Shop our Airplanes Products Drone Products Sales
Thread Tools
This thread is privately moderated by mnemennth, who may elect to delete unwanted replies.
Mar 14, 2014, 03:24 PM
AMA 1033652
mnemennth's Avatar
Discussion

APM 2.5.2 3.3V Regulator Mod - TEST RIG


[EDIT]

WARNING: THIS MOD IS STILL IN BETA!


I'm beginning to see lots of folks talking about how this mod (seen here and elsewhere) has saved their APM, but really not a lot of talk about what testing they've done to verify the validity of their repairs. With that in mind, I feel it's necessary to place this warning conspicuously:

Please, PLEASE, don't go flying in any autonomous modes without FIRST verifying that EVERYTHING works perfectly on the bench, and then when you deploy in an actual aircraft, thoroughly test in ALL manual modes before trying anything where the aircraft is thinking for itself.

Treat it with suspicion just like you would any Flight Controller that had survived a catastrophic crash; because it HAS. The difference is that it survived a mazillion-electron pileup on the 3.3V Bus, rather than getting flipped upside down or bounced off grand-dad's grand old Elm tree.

The potential for epic, gruesome failure is not one whit less.


Please, for your sake and mine, treat it like it's possessed by demons until you have proven otherwise with careful, methodical testing; for this is our recipe:

1 part Ginormous LiPo battery with incendiary potential of a gallon of gasoline

1-8 parts high-powered Electric Motor with "Whirling Blades of Doom"

1 part Ultra-light, ultra-strong airframe equipped with a brain that can think for itself

Umpteen parts uber-complicated radio transmitter and electronics to connect it all together

I don't know how Murphy manages to stay away long enough that EVERY FLIGHT of every machine ever built doesn't end in major litigation.


With this in mind, proceed with great care. Then, please come back and report your results in as much detail as you can manage on my Mod page; so we can collect the results and pass them on to others, that they too can make an informed decision.


mnem

"Bring me ten propeller hubs of matching dimensions and weight, that we may form the sacred symbol and start the incantation..."

[/EDIT]

We now return you to our show...



APM 2.5.2 3.3V Regulator Mod - TEST RIG

I've spent a lot of time reading on RCG and DIYDrones regarding the 3.3V regulator issue; enough to know that there are varying opinions on how best to resolve it. I too have one of these boards; an HK APM 2.5.2 clone.

After reading and reading, my impulse was mostly "Okay... why did we choose the TPS79133, when the LM1117-3.3 is a tried and true device which is known to be nearly grenade-proof in this kind of application?" Yes, it is approx 10x as much capacity as we need, but it is also 10x cheaper and readily available anywhere. I have a dozen in my parts bin at almost any time.



So... after looking at the problem for a moment, I figure out this: If you take the LT1117CTS-3.3 and bend the pins and tab upward as shown with the specimen on the right, you can flip it upside-down and it will fit in the OEM location securely without requiring any additional leads and will not interfere with access to any pin headers.



Here is my prepped board with the old regulator removed and identifying my solder points.



Here it is with the +5V IN leg soldered to the VCC side of C21 and GND leg soldered to GND side of C11...



...and +3.3V OUT TAB soldered to (+) Side of C22/C23.

[EDITED TO CLARIFY]

Here I've taken the original circuit from the APM2.5.2 RELEASE and turned it sideways so the pinout aligns with the solder pads as seen in my photo above.



THEORY OF OPERATION

The TPS79133 needs a few discrete components to operate properly; those are the decoupling capacitor at C21 and C11, and the brute-force filtering cap bank at C22/C23.

The decoupling cap at C11 connects to an internal low-pass filter at PIN 4 for ripple rejection; the LM1117-3.3 has this internally so an external capacitor is not needed. It can be ignored as it only connects to ground on the opposite side.

The TPS79133 is designed for use in cell phones and tablets where every milliamp of draw is crucial; so it has an ENABLE circuit at PIN 3. This needs to be held LOW for the regulator to turn ON, so it is connected to GND.

The LM1117-3.3 is always ON so it doesn't have this PIN, but its quiescent current draw is small enough as to be inconsequential in our application.

As a result, we only need connections to the circuit at PIN 1, PIN 2 & PIN 5. These connections are more readily available at the points I've labeled in my photo.

