[info] Power supply - kb1gtt

Hardware inside and outside of the ECU
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

I'm looking at that sky-macau company as they offer this inductor http://www.sky-macau.com/Keyword-Search/C-0/k-3817361683/ the design uses 2X of them. Which costs about $1. The design should be posted shortly, and I'll be looking for people to review it.

That company didn't have the catch diode I was looking for, so I chose their best option. When used it limits the output to 2.3A, and if you replace that diode with the one in the BOM, it would get back to the 3A output. However most people don't need the full 3A.

Yes datasheet claims 80% at 5V output. I'm not completely sure how that changes over the range of voltages. I think it get's closer to 90% as you get closer 7V input. But I won't know until it can be measured.

About the thermal coeff, I was referring to the case to ambient. I'm assuming reasonably poor air flow. To compensate for poor air flow, you have to make a large alum case. When we drop the watts in the controller, we decrease much aluminium we need in the enclosure heat sink.
Welcome to the friendlier side of internet crazy :)
User avatar
mobyfab
Posts: 139
Joined: Tue Oct 29, 2013 10:09 am
Location: Versailles, France

Re: Power supply - kb1gtt

Post by mobyfab »

kb1gtt wrote:I'm looking at that sky-macau company as they offer this inductor http://www.sky-macau.com/Keyword-Search/C-0/k-3817361683/ the design uses 2X of them. Which costs about $1. The design should be posted shortly, and I'll be looking for people to review it.

That company didn't have the catch diode I was looking for, so I chose their best option. When used it limits the output to 2.3A, and if you replace that diode with the one in the BOM, it would get back to the 3A output. However most people don't need the full 3A.

Yes datasheet claims 80% at 5V output. I'm not completely sure how that changes over the range of voltages. I think it get's closer to 90% as you get closer 7V input. But I won't know until it can be measured.

About the thermal coeff, I was referring to the case to ambient. I'm assuming reasonably poor air flow. To compensate for poor air flow, you have to make a large alum case. When we drop the watts in the controller, we decrease much aluminium we need in the enclosure heat sink.
The efficiency of non synchronous converter gets worse with lower output voltages because of the diode forward voltage.
ie: is Vf is 0.6V, it's 18% of loss at 3.3V, and 8% at 5v, just for the diode.

LM2596's efficiency is actually worse at 7v input, about 77% to 5v.

For the rest it's a compromise between mosfet (RDSon and switching losses) and inductor resistance.
Higher frequency -> Smaller inductor, less resistance. More switching losses. Smaller output cap.
Lower frequency -> Bigger inductor, more resistance. Less Switching losses. Bigger output cap.
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

With an output of 5V the diode losses are likely around (5-.7)/5*100=86%. However it's switching, so I expect the catch diode is not fully loaded. Assuming it's fully loaded, we have around 6% from switching losses and such. As we decrease the upstream voltage, you'll likely slow down how fast you charge the output as the voltage is lower so it will take longer to get current flowing in the coil. So the switching frequency will probably decrease which will decrease the losses while that's in the transition state(s). Also if you aren't running that full 3A, I don't know how much that will change things. Ultimately I don't trust my predictions until I make it and run a bunch of tests. For all I know magnetic influences could help or hurt the entire thing.

Any how, see schematic and PCB layout found here. Please review letting me know strengths and weaknesses. Of course KICAD files are posted there as well.
http://daecu.googlecode.com/svn/Hardware/trunk/KICAD_Project_PWR_buck_12V_switcher/PWR_buck_12V_switcher_schematic.pdf
http://daecu.googlecode.com/svn/Hardware/trunk/KICAD_Project_PWR_buck_12V_switcher/PWR_buck_12V_switcher_PCB.pdf
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Power supply - kb1gtt

Post by AndreyB »

I've just pushed your current version to rusEfi svn.

