206 replies to this topic

[mcinnes01]

Hi, Thanks for all the help, so the ATtinys will allow me to control the current and will provide smoother control of the dimming that may get some issues if ran directly from the managed code through the TLCs right? I am just looking for the ATtiny461a, they are available but they don't seem a very common chip, are there any similar atmel chips that provide all the functionality I will need? Maybe cheaper? Are there any where multiple outs from the TLCs can go in and seperate outs for multiple LEDs come out. E.g. 4 TLC pwm go in and then each of these have an out? I'm just thinking to keep the number of chips and cost down? Cheers, Andy and thank you so much for the help and the diagrams, fantastic!

[Magpie]

Hi The attiny461a should be a drop in replacement for the attiny461. So you should be able to use either. You can also use the 861s as they have a little more memory, but are otherwise exactly the same. You cant use the 261s as they have less memory and so my program wont fit in. I think they cost $3 each from digikey. Each attiny will do 3 channels they have 3 inputs for current sense. 3 inputs for setting the pwm. 3 outputs for the pwm. so for 16 channels you will need 2 x tlc and 6 x attiny

[mcinnes01]

Thanks magpie, I am just ordering the parts at the moment. I've found them for £1.70 + VAT for 10, which is how many I will need for the 30 channels right? I thought it would be 2x tlc and 10x 461 for 30 channels is this correct? (each tlc has 16 pwm outs and each 461 has 3 right?) My blue leds have just arrived so I will test with those for now, they are more or less the same ratings as the cree's and are cheaper if I kill a few. Looking at your schematic, to provide the 30 channels, how many of what other componets would I need to order? I thought I might as well get it all at once since I will save on postage etc. Another thing, as you know different types of 3w Leds have slightly different power ratings, is it just the resistors I need to swap to match the power needs of the Leds? And/or is it possible to make it adjustable for the leds? I have an avr isp mk2, so this should suffice for flashing the chips, is there any advice on connecting this up? or is there a way I can design the boards to incorporate the 6 pin headers if I wanted reprogram them in circuit? Thanks again for all your help. Andy

[Magpie]

Hi

Looking at your schematic, to provide the 30 channels, how many of what other componets would I need to order? I thought I might as well get it all at once since I will save on postage etc.

I gave you a bom for single 3 channel led driver. You will have to multiply this by 10 to get 30 channels.

So something like 60 pnp transistors, 30 npn, 30 fets, 30 diodes etc.

Remember I did say that you should do a proof of concept, I was actually thinking that you should be of doing the proof of concept before ordering enough for 30 channels.

So the 2 x tlc should still be fine.

You should also use your wago plugs, try and arrange it so that you cant reverse plug anything.

Another thing, as you know different types of 3w Leds have slightly different power ratings, is it just the resistors I need to swap to match the power needs of the Leds? And/or is it possible to make it adjustable for the leds?

The leds will be dimmed by the TLC's duty cycle. The only other things you have to decide on are the fuse ratings and the maximum current through the leds.
The maximum current can be adjusted by either changing the sense resistor or changing a value in the firmware.

In the light that I made (see photos) the sense resistors were either 1 ohm or 0.5, then you can see in the photo that I paralleled them up to get a bit more efficiency. (This value dwarfs the rds on value unfortunately). It might be good to leave some space near the sense resistors, so you can parallel them up.

I have an avr isp mk2, so this should suffice for flashing the chips, is there any advice on connecting this up? or is there a way I can design the boards to incorporate the 6 pin headers if I wanted reprogram them in circuit?

You could reprogram them in circuit possibly, but I just used 20 pin dip sockets and pull them out to program them. Once your program is ok you wont have to change it.
Go to avrfreaks forum, for advice on the avr isp mk2, lots of people there.

No problems about help, I just hope that your not biting off more than you can chew.
Get one 3 channel device going first. Use it as you reference.

30 channels is a lot of soldering and a lot of work.

Things that would be useful:
A current limiting supply.
Oscilloscope.
Half decent multimeter.
Good soldering iron.

[mcinnes01]

Hi Magpie,

I think you're right, start small (proof of concept) then move to a batch production once I am happy with the outcome.

I'm just going through the BOM and had a few ideas and questions in terms of "the nitty gritty" with the parts order (some details are outside of my electronics knowledge currently).

Ok so here is your bom for a fader controlled version, if you don't think its too much of a stretch do you think it is possible I could do my proof using your tlc controlled schematic rather than the faders or do you think it best to go with the fader route first?

