www.robowars.org

RoboWars Australia Forum Index -> Technical Chat

Electronics design for cash
Goto page 1, 2  Next

Post new topic   Reply to topic
  Author    Thread
Nick
Experienced Roboteer


Joined: 16 Jun 2004
Posts: 11802
Location: Sydney, NSW


 Reply with quote  
Electronics design for cash

Want to make some money using your knowledge of power MOSFETS? I need a replacement for the Jaycar SSRs that have started breaking; nothing on the market seems worthy and using a Victor to control a Mag motor is expensive and probably just a mechanically lacking.

If you think you can design or adapt an opto-isolated FET driver circuit to control two or three large FETS, then its worth $$$. I will source all the components for you and fabricate the casing PCB etc. Just to start things off, I like the specs of this power FET: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=IXFN340N07-ND The current is limited to 100A by the case design, so at least two will be needed.

If you would like a crack at it, post here or PM me.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Post Sun Jun 22, 2008 6:53 pm 
 View user's profile Send private message
dyrodium
Experienced Roboteer


Joined: 24 Aug 2004
Posts: 6476
Location: Sydney


 Reply with quote  

I'll talk to conrad at work, he's got some experience with power circuits. You're pretty much after a FET brick, right? Laughing
BTW how much shockmounting did you give the ssr's? It looked like in the real big hits they moved enough to hit something solid, hence the cracking... maybe not.
_________________
( •_•)

( •_•)>⌐■-■

(⌐■_■)

YEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAH

Post Sun Jun 22, 2008 7:54 pm 
 View user's profile Send private message Visit poster's website MSN Messenger
Nick
Experienced Roboteer


Joined: 16 Jun 2004
Posts: 11802
Location: Sydney, NSW


 Reply with quote  

That's right, a custom high current SSR - I have a few other specifications to meet and I want to cast the parts in a pocketed aluminium block, partly for cooling and mostly for strength.

The SSRs were completely wrapped in foam for RG08 and the rest of the enclosure was foam-filled so they could not move far - it still didn't help in the finals match. A big part of the problem is the heavy motor wiring; its pulling the terminal posts out of the SSR cases!
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Post Sun Jun 22, 2008 8:03 pm 
 View user's profile Send private message
Knightrous
Site Admin


Joined: 15 Jun 2004
Posts: 8511
Location: NSW


 Reply with quote  

Looks like the way your solidly mounting them to the chassis to heat sink is subjecting them to the 200+ G's on impact, the results are evident. Maybe you could cast the SSR's into a block of urethane to shock protect. I'm sure you've got plenty of urethane there to test with Smile
_________________
https://www.halfdonethings.com/

Post Sun Jun 22, 2008 8:06 pm 
 View user's profile Send private message
Nick
Experienced Roboteer


Joined: 16 Jun 2004
Posts: 11802
Location: Sydney, NSW


 Reply with quote  

True, that was my first idea. It ends up being pretty bulky and the wiring could be simpler. I figure that an aluminium cased custom SSR will have double the current capacity but only half the size and weight of dual Jaycar units and the terminals can be made much stronger.

Those FET modules I linked to will take up to 70V, so the custom SSR could easily power a larger, higher voltage motor in a lightweight in the future.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Post Sun Jun 22, 2008 8:27 pm 
 View user's profile Send private message
Spockie-Tech
Site Admin


Joined: 31 May 2004
Posts: 3160
Location: Melbourne, Australia


 Reply with quote  

Chris Baron from Robot Power would probably be the man to ask - You can find him on the OSMC mailing list at http://tech.groups.yahoo.com/group/osmc/

There are lots of other power fet gurus on that list as well (its where I learned most of my Fet knowledge) who might take you up on it - but Chris has more experience at designing things suitable for combat environments, most of the other gurus are more industrial control guys who might not take into account shock loading and regular overloads as much.

Last time I talked to Chris, he had just landed an OEM motor controller contract so was fairly busy and might not want to take on a new project, but hes usually willing to chip in with some ideas, and if you can convince him it would be a saleable product, he might be interested.

Instantanous power dissipation is the tricky bit to deal with when switching high current loads - much like Jolt has to deal with harsh shock loads because of its rigidity that causes unexpected areas to break - having a "stiff" power supply/batteries that can near-instantly deal out large chunks of watts makes things go boom..

if you can soften up the "hit" when the switch changes state at the source of the power (not at the switch which would cause fet heating) then you can get away with a lot less precision power drive cleverness required in the fet drive circuit - limiting the batteries ability to deliver hundreds of amps with a thin wire or other resitive element will makes the controllers job a lot easier - it might add a small amount to your spin up time but shouldnt reduce the final energy level reached majorly.

I have enough on my plate at the moment that I can only offer some general advice, but not a complete design service at this time sorry.

