V wall-wort. Initial testing will be with a single 14.8V LiPo pack for Vcc which should drastically reduce the short-term semiconductor carnage hehe.

The motor I will use for initial testing has a DCR of ~0.5 ohms per phase, an inductance of ~12uH per phase and a power rating of 800W (54A @ 14.8V !!). The LiPo packs have a max discharge rate of ~30A suggesting a source impedance of ~0.5 ohms, so, no matter how crazy the numbers sound, the source impedance combined with the motor impedance + DCR + RDS on should limit the current to less than 15A per phase. Assuming 15A per phase, the maximum duty cycle of any given Mosfet will be limited to < 17%, giving a maximum power dissipation of 0.17 * ((15A)^2 * 0.1 Ohm RDSon) ==> ~3.8W/Mosfet, certainly not a sustainable dissipation rate for a TO-220 package w/o some serious heat sinks, but for proof-of-concept with << 20 second runs ..... maybe the smoke will stay in, hehe.

Anyway, please point out any stupidity in my schematic, I hope to do some actual testing Wed (still have to write the code, lol).

Fish

 

[KrisBlueNZ]

Hi Fish

I think there are a few problems with that design.

1. The transistors in the optocouplers seem to be connected backwards. They're NPNs, right?

2. I think it would probably be better to arrange the gate driving circuits so that the MOSFET will be OFF by default, and only turned ON when the optocoupler is ON. With the way you have it shown at the moment, if the control signals fail, both MOSFETs will turn ON and short out the VCC supply.

3. In any case, the gate voltage to the high-side MOSFETs cannot switch between VCC and VCC+12; if the low-side MOSFET is ON and the high-side MOSFET has VCC on its gate, it will also conduct. The high-side MOSFET gate needs to swing between VCC+12 and the high-side MOSFET's source (which is where I would connect the pulldown resistor).

4. I wouldn't name a voltage rail "VCC+12". I would rename VCC and VCC+12 to something like V12 and V24 or +12V and +24V.
Edit: Oh, VCC isn't 12V. In that case I'd call them VCC and VGG (gate supply voltage).

5. There isn't really any need for optocouplers for the low-side MOSFETs.

6. Your drive signals need to allow a dead time between turning one MOSFET OFF and turning the other MOSFET ON. The dead time needs to allow for the propagation delay in the optocoupler and time for the gate-source resistor to dischare the gate-source capacitance.

7. Because there will be a dead time and you're running into inductive loads, it's normal to include Schottky diodes across the output MOSFETs. You've included D1~3 but you should have them for the low-side MOSFETs too. Normally Schottky diodes are used - for speed, I guess, and because they have a lower forward voltage than the parasitic diodes in the MOSFETs, so they will carry the current caused by the back-EMF from the loads.

8. I don't see the reason for C1~3. I'd suggest removing them.

9. The MOSFETs should have gate-source overvoltage protection, especially the high-side ones. Normally just a 15V zener is used, although this increases the gate-source capacitance. That's unless the MOSFETs include a zener; in that case, I would include it in the circuit symbol. I also include the parasitic drain-source diode in the circuit symbol.

H-bridge MOSFET drivers solve ALL of those problems. The only limitation they have is that you can't leave any half-bridge permanently high because the bootstrap capacitor will discharge. I would use an IC and address that problem specifically. For example, you could generate a VCC+12V rail using a charge pump (it doesn't need to supply any significant continuous current) and couple it into the bootstrap capacitor connections of the driver ICs through a diode and series resistor, or something like that. Have a think about it; there's probably a cleaner option.

 

[Fish4Fun]

TY Kris.