Jared, do you want me to order the PCB and send it to you? Did you do a paper validation already?
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

I was going to look for pear review here, then if all looks good I was going to do the paper validation with what I can, then order the assembled boards. That was my plan. I think I should give people until the weekend to do a review. Then I'll have to figure out how to get my printer working. Seems those ink's are always drying up.
Welcome to the friendlier side of internet crazy :)
User avatar
Sergey89
contributor
contributor
Posts: 839
Joined: Wed Sep 25, 2013 5:30 pm
Location: Russia, Velikiy Novgorod

Re: Power supply - kb1gtt

Post by Sergey89 »

I think that the pair of the D2/D3 diodes should have a much larger limit of the maximum power dissipation according to the ISO. Something like a 5KP series TVS diode.
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

I'm probably less than clear about the BOM. My original choices are in the first columns and currently start with "ref_" and those are currently for reference only. I then chose components from the china list which are the footprints I designed with.

One item worth noting, is that the fuse is a polly fuse. I have no idea how much current that will allow, or how long it will allow it. I'm assuming it will trip before the diodes break, and that it will reset reasonably quickly. When it trips I'm assuming the bulk cap will ride out the voltage dip. I guess I could put a parallel resistor to allow some current to flow during the 100V spike. However with a lack of data sheets, it's all a bunch of guess work. Basically I expect the diodes don't need to be so large.

Argh, that fuse is 6V rated. I need to change that, probably should remove it, or change it for a good voltage rating.
Welcome to the friendlier side of internet crazy :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

quote request sent. Lets see what kind of pricing they come back with.
Welcome to the friendlier side of internet crazy :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

Oh, I almost forgot to thank Sergey89. Looking closer at the diodes and PCT, I agree it was significantly flawed. I have removed that PCT, and I have changed to the larger thru hole zener(s) found here http://www.sky-macau.com/Products/Zener-Diode-C80/10-pcs-Zener-Diode-20V-20V0-200V-200-V-1W-1000mW-DO-41-P8262743708.html Hopefully those zener(s) can blow an upstream 5A fuse. Fingers crossed I guess.
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Power supply - kb1gtt

Post by AndreyB »

Did you have a chance to print the PCB & make a visual check with actual components?
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

I was mostly successful with that approach. I couldn't get a real 100% scale, it was around 95% so I had to kind of eyeball that it was OK, but as far as I could tell, it was OK.

I also got one e-mail back from the company. I still don't know how much it will cost, or how long it will take to assemble.
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Power supply - kb1gtt

Post by AndreyB »

what about the fuse - is if the same package?

For a prototype is not that easier to just solder an oshboard ourselves? I can order a set from oshpark if the fuse package is fine
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

They didn't have a suitable fuse, so I just put in two vias side by side. We can install a fuse later on if we want, or we can simply make that a 0R. The issue with most of there fuses are that they are for low voltage, so when the fuse trips, it would have to hold off 12+ V, but they were only rated for 6V. Also the reset-able fuses don't change resistance as much as I would prefer. It really should have an upstream fuse.
Welcome to the friendlier side of internet crazy :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

I have e-mailed those folks a couple times, but I still do not have a quote or lead time. So I still do not know how much it costs or how long it will take to assemble. I get a generic e-mail noting they got the e-mail, but haven't gotten an e-mail from an actual person. So it's not looking very good for this approach. Seems I have gotten lost in an automated system.
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Power supply - kb1gtt

Post by AndreyB »

These guys in this context are sky-macau.com?

Would oshpark be an option, especially since the guy behind has registered here to confirm these are made in the US? :)
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
Tomin
Posts: 39
Joined: Fri Oct 18, 2013 8:03 pm

Re: Power supply - kb1gtt

Post by Tomin »

Hi,

do you really want to go with switching power supply in automotive ? I think it is a bigger work to design it well than it seems.
Every engine ECU what I saw until now had linear power s. I saw switch. p. only in clusters.
What about transient peaks (100V ?) in +12V ? How good will be response, for example, at +5V side ?
Especially with China chips, I don't think so ....

What about reliability ? In case of testing on a table it is ok, but for my car, for almost every day using, it is bad path, definitelly.
Switch. power are notorious known with bad capacitors ... almost everything what I had to fixed in my home/my friends/... was switch. power supply.
After 2 years of using ... dry el.capacitors.... shorted diodes.... in my car ? No, thanks. ;)
Tomas
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

Correct I'm getting a slow reply from sky-macau.com who seems to have lost things in the automated world. I got a real reply back last night, but it appears they did not look at the BOM, as I specified all their components, but they suggested I obtain the components, then provide those components to them. The assembled price wasn't to bad, it was 5 PCB's for $93 - component costs. I have replied to the e-mail asking if they can provide the components found on their web page. I'll see what kind of reply I get.

