Teensy 4.0 external power burning hot

Status
Not open for further replies.
My multimeter seem not to want to measure current... it's a cheap AstroAI AM33D, when I set it on "200m" it's going over the limit, if I set it to "10" it's reading 0.00... so between 200mA and 10A...

Most multimeters have 1 or 2 fuses. If you ever mistakenly attempt a voltage measurement across a power supply while the meter is accidentally set to current mode, the fuse blows. That's nicer than massive sustained current damaging your gear, and maybe even starting a fire if the power source is huge. But then it leaves your multimeter forever unable to measure current until you replace the fuse, which you may not even know is inside the meter and you probably wouldn't even know it's blown if it happened by accident when attempting a voltage measurement.

Alternately, you could replace that 33uH inductor before the regulator with a resistor in the 1 to 10 ohm range. Then just measure the DC voltage across the resistor and use Ohm's law to compute the current.

You really should measure the current.

Besides this i would suspect something wrong on the teensy side, most probably some bad soldering at the pins?

Indeed, there could be another problem causing the too-high current. You're really flying blind without a way to measure the actual current.
 
Besides this i would suspect something wrong on the teensy side, most probably some bad soldering at the pins?

Well I use this Teensy every day, it seems to work fine.

Checking the Board Pic from above where are the usual huge Caps for the regulator? I just can spot 2 tiny SMT ones, i doubt they will be the needed 22uF+ ones. 0,1uF will certainly get the LM78 going but wont allow for any significant load.

I use the attached schematics to power 3.2s in most of my modules, having a OLED on the teensys 3,3V rail with no issues regarding noise or temperature (gets barely warm).
View attachment 24768

I was only following the datasheet:
https://datasheet.lcsc.com/lcsc/1811052003_Changjiang-Electronics-Tech-CJ-CJ7805-3_C21706.pdf

It shows a 0.33uF and a 0.1uF capacitors.

But for the next rev of this board since I want to replace the LM7805 with this https://www.digikey.ca/en/products/detail/cui-inc/VX7805-500/7350283 I'm going to replace them with the 10uF and 22uF they show on the datasheet.

Or do you advise differently?


Most multimeters have 1 or 2 fuses. If you ever mistakenly attempt a voltage measurement across a power supply while the meter is accidentally set to current mode, the fuse blows. That's nicer than massive sustained current damaging your gear, and maybe even starting a fire if the power source is huge. But then it leaves your multimeter forever unable to measure current until you replace the fuse, which you may not even know is inside the meter and you probably wouldn't even know it's blown if it happened by accident when attempting a voltage measurement.

That is probably the case.

Alternately, you could replace that 33uH inductor before the regulator with a resistor in the 1 to 10 ohm range. Then just measure the DC voltage across the resistor and use Ohm's law to compute the current.

You really should measure the current.

Indeed, there could be another problem causing the too-high current. You're really flying blind without a way to measure the actual current.

Are you talking about @fdaniels schematics? Because those are not my board.
But I just did that test though, so connecting the Teensy on the headers except the 5v pin and adding a 10ohm resistor in series on the 5v pin and the 5v from the board. And then measuring the voltage on the resistor. So the result is 1.2v/10.3ohm=0,1165048544 A = 116mA That doesn't sound crazy actually?
 
@ghostintranslation

I cant read your schematics well, as the chosen color schmeme black/red doesnt work very well for me. I'd try the folowing:

- replace the 2 tiny SMT Caps C1, C2 (?) at the regulator with something like 22µF or even bigger (i have nice 100µF Subminature Caps) - the bigger the better. Worst case short the SMT pads, solder directly to the regulators pads to try.
- replace the LM7805 with an MC7805CDTG (same footprint as far as i can see) https://datasheets.diptrace.com/on_semi/mc7800-d.pdf

and see if that works.

Generally you should buffer the rails around the regulator generously, helps with transients.
Important is a 33µH inductor to isolate the Rails from incoming and - much more important as this effects every other module - outgoing noise. At least the Teensy 3.2 + OLED isnt "quit" if you dont use that inductor.

116mA is a joke, an 0,96" OLED needs this alone. My guess would be the buffering caps are to small for the transients the 4.0 produces.
 
@ghostintranslation

I cant read your schematics well, as the chosen color schmeme black/red doesnt work very well for me. I'd try the folowing:

- replace the 2 tiny SMT Caps C1, C2 (?) at the regulator with something like 22µF or even bigger (i have nice 100µF Subminature Caps) - the bigger the better. Worst case short the SMT pads, solder directly to the regulators pads to try.
- replace the LM7805 with an MC7805CDTG (same footprint as far as i can see) https://datasheets.diptrace.com/on_semi/mc7800-d.pdf

and see if that works.

