 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : Flash 3 common catode LED lamps.
  Page 10 of 12   |
|||||
| Author | Message | ||||
Mixtel90 Guru  Joined: 05/10/2019 Location: United KingdomPosts: 7820 |
hehe... Did some more sums for the TBD62783APG (the through-hole DIP). Package dissipation is 1.47W max. 900mA split over 3 channels is 3x0.342 = 1.026W 2x 100mA is 2x0.0325 = 0.065W Total is 1.091W This thing should handle it easily (if the Vcc pin is ok for 1.1A). It scores because the DMOS output has a maximum of 1.4v drop at 350mA, and a lot less as the load drops. In theory you only need 2 outputs for the 900mA load, but it's unnecessarily pushing them towards their top limit. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 5023 |
Why use 5 channels in a 8 channel chip? Use 6 channels for white, and 1 for each color. Edited 2021-08-11 05:48 by Volhout PicomiteVGA PETSCII ROBOTS |
||||
| Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
I was thinking of spreading out the "white" to 1,3,5,7 as it's a thermal concern but I can try all options  |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 7820 |
Try working out the power dissipation. I must admit that I didn't work it all out in this case, I just picked a number that should work. :) The total doesn't necessarily go down as you use more channels as Rds doesn't have to be absolutely proportional to current. There's probably an optimum number of channels after which you get no improvement - it may even get worse as you have to have a minimum drop per channel. Unused channels dissipate virtually no power, of course. EDIT: Just did some more sums. Spreading the 900mA over 5 channels definitely helps a lot. :) Now it's only a matter of whether the Vcc pin is ok. I think it probably will be, but I'd love to see some documented verification. 1,3,5,7? Are you sure you aren't just trying to make the pcb more complicated? ;) Edited 2021-08-11 06:38 by Mixtel90 Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
Well, I suspect that the drivers may not be spaced along the chip, the way the pins are and therefore, no guarantee of spreading the thermal dissipation but what-the-hey, this will be breadboard and so I'll try everything  Edit: Will include the TLP350H devices as a separate experiment...sure to get some white smoke, one way or another Edited 2021-08-11 07:10 by Tinine |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 7820 |
Poor little TLP350Hs. I'm sure they won't die in vain. I'm pretty sure you'll get at least 1.5A through the Vcc pin of the driver chip though. Whether it will go much higher I'm a bit doubtful. On the low side drivers the ground pin is rated at 3A, but that'll be to the substrate so it can probably be a very short and relatively heavy bonding wire. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
When designing a board for this, a large copper area can really help to suck the heat out from the chip through the output pins. This is especially easy to do when many of the outputs are paralleled. No reason why you could not use two of these chips, and spread the load across sixteen outputs. Cheers, Tony. |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 7820 |
I don't think there's any need for two, Tony. According to my calculations the package would be dissipating around 0.6W with the 900mA white LED split over 5 channels and another 2 channels loaded up with 100 mA each. It's still possible to add a third 100mA LED and stay within package limits. That may be useful if Bob decides he wants a dimmer white channel without having to use PWM or just add another colour. The LEDs are flashing and probably not all lit at the same time for any length of time so actual package dissipation should be less than 0.6W. I'd be happy with it now if it's proved that the Vcc pin is good for the current. I can't understand why *no* manufacturer seems to be specifying a max value for that, other than that if you stay within the dissipation levels and don't exceed the maximum chip voltage then the things will be ok. Edited 2021-08-11 16:38 by Mixtel90 Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
I am sure you are right Mick. I have used plenty of the UDN chips in past projects, and they are pretty tough. The newer versions are possibly even better, but have had no practical experience with them. Cheers, Tony. |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 7820 |
I based the calcs on the DMOS version from Toshiba. The UDN family have a higher volt drop and hence higher dissipation. Mind you, the package rating for those is also specified as slightly higher so that's some compensation. They'll just run a bit hotter, that's all. A bit of aluminium angle and some silicone heatsink glue can work wonders. :) Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1162 |
Pulsing those Led lamps may involve quite high peak currents as you are effectively pulsing the lamp driver circuitry inside the device. Personally I wouldn't use those particular DMOS high side drivers as they are running too close to rating. Cheers Mike |
||||
CaptainBoing Guru Joined: 07/09/2016 Location: United KingdomPosts: 2170 |
