[info] Power supply - kb1gtt

[kb1gtt]

Never mind, it would appear that there are 1N SMT diodes.

[kb1gtt]

Looks like the goal was to get low cost ebay diodes. I see SMT 7.5V zeners on ebay for about $0.03 each in qty 100. If the goal was low cost ebay diodes, see these options.

http://www.ebay.com/itm/New-Brand-ST-ZMM-SMD-SMT-Zener-Diode-3V-39V-LL34-1206-Multi-Variation-Values-/252689538966?var=&hash=item3ad5785f96:m:mtoQgIxNSQMQu6AvXKk0ODA

http://www.ebay.com/itm/On-Semi-7-5V-500mW-Zener-SOD-123-MMSZ5236BT1-20-pcs-/252772844859?hash=item3ada6f853b:g:NvcAAOSwDN1UOUk2

[AndreyB]

What is the desire for a 1NXXX designation?

The desire is to use wider available cheaper parts where possible. I was afraid of custom BZT52C7V5-7-F being expensive which turned not to be the case. Considering how cheap BZT52C7V5-7-F is, we will be using BZT52C7V5-7-F and not 1N5343B

[AndreyB]

@ has all the components to populate two boards - one with Mouser-sourced components another board with eBay sourced components.

136) D1003 & D1006 one of pads needs larger drill, same as other
137) D1001 needs polarity silkscreen

[kb1gtt]

Code: Select all

138)  Verify diode polarities, new KICAD libs flipped pin 1

[kb1gtt]

On R0.1, the diode silk screen is backwards for D1001, D1002, D1003 and D1006. The diode D1004 is OK.

[AndreyB]

Ready to be shipped north.

[kb1gtt]

Hmmm, looks like Q1002 footprint should have used the same as the D1006 diode. Oops.

Do I recall hearing they make 5V? If so that's a good start.

[AndreyB]

VS-50WQ04FNPBF is only available in qty 75 and up on Mouser now, but they also suggest Similar Product VS-50WQ04FN-M3 available cheaper and in qty 1+

Do we care for Ifsm - Forward Surge Current drop from 550 A to 340 A? Do we understand the relationship between these two part numbers? Does VS-50WQ04FN-M3 work for us?

[kb1gtt]

I see the below 2 datasheets.

Suggested
http://www.vishay.com/docs/93312/50wq04fn.pdf

Current
http://www.vishay.com/docs/94233/50wq04fn.pdf

The only difference I see in these 2 datasheets is that the suggested diode is Halogen Free. I would say the VS-50WQ04FN-M3 is a suitable alternative. Thinking about it, I guess the original is a valid alternative. Basically the VS-50WQ04FN-M3 is what will be popular in the future, so the original is basically going obsolete.

[AndreyB]

These smaller round inductors are a bit less available on eBay.

There are actually at least two kinds:
Radial Lead Inductor 68uH 700mA 5x7mm
68uH 8x10mm Radial Leaded Power inductors 68 uH 2A

Would any of these work? Do we care for 0.7A vs 2A?

Why two boards on the picture above? The one on the left - both terminals blue - has a larger percentage of mouser, more expensive parts.

The board on the right - one terminal blue one green - has a larger percentage of eBay parts. It would be cool to see if we can see any objective differences in the behavior of these two. The inductors are 700mA 5x7mm.

[kb1gtt]

I got to do some basic tests. The china board is noisy, it's making a 1 kHZ hum which is annoying. It's dropping some voltage across the inductors, so the output voltage is a bit below 5V. I think the cheap inductors aren't so good, and I suspect the noise and voltage drop will get better if the inductors were replaced. I have some of the better inductors installed on the last board, so I'm tempted to switch them out, then start more in depth testing. Some results from my basic testing.

The Q1002 high voltage surge protection transistor drops about 0.4V under normal conditions. It limited the LM2596 input voltage to about 17V to 18V with a power main voltage of up to 62V. I have another power supply, so I can go higher in voltage. However it get's hot fairly quick as it's dissipating watts to provide this surge protection. So I'll need to do this test fairly quickly to prevent it from overheating.

It maintained 5V until about 6.5V, and it still produced 4V at about 5.5V. I was testing with a 10 ohm resistor for the load, so output of 0.5A load.

All in all, it's looking to function OK. Lets see if it snubs some of that higher frequencies.

[AndreyB]

Hold on, one board has mouser inductors another one has eBay inductors. Did you try both?

[kb1gtt]

Yes, I tried both. The China / Ebay board had some minor issues including an audible noise and slightly lower than 5V output. Both minor issues on the China board will likely go away if we use the better inductors. The Mouser board did not have these issues.

[kb1gtt]

Both the ebay inductor and the mouser inductor are rated for less than 1A. This means that as I increase the current draw they start to get hot. I can find higher amp rated devices, but I don't like the costs.

$2.90 to $4.18 https://octopart.com/1120-680k-rc-bourns-39573188
$0.69 to $1.51 https://octopart.com/5248-rc-bourns-39573028

There are many options for around 1A that are $0.70, to $1.20. The currently used coil is $0.69 to $1.10. So I guess the vertical mounted 5248 above is basically the price to expect expect from official channels, but with better amp ratings. I forget how I came to the one we currently have. My guess is it was a short coming of my 10 minute work windows. I think we want a higher amp rated coil to avoid lost watts.

I'm not good at finding these kinds of things on ebay. If someone finds a coil on ebay that's between 68uH and 100uH and has a current rating of more than 2.5A, we could consider those coils as well. We are also looking for a small PCB size, we can go tall with out any issues, but we have limited PCB area to work with.