1 x attiny 461a
1 x 12 or 24 volt psu. (choose current mainly for the leds, the attiny only takes about 20ma I think.
1 x 5 volt regulator ( for the attiny whatever it needs) 7905 if you want to go upside down.
1-3 x 10k fader (for testing.)
3 x 100uH inductor or there abouts.
3 x Fet 40-70 volts, low rds, n type enhancement. IRF1010e
3 x 5819 shottky diodes
3 x 0.5 ohm resistors
3 x 470 ohm resistors
3 x 180 ohm resistors
3 x 9 ohm resistors (attiny pwm out ?? into base)
6 x bc327
3 x bc337
3 x 10uF tantalum 35v caps
1 x 100uF or so 35v or greater electrolytic cap

So questions top down, I have a few power supplies, I have some 12v 1amp wall warts, a few PC PSUs, and a lab pack with adjustable voltage (no idea current it produces), would any of these surfice?

In terms of the 5v for the 461s, I was wondering rather than regulate the 12v down could I just use the 5v from the netduino? My main board that powers the TLCs is soldered up and I will be passing some cables between that enclosure and the enclosure the drivers will go in (maybe I will use an old printer cable they seem to have a fair few pins) any way the point is I could pass the netduinos regulated 5v to be used to power the attiny461s, what do you think and would that get rid of the 7905s?

In terms of the inductors you said this :

The inductor L1 can either be a lossy inductor with thin wire, or a resistor, or a resistor + an inductor. With the cap it will make a low pass filter. I was thinking of one of those inductors that look like resistors. The 3db point should be very low, say 10-50 hz.

I know the resistor look ones you are talking about but in terms of the 3db point I'm not sure what this means? And the 10-50 hz all the ones I've found seem to be in the Khz range or am I looking at the wrong thing? Here are a few examples is there anything I should be looking for i.e amp rating etc farnell. Do the inductors help remove some of the RF that is produced by PWM?

With the schottky diodes does it matter if the are 1a or 3a and does the mv rating matter either?

With the bc327 does the transition frequency need to be 100 or 260 and does the current gain matter?

Same with the bc337?

And that just about concludes my lack of understanding with all the value choices I am faced with . Again appologies you are dealing with someone who has some overall understanding of electronics, but very little specific knowdledge, but I'm definately learning as I go along.

Cheers,

Andy

[Magpie]

Hi

Ok so here is your bom for a fader controlled version, if you don't think its too much of a stretch do you think it is possible I could do my proof using your tlc controlled schematic rather than the faders or do you think it best to go with the fader route first?

Just get a fader(potentiometer), 5k - 100k doesn't matter, grab one off something old. It's only for testing. New they are only 2 dollars and you only need one for the whole job.

1 x attiny 461a
1 x 12 or 24 volt psu. (choose current mainly for the leds, the attiny only takes about 20ma I think. ( for 12 volts I would have to change a couple of resistors around.

1 x 5 volt regulator 7905 is a must.
1-3 x fader (for testing only .)
3 x 100uH inductor or there abouts.
3 x Fet 40-70 volts, low rds, n type enhancement. IRF1010e
3 x 5819 shottky diodes
3 x 0.5 ohm resistors
3 x 470 ohm resistors
3 x 180 ohm resistors
3 x 9 ohm resistors (attiny pwm out ?? into base)
6 x bc327
3 x bc337
3 x 10uF tantalum 35v caps
1 x 100uF or so 35v or greater electrolytic cap

+ 2 x 0.1 uF Ceramic to decouple the chips power supply( you might need some for the TLCs.)

I know the resistor look ones you are talking about but in terms of the 3db point I'm not sure what this means? And the 10-50 hz all the ones I've found seem to be in the Khz range or am I looking at the wrong thing? Here are a few examples is there anything I should be looking for i.e amp rating etc farnell. Do the inductors help remove some of the RF that is produced by PWM?

We need a filter to turn the square wave into an analogue signal. You calculate a filters frequency by 1/(2* pi * RC) I thought if it started rolling off at around 10-50 hz then the input to the attiny wont jump around too much.

Just get the 5819 shottkys they wont be conducting for long anyway. Lower millivolts in forward voltage is good but I used the 5819s and they are working fine. They should be very cheap. Farnell is very expensive. If you have the time 2-3 weeks Futurlec is very cheap but they tend to have older models of everything.

