DIY Brushless Speed Control - Page 10 - RC Groups
Thread Tools
Jan 28, 2004, 04:55 AM
foamforce pilot
dicker's Avatar
nonono....

only low or high side fets should be triggerd by pwm. Therefore onle 3 diodes are required....
And I never would put them on the Fets, but on the driver to the fets. Look:



Ah, D1, D3 and D5 are the PWM diodes, the others are for the driverIC to drive the floating gate

greets Julian
Sign up now
to remove ads between posts
Jan 28, 2004, 05:09 AM
Impossible? Hah!
KreAture's Avatar
Here's a baby I can use:


It's a 6-bit wide buffer/inverter with strobe and tristate. It should be great for disabeling the output when it should NOT commutate (as when one looses synch...) and it will allow turning on/off all 6 channels directly to feed the N-transistors (3 of wich then would control the P's for the upper side)

The tristate function is rather useless though.
Jan 28, 2004, 05:19 AM
Impossible? Hah!
KreAture's Avatar
The whole point was avoiding the drivers...
And the current-drain in the diode solution won't be nice either.

I know you only need to do 1 sided PWM, but for noise and the trigger-balance 2-sided should give less problems.

The cost would be minimal too.


But, maby you are right. only N-sided PWM should do nicely. I would just need the circuit to compensate for the slope offset at other than 100% duty...
Jan 28, 2004, 05:21 AM
foamforce pilot
dicker's Avatar
no, imho this is no good Idea...

Because its not reasonable pulling both Fets up and down, by that your doubling your switching losses.

On the other Hand a both side PWMed esc is regulating in a very special and not linear manner...

We've tested that and the Motor is taking Gas exponential instead of linear.

If you're taking only low or high side fets for PWM (doesn't matter which side) it regulates more smooth and linear.

The last thing is that 3 diodes are much more easy to layout and rout than this little bugs with 10 or more feets...

Greetings, Julian
Jan 28, 2004, 06:15 AM
Impossible? Hah!
KreAture's Avatar
True dicker...
The small 150 mil chips are a pain, but I like them (My main controller is a 208mil)

I agree with you on the upper / lower issues. I have not yet decided what to do.

I was supposed to power an op-amp from B+ and B- and then use diodes from the phases after the FET's but before the motor and a simple voltage-divider between them to create a fully 3-phased center estimate. This would be the threshold for making the EMF into a square wave then clamping the output to TTL. It should work and be self balancing if I am not too much mistaken.
Jan 28, 2004, 06:16 AM
Impossible? Hah!
KreAture's Avatar
Dho! I forgot that the motor phases aren't wired inbetween the FET's like a brushed one.

Hmm, how can I still get this to work...
Jan 28, 2004, 06:47 AM
Registered User
mmormota's Avatar
There is a problem with the 4502 driver. A uC usually works at Vcc: 3...5V, while the motor on 6...15V. The 4502 inputs treshold level too high in most cases for the uC output level.
Jan 28, 2004, 08:08 AM
Impossible? Hah!
KreAture's Avatar
Err... The 4502 is a 5v ttl circuit from the sheets where I selected it...

I will be running my chip at 5v and it has 4.8v or so minimum high output and 0.2v or so low output.

My FET's trigger at about 1v and -1v so they should have no issues with the TTL-levels.

I do however look for ideas on how to balance the comparator so it will detect 1/2v crossing correctly...
Jan 28, 2004, 08:43 AM
homo ludens modellisticus
Ron van Sommeren's Avatar
Quote:
Originally posted by dicker
...The german collection office has written me some ugly letters , so I'll have to do some hard stuff first...
Julian, what's this 'collection office'?

Cu Ron
Jan 28, 2004, 08:48 AM
foamforce pilot
dicker's Avatar
Its about some financial matters from my self-employed incomings....

greets Julian
Jan 28, 2004, 12:20 PM
Impossible? Hah!
KreAture's Avatar
Great news... Via the university I am now borrowing Multisim
It is a analog and digital simulator that is supposed to be very good and accurate. It has most of the components I will be using so I can simulate some of the pwm control, filtering and signal management. I believe it even has the ability to simulate the motor.

I'll do some experimenting to see what I will be using, but I am actually leaning towards a active low-pass filter followed by a high-pass filter to remove the DC component alltogether. Then Another amp with self-adjusting bias that will prepare the signal for TTL reading. This should be fun...
Jan 28, 2004, 02:58 PM
Registered User
mmormota's Avatar
Quote:
Originally posted by KreAture
Err... The 4502 is a 5v ttl circuit from the sheets where I selected it...

I will be running my chip at 5v and it has 4.8v or so minimum high output and 0.2v or so low output.

My FET's trigger at about 1v and -1v so they should have no issues with the TTL-levels.

I do however look for ideas on how to balance the comparator so it will detect 1/2v crossing correctly...
I mean that the PFet's source is on B+, the battery + voltage. To switch off the PFets the 4502 has to work on this supply voltage too (or the Pfets will be on regardless of the output). If the 5V is ok for the motor, it is no problem. If you need the B+ voltage higher, you have to increase the driver supply to the same voltage (Pfets!). Then the treshold of the inputs of the 4502 will be B+ / 2, it could be too high for the uC outputs.
Jan 28, 2004, 05:30 PM
Impossible? Hah!
KreAture's Avatar
No, there is a N-transistor doing pulldown there. Look again...
Jan 28, 2004, 05:36 PM
Registered User
mmormota's Avatar
Quote:
Originally posted by KreAture
No, there is a N-transistor doing pulldown there. Look again...
Holy cow, I missed 3 fets on a drawing with only 9 of them... Sorry.
Jan 28, 2004, 06:25 PM
Impossible? Hah!
KreAture's Avatar
Hehe. Oh, and I now offichially hate Multisim...
I slapped up a circuit with the 6 main power Mos-FET's in a 3-phase H-bridge. I powered the bridge with only 5v and used ttl-source from the data-word generator to step the FET's. I added 3 resistors in a star topology to create some sort of load for the FETs without inductance. I also added some series resistors between the gates and the generator. And, I did not forget grounding.

Still I am getting "Timestep too small" errors during transient analysis. It is driving me bonkers!

Are there any free tools that would do some simple simulations?
This huge bucket of bolts isn't worth the bits it occupies in my computer...