I truly hate my dyslexia :-( It is late here, my head is a bit fuddled, and the efficacy of my meds has expired, so I need some sleep before I can ferret out if I mis-drew the schematic or dropped the ball completely....but it is definitely drawn wrong. I was in high school before I could confidently read or write sentences that included common three and four letter anagrams like god and dog. To this day, I have terrible trouble with such simple concepts as "anode" and "cathode" with respect to correct translation to schematic symbols; in this particular case, I "get" the schematic symbol but am all but incapable of properly identifying/describing them properly. With MOSFETs//Transistors I fundamentally understand how they work and their various specifications, but 99 times out of 100 would fail to see them misrepresented schematically. If I looked at two schematics, and one had the drain and source of all of the Mosfets reversed I would swear the schematics were identical no matter how long I stared at them. Even if I have the schematic correct, there is only a 50/50 chance I will get source/drain correct in a PCB layout if I don't use a cad program that utilizes schematic capture to enforce proper pin mapping for ALL the parts. Sadly, in general, my work-around solution is to physically build the circuit and then "test" it with high impedance to verify proper function via a DSO. Then, once everything works properly, re-translate the working circuit to a PCB layout using the bread-board prototype pin out. I frequently work out highly complex problems with the certain knowledge that the end result only has a 50/50 chance of "being the right sign", and am perfectly happy simply negating the sign at the end. This problem is pervasive in my life; for example I have designed/built hundreds of molds over the past 20 years, only to find when I am finished that one or more of the "reliefs" were 180 degrees out, and once the part was cast, it was permanently locked in place, giving me the opportunity to build a second mold almost exactly like the first ;-)

So, tomorrow I will check to see if I "did it right" and "drew it wrong", or if I in fact "drew it right" (ie "wrong") and simply had it wrong, lol.

With reference to C1-3...you are almost certainly correct, and I had reservations about them when I placed them. My intention was to "quiet" the output noise for zero-crossing detection, but I really need to see what the signals look like before I try to "fix" them; in any case I will remove them prior to power-on.

I will add a diode across the low side mosfets as suggested. I plan to use a uC as the signal source, and the dead-time was to be enforced in firmware; probably a bad design parameter, but it does simplify the design. To prevent turn-on issues (while NOT shown) my plan was to use a N.O. relay controlled by the uC to enable Vcc to the power circuit, thus eliminating the chances of spurious transients during start-up from translating to the outputs. Again, likely a wasteful design parameter, but it is simple and should be effective.

Isn't the high-side gate voltage clamped to ~12V by the battery? Is this not sufficient? As Vcc "floats" is it probable that the battery could over-charge? I just can't see this as probable, but if you think it is important to add zeners to each gate, I will certainly do it, though I would have thought with a High-side rail, simply clamping the rail with a single zener would be sufficient, but if each gate needs a zener, I will certainly add them.

Low-Side optocouplers...maybe I haven't thought this all the way through, but the two reasons they were included were 1) Vcc could be 80V or more, an optocoupler seemed like a good way to ensure isolation from the uC circuit. 2) In theory having the same gate drivers on the low-side and the high-side would simplify timing issues, though in the case of a 3-phase BLDC motor this really isn't an issue.

Gate Driver ICs....I have several different ones on hand. To date I have had trouble making them function reliably in low frequency applications. I have read the data sheets and application notes and, as you suggest, in low frequency circuits the addition of a charge-pimp is discussed. From my current perspective, the addition of a charge pump when combined with the driver is less cost-effective and more complicated than simply adding a Vcc + 12V rail. Ultimately my goal is to build a "universal" h-bridge for driving BLDC, Stepper and Brushed DC motors in a wide range of DC voltages/currents. The most recent schematic is simply a "proof-of-concept" output-driver for a 3-phase BLDC using batteries as Vcc and **VGG**, but with-respect-to the final project goal is merely part of the prototyping phase.


THANK YOU for taking the time to look/review. I have now muddled around long enough it is "time to be up for the day", so I will take my meds and figure out if I drew it wrong or did it wrong (the optocouplers).

THANKS!

Fish

 

[KrisBlueNZ]

Whoa! Fish, thanks for your explanation of your dyslexia! It sounds like nothing I can imagine.

I don't have any problem with letters, or left/right. Sometimes I'll get doubled-up numbers wrong - like I might write 3360 when it should be 3660 or something like that. But full-on dyslexia must be so frustrating!

Your written language is very good. But I don't know how much effort it takes you to write properly.

I plan to use a uC as the signal source, and the dead-time was to be enforced in firmware; probably a bad design parameter, but it does simplify the design.