With the bc327 does the transition frequency need to be 100 or 260 and does the current gain matter?

you can substitute 2n2222 and 2n2907 and many more. They just need to be bog standard. If they get warm then they probably aren't fast enough.

I realised I missed half your questions, I give you a more detailed response tomorrow.

Read up on passive filters on wikipedia, 3db point, and bandwidth.

[Magpie]

Hi

So questions top down, I have a few power supplies, I have some 12v 1amp wall warts, a few PC PSUs, and a lab pack with adjustable voltage (no idea current it produces), would any of these surfice?

I would add this circuit. This will give you a current limited supply.
Add on current limiter
It works in the linear region and so burns of the excess power as heat, good for the test bench, bad for the final project.


The circuit for the converter is based around 15-24 volts.
You can probably use it straight off with 12 but the Fet has to be a type that is totally switched on at Vgs 8v.
I checked this out, it seems ok. see Image.
The linear region is the diagonal region,
where it flattens out the fet is the fully on region.

With the bc327 does the transition frequency need to be 100 or 260 and does the current gain matter?
Same with the bc337?

Current gain does matter but it only matters if the lack of current gain takes you into the linear region.
The easiest approach is to get more current gain if it doesn't cost, anything over hfe 150 will be fine, less than 100 dont bother. you should be able get bog standard to92 package transistors anywhere, give them a go. If something gets hot then you may need to change them. it probably needs a frequency response of about 1Mhz or greater, but you dont need to check this just get standard to92 npns and pnps.

Attached Files

•  irf1010e operating region.png   13.54KB   7 downloads

[mcinnes01]

Hi Magpie,

I have read up on 3db point and passive filters and kind of have a better understanding of that now. However I'm still a little unsure as what to look for in an inductor and how to check if the one I choose will produce the desired roll off around 10-50Hz.

You gave this formula: 1/(2* pi * RC)

So I tried to use the details off the datasheet for this inductor, but I couldn't find or workout RC.

I then read this which further added to my confusion (I haven't done this sort of maths for about 7 years and I can't say I was fantastic at it then, so I find I get quickly confused).

Any way this is the last component I need to add to my order, I have gone with attiny861a as they are in stock currently.

I am using farnell, they may not be the cheapest but they seem to be the only people with all the components in stock so I'm not too bothered.

Also the fets (these) I've never bought fets before, are they generally quite expensive? They are £1.49 on farnell for the specific ones you mentioned and even on ebay from china they are a similar price. This isn't an issue, it just seemed strange to me that they cost so much? Does it have to be those specific fets or are there any others you would recommend? As I say pricewise I'm not really bothered but I just wondered if there is a comparable component?

Thanks again, I'm looking forward to getting a test driver up and running!

In terms of the power supply, I have some wall warts I thought I would use for my test circuit, they are 12v 1amp will this be ok to use as is or would you recommend using the current limiting circuit as well that you suggested?

Also in terms 7905s will I need 1 per LED, 1 per ATTINY861a or just 1 when I build all my drivers?

I wonder when I do come to building the 30 drivers if there is a way to kind of have 1 driver board that provides 30 channels, i.e. connect each of the drivers circuits together where possible and if this will mean for example I would only need 1 of certain componets like the part of the circuit that provides power to the attinys etc?

Many thanks

Andy

[Magpie]

Hi

For the Low Pass filter I can see a few options (which I didn't really spell out properly).

LC, RC, CLC, RCRC. (eg. LC means inductor then capacitor to ground)

Upon a closer look I think the ones with an inductor are impractical as the inductor will be too large.
There is also an option for RCRC which is second order, which gives a sharper rolloff. This is probably uneccessary.

Which leaves us with RC, plenty of people have had the same problem.
http://forum.allabou...ead.php?t=22222
http://www.ontrak.net/pwm.htm
Their answers are probably good for you, so 4.7k resistor and 1uF cap. gives around 30 hertz.
There is also a pdf from microchip which has a bit of useful info too.


Attiny 861s will be fine.

Does it have to be those specific fets or are there any others you would recommend?

No there are many alternatives that you can use and millions of ones that you couldn't use.
I chose a T220 package because they were large and can emit a lot of heat but normally you could get away with a small surface mount package, for the output currents we are talking about.