[mobyfab]

No sure if this is the right place but...

Have we checked if using a separate LDO for analog is actually better than trying to filter VDD?
I'm pretty sure you can reduce noise using a cheap low noise SOT23 LDO.
This is what I did, went from ~30mV to ~2mV ripple on ADC values using a LDK120DM33R.

One other thing you can do is reduce the pin output speed on the MCU as these are the major source of ripple on VDD.

[Abricos]

Inductance: 100uH
Wire Diameter: 0.55mm
Inner Diameter: 6mm
Outer Diameter: 13mm
Thickness: 5mm
Rate Current: 6A


Look at this on eBay http://www.ebay.ca/itm/142148453127

[AndreyB]

This Toroid Core would take even more space than the original square we were using. Well, maybe not more but pretty close. Remember we want three of them.

[mobyfab]

You can also change to a modern regulator that uses a higher frequency.
ie: http://www.ti.com/lit/ds/symlink/lm22670.pdf

Full range of replacements here:
http://www.ti.com/ww/en/simple_switcher/regulators.html

[kb1gtt]

We only use the LM2596 because it can be easily obtained on e-bay for very low $. I would expect higher frequency regulators would allow for a physically smaller size. I've done a couple other designs which use more modern chips and higher frequencies, one that was specifically for automotive applications. However those cost much more. When purchased from ebay the LM2596 is about $2-ish. If we could fit the slightly larger square inductors, the cost's would not be very bad. The problem is that I'm trying to add filtering and high voltage protection, all on the same side of the PCB. The problem is that there are several other circuits around this which makes it hard to get the PCB real-estate.

[mobyfab]

These are starting to appear on aliexpress:

1.5Mhz, synchronous, small, very cheap
https://www.aliexpress.com/item/Smart-Electronics-XD-45-Mini-360-Model-Aircraft-DC-DC-Step-Down-Power-Supply-Module-Better/32482324438.html

Datasheet
http://www.haoyuelectronics.com/Attachment/MP1584/MP1584.pdf

I ordered a bunch, I will test them with various loads and check temperature/ripple.

[Abricos]

I'm totally confused ...
On Frankenzo R0.1 v first one ...
Power supply have 5 volts ... is it have 3.3 volts ???

Or discovery make 3.3 volts ???

[kb1gtt]

Discovery makes 3.3V.

[AndreyB]

We have a problem

Jared, you still have these test populated boards? Can you run them for 10 minutes? 30 minutes?

PS: I am a bit confused. While @ has recently reported something similar at https://rusefi.com/forum/viewtopic.php?f=5&t=1301, his board is Frankenso 0.4?! And https://rusefi.com/forum/viewtopic.php?f=3&t=1282 shows a 0.5 board without this problem. So my assumption that this is 0.5-specific does not hold ground. Yet I have two assembled boards with this issue?

Let's keep this in the power supply topic and I will move later if the problem would be about something else.

[video]https://www.youtube.com/watch?v=YOD3ITIhWkk[/video]

[kb1gtt]

-- Can you do a similar video while a meter is in view, showing the 5V?
-- Do you have 2 meters? Can you put one meter in series and shows the current consumption? Can you show the current consumption of a R0.4 board?
-- Now if you had a thermal camera, you could also show me how hot it gets. Does anything seem to get hot around the power supply? Specifically U1001, any of those diodes, Q1001 or Q1002.
-- How about the voltage at P902? Do we see the voltage drop there when it dies?
-- Can you verify R1003 is 100k?

I'll see if I can find those test boards. The better half has cleaned up a couple of my areas lately. I'll see if I can find the box which those were put into. I'm sure I've run those boards for more than 10 minutes though. So I doubt they would show the same problem.

The 5V appears to be trailing off like it started at 5V, then starts to slowly drop. The thing where turning it off for a short period of time seems to reset it for another 300 seconds is odd. I can see thermal issues causing it to shutdown, but I would expect the second run to turn off quicker as it has already gotten hot. I've seen some issues with LED's which run for several minutes then stop making light. I wonder if similar is happening with D1002. If that zener voltage is changing similar to how I've seen china LED's fail, then that could choke U1001. If U1001 is choked off then I could see how it would have trouble making 5V.

[kb1gtt]

Also what is your 12V power supply voltage while this is happening? AKA perhaps the 12V supply is failing. I could see how that could have a 300 second issue.

How did https://rusefi.com/forum/viewtopic.php?f=3&t=1282 get an R0.5? Was that a first run board, which you sold him? Was that a self populated board? Do we know for sure it's an R0.5 and not an R0.4?

[AndreyB]

-- Can you verify R1003 is 100k?

Problem solved - there was no R1003

More details on https://rusefi.com/forum/viewtopic.php?f=4&t=569&start=390#p27980

[kb1gtt]

Ah they installed one of those giga-ohm spark gaps. Got it and good news it's running now.

[puff]

minor problem with 3.3 rail. Right now, with 5V input I have 2.945V on 3.3V rail - the rail which powers up our canbus transceiver (one of the early revisions of frankenstein).
Because of that, when I connect TJA1050 to our canbus, I don't see anything, since CANH in dominant state outputs 2.8V at maximum, while the TJA1050 datasheet states, that minimum valid value for that state is 3V.

[AndreyB]

minor problem with 3.3 rail.

My understranding is that this topic is about the specific 5v power supply schematics - that schematics has nothing to do with 3.3v I assume. What is the context of your question?