I'm game for OSHPark, I've had good success with them before and I've been very happy with the quality and service and such. My issue is assembly time, It's hard for me to get a period of time where I can sit down with the iron long enough to assemble a board.
Welcome to the friendlier side of internet crazy :)
puff
contributor
contributor
Posts: 2961
Joined: Mon Nov 11, 2013 11:28 am
Location: Moskau

Re: Power supply - kb1gtt

Post by puff »

iron for soldering or for etching?
oshpark is way too expensive in comparison to chinese fabs when speaking about something bigger than inch-by-half-inch boards. this poorly correlate (if at all) with the whole purpose of this project (building cheap ecu)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

My constraints are less about $ and more about time. OSHPark has reasonably low prices, with a proven good quality record.

About the use of switch mode power supplies, I have not see any modern ECU's that do not use switch mode devices. The issue with Linears is that you need a very large heat sink, as linear are very inefficient. Both linear or switch caps designs may or may not use electrolytic caps. If a design uses more expensive tant's instead, you also have to remember that if you don't properly protect them, they like to explode violently. So it's a bit of a trade off which type of cap you use. I generally prefer Tant's when I can.

Explode cap (note mote electo's have the vent options)
general tutorial about caps.
Welcome to the friendlier side of internet crazy :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

Seems I have gotten past the automagic issues, they are being very responsive now. And they have noted they can populate with components from their web page. I inquired how the price would change. I'll see what they come back with for a total populated price.
Welcome to the friendlier side of internet crazy :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

It appears the total price for 5 of these PCB's is going to be $118.45, so about $23.69 each. Not bad for QTY 5. They noted "The component price is about US$25.45" Looking at the web page I see a min and max price listed for each component. This gave me a range of price at a min of $17.50 and a max of $54.20. So increased qty could decrease the component price some, but it look like the components cost about $5, and increased qty would drop the price significantly in terms of PCB and setup fee's. All in all it's looking quite good. I have inquired how I pay, and I should have an order placed shortly.
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Power supply - kb1gtt

Post by AndreyB »

I am not getting it, why would you need 5 of these? I vote for an oshpark board + I can solder it and ship it to you for testing, should not that be WAY cheaper?
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

5 boards was the min qty. You've got enough on your plate with software. I've also wanted to try one of these proto houses just to see how it goes. I've got many projects and having a low cost proto house would be handy. I don't think we are in a rush for time on this one, as we can buy near equivalents for low $.
Welcome to the friendlier side of internet crazy :)
puff
contributor
contributor
Posts: 2961
Joined: Mon Nov 11, 2013 11:28 am
Location: Moskau

Re: Power supply - kb1gtt

Post by puff »

I didn't get how that would bit my $2.98 power supplies (russian pointed to the store; i ordered two)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

That little black box may not survive voltage spikes up to 100V. We should break one apart some time and see what's inside. Perhaps there is a lower cost chip that could be used.
Welcome to the friendlier side of internet crazy :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

This looks cool does + or - regulation with lots of protection's Can handle up to 42V in, and output 2A. My pledge of $18 is getting me an assembled unit, so materiel costs have to be lower than that.
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Power supply - kb1gtt

Post by AndreyB »

Image
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

Very nice very nice. Did you get the component's I sent?
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Power supply - kb1gtt

Post by AndreyB »

BTW at least some of the back side silk screen is off.

Yes I've got the parts you've mailed me and I have just ordered the rest from digikey - I hope I've ordered the right stuff.
Actually, now I am looking at the schematics and I think I am still missing at least D1 - the 1206. Let me email you my digikey order.
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: Power supply - kb1gtt

Post by kb1gtt »

It would appear I had the back silk turned off when doing the layout. Oops. It's a minor first spin issue. Thanks for reporting I'll aim to update it when I get home.
Welcome to the friendlier side of internet crazy :)
Post Reply