/diy/ - Do It Yourself

the fuck is with people here and 555 timers and relays? it's 2012. We have technology now

>>349317
voltage divider will be too slow. You will need a proper level shifting chip.

>>345056
>with a square wave output from the MCU
well rather the the current the bridge sinks will be a square wave- the MCU output is a constant duty PWM

>>342362
there's probably some asic that does pretty close to what you're after already. Otherwise, start by making a UART module on whatever MCU etc you want.

>>345021
regardless of single or 3 phase source from wall outlet- rectify with full diode bridge and add smoothing caps/ decoupling caps to get a rough DC supply.

next 2 possible ways to drive the motor: add hall sensors or encoder to motor, and commutate it as a BLDC motor

or drive the motor with a conventional 3 phase sine wave, as it normally would be (open loop)

I'd go for the 1st option (especially if using a battery) since there will be no issue with rotor slip. If you are using a battery and motor is wired for 240/440v you'll need to make a charge pump as well. You'll also need to make a 3 phase full bridge drive (mosfets or IGBT's)

you can then drive the fets via gate drives with a square wave output from the MCU you use, based on the combination of hall sensor inputs into the MCU (6 step commutation for BLDC). The high side of the mosfet bridge is driven by PWM- change duty cycle to vary motor speed.

Or, if you want to drive it conventionally, you will have to vary the PWM duty cycle applied to high side transistors, so the current wrt time produces a sine wave. Each sine wave is 1/3 of a period out of phase wrt the preceding wave. Change the period of the wave (i.e. change frequency) to change motor speed.

>>328934
for example, take a 1" measuring standard from way back in the day (b4 SI) and measure it against something modern- it will not be 25.400 mm. Take a brand new, 2012 1" measuring standard and measure it- it will be 25.400. 1" is now defined as exactly 25.400 mm- obviously arbitrary standards cannot match exactly (no 2 physical quantities are the same), so the inch as you know it, has been changed to fit with the metric system. it's technically not even an inch anymore, but they kept the name to stop a certain group of people becoming significantly buttflustered.

despite OP's thread being intended to troll the educated:

how many chains to a furlong you arbitrary morons??

also, feel jelly that your imperial system is now regulated by metric ISO standards. Inch etc now defined in terms of the meter

>>328923
>like using higher voltage rated caps

wrong- you can always read about why if you're actually interested

>>326379
>Get old cnc stuff for cheap

no, don't do this. It's a terrible idea

>2012
>using anything other than Mazak or Mori Seiki

how's being an entry level hobbyist working out for you?

>>293260
I usually set up a 32 (or 64 bit register if I want to be really precise) and increment the register count by the closest integer to n where f =2^32/n, and use the MSB as my scaled clock signal. not very resource intensive and very accurate. 50% duty cycle too. Pity arduino is all the rage with kids these days :( they will never appreciate 1's and 0's on the same level.

>>293036
*impractical

>>292972
>just pure verilog

really? I wouldn't have thought this would be practical, considering how much of a pain it is to actually write synthesisable division for fixed point numbers, or even make exact given frequency clock signals that are real number divisions of the system clock. I mean it can be done, but I'd think at RTL it would take months and months? Tell me your secrets!!

What university is this at?. At the bare minimum I'd imagine you'd include current sensing, variable PWM generation and ADC for throttle in an MCU. Beyond that its pretty much just gate drives and suitably heatsinked mosfet bridge. You don't even need to make a floating PSU for the bridge high side with most IC gate drives. I feel bad for your traces though.........you will need to use to-247 packaged fets just so you can get enough track width to handle the current.

>although if a had one i probably wouldn't share all of it, due to the fact that I might be able to sell it

if it were a servomotor system, probably, but that's about 3 orders of magnitude difference in complexity.

>this thread

holy fuuuu why are there XILINX aaaaaaaand ALTERA chips on that boad OP? it's like gluing an AMD to ur intel cpu. There can be only 1.

>>260516
>Everyone I talk to agrees that you can't go wrong if you get a bridgeport

it sounds like everyone you talk to is a hobbyist. chances are, they are comparing a bridgeport mill to a small, generic chinese made machine they bought as their first mill or something. In reality, these are massively over hyped- especially on the internet where hobbyists declare to the world they are the best mills ever (comparing to aforementioned chinese stuff) . Similarly, one level up, small engineering firms purchase their first HAAS or similar quality cnc, or toolroom lathe, and think they're great compared to everything else, when in reality they're cheap entry level. They might work fine for what is required, and can decent value, but certainly not the best.

>When would you say their tools got better? And what brands are you talking about?

start of 80's. Of course they've always been ahead for motion control to begin with- pretty much every cnc made worldwide uses japanese electronics -except a few cases where its ABB or siemens (e.g. think fanuc.) Best brands (expensive) Mori Seiki, Yamazaki Mazak (largest machine tool producer in the world atm I think) and Okuma.

of course given your budget, "best machine tools in the world" is fairly irrelevant. pretty much you're limiting yourself to "cheapest machine tools in the world" at less than $1500. Realistically any beat up old 2nd hand mill will work if you don't have certain requirements for accuracy/capacity/performance. A deckel/maho/ most other european machines would be better than an old j-head type bridgeport though.

>>259296
because they're cheap and there's lot's of them (because they were cheap). As far as a machine tool goes, they're pretty sub standard in terms of fit, accuracy and rigidity.

The Japanese have been making the best machine tools in the world for a while now. Although there's not many around, a used japanese machine is likely the best. Failing that, huron, deckel maho are decent european mills for cheap 2nd hand. if you're buying cheap stuff new, generally speaking taiwan>china>india

And go as big as possible. On used mills especially the increased mass helps compensate for wear that reduces rigidity in damping chatter etc. And get a self releasing taper- preferably iso- spindle.

>>236391
not convinced- on the basis that is moronic-as-fuck.

>>230126
they are. There's a reason that even though an FPGA is much more capable than a MCU - and comparable in price for capability that you see a dissimilar amount of hobby level projects using mcu's/FPGA's. They are a lot more involved. Microcontrollers are simple business. What design software are you using?

>>230114
constructing the logic is the least of your worries. How much experience with electronics do you have? It's more relevant to understand electronics than being able to program with FPGA's- making a program which theoretically does what you want is pretty simple- or at least not much more difficult than programming the same thing in a higher level language like C. when it comes to simulation/testbench/pin fitting/pin output/input modes and timing analysis is when things (we're still only on the software side) get tricky

>decode and sample the HDMI signal in real time.
>I'm pretty sure I'll need an FPGA to do this
>I want to learn how to use them.


From my experience with making CPLD/FPGA based systems- I don't think this is going to be as easy as you think it will be. What sort of engineering degree do you have OP?

>>227679
>You basically need a bachelor's in engineering PLUS you need to know how to do CAD and CNC programming,

Not really. I have qualifications as CNC programmer, machinist and fully qualified engineer.

As a machinist you don't need knowledge touching what an engineer learns- for CNC you need to be proficient with CAM software, as well as the machine controller and general machining (feeds, speed, tool geometry, chip loads other basic stuff). That pretty much has you set.

A machinists role isn't supposed to include modeling with CAD utilities either. This is a draftsperson, which is a separate specialised trade. But, its a pretty relevant skill, and most trade courses include the basics of it to at least give an appreciation for drawing interpretation.

Depending on where you are in the world, rates can be pretty good. Here in Australia, depending on the special skill set/competence you have you can make $90/hr as a CNC programmer/machinist. It can easily be more profitable than being an engineer- but you have to be good at what you do.

>>227340
you don't need a cnc machine

>>227391
you don't know what you're talking about

>>219264
yes
>>219625
>CNC master race
where is your cnc machine? I see a lot of handwheels.

also:
>tb6560
>R8

were the first 2 mistakes I saw. At this rate you qualify for garage tinkerer tier at best.