So choose a fet that has the following.
N channel enhancement MosFet
VDS > 40v and
Vgs threshold around 4 volts ( not logic level)
Suitable package for easy soldering onto vero board, (ie. probably not surface mount)
At least 4 amps Ids. probably more than 10.
Lowish Rds on less than 50mOhms.
suitable thermal characteristic. What I mean here is some chips quote an Ids that is suitable only at liquid nitrogen temperatures. If you have a low rds on and plenty of margin in Ids, this wont be an issue.

If you give a link to a preferred supplier I could suggest something.

In terms of the power supply, I have some wall warts I thought I would use for my test circuit, they are 12v 1amp will this be ok to use as is or would you recommend using the current limiting circuit as well that you suggested?

The wall warts are fine, I recommend also making the current limiting circuit, it is really easy and if you make 30 channels of fet drivers I guarantee it will save you at least once.

Also in terms 7905s will I need 1 per LED, 1 per ATTINY861a or just 1 when I build all my drivers?

You need at least one, but also if you split the board onto multiple boards then you might have one per board. At a guess 2-6.

I wonder when I do come to building the 30 drivers if there is a way to kind of have 1 driver board that provides 30 channels, i.e. connect each of the drivers circuits together where possible and if this will mean for example I would only need 1 of certain componets like the part of the circuit that provides power to the attinys etc?

I think if you build the 3 channel prototype, and can get it working, you will be in a position to answer this yourself. If you can build the prototype and get it working, things will seem a lot clearer.
It is very hard to design complicated things, perfectly, in one go.

Also you will have a lot of wires, use your wago plugs to make sure that things can't be attached incorrectly, even on your prototype.

[mcinnes01]

Hi,

I've put together a basket based on 30 drivers, I tend to do this to get an idea of final cost but obviously I will get enough for a proof at first. Somethings like caps and resistors and the smaller transistors I will be ordering for other things like my relay control board.

In terms of the fets are these ok or are there any you would prefer? fets

I already have a few bit so instead of the faders I have some pots (although I can rip a few faders from an old mixer).

For the cuurent limiter I have some 1k resistors, on my calc to get a 700ma supply I want the 1R resistors right? (do you think 700ma is the right supply for my cree xp-e's?)

What are the 9ohm resistors you mentioned in your bom? I can't find any 9ohms would you suggest 2 18ohms or another more available value or can you see any 90hms on farnell perhaps I'm just blind?

The info was very useful I feel a bit more confident with the idea behind the filter now.

Here is my checkout:

MULTICOMP - MCMHR50V104M4X7 - CAPACITOR, 0.1UF, 50V
STMICROELECTRONICS - 1N5819 - DIODE, SCHOTTKY, 1A, 40V
TAIWAN SEMICONDUCTOR - TS7905CZ - IC, V REG -5V, 7905, TO-220-3
ON SEMICONDUCTOR - BC327-25ZL1G - TRANSISTOR, PNP, TO-92
ON SEMICONDUCTOR - BC337-25ZL1G - TRANSISTOR, NPN, TO-92
MULTICOMP - MCGPR50V107M8X11 - CAPACITOR, 100UF, 50V
MULTICOMP - MCKNP01SJ050KA10 - RESISTOR, WIREWOUND, 0R5, 5%, 1WS
MULTICOMP - MCRE000028 - RESISTOR, CARBON FILM, 125MW, 180R
AVX - TAP106M035CCS - CAPACITOR, 10UF, 35V
ATMEL - ATTINY861A-PU - IC, MCU, 8BIT, 8K FLASH, 20PDIP
MULTICOMP - MCRH100V105M5X11 - CAPACITOR, 1UF, 100V
MULTICOMP - MF50 4K7 - RESISTOR, 0.5W 1% 4K7
MULTICOMP - MF50 470R - RESISTOR, 0.5W 1% 470R
NTE ELECTRONICS - 1N4148 - DIODE, STD RECTIFIER
MULTICOMP - MCKNP03SJ010JA19 - RESISTOR, WIREWOUND, 1R, 5%, 3WS
MULTICOMP - D44H10 - TRANSISTOR,NPN,10A,80V,TO220
INTERNATIONAL RECTIFIER - IRFZ44VZPBF - MOSFET, N, 60V, 57A, TO-220AB



Thanks again,

Andy

[Magpie]

I already have a few bit so instead of the faders I have some pots (although I can rip a few faders from an old mixer).

sounds good.

For the cuurent limiter I have some 1k resistors, on my calc to get a 700ma supply I want the 1R resistors right? (do you think 700ma is the right supply for my cree xp-e's?)

Dont confuse the power supply current with the led current.
Yes the calculation is 1R for a 700mA current. This will suit your purposes. You could also switch in/out a few more ohms to make it 100mA as a sort of settable current limit. 100ma initially and if that is good then 700mA.
700mA is a nice running level for the crees, but you could go up to 900mA with a 10% margin for error. You dont need to make the current limiter bigger to get 900mA as your buck convertor will do this. eg. If you have 2 leds and 12 volts in your power source current might be 400mA and your led current might be 700mA.

MULTICOMP - MCMHR50V104M4X7 - CAPACITOR, 0.1UF, 50V

use ceramic

What are the 9ohm resistors you mentioned in your bom? I can't find any 9ohms would you suggest 2 18ohms or another more available value or can you see any 90hms on farnell perhaps I'm just blind?

You are right. I think I used 8.2ohm on my light. But I don't think this is critical. I think that I could even have made it zero ohms. I was just playing around with it because the small transistors were getting hot. When I put decent transistors in it was fine anyway.
In the photo you can see some transistors have a metal can. and on one channel I have t0-92. the metal can ones were running hot, I got them from futurlec.

STMICROELECTRONICS - 1N5819 - DIODE, SCHOTTKY, 1A, 40V
TAIWAN SEMICONDUCTOR - TS7905CZ - IC, V REG -5V, 7905, TO-220-3
ON SEMICONDUCTOR - BC327-25ZL1G - TRANSISTOR, PNP, TO-92
ON SEMICONDUCTOR - BC337-25ZL1G - TRANSISTOR, NPN, TO-92
MULTICOMP - MCGPR50V107M8X11 - CAPACITOR, 100UF, 50V

All these are right.

MULTICOMP - MCKNP01SJ050KA10 - RESISTOR, WIREWOUND, 0R5, 5%, 1WS

1 ohm but put 2 in parallel to make 0.5 ohm, these are more accurate.

MULTICOMP - MCRE000028 - RESISTOR, CARBON FILM, 125MW, 180R
AVX - TAP106M035CCS - CAPACITOR, 10UF, 35V
ATMEL - ATTINY861A-PU - IC, MCU, 8BIT, 8K FLASH, 20PDIP

All these are right.

MULTICOMP - MCRH100V105M5X11 - CAPACITOR, 1UF, 100V

use ceramic or film

MULTICOMP - MF50 4K7 - RESISTOR, 0.5W 1% 4K7
MULTICOMP - MF50 470R - RESISTOR, 0.5W 1% 470R
NTE ELECTRONICS - 1N4148 - DIODE, STD RECTIFIER
MULTICOMP - D44H10 - TRANSISTOR,NPN,10A,80V,TO220
INTERNATIONAL RECTIFIER - IRFZ44VZPBF - MOSFET, N, 60V, 57A, TO-220AB

All these are right. the fet looks good.

MULTICOMP - MCKNP03SJ010JA19 - RESISTOR, WIREWOUND, 1R, 5%, 3WS

is this for the current limiter? If so it is fine. You might want to use some of these as a dummy load to replace some of the leds in initial testing. they come in a pack of 10.

You haven't got the 100uH inductors. These are integral to the buck converter.
http://uk.farnell.co...10-3/dp/1692925 are a minimum
these would be better

Obviously Farnell is convenient but pricey.

[mcinnes01]

Hi,

I've found these I'm not sure if they are chokes or inducturs and if it even matters but what do you think?

My link

Andy

[Magpie]

http://uk.rs-online....leaded/7157226/ should be ok but costs more.

these might be better but I am not sure what material they are or their frequency response.

We dont want a choke, as it is a sub class of inductor that often performs poorly at high frequencies, but uses a cheap magnetic material and is therefore inexpensive. Although a lot of inductors with high quality materials can be successfully used as chokes and hence are grouped in inductors/chokes.


We need an inductor that performs well up to 1Mhz or so.
We will switch at either 64khz or 128khz ( I cant remember which). But a square wave at 128khz has some reasonably strong harmonics at 10x the frequency. So to keep the power transfer efficient we need to have and inductor that performs well at about 1Mhz.


What we are looking for is good performance at 1Mhz, 2A max current (approx), small series resistance as possible. Small cost. and obviously somewhere near 100uH.
By the way 100uH is a number I pulled out of the sky, I found some on some old switch mode supplies and they seemed to work.
As you can see inductors can be very complicated, especially if you start looking into core types.
Inductors is where many manufacturers of switch modes skimp, but usnig a good one it should improve efficiencies.

CoilCraft have a good calculator for their products

coilcraft calcs,

they should have suppliers in the uk.

That said the most inefficient part of my circuit is actually the current sense resistor, any volts lost here are pure waste. I should make it smaller so it wastes less power and use a transistor to amplifiy the voltage so the Attiny can read it more accurately.

[mcinnes01]

Quick update, I finished soldering up my prototype control board, this isn't the drivers but it does include the TLCs which are a key part. So far I haven't done much testing, but I have powered the entire board up for the first time and no magic blue smoke or fire, so that can only be a good thing... In terms of the drivers for my LED controller I am still deciding on a parts list, I will update when I have more news.

[Magpie]

Hi Andy
I have been looking at suitable components for my Steff shield project, they will probably be useful for you too if we end up doing the same driver stage.

For a shottky diode, MBRS240LT3G might be worth a try.

for an inductor either the Wurth 744770133 or the 7447709330 would be excellent.
They may be bit pricey but they wont saturate, wont have much core loss and wont have much resistance loss.

[Magpie]

Hi Andrew How is the aquarium going? I have some of the NCL30160's but I have blown 4/5 of them. They are static sensitive. But seem to work as specified. I used the circuit straight out of the datasheet. I may soon try and rejig my attiny version for my domestic lighting project. I have a book on PID control and and some better surface mount fets. But I would recommend you continue with something like the NCL30160, to keep the part count low and to interface nicely to your tlc5940s.

[mcinnes01]

Cool so your happy with the results? Do you have a BOM I will get everything ordered as soon as. It is a little worrying re the static and turning on too quickly, what exactly do you mean turning on too quickly do they need a soft start or something?? Thanks for your long term input on this topic I really appreciate it! Andy

[Magpie]

I take back anything I ever said about the NCL30160s

They are Junk!

I had my last one running this afternoon, and for one hour it was fine.
I went away and came back in another hour, my workshop was smokey and chip number 5 had gone short circuit. It was only running at 130mA and the chip was running cool. It did actually run very cool for the work it did. I once had it up to 800mA.

I initially had 29 volts, 130mA running into the chip but now I had 450mA running into it (I didn't read the voltage but It would have dropped to about 3-6 volts at a guess).
My current limiting supply had gone into limit mode and I had 450mA running through a breadboard to my chip. The heat in the chip was making the smoke and now it is stuck to the breadboard, I will send a photo, soonish. My smoke alarm didn't go off, it only goes off for normal cooking smells. sigh.

The only excuse I could find for this chip is that the circuit was breadboarded, and you shouldn't do this at High frequencies, but I dont actually believe the breadboarding of this circuit was a problem.

In short the NCL30160 is junk. Dont use it for your own safety.

If you look at the block diagram the short seems to be in the built in voltage regulator.

A timely reminder to take precautions when doing power electronics,

• smoke alarm
• current limited supply
• fuses
• watch closely

[mcinnes01]

Sorry to hear the bad new Magpie. On the plus side at least it didn't burn your workshop down So I guess its back to the initial driver circuit... What do you rate the efficiency at for your normal driver especially if I am doing a single LED per channel? Do you think there may be some other chips with a similar functionality to the NCLs but that are more robust and more reliable or do you think I will be best following suit and going for the micro controlled driver? Thanks for all your help on this subject, its been illuminating (litterally!). Andy

[Magpie]

I am thinking for your purposes, you can get away without current sensing.
You just need over current protection.
This will give you
less losses
very simple circuit
excellent linearity, good for colour.
but at a cost of

fair bit of software fiddling if you change the leds per circuit, or supply voltage, if this is the case you will have to reset the max pwm
you will need fuses too, to protect your leds. make sure the fuses blow before the leds.

Have you worked out how many channels of how many leds?
Is it 32 channels of 1 led?
For the single led channels I don't think I would bother getting an inductor either.

just something like.

tlc -> PMD3001D -> nfet gate.(logic level fet, or maybe STN3NF06L) 

and VCC - fuse - leds - nfet drain and nfet source to ground.


or VCC - fuse - leds - inductor - nfet drain and nfet source to ground.
              |              	|
              |--- diode --------|

Obviously prototype first.


For my purposes ( domestic lighting 10 leds in series) I am going to rejig the Attiny461a, finish, improve the code, and choose the best transistors, and fet.
Maybe even go synchronous.