I think the design would be most simplified if you used MOSFET driver ICs. That would solve nearly all of the problems I listed. I think the worst problem is the optocoupler delay, because it is not well-controlled. I really think you would be much better to use MOSFET driver ICs.

To prevent turn-on issues (while NOT shown) my plan was to use a N.O. relay controlled by the uC to enable Vcc to the power circuit, thus eliminating the chances of spurious transients during start-up from translating to the outputs. Again, likely a wasteful design parameter, but it is simple and should be effective.

That's just another workaround that wouldn't be needed if you used MOSFET driver ICs, and it's not guaranteed to solve all of the problems. I really don't think it's a good idea to arrange the drive circuitry so that an optocoupler HAS to be ON otherwise the driver shorts VCC to 0V.

Isn't the high-side gate voltage clamped to ~12V by the battery? Is this not sufficient? As Vcc "floats" is it probable that the battery could over-charge? I just can't see this as probable, but if you think it is important to add zeners to each gate, I will certainly do it, though I would have thought with a High-side rail, simply clamping the rail with a single zener would be sufficient, but if each gate needs a zener, I will certainly add them.

Think about what happens during the dead time. If you have a load that wants to pull a MOSFET driver output towards VCC, and the top MOSFET's gate is sitting at 0V during the dead time, the load will pull the top MOSFET's source up to VCC and the top MOSFET will experience a Vgs voltage equal to negative VCC!

I think MOSFET driver ICs include diodes to prevent the gate drive outputs from going too far away from the source voltages, but if you don't use them, I think you should add zener diodes externally - at least on the top MOSFET. A zener on the bottom MOSFET might not be needed; I would just put one on every MOSFET, "just in case".

Low-Side optocouplers...maybe I haven't thought this all the way through, but the two reasons they were included were 1) Vcc could be 80V or more, an optocoupler seemed like a good way to ensure isolation from the uC circuit. 2) In theory having the same gate drivers on the low-side and the high-side would simplify timing issues, though in the case of a 3-phase BLDC motor this really isn't an issue.

The VCC voltage isn't relevant to the bottom MOSFET, as far as I can tell, as long as it can withstand that much drain-source voltage. And I don't think that adding an unnecessary, poorly controlled delay on top of an existing poorly controlled delay is helpful!

Gate Driver ICs....I have several different ones on hand. To date I have had trouble aking them function reliably in low frequency applications. I have read the data sheets and application notes and, as you suggest, in low frequency circuits the addition of a charge-pimp is discussed. From my current perspective, the addition of a charge pump when combined with the driver is less cost-effective and more complicated than simply adding a Vcc + 12V rail.

I can't believe that. You want to add a 12V battery to the design, and to worry about the extra hassle caused by driving the top MOSFET gate from a fixed voltage rail instead of a rail that's relative to the source voltage?

At the very least, you can replace the 12V battery with a charge pump, or a switching supply (inductor-based) if you prefer. It doesn't need to supply any significant long-term current.

Ultimately my goal is to build a "universal" h-bridge for driving BLDC, Stepper and Brushed DC motors in a wide range of DC voltages/currents. The most recent schematic is simply a "proof-of-concept" output-driver for a 3-phase BLDC using batteries as Vcc and **VGG**, but with-respect-to the final project goal is merely part of the prototyping phase.

And I admire the idea. But I really think you're creating much more trouble for yourself by refusing to use gate driver ICs. If there are application notes that cover using these ICs with 100% duty cycle control, I can't see how you can justify any other approach.

I know it can be hard to return to a device or technique that you've had significant problems with in the past. Really bad experiences have a big psychological effect and can colour our attitudes for years or decades into the future. But MOSFET driver ICs are widely used and are not generally a weak point in designs.

Whatever you decide, I hope it goes well

 

[Fish4Fun]

Hrmmm....

