BBB startup current

According our mesurment BBB has over 2A startup power peek just for 0,1ms.

This make it incompatible with some powersupply has advanced overquerrnet protection.

This power peek is normal?
There is any way to change startup current “policy” of board?

Thanks fo any advise.

All capacitors look like a short when they are discharged. What you are seeing is the capacitors on the board charging up. The board is not even powered up after .1ms.

Gerald

You are sure “all capacitor”?
I think TPS has no capability to flow trough more than 1A even in peek, therefore capacitor after TPC cannot generate 4A peak!?
Only capacitor charged directly in this phase is C2.

But only capacitor which can generate 4A peak is C34.
So TPS can flow trough more than 4A for a milisec?

I think there is missing som kind of FBB from P1 pin 1 beacuse the devlopment focused to USB power.

Sounds like you have it figured out.

Gerald

Before I forgot.

TPS not limit output capacitors, but recommend 22+10+10+10+10=62 uF. BBB use 100+10+10+10+10+10=140 uF. This is more than twice than recommended.

Furthermore TPS recommend not only capacitance but ESR of capacitors too (as 20mOhm). This is determine start-up current too.

BBB use C34 100uF as puffer to feed USB host connector. But it is not isolated from sys power line (need FBB?). Looks like this is where current peak originated from.

TPS has soft start feature but only for DC-DC converters and LDOs. Therefore no soft start for capacitive load connected to SYS power line.

It is advisable to add some part to SYS power line for further models (e.g. Green).

But up to this moment I think it is recommended to use power supply with <=2.5A current limit to prevent damage of TPS. TPS has 3A absolute maximum current limit.

OF all the board we have shipped, we have never lost a TPS5217C due to this issue. In fact, I think we have only lost maybe 5 TPS65217C devices total in 4+ years.

And as i said, you have to charge up those caps. I am not sure how to prevent caps from needing to be charged.

And as we also say, use a good power supply. That sometimes means not the cheapest ones.

The makers of the Green are certainly free to add whatever they like to their design.

.
Gerald

“And as we also say, use a good power supply.”
This was originator of the problem. At first time we try to use medical approved 4A PWS.But this type has a fast current protection circuit, which detect 1ms 4A current peek as short circuit.
So our advise is to use middle categhory PWS (we have success with MW GS15B-1P1J).

“we have never lost a TPS5217C due to this issue”
This is understandable beacuse average application is one BBB alone. This require 2A PWS which have no chance to harm TPS in this way.
But in case of our special application more 5VDC powered device work together. This can be 2 or more BBB and more other device. This is why we try to use 4A or higher current PWS. This can demage TPS at startup.

“I am not sure how to prevent caps from needing to be charged”
You dont need it!
First what need to be learn by power rail diesigner is the rule: capacitive load on output of power regulator is very sensitive area. This is why power regulator manufacturer (including TI) using pages to explain its calculation.
Loading reagulator with high/low/good/bad capcitor have diffrent effect to stabilized power rail.
At first impression 100uF at position C34 is very high for regulator output (ext PWS). The 22uF with ESR=20mOhm (TI rec.) limit startup peak to normal value.
Moreover I cant known philosophy of USB host interface designer, why need 100uF at IN of PDS? All AN of TPS20x1 PDS are recomend 0,1 to 10uF at this position except “Typical Hot-Plug Implemantation” where 1000uF applied. But this is not match, beacuse you cannot Hot-Plug USB host interface to BBB beacuse it is hardwired.
I think it is some missplacing of this capacitor. USB 1.1 require ~100uF at downsteram port. But this capcitor must be connceted at output of TPS20x1 not at input of. This provide a soft charge for system power rail. At input need just <=10uF.

The USB specification requires 100uF on the USB ports. TI did not write that specification. Trust me. I know.

If you put the CAP after the switch then it tends to pull down the 5V input when you activate the USB power, something about charging that cap. It works better before the switch as it prevents a dip on the 5V rail, which caused issues with the PMIC, power rail going low after power up.

I am not going to make any changes to the design. Fell free to convince the Green people to change their design.

As to your specific application, I did not design the board for your application. I did however give it to you for free. Feel free to make whatever changes you need for your application and build you own board. I think the statment on the first page of the schematic is self explanatory.

Gerald

First of all making changes on design “tomorrow” is irresponsible, so I never request it. But good to know where is some “leak” in design. For example it is help to make workaround.

“TI did not write that specification”
No, but use it in all reference design. See TI TPS20x1 PDS application information. See for example TPS2051 docu Fig 33.

“If you put the CAP after the switch then …”
Then why CAP placed OUT of PDS in all TI application information?
Because PDS has soft start feature which prevent overload IN (BBB SYS power rail). See for example TPS2051 docu Fig 4 and 8.
Fig 8 is perfect draw for this. The soft start feature limit charge of 100uF to 0,5A, therefore current never exceed USB1 and 2 current limit, therefore no dip on IN.
This is one main function of PDS.