Generally you should buffer the rails around the regulator generously, helps with transients.
Important is a 33µH inductor to isolate the Rails from incoming and - much more important as this effects every other module - outgoing noise. At least the Teensy 3.2 + OLED isnt "quit" if you dont use that inductor.

116mA is a joke, an 0,96" OLED needs this alone. My guess would be the buffering caps are to small for the transients the 4.0 produces.

What's the benefits of the MC7805CDTG over the LM7805?

I just checked to make sure, according to Teensy specs, it is supposed to consume 100mA, so it looks right, especially because I'm not using any of the other chips on that test: https://www.pjrc.com/store/teensy40.html, so yeah nothing crazy.

2 thoughts about the capacitors now that you mention it:
- I used capacitors described as "330nF ±10% 25V X7R 0603 Multilayer Ceramic Capacitors MLCC - SMD/SMT RoHS" and "100nF ±10% 50V X7R 0603 Multilayer Ceramic Capacitors MLCC - SMD/SMT RoHS". Is it the ceramic type ok and is the 0603 package ok?

- You mention isolation, I'm thinking I shouldn't worry that much about noise since the power would be coming from a eurorack rail that is supposed to have big capacitors and be very stable, and since all I'm doing is digital I'm not so concerned by noise in my modules, and I'm guessing any other module that would be concerned would implement anything necessary to avoid noise. No?

I'm also thinking I could do 2 more tests:

- Unplug the teensy completely and just run 2 wires to 5v and GND with the 12v power connected, so that I make sure it's not something else on the board that makes it heat when all the pins of the Teensy are connected.

- Unsolder the 2 capacitors and try again, because I'm not actually 100% sure it's the 7805 that is heating and not the capacitors because it's so tiny and they are close...
 
116mA average current seems about right, if you're powering up some other stuff besides the Teensy.

Did a quick check here with a Teensy 4.0 on my desk. At 600 MHz is draws about 90mA. Stepping down to 528 MHz lowers it to about 80mA.


- Unplug the teensy completely and just run 2 wires to 5v and GND with the 12v power connected, so that I make sure it's not something else on the board that makes it heat when all the pins of the Teensy are connected.

This is a good test. But since you measured a pretty reasonable 116mA flowing into the regulator, you probably need redesign with a heatsink or more efficient switcher rather than a simple linear regulator which gets hot.


- Unsolder the 2 capacitors and try again, because I'm not actually 100% sure it's the 7805 that is heating and not the capacitors because it's so tiny and they are close...

I wouldn't bother this this.

Capacitors are almost never a source of heating. It takes a lot of high frequency current, which just isn't likely when you have a linear regulator as your current source and a board like Teensy (which has lots of ceramic capacitors close to the chip) as the current sink.
 
What's the benefits of the MC7805CDTG over the LM7805?

Its a much newer design that performs better in every aspect.


2 thoughts about the capacitors now that you mention it:
- I used capacitors described as "330nF ±10% 25V X7R 0603 Multilayer Ceramic Capacitors MLCC - SMD/SMT RoHS" and "100nF ±10% 50V X7R 0603 Multilayer Ceramic Capacitors MLCC - SMD/SMT RoHS". Is it the ceramic type ok and is the 0603 package ok?

Package doesnt matter. But theres a slight difference between 300nF and 22µF. Not that much, just about 3 orders of magnitude.



- You mention isolation, I'm thinking I shouldn't worry that much about noise since the power would be coming from a eurorack rail that is supposed to have big capacitors and be very stable, and since all I'm doing is digital I'm not so concerned by noise in my modules, and I'm guessing any other module that would be concerned would implement anything necessary to avoid noise. No?

Your Module exists in coexistence with all the others in your modular, so you should care as everyone else keeping the power rails as clean as possible. A Big Cap doesnt help with high frequency noise and the Inductor just attenuates both ways so you are not "isolated". Its the same story as with polluting the ecosystem you live in and not caring.

If you dont care about polluting the rails i'll just stop here and watch people rip your module apart because of emitting digital noise to their system filled with oldschool full analog modules.

For the noise IN your modules you should well be concerned, latest if you have hiss in your audio out and cant read any A/D better than 6 Bits resolution.....

One last thing: did you consider reverse voltage protection?
 
Its a much newer design that performs better in every aspect.




Package doesnt matter. But theres a slight difference between 300nF and 22µF. Not that much, just about 3 orders of magnitude.





Your Module exists in coexistence with all the others in your modular, so you should care as everyone else keeping the power rails as clean as possible. A Big Cap doesnt help with high frequency noise and the Inductor just attenuates both ways so you are not "isolated". Its the same story as with polluting the ecosystem you live in and not caring.

If you dont care about polluting the rails i'll just stop here and watch people rip your module apart because of emitting digital noise to their system filled with oldschool full analog modules.