Psychologically difficult or not, I guess I had best at least give your advice another go. I have HIP4082's ( http://www.digikey.com/product-detail/en/HIP4082IPZ/HIP4082IPZ-ND/821447 ) , FAN73711's ( http://www.digikey.com/product-detail/en/FAN73711MX/FAN73711MXCT-ND/2115888 ) and IR2184's ( http://www.digikey.com/product-detail/en/IR2184PBF/IR2184PBF-ND/812271 ). Ever used any of the above, or have another recommendation? I don't mind ordering different drivers if there is one you are familiar with and/or suggest.

Writing, thank you, and yes, I have a degree in creative writing but would likely have failed out of college over English had it not been for the timely invention of the word-processor. Effort? A lot. For instance my previous post I began @ 02:45 and "finished" (prior to editing a typo) ~07:00, so ~4.25 hours for that post, some take considerably longer, others are "off the cuff" and might be completed in 20-30 min. Rarely do I post anything I have not fully considered, edited and re-read. It is a good thing that I require very little sleep (typically 3-5 hours a day) hehe, or I would never have time to do anything!

Since most of your post concerns problems involving driving the mosfets, I will wait for any recommendations you might have before I proceed (it's fine if you don't have any, I will re-read the datasheets and applicable application notes and "flip a coin" ;-) ).

For now....Thank You for pushing me to use driver ICs....

Fish

 

[KrisBlueNZ]

I've never actually used H-bridge driver ICs myself! I just understand how they work.

Of the three you listed, the IR2184 is the only one that has the important features - high-side and low-side drive and dead time enforcement. It looks pretty good to me.

Can you point me to any application notes (or data sheets) that address the problem of driving the half-bridge high continuously, i.e. something in addition to the usual charge bootstrap arrangement? From any manufacturer. I'm interested to know what method(s) they suggest.

That's amazing that you would spend four hours over a forum post! The only time I ever spend that sort of time is when I'm writing a tutorial such as my recent post https://www.electronicspoint.com/led-flasher-help-t263649.html#post1576930 which is 11 screens long! Normally I leave the proofreading until after I've posted, which is why so many of my posts have the "Last edited by KrisBlueNZ ..." line at the bottom :-/

Do you write fiction? Is any of your work internet-accessible?

 

[Fish4Fun]

Kris,

IR2184 it is then, lol.

IRF's AN-978 Section 8 "HOW TO PROVIDE A CONTINUOUS GATE DRIVE" (page 17&ff) details the use of a 555 timer in conjunction with an IR2125 and an IGBT to maintain the voltage in a bootstrap capacitor.

I haven't written any fiction in quite some time, though I have toyed with the idea on more than one occasion. All of my writing the last decade or so has been non-fiction, and is absolutely internet accessible, lol, it is all forum posts, typically on technical sites I graduated from university and began a business in December of 1986. While in the long term the business has done very well, it consumed 12+ hours a day 7 days a week for the first ~20 years. I missed a lot of things getting myself into a financial position where I could do the things I want to do, but the last ~7 years I have backed way off of work and spent time pursuing other things. Last December my son graduated from college which ended 9 consecutive years of $25k+/yr in education expenses, and the future is looking pretty good. I still do 30-40 hours a week @ the store, but comparatively it seems like I am on holiday. My business is seasonal and the Winter is the slow time, so perhaps one of these Winters I will again pursue fiction (there is a story I have been shaping for a decade or so, but it is not quite ready to come out yet.) TY for asking.

Fish

 

[KrisBlueNZ]

IRF's AN-978 Section 8 "HOW TO PROVIDE A CONTINUOUS GATE DRIVE" (page 17&ff) details the use of a 555 timer in conjunction with an IR2125 and an IGBT to maintain the voltage in a bootstrap capacitor.

Right, thanks. That looks like a reasonable solution. You'll need to change the 100k resistor from the 555's ground rail to the 0V rail.

They say that 100k was chosen for a 500V +HV rail. It would have about 485V across it, so 100k will give a current of 4.85 mA. That seems reasonable provided that you're using a CMOS 555. Multiply that by the voltage, to determine the power dissipated in the resistor, and you get 2.4W. Much more than the 1W rating suggested on the schematic.