The solder pads for the TPS79133 are of very small metal mass and are really only intended to be soldered once - at the time of manufacture. Reworking the board requires applying heat several times; once to remove the old part, once to clean up the pads, and then another time to solder the new part in.

This makes the likelihood of damaging one of the pads due to heat delamination high. This is why soldering to the capacitors as I have shown is a preferable means of connecting for this Mod; the capacitors add mass which helps prevent heat delamination while you are working with the board.


[/EDIT]


I've done testing of this mod as outlined here on the DIYDrones 3.3V Reg thread: http://diydrones.com/xn/detail/705844:Comment:1591737

And my preliminary testing of this simple mod is very promising; I've hot-plug abused it with some pretty large loads, and while the 5V rail does drop enough to make the APM reset when the load is applied, no harm befalls the regulator and 3.3V output never wavers.

[EDIT]


I think this mod will be an acceptable substitute for the TPS79133; I am quite satisfied that it is a much more durable substitute, and continuing testing in Mission Planner after performing the buffer capacitor mod (100uf CERAMIC capacitor across the Analog +5V and GND rails; A2 is suggested as a plug-in location because it is rarely used) recommended by the Devs in the above thread indicates the power output is stable enough and clean enough for the sensitive sensors that will be attached to it.

I've taken that approach and expanded it to include all 4 main plug-in headers with some low-ESR 150uf Tantalum caps (because I didn't have any single ceramic capacitors bigger than 47uf on hand) at the the I2C, INPUT, OUTPUT and ANALOG headers. I know the one on the INPUT header isn't exactly necessary as I already have one on the ANALOG header; but the board wouldn't sit level on my bench without it.

I'll probably need to move the caps once my case gets here anyways; I can put ceramics in place then.



So far, so good; the LM1117 easily handles the inrush current and it appears the additional low-ESR caps have cleared up some PWM noise I was seeing on the USB power in the 200mv range that was passing through the LM1117. Now all I'm seeing is the heartbeat down in the 10mv range. I've tried powering with 3 different PCs and 2 different USB wall-charger power packs; I can't make the APM reset or even lose comms from plugging in my big ceramic cap bank, or from any RX I have on hand.

I think this mod could become an upgrade fix that will be as close to grenade-proof as possible, as a parallel development to the existing protection zener/buffer capacitor device which is being developed to protect boards that haven't failed yet or have already had the TPS79133 replaced.

As you can see, the regulator mod is easily and cheaply implemented; and easily reversible if it doesn't work out. I'm continuing testing to make sure the LT1117CST-3.3 I've used provides clean enough power. I'm already 99% certain it's tough enough.

I believe the multiple buffer caps are better; it's not the times when everything works RIGHT that they're a good idea, it's when things go WRONG... like Joe's son's doing crash-testing with his quad... that they can save yer bacon. That's why I put them at all the rails, not just the one we're having a problem with today.

You never know when or where something is going to get unplugged in a crash, or where corrosion will strike... THAT is the reason it's most important to have something that's SOLDERED in place, not like a plug-in "brownout stopper" cap.

Yeah, maybe I'm being paranoid; a bit silly even. but each buffer cap is another layer of protection, so maybe that will give the gremlins a bit more to gnaw on before they bring my beast down.

[/EDIT]


[UPDATE 3/28/2014]

Okay; so my APM Power Module and my APM Case have arrived; as I expected, my capacitor mod doesn't exactly fit inside the case. What I DIDN'T expect was that my LT1117-3.3 ALSO doesn't fit. I was able to make clearance by Dremeling out about a 5mm x 5mm x 2mm deep cavity in the upper case; but I decided to see if it was possible to make it fit in a stock case.



I have another 2.5.2 board on my bench; so I decided to try rearranging the layout a little differently. As a bonus, no funny bending of leads is necessary; however, we WILL need to add 1 wire for the +5V IN.



We still solder the big 3.3V OUT tab directly to the (+) Side of C22/C23, but this time we don't flip the LM1117; we just turn it a quarter turn so the GND lead aligns directly over the GND side of C11, and the +5V IN lead rests on top of the Atmel 45DB321 Flash Memory IC.



Here's a detail view of the inboard side...