For the noise IN your modules you should well be concerned, latest if you have hiss in your audio out and cant read any A/D better than 6 Bits resolution.....

One last thing: did you consider reverse voltage protection?


Alright I see,

No I didn't consider reverse voltage, first time I'm doing anything with power really. Should I add a diode for that?

Also I was looking at the Arduino Nano that actually can take 12v natively, it uses the LM1117IMPX-5.0 with just one 1uF capacitor:
https://content.arduino.cc/assets/NanoV3.3_sch.pdf (this is from https://store.arduino.cc/usa/arduino-nano)

I'm guessing that if they chose this chip it can handle 12v without getting dangerously hot.

So looking at LCSC (I get my PCBs from JLCPCB which LCSC is the list of components for in factory assembly), there is this chip that seems similar:
https://datasheet.lcsc.com/lcsc/1810231832_Advanced-Monolithic-Systems-AMS1117-5-0_C6187.pdf

Now comparing back with CJ7805:
https://datasheet.lcsc.com/lcsc/1811052003_Changjiang-Electronics-Tech-CJ-CJ7805-3_C21706.pdf

What I note is the thermal resistance from junction to air (with similar packages):
- CJ7805 80ºC/W
- MC7800 40~50 ºC/W
- AMS1117 45 ºC/W

I guess that depends on many things like how much power to dissipate and how the PCB under the component is done but that just shows that the CJ7805 is getting about twice as hot at the others if I understand that correctly?
 
What I note is the thermal resistance from junction to air (with similar packages):
- CJ7805 80ºC/W
- MC7800 40~50 ºC/W
- AMS1117 45 ºC/W

I guess that depends on many things like how much power to dissipate and how the PCB under the component is done but that just shows that the CJ7805 is getting about twice as hot at the others if I understand that correctly?

Keep in mind that every datasheet is a sales pitch. It's main purpose is to convince you to buy the chip. Complete & accurate technical documentation is a secondary concern. Datasheets are written by marketing departments whose only mission is to sell the chips.

Most specs are pretty clear, but you need to be careful with some specs like "junction to ambient". The IC vendor isn't selling you the ambient environment, nor lots of things than come between the ambient air and the semiconductor junction and package. If you look at TI's LM1117 datasheet, there's a footnote under section 7.4 on page 5. The hyperlink takes you to an app note called "Semiconductor and IC Package Thermal Metrics", which remarkably candid advice on page 2.

screenshot.png

They admit these datasheet specs are pretty much meaningless. They don't quite confess to gaming the spec by testing with an utterly unrealistic PCB design, but the deeper you dive into all this, the harder it is to see those optimistic specs in the datasheet as objective info. Remember, it's not. It's a sales pitch. Buyer beware. (and remember 70-95% of the actual power dissipation depends on your PCB design, not the chip)
 
The following pages from that shown above went on to describe suggested PCB design in an attempt to alleviate the heating effect.
 
I've had great luck with this switcher:
https://www.digikey.com/en/products/detail/recom-power/R-78E5-0-0-5/2834904
$2.92 + 3 tiny caps and an inductor. I'm using on a portable device so efficiency of %90 is great for battery life. Currently running on 12V input dropping to 10.4V as battery discharges. Experimenting with running 2 lithium batteries instead of 3 with drops down to 6.8V and device seems to still perform well even though out of manufactures input spec. Using on T3.2 and T4.0 with audio board and on board microphone and headphones for testing.
 
Keep in mind that every datasheet is a sales pitch. It's main purpose is to convince you to buy the chip. Complete & accurate technical documentation is a secondary concern. Datasheets are written by marketing departments whose only mission is to sell the chips.

Most specs are pretty clear, but you need to be careful with some specs like "junction to ambient". The IC vendor isn't selling you the ambient environment, nor lots of things than come between the ambient air and the semiconductor junction and package. If you look at TI's LM1117 datasheet, there's a footnote under section 7.4 on page 5. The hyperlink takes you to an app note called "Semiconductor and IC Package Thermal Metrics", which remarkably candid advice on page 2.

View attachment 24780

They admit these datasheet specs are pretty much meaningless. They don't quite confess to gaming the spec by testing with an utterly unrealistic PCB design, but the deeper you dive into all this, the harder it is to see those optimistic specs in the datasheet as objective info. Remember, it's not. It's a sales pitch. Buyer beware. (and remember 70-95% of the actual power dissipation depends on your PCB design, not the chip)

I see, thanks, at the time I made this PCB I didn't think about heat so I didn't look into that subject at all, now I know a few things.

The following pages from that shown above went on to describe suggested PCB design in an attempt to alleviate the heating effect.

Interesting subject for sure.