Another possibility might be the cafe electric guy who makes the "Zilla" controllers - http://cafeelectric.com/shop/index.php?main_page=index&cPath=1 - the most powerful consumer ESCs in the world that I know of - 300v at 2000amps - around 600Kw or 800Hp ! - I dont know if he is available for custom electronics design, but if anyone knows how to spin a high power SSR together, he should.
_________________
Great minds discuss ideas. Average minds discuss events. Small minds discuss people

Post Mon Jun 23, 2008 1:42 am 
 View user's profile Send private message Send e-mail Visit poster's website MSN Messenger
Daniel
Experienced Roboteer


Joined: 30 Aug 2005
Posts: 2729
Location: Gold Coast


 Reply with quote  

How about just making your wires slightly longer so they arn't constantly pulled tight. As soon as something bends in your robot the wires can't stretch and end up pulling the components apart. Not many places in the world where you'd find wiring as tight as yours and an extra 10mm won't destroy the neatness.

Post Mon Jun 23, 2008 6:53 am 
 View user's profile Send private message Send e-mail
Nick
Experienced Roboteer


Joined: 16 Jun 2004
Posts: 11802
Location: Sydney, NSW


 Reply with quote  

@ Daniel: that area actually has loose wiring - I can pull the SSRs almost out of the frame with the wiring attached. I didn't want to cut the Mag motor wires short and the only short wiring are the links between contacts on the SSR modules. The problem is the weight of the multiple loose 8g wires on poorly supported contact points.

All: What I found last night is that the control PCB in these modules is not potted at all; it just floats under the flimsy top cover with nothing much to stop it bending. there would be at least a 2mm air gap underneath it Mad. That gives two options: hack the case down and re-pot the thing properly, or cut out the PCB and use it to drive several larger FETS (assuming it can).

Would I be correct is saying that a) virtually all power control FETs are N channel and b) if I don't need high switching speeds (like PWM), then capacitance and drive current are less finicky? I know that is a very broad area, but it sounds like it is worth venturing a few dollars on some IBC sized FETs for a test circuit. If I can get that working, up-sizing to a couple of 300A FETs is feasible.

@ Brett: thanks for the tips; I will definitely email Chris Baron if the above hack does not work.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Post Mon Jun 23, 2008 7:24 am 
 View user's profile Send private message
Spockie-Tech
Site Admin


Joined: 31 May 2004
Posts: 3160
Location: Melbourne, Australia


 Reply with quote  

P-Channel Power Fets do exist, but they are usually not available in as high a power rating as N-Channels for some reason, and are usually much more expensive for a given power as well.

Which is why most H-Bridge controllers do all the jumping through hoops necessary to drive N-Channel fets on the high side of the bridge as well as the low side rather than use P-Channel for the high side (which would be much simpler to drive, no Vs sensing and gate voltage charge pumps would be required for P-types on the high side)

Repeated high speed switching (like PWM) does mean you spend more time in the "transitional zone" (where both the current flow and voltage drop across the Fet are high which means high power dissipation and booms), than a single-switch event.

But, even a single poorly-driven switch event is enough to destroy a fet if the gate drive is weak and more than fractions of a microsecond are spent transisitioning from off to on, so you still need to be able to charge and discharge the Fet gate capacitance quickly, even although you're only switching occasionally.

As an analogy (maybe not a great one, but it will do for now), you cant make a gun barrel out of cardboard, even if it does only need to fire occasionally.. the high energy event still needs to be contained and directed for it to work even once.

Higher power fets tend to have higher gate capacitance and need a correspondingly higher current gate drive to switch rapidly. So make sure you check the Gate Capacitance spec for your 300amp fets and rate your drive circuit appropriately..

For example - The IBC's HIP chips can supply a momentary current of 2.5 amps to charge the 1405 Fets 5 Nanofarad gates in under a few nanoseconds.

Putting in higher power rated fets (or more fets in parallel) increases the capacitance to be charged and thus slows down the switching time (causing more heating) correspondingly so you dont get as much benefit as would first appear by just banking up fets or putting bigger ones in. The 8-Fet controllers (40nf total) are about the limit for the 2.5a Hip drive drives afaik.
_________________
Great minds discuss ideas. Average minds discuss events. Small minds discuss people

Post Mon Jun 23, 2008 11:05 pm 
 View user's profile Send private message Send e-mail Visit poster's website MSN Messenger
Valen
Experienced Roboteer


Joined: 07 Jul 2004
Posts: 4436
Location: Sydney


 Reply with quote  

switching is where you have the difficulty.
I can probably hack something up out of 1405's
The biggest issue is a power supply for the thing.
you need a fet drive voltage of ~12V. If you can guarantee a supply of 12 volt regulated or > 14 volt regulated power then it becomes fairly simple.
If however the power supply voltage drops (say at peak current during spin up) then your fets leave saturation and quickly turn into SiO2 rather than Si.
I might have something provided input power is > 5 volts.
_________________
Mechanical engineers build weapons, civil engineers build targets