has anyone considered a MOSFET hi-side driver? ... woosh... and the bait is cast  |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 7820 |
Lol! How about a changeover relay to select white or blue anodes then a single n-channel mosfet from the common cathode to ground? Use the relay to select the colour then pulse it with the mosfet. You can't have both on at once, of course... Red could have a second n-channel mosfet as it isn't a common cathode one. @Mike If there are big surge currents I'd simply add series resistors up to about 22R and accept the volt drop. The trouble is we have no idea what's in the LEDs. Do we over-engineer? If so, by how much? I'm betting on them only having the cheapest, simplest current limiting. Possibly only a resistor driving some parallel strings of LEDs, although it could be a simple regulator. I doubt if it's a buck converter, although it's possible. Paralleling 5 channels for the white LED allows a possible 2.5A surge while staying within the ratings. That's over 2.5 * full load, which is quite a surge. The Red and Blue channels are 100mA on 500mA rated channels so I suspect they will be ok. There's not enough specification to be certain of anything. It's worth a try though, but it's one of those things where the loads should be checked for surge inputs anyway - no matter what approach might be made. The load is absolute. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
Bear in mind that the 2981 is suited for inductive loads such as motors and solenoids. |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 7820 |
If the load characteristics (surge etc.) aren't specified then it's completely pointless attempting to design a driver. We *do* know, in this case, the full load currents for the LEDs. We can't know anything else as we don't have the LEDs and Bob probably couldn't give us any more information than he has. We know that they aren't voltage critical and a couple of lost volts is nothing to get upset about so at least some surge protection could easily be done with a resistor. The TBD62783APG driver isn't a 2981. It has lower dissipation for the same output current. It also has the advantage that, being DMOS output, there is far less problem in paralleling the outputs. Toshiba say it's suitable for inductive loads. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Only way to tackle a problem like this is to put together a prototype and carry out some real world tests and measurements. Then compare the actual figures to any published specifications that are available. None of that is ever going to happen of course. Cheers, Tony. |
||||
| Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
`Scuse me? I am putting my money where my mouth is; RS order is due to arrive today. I have 2981's TLP350's and SSR's...and the OP still wants to use the wrong crap, just because it's what he has?!?!?! "hey I need to fly from Detroit to Chicago but I already have a round trip ticket to Japan...I wanna use this ticket" This has to be the dumbest thread ever but enough of us are curious about datasheet vagaries that I consider this to be a fun experiment. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
There are quite a few of us here, and on the other thread, that could have made all this work quite easily and without any great effort. But as you say, the original poster wants to use a whole lot of the wrong crap because that is what he has. Now I am all for using stuff I already have to make some kind of interesting obscure gadget for personal use. But manufacturing a whole batch to sell commercially for profit is a whole different ball game. Parts need to be purchased anyway, so might as well buy the APPROPRIATE PARTS and make a proper job of it. Cheers, Tony. |
||||
| Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
All I have ever done is for in-the-field use and this includes big-auto in Detroit and aerospace in Everett, WA. I don't make flaky products. Flashing lights are pretty fricken trivial. My market right now is replacing CNC controls that I supplied 30 years ago which, incidentally, have out-lasted Siemens products. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Before retirement I was in the hardware design game too. One project was a multi million dollar stage automation system now fitted to several of the most famous state theaters around the world. As you are in the UK, you will know of Covent Garden Theatre in London, one of our systems is installed there. For us Aussies, its the Sydney Opera House, and in Melbourne the State Theater. Also large systems in America, China, France, Malaysia, and possibly others since then. Each system uses about forty of those UDN chips to light small 24v incandescent indicator lamps on panels near the machinery. Each chip has a one amp fast blow fuse to the supply pin. That seems to have worked fine as protection, the fuses always blew first never a UDN chip. So the bonding wires seem to be up to at least one amp. Cheers, Tony. |
||||
| Page 10 of 12 |
|||||
 Print this page |
 |
The Back Shed's forum code is written, and hosted, in Australia. | © JAQ Software 2025 |