I've had great luck with this switcher:
https://www.digikey.com/en/products/detail/recom-power/R-78E5-0-0-5/2834904
$2.92 + 3 tiny caps and an inductor. I'm using on a portable device so efficiency of %90 is great for battery life. Currently running on 12V input dropping to 10.4V as battery discharges. Experimenting with running 2 lithium batteries instead of 3 with drops down to 6.8V and device seems to still perform well even though out of manufactures input spec. Using on T3.2 and T4.0 with audio board and on board microphone and headphones for testing.

Nice thanks for the feedback.

I have ordered this one https://www.digikey.ca/en/products/detail/cui-inc/VX7805-500/7350283 earlier this week following this thread to give it a try. It looks similar to the R-78E5-0-0-5 you use. So I'll be able to test next week with my current PCB by just desoldering the SMD 7805 to solder wires and connect the dc converter on a breadboard. Additionally I ordered a through hole 7805 to see if this one acts differently.

Now that I've seen the Arduino Nano schematics and board design I'm also tempted to try the same LM1117IMPX-5.0. I noticed that they added just 3 vias under it, with only 1 being actually used and the 2 others are there only for heat transfer. Since it's a proven design I could just go with that and not try to reinvent the wheel.
 
So I just got my order and tested both MC7800 and VX7805-500:
https://www.digikey.ca/en/products/detail/on-semiconductor/MC7805CTG/919333
https://www.digikey.ca/en/products/detail/cui-inc/VX7805-500/7350283

Result is, the 7805 is getting as hot as the smd one I had before, I burn my fingers on it, but the VX7805 stays completely cold.

I also tried the MQS audio while using the VX7805 to see if I would get any hearable noise but no. I don't have an oscilloscope to see better unfortunately.

So now the DC converter seems a good option to me, I'm still curious about the LM1117IMPX from Arduino Nano but I also just noticed that the revised version the Arduino Nano Every is using a DC converter, the MPM3610, so I'm not sure I want to try the LM1117IMPX anyway.

I guess I will look at what JLCPB is offering as SMD DC converters with minimal components required around since I would not need to be able to switch from a linear regulator to a pin compatible DC converter since the linear gets way to hot even in through hole package.

So something like those:
https://datasheet.lcsc.com/lcsc/1809050422_XLSEMI-XL1509-5-0E1_C61063.pdf
https://datasheet.lcsc.com/lcsc/1806140314_Texas-Instruments-LM2576SX-5-0-NOPB_C34465.pdf

I'll update this feed when I have a new prototype of my board with the change.

Thank you for the help.
 
My multimeter seem not to want to measure current... it's a cheap AstroAI AM33D, when I set it on "200m" it's going over the limit, if I set it to "10" it's reading 0.00... so between 200mA and 10A...

Even with 10A it should measure something. Usually there are two terminals for measuring currents - one for small currents (200mA) and second one for high currents - so maybe you were measuring it on wrong terminal
 
So I just got my order and tested both MC7800 and VX7805-500:
https://www.digikey.ca/en/products/detail/on-semiconductor/MC7805CTG/919333
https://www.digikey.ca/en/products/detail/cui-inc/VX7805-500/7350283

Result is, the 7805 is getting as hot as the smd one I had before, I burn my fingers on it, but the VX7805 stays completely cold.

I also tried the MQS audio while using the VX7805 to see if I would get any hearable noise but no. I don't have an oscilloscope to see better unfortunately.

So now the DC converter seems a good option to me, I'm still curious about the LM1117IMPX from Arduino Nano but I also just noticed that the revised version the Arduino Nano Every is using a DC converter, the MPM3610, so I'm not sure I want to try the LM1117IMPX anyway.

I guess I will look at what JLCPB is offering as SMD DC converters with minimal components required around since I would not need to be able to switch from a linear regulator to a pin compatible DC converter since the linear gets way to hot even in through hole package.

So something like those:
https://datasheet.lcsc.com/lcsc/1809050422_XLSEMI-XL1509-5-0E1_C61063.pdf
https://datasheet.lcsc.com/lcsc/1806140314_Texas-Instruments-LM2576SX-5-0-NOPB_C34465.pdf

I'll update this feed when I have a new prototype of my board with the change.

Thank you for the help.

So I said I'd update... Since then I had to redesign entirely my PCB due to chip shortages...

On the new design I'm using shift registers in place of the LED drivers and I changed the DACs, the LM7805 gets very hot but not as much as before, and the dc-dc converter (VX7805) works perfectly without any heat. I got a cheap oscilloscope, so not super precise perhaps but enough to check for noise in the circuit I think and I couldn't see anything at the out of the converter after the capacitor nor at the DACs outputs.

I find it's good to have the option of both LM7805 and VX7805 so I'll stay with that.
 
Status
Not open for further replies.
Back
Top