05-03-2019, 11:12 PM
Hi, so I've been busy with the power bars and finally got some boards ordered yesterday and should have them soon.
As I mentioned instead of SSR's the switching is done with Triacs. I've settled with some that handle 16amps RMS, unfortunately these still put out a lot of heat under high loads and using a larger Triac doesn't help as it still needs to dissipate the same power. This was a bit of a challenge because I discovered it takes a large heat sink to keep it cool to my liking when pushed to 7 amps which I've settled on for the limit on an outlet. There's no way each Triac could have it's own heat sink as they would be too large, expensive and a waste overall. I needed something they could all share but couldn't find a pre-made sink so I went to the metal shop, got all shapes and sizes of aluminum and started testing.
In the end from what I had the best was a solid chunk of aluminum 16mm x 16mm x 150mm, this worked better than tube or angle. With this it only starts to warm up if a Triac is pushed to 6.9 amps. If the Triac is pushing 4.8 amps the heat sink basically stays cool. If 2 Triacs are being pushed to these levels the total is 11.7 amps which is just under the max 12 amps the heat sink does warm up but still can be easily touched. With the board finished I can actually fit a 200mm long heat sink so there will be an extra 25% which will go a long way. I still may use channel with the longer length as it might get warmer but should cool quicker than the block but in this case the block might win as it might never get above a temp which it's looking like. This is also the worst case scenario, as more current goes through the Triac the heat really adds up and thankfully it's only possible one Triac can actually hit 7 amps as two would be 14 amps and blow the main 12 amp fuse. When I push it with 3.6 amps the heat is manageable with a small sink so I'm confident if 3 Triacs was running this which would be 10.8 amps the heat would be ok. And when you break it down into amps per outlet, 12amps / 8 outlets = 1.5 amps each the heat is much lower. It's not really possible to setup 8 Triacs on my desk but will test more when I have a board assembled. Sorry for rambling but thought someone might find this interesting.
The rest of the components are over sized and have been carefully selected, they have the highest safety ratings and from trusted manufactures ordered locally. Unfortunately with this design the overall cost did go up compared to the SSR's but this should be a nice power bar that will last. Each Triac has been isolated from the DC side up to 6700 volts and there's a heavy duty snubber that will protect from roughly the same over voltage caused from inductive loads. The current power bar has neither of these features and I believe the reason why a relay can die prematurely and because this is overdone it should handle the nastiest pumps out there.
The energy monitoring power bar has an ATmega32U4 microcontroller running at 16 Mhz (equivalent to Mega 2560), this is dedicated to reading power levels on each outlet. Reading consumption from 8 outlets does take some time, it takes a minimum of 5 seconds to read all 8 outlets with some stability. Set like this the numbers do drift slightly, for example you might have a range of 25.0w - 25.3w for a given load but if the number of samples for the reading is increased it can be rock solid at say 25.1w. Obviously I like that better but then it takes maybe 10 seconds to read all 8 outlets which I feel is a little long. Because of this I'm going to have it adjustable if you want to increase accuracy. Now don't let that scare you, it works good but if you plug in a device and expect an instant reading you will be disappointed.
Last update I mentioned anything that was less than 5w didn't work, this has been improved quite a bit using a different transformer, resistors and code. I have a couple small lights that have a 3 watt bulb and now they shows up the same as my kill-a-watt knockoff. I also like the power factor numbers better, they feel more accurate now. I also expect things to be even better once everything is mounted on the PCB instead of using jumpers and filtering caps not ideally placed on breadboard and other boards.
And finally I can't post without some pics so here's the updated board.
![[Image: ct_Front.png]](https://www.robo-tank.ca/pictures/ct_Front.png)
![[Image: ct_back.png]](https://www.robo-tank.ca/pictures/ct_back.png)
As I mentioned instead of SSR's the switching is done with Triacs. I've settled with some that handle 16amps RMS, unfortunately these still put out a lot of heat under high loads and using a larger Triac doesn't help as it still needs to dissipate the same power. This was a bit of a challenge because I discovered it takes a large heat sink to keep it cool to my liking when pushed to 7 amps which I've settled on for the limit on an outlet. There's no way each Triac could have it's own heat sink as they would be too large, expensive and a waste overall. I needed something they could all share but couldn't find a pre-made sink so I went to the metal shop, got all shapes and sizes of aluminum and started testing.
In the end from what I had the best was a solid chunk of aluminum 16mm x 16mm x 150mm, this worked better than tube or angle. With this it only starts to warm up if a Triac is pushed to 6.9 amps. If the Triac is pushing 4.8 amps the heat sink basically stays cool. If 2 Triacs are being pushed to these levels the total is 11.7 amps which is just under the max 12 amps the heat sink does warm up but still can be easily touched. With the board finished I can actually fit a 200mm long heat sink so there will be an extra 25% which will go a long way. I still may use channel with the longer length as it might get warmer but should cool quicker than the block but in this case the block might win as it might never get above a temp which it's looking like. This is also the worst case scenario, as more current goes through the Triac the heat really adds up and thankfully it's only possible one Triac can actually hit 7 amps as two would be 14 amps and blow the main 12 amp fuse. When I push it with 3.6 amps the heat is manageable with a small sink so I'm confident if 3 Triacs was running this which would be 10.8 amps the heat would be ok. And when you break it down into amps per outlet, 12amps / 8 outlets = 1.5 amps each the heat is much lower. It's not really possible to setup 8 Triacs on my desk but will test more when I have a board assembled. Sorry for rambling but thought someone might find this interesting.
The rest of the components are over sized and have been carefully selected, they have the highest safety ratings and from trusted manufactures ordered locally. Unfortunately with this design the overall cost did go up compared to the SSR's but this should be a nice power bar that will last. Each Triac has been isolated from the DC side up to 6700 volts and there's a heavy duty snubber that will protect from roughly the same over voltage caused from inductive loads. The current power bar has neither of these features and I believe the reason why a relay can die prematurely and because this is overdone it should handle the nastiest pumps out there.
The energy monitoring power bar has an ATmega32U4 microcontroller running at 16 Mhz (equivalent to Mega 2560), this is dedicated to reading power levels on each outlet. Reading consumption from 8 outlets does take some time, it takes a minimum of 5 seconds to read all 8 outlets with some stability. Set like this the numbers do drift slightly, for example you might have a range of 25.0w - 25.3w for a given load but if the number of samples for the reading is increased it can be rock solid at say 25.1w. Obviously I like that better but then it takes maybe 10 seconds to read all 8 outlets which I feel is a little long. Because of this I'm going to have it adjustable if you want to increase accuracy. Now don't let that scare you, it works good but if you plug in a device and expect an instant reading you will be disappointed.
Last update I mentioned anything that was less than 5w didn't work, this has been improved quite a bit using a different transformer, resistors and code. I have a couple small lights that have a 3 watt bulb and now they shows up the same as my kill-a-watt knockoff. I also like the power factor numbers better, they feel more accurate now. I also expect things to be even better once everything is mounted on the PCB instead of using jumpers and filtering caps not ideally placed on breadboard and other boards.
And finally I can't post without some pics so here's the updated board.