Post Mon Jun 23, 2008 11:52 pm 
 View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger MSN Messenger ICQ Number
Nick
Experienced Roboteer


Joined: 16 Jun 2004
Posts: 11802
Location: Sydney, NSW


 Reply with quote  

The power supply should be fine - the data logger shows it never gets under about 19V when the Mag spins up. I wouldn't have a problem sticking in a small dc-dc converter if it guaranteed reliable performance. I am sure I came across a nifty automotive FET controller last year - it had all the features I could possibly want and now I can't find the little bugger Sad
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Post Tue Jun 24, 2008 12:34 am 
 View user's profile Send private message
Valen
Experienced Roboteer


Joined: 07 Jul 2004
Posts: 4436
Location: Sydney


 Reply with quote  

Yeah we can probably hack something up
Those isotop fets be nice but they lack some wang in terms of bang / $
For a non-PWM drive setup I think we can hang 8-16 1405's off the driver I have if we add some nice big tantalums to bypass the driver chip. (13A driver chip, 2A continuous)
16F876 for drive or some such, You can probably get some PWM out of it too if you drop it back to 6 FETs (~300A rating roughly with a decent heat sink). So that should let you do temperature limiting and the like and soft start.
_________________
Mechanical engineers build weapons, civil engineers build targets

Post Tue Jun 24, 2008 1:40 am 
 View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger MSN Messenger ICQ Number
Nick
Experienced Roboteer


Joined: 16 Jun 2004
Posts: 11802
Location: Sydney, NSW


 Reply with quote  

@ Jake: it sounds like you are well on your way to a couple of hundred $$$ Smile

The main reason I was looking at the big FETs was a) the package can take a heap more current than a TO220, B) the package is isolated, making it really simple to use, and C) the fewer FETs, the simpler the development is (according to the IRF tech documents). The particular model was the cheapest overall as the minimum quantity was 10.

If we go go smaller FETs, could I suggest a package like a TO247 or something similar? the pack size and lead spacing will make construction heaps easier and worth some extra cost. I can order something from digikey or mouser as soon as we settle on the parts.

If you want to use a familiar HIP chip. we could cheat a bit with the drive. The motor has two windings so we could drive two smaller banks of FETs from separate outputs. I don't know anything about HIP chips (yet); can some drive outputs be ganged for more current? Perhaps we can just use one or two parts of the OSMC H bridge design?
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Post Tue Jun 24, 2008 4:11 pm 
 View user's profile Send private message
Valen
Experienced Roboteer


Joined: 07 Jul 2004
Posts: 4436
Location: Sydney


 Reply with quote  

TO247 has the same silicon area as the other 220 package fets i'm pretty sure
the legs on a 220 are "good" for 75A apparently
If your going smaller then you go to some of the magical metal can surface mount jobbies if you can get them in the voltage.
I really don't like hip chips and for this application they wont work (well most of them), HIP chips cant run at 100% on time because they use the switching to run their boost converters for high side drive.
You don't need high side drive for a one directional controller.
If you have 14V minimum input you just need an automotive rated voltage regulator to take that "24" volts (with spikes up to 50 or so) to take it down to 12v for fet driving, run it into the magic high power fet driver, and another 7805 to run your logic level stuff, (to understand servo pulses etc). Just need to select an opto isolator, the 10mhz ones seem to work well without needing too much from the transmitter.

Then you need to blow up 3 or 4 sets of fets etc to find the errors ;->

Probably a good idea for testing to run with a big relay in series so you can turn it off when the fets blow and lock on.
_________________
Mechanical engineers build weapons, civil engineers build targets

Post Tue Jun 24, 2008 4:22 pm 
 View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger MSN Messenger ICQ Number
Nick
Experienced Roboteer


Joined: 16 Jun 2004
Posts: 11802
Location: Sydney, NSW


 Reply with quote  

Ah, I thought you were suggesting a HIP driver, I was thinking it was overly complex and we definitely would not want something that could not switch the FETS on all the time. Would you get any more current drive if the switching frequency was lower than usual? The 'soft start' will only need to run at less than 100% duty cycle for a few seconds, so perhaps a lower freq would let your chip charge up a bit more between cycles?

I was thinking the TO247 pack had more spaced out, thicker leads and would be less demanding on PCB design, plus the thermal ratings are a bit better. I ordered a sheet of 1.5mm copper sheet for bus bars and we could just sweat that onto a regular PCB to avoid the expense of a high current PCB. I guess we could try soldering the FET leads direct to bus bars and let the potting hold everything in place; a bit risky for shorts but it might simplify the design.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Post Tue Jun 24, 2008 10:35 pm 
 View user's profile Send private message
  Display posts from previous:      

Forum Jump:
Jump to:  

Post new topic   Reply to topic
Page 1 of 2

Goto page 1, 2  Next

Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum

 

Last Thread | Next Thread  >
Powered by phpBB: © 2001 phpBB Group
millenniumFalcon Template By Vereor.