“I did not design the board for your application”
It is not required. But during research work to specify our problem I found many topic where users discover mysterious problems with power supply, and try to found a right one for BBB. This can be originated from startup current peak.

Thank you for your feedback.

Gerald

kzsoltkzsolt,

I would like to point out to you that you’re talking to the person who designed the beaglebones, who also used to work for Texas Instruments at some point in his career. Someone who has made his designs free of charge to the public, which he has made perfectly clear to you in these post that you’re free to change and use for your own personal use.

So, telling him things, he probably already knows, in hopes of making yourself looks good. Actually make you look like a “know it all”. e.g. it doesn’t make you look good.

SO perhaps you should realize that Gerald is probably well aware of what you’re trying to discuss here, but is unwilling to change for various reasons. Reason, that you, I, or the next person do not need to understand. Because we can change to designs to our own liking if we so wish.

Pay no attention to William. You comments are welcome and Gerald has accepted your comments as valuable input by thanking your for your feedback. Now, let me address your concerns:

  1. The power supply used to power the BBB should be selected so that it does not damage the BBB, so a 2A power supply was specified. If you wish to change that specification, then the onus is on you to verify that a 4A power supply will not damage the BBB. Your conclusion that is may damage the BBB means that you should not use a 4A power supply. In addition, a power supply that is spec’d at 4A should not shutdown when it sees a 4A load, but rather, it should current limit at 4A. If the power supply is spec’d at 4A, then 4A should not be treated as a short circuit.
  2. The TI spec for the TPS65217C is a general recommendation as they are unaware of how you are going to use the part. The BBB SYS_5V powers several subsystems, including HDMI, I/O (VDD_3V3B) and USB. Clearly you could move the 100uF to the other side of the TPS2051, but then you need an additional capacitor on the SYS_5V which increases the cost and doesn’t provide any clear benefit, if you choose the correct power supply.
  3. As Gerald has pointed out, the BBB is just a reference design. It was designed as a low cost solution which meant that tradeoffs were required to keep the price low. Clearly things could have been done differently, but then the BBB price would have been much higher and the board larger. Given that most users would probably not need these extra features, they were not incorporated into the current design. There are several spinoffs of the BBB, some with wifi, some with more RAM, etc, but none have been as successful as the BBB.
  4. While I have provided Gerald input into both the BBB and BeagleBoard-x15 designs, I ultimately defer to his judgement because he has the track record or having designed several products that are very successful.

Regards,
John

Pay no attention to William. You comments are welcome and Gerald has accepted your comments as valuable input by thanking your for your feedback. Now, let me address your concerns:

vary):

1) The power supply used to power the BBB should be selected so that it does not damage the BBB, so a 2A power supply was specified. If you wish to change that specification, then the onus is on you to verify that a 4A power supply will not damage the BBB. Your conclusion that is may damage the BBB means that you should not use a 4A power supply. In addition, a power supply that is spec’d at 4A should not shutdown when it sees a 4A load, but rather, it should current limit at 4A. If the power supply is spec’d at 4A, then 4A should not be treated as a short circuit.

I would have designed the power supply circuitry so that with a power
supply of appropriate minimum rating, the maximum rating would not
have mattered. Using a power supply with a maximum current rating to
avoid damaging circuitry is not (again, IMHO) the best solution. If,
because of economic considerations, that decision is made, then it is
imperative of the designer to put this information specifically in the
power supply recommendations. Not doing this leads to damage, doing
this puts the responsibility on the user. Is this a "before the
design/after the design"? I don't know, and I don't remember (either
way) if this warning was ever in the power supply requirements.
Hindsight is 20/20, of course. If it's that important, then perhaps
the documentation needs to be changed. Decision not up to me.

2) The TI spec for the TPS65217C is a general recommendation as they are unaware of how you are going to use the part. The BBB SYS_5V powers several subsystems, including HDMI, I/O (VDD_3V3B) and USB. Clearly you could move the 100uF to the other side of the TPS2051, but then you need an additional capacitor on the SYS_5V which increases the cost and doesn’t provide any clear benefit, if you choose the correct power supply.

"correct power supply" bothers me. I'm familiar with minimum current
capacity, voltage limits, short circuit current limits (infrequently
applied). Again, "a 4 amp power supply will allow the board to damage
itself, so we depend on a 2 amp maximum supply to avoid damage." This
could be discussed a bit....

3) As Gerald has pointed out, the BBB is just a reference design. It was designed as a low cost solution which meant that tradeoffs were required to keep the price low. Clearly things could have been done differently, but then the BBB price would have been much higher and the board larger. Given that most users would probably not need these extra features, they were not incorporated into the current design. There are several spinoffs of the BBB, some with wifi, some with more RAM, etc, but none have been as successful as the BBB.

Hmmm, well, perhaps (although not required) it might be nice to know
what the engineering limitations are of the design.

I've seen 1) the ones I know about, and 2) the ones I haven't found
out yet... and 3) the ones people are