...and the outboard side.



This view shows the height of the replacement regulator; to fit inside the stock case, it needs to be lower than the top surface of the GPS/UART1 plug. Mission accomplished!

Personally, I feel the original solution is probably a bit more solidly mounted as it's soldered at 3 points, but the second solution is more symmetrical looking and fits inside the case. Once it's inside the box, I really don't believe it matters a whole lot.

[/UPDATE]


[UPDATE 3/28/2014]

Okay... so, after working on the regulator issue, I decided to tackle the issue of soldering in a 100uf buffer cap for the 3.3V regulator (As developed by Philip at DIYDrones.com: 100uf Buffer Cap Mod) , only making it so it would fit inside the stock case. After vacillating over several locations on the bottom (none of which had enough clearance), I decided to try the top side.



I had considered these unused ports on UART0 right in front of the regulator; they actually share the same trace as the pads for the regulator, so are a very low-resistance connection point.



First, we tin the ports to get a good solder ball; this will meld with the tinning on the capacitor when we heat it.



Then, just apply a hot iron and solder across both ends at the same time; the capacitor will sink as the solder balls meld with the tinning of the capacitors. I could put up before and after pics; but they won't look any different.



As you can see, it's almost exactly the same height as the GPS/UART1 socket; so should fit nicely inside the enclosure. w00t!



Here's what it looks like on my ESR meter; slightly higher than expected capacitance. This cap actually measured 111.9 uf; the rest is from smaller ones already part of the circuit. What's more important is that super-low ESR!

Testing shows similar behavior to my first experimental board; as above on USB ONLY & Mission Planner running. Hot-plugging with several different RX, including my ORX RX3S W/DSM2 (RX/Gyro Combo) with big caps inside, no stumbles or resets and it never loses comms with MP. Alright!

We'll be shipping this one back; the original owner has promised to put it up on a quad and give it a good workout and report. I'll update when I get the update!

[EDIT]

On choosing a capacitor for this Mod:

There are reasons for choosing the 100uf Ceramic capacitor...

By nature, all ceramic capacitors of this sort (known as Monolithic Ceramic Capacitors) are VERY LOW ESR, so they're a known quantity we can recommend to anyone. The super low ESR means we can use a smaller value (100uf) to provide the fast response buffering we want, yet still have a low enough capacitance not to damage power supply components (like the regulator we're trying to protect) when they first charge up . Plus, they are very compact, so a good choice for small flying things.

Before you PM me asking "Will my XYZ type capacitor offer the same protection?" I'll pass along this:

I have equipment for testing the ESR of a capacitor; as you can see above. I can tell that the Tantalum Capacitors I've used are of a Low-ESR design; I can't tell for you if the ones you have are or aren't.

Unless you have a similar tester and can tell me what the ESR is of the caps you want to try, I really can't answer your question.

If you do have one, then the answer is: if you can get your TOTAL ESR down around 0.01Ω-0.02Ω (I'd guess 0.05Ω at the highest), you should be good. But if you own an ESR tester like mine, you probably already know that.

No, your great big umpty-hundred uf plug-in electrolytic capacitor IS NOT a suitable replacement; it isn't fast enough at clamping down the current surge because it's a cheap electrolytic capacitor. LOW-ESR for a Radial Electrolytic Capacitor is NOT the same as LOW-ESR for a Ceramic Capacitor. PERIOD. But if it doesn't make the USB port on your PC gag and shut down when you plug into it, it won't hurt either.

Yes, you can put a 100uf Ceramic capacitor on a servo connector and plug it in at A2; this is the gist of Philip's Mod: to get this on there for those of us who don't have the level of Solder-Fu I have.

My contention is that this is too important an upgrade to trust to a plug-in connector, so really needs to be soldered-in. That way we don't have concerns over corrosion, poor connection from a faulty connector, or just plain forgetting to plug the damn thing in before we hook up the board and work with it.

This Mod is for folks who are prepared to get the right part, and are prepared to do some reasonably delicate soldering to provide the best protection for their APM, not those who are looking for a quick, easy substitute.

Suitable capacitors are readily available from DigiKey, Mouser, Newark/MCM Electronics. You can also find several vendors selling lots of 10 to 100 pieces on eBay; they may be a bit cheaper than a few items plus shipping from the big suppliers listed above. Search for "100uf Ceramic Capacitor" and choose the lowest voltage rated capacitor you can find that's higher than 5V.

[/UPDATE]



mnem
"Scotty! We need... more... power!!!"

"I'm givin' ye all she's got Cap'n, but we're just goo-in aroo-nd in cirr-cles; and at Warp Ten, that's gettin' noo-where mighty faa-st..."
Last edited by mnemennth; Apr 05, 2014 at 11:59 AM.
Sign up now
to remove ads between posts
Mar 26, 2014, 01:43 AM
Registered User
Nice
Mar 26, 2014, 11:33 AM
Went to mow a meadow
frank48's Avatar
What a brilliant idea, so easy to flip ...

Thanks
Frank
Mar 26, 2014, 06:13 PM
The king of mediocre flyin'!
rikybob's Avatar
Odd, that looks like my old board?
Mar 28, 2014, 09:33 PM
AMA 1033652
mnemennth's Avatar
proficnc, Frank -

Thanks , guys... I'm trying to find a better fix. Hopefully I'm on the right trail.

b- I have no idea why you might say that.


mnem
Grunk!
Mar 30, 2014, 12:25 AM
FPV in Hawaii
iskess's Avatar
Thank you!!!
That's going to work great.
I'll give her a work out.
Mar 30, 2014, 10:56 AM
Registered User
charosenz's Avatar

request to label the 5 pads on the board.


Mnem,

Great work, and thanks for sharing. That is the super part of these threads - helping each other out.

I would like to ask a favor. Would you mind taking that one pic of the board that has the old regulator removed and label each of the 5 solder pads.

Also, do you have a good source for those regulators you are using?

Thanks again.
Mar 31, 2014, 10:29 PM
AMA 1033652
mnemennth's Avatar
Quote:
Originally Posted by charosenz
Mnem,

Great work, and thanks for sharing. That is the super part of these threads - helping each other out.

I would like to ask a favor. Would you mind taking that one pic of the board that has the old regulator removed and label each of the 5 solder pads.

Also, do you have a good source for those regulators you are using?

Thanks again.
The TPS9133 Datasheet is HERE. It shows the pinout.

I buy them from eBay unless I'm already making a big enough order to warrant buying from DigiKey, Mouser or MCM/Newark. Just search for LM1117-3.3 and choose a vendor with lots of positive feedback, very high percentage feedback, and who offer ePacket shipping, which is typically around 8-10 days to the continental US.


mnem
Tuesdog.
Mar 31, 2014, 11:35 PM
Registered User
charosenz's Avatar
Quote:
Originally Posted by mnemennth
The TPS9133 Datasheet is HERE. It shows the pinout.

I buy them from eBay unless I'm already making a big enough order to warrant buying from DigiKey, Mouser or MCM/Newark. Just search for LM1117-3.3 and choose a vendor with lots of positive feedback, very high percentage feedback, and who offer ePacket shipping, which is typically around 8-10 days to the continental US.


mnem
Tuesdog.
Thanks for the link. I just want to confirm that the values for the APM are as labeled as in the pic that I am attaching.

I also wanted to confirm what that you did not connect to the tab on the board where the arrow is pointing that I have labeled with question marks. It looks like this has a line going over to a component to the right side of the tab and I jus wondered about not connecting to this pad.

I also am taking an educated guess () at the middle tab being a ground as it looks like it is connected to the ground.

Thanks.
Apr 01, 2014, 06:39 PM
AMA 1033652
mnemennth's Avatar
If you're doing MY Mod, then NO. You won't be soldering to ANY of the original solder pads. You just need to get the old regulator out out there so it can't interfere.

You will be soldering to the ceramic capacitors AROUND the solder pads; look closely at the photos I've provided, particularly the finished product from multiple angles. Click on the black bar to view the full-size photo for clarity.

My lines point directly to the points you want to be soldering to; the solder lands ON the capacitors themselves. They are MUCH stronger than the tiny pads where the old regulator used to be.

Also, NO. The big TAB on the opposite side on the LM1117-3.3 is 3.3V OUT, as well as the center leg. Look at my hand-drawn pinout above.

Hope that helps!


Paul
Apr 02, 2014, 12:12 AM
Registered User
charosenz's Avatar
Quote:
Originally Posted by mnemennth
If you're doing MY Mod, then NO. You won't be soldering to ANY of the original solder pads. You just need to get the old regulator out out there so it can't interfere.

You will be soldering to the ceramic capacitors AROUND the solder pads; look closely at the photos I've provided, particularly the finished product from multiple angles. Click on the black bar to view the full-size photo for clarity.

My lines point directly to the points you want to be soldering to; the solder lands ON the capacitors themselves. They are MUCH stronger than the tiny pads where the old regulator used to be.

Also, NO. The big TAB on the opposite side on the LM1117-3.3 is 3.3V OUT, as well as the center leg. Look at my hand-drawn pinout above.

Hope that helps!


Paul
Paul,

Thank you for responding, I really do appreciate it. To be honest, I was hoping to learn what the function are of each of the 5 pads that the original regulator connected to. With all respect, I am just a bit reluctant to abandon functionality of pads without learning more about the decision. I really hope that is not offensive, it is just my effort to learn and to make an informed decision.

If you know of a reference to a link that identifies the components of the APM board would be a appreciated.

If I am getting too OT then I apologize and I will move on. No worries.

Thanks again.

Charlie
Apr 02, 2014, 02:01 AM
FPV in Hawaii
iskess's Avatar

APM 2.5.2 3.3V Regulator Mod - TEST RIG


Charlie,
Take a look through the DIYDrones 3v3 regulator thread. I remember Jabram (Joe) posted the schematic on there a while back in response to a question Randy had. Sorry I can't dig for it now, I'm on my mobile phone.
Apr 02, 2014, 09:14 AM
AMA 1033652
mnemennth's Avatar
Quote:
Originally Posted by charosenz
Paul,

Thank you for responding, I really do appreciate it. To be honest, I was hoping to learn what the function are of each of the 5 pads that the original regulator connected to. With all respect, I am just a bit reluctant to abandon functionality of pads without learning more about the decision. I really hope that is not offensive, it is just my effort to learn and to make an informed decision.

If you know of a reference to a link that identifies the components of the APM board would be a appreciated.

If I am getting too OT then I apologize and I will move on. No worries.

Thanks again.

Charlie
Sorry; I forget sometimes that other folks don't have my electronics engineering background and have trouble figuring this out from reading the documentation I've provided links to. I've edited the original Mod above to clarify the whats, wheres and whys.

I hope this helps.


mnem
No, LEFT there. RIGHT goes over HERE.
Apr 08, 2014, 09:46 AM
Director of Esoteric Projects
clough42's Avatar
Been reading this and thinking...why try to shoehorn in such a large regulator? Is there a reason you decided not to go with something smaller, like the MIC5219? The 5219 has a -20V tolerance on the input, which appears to be the sensitive parameter here. I think it will also fit on the original pads, requiring only that the enable pin be bent up and connected to Vin with a small piece of wire.

Am I missing something? Is there another reason to not use the 5219?
Apr 08, 2014, 11:10 AM
AMA 1033652
mnemennth's Avatar
Mostly, just as I said... I have these in my toolbox; so I started with them. By using this layout, we avoid multiple heating events with the original solder pads, thereby reducing the chance of destroying the APM board for the sake of experimentation. This keeps the option of replacing with the original part open if the Mod doesn't pan out.

The original part was chosen because of a very low noise threshold, high ripple rejection and very high stability under light loading. This is a real concern with the very sensitive sensors this regulator powers. So yes, it REALLY is an experiment, not just a "drop it in and run with it" replacement.

Also, it appears a big issue for a lot of folks... the micro-soldering involved in installing this size of component is daunting; though most seem to be able to get the old one out just fine.

So, we can make the Mod accessible to ordinary hobbyists with ordinary soldering skills and electronics tech knowledge, instead of just those of us with the knowledge and skill to solder these tiny devices.

Most FPVers with any experience are going to have a few LM1117-3.3s lying around for other projects, so parts availability too, though much less of a concern.

Also, I have other Mods in the works; this is just my first one.


mnem
*Fuel*
Last edited by mnemennth; Apr 08, 2014 at 11:16 AM.


Quick Reply
Message:
Thread Tools