If your +HV rail may vary over a significant range, you should probably use a 5 mA current sink instead of that resistor. If your maximum +HV is 80V, the current sink transistor will have up to 65V across it, so its dissipation will be no more than 325 mW. You could use a 2N5551 or a KSC1845/2383/2690, or for Darlingtons, MPSA29, ZTX614 or various other larger ones like MJE270.

You'll need one circuit for each half-bridge output, unfortunately. There might be ways to reduce the component count but it's probably not worth bothering.

I haven't written any fiction in quite some time, though I have toyed with the idea on more than one occasion. All of my writing the last decade or so has been non-fiction, and is absolutely internet accessible, lol, it is all forum posts, typically on technical sites I graduated from university and began a business in December of 1986. While in the long term the business has done very well, it consumed 12+ hours a day 7 days a week for the first ~20 years. I missed a lot of things getting myself into a financial position where I could do the things I want to do, but the last ~7 years I have backed way off of work and spent time pursuing other things. Last December my son graduated from college which ended 9 consecutive years of $25k+/yr in education expenses, and the future is looking pretty good. I still do 30-40 hours a week @ the store, but comparatively it seems like I am on holiday. My business is seasonal and the Winter is the slow time, so perhaps one of these Winters I will again pursue fiction (there is a story I have been shaping for a decade or so, but it is not quite ready to come out yet.) TY for asking.

What sorts of fiction have you written in the past?

 

[Fish4Fun]

Kris,

For now I am going to forgo the LF/Constant on, and focus on prototyping a BLDC driver with 3 * IR2184 and some mosfets for a relatively low voltage/current motor (<15V/10A), just so I can get some experience with how these ICs actually work. I understand the theory just fine, but I learn a lot more when I attach my DSO and WATCH what is actually happening.

I took a couple of hours this evening and bread-boarded the driver:

I was going to draw it up, but silly Altium doesn't have an IR library with the 2184's and I don't feel like battling through creating the part tonight. I have to work tomorrow through Monday, and I have several other projects screaming for attention, so it will likely be a while before I get back to this. (I will prolly write the ASM and do some testing before I actually get around to drawing it up ;-) ) In fact knowing how I do things, if the bread-board version works like it is suppose to, I will almost certainly do a PCB before I do the schematic, hehe.

Writing....In school I was a big Hemmingway fan, so I worked hard to write EHH flavored short stories. While I still love EHH's writing style, I have read too much psychology and, perhaps, lived a bit too long to wallow in his subject matter. The most recent fiction I wrote (~ 20 years ago) was a ghost story (oddly enough for a class my wife was taking, lol. She went back to school to get a teaching degree to go along with her lit degree). The story was good enough the teacher insisted my wife submit it to a state literary magazine to be considered for publication. Of course it was not submitted because it was not her work, lol, and it would not have been a 'good thing' for me to submit it for fear of getting her in trouble. I doubt I even have a copy any more, lol.

I am a sucker for Sci-Fi, and the the idea that has been roaming around in my head for the past decade is in the Sci-Fi venue. Like most Sci-Fi writers, I have some societal laundry to air, and the best vehicle for the job is a future that is bright and shiny, devoid of the cancers of the present. The basic idea of the story is to demonstrate how education can be used to fundamentally shape the future. The story would follow several groups of characters from different times on a 150 year time line; a collection of vignettes if you will. There would be some continuity in the groups as they recount/experience some of the same events from significantly different age perspectives. I certainly don't have it all worked out, and I have concerns about attempting to develop too many characters/groups for a reader (or me, lol) too keep up with.

I expect it will be a while before I am ready to revisit writing fiction, and I will likely start with character sketches and perhaps some short stories as primers. (Counter-intuitively most authors agree that a short story is more difficult to write than a novel. Having never written anything novel-length, I cannot attest to the veracity of the statement, but from past experience I know that a well-written short story requires an efficiency of language and attention to detail well beyond what one might think.

Any way, enough rambling. I will leave you with my favorite quote: "I stand at the chasm's edge, knowing, never in ten-thousand years, will it reach out to snatch me unless, in a lunatic fit of religion, I jump." (From John Gardener's Grendel . A fabulous re-telling of Beowulf , from Grendel's point-of-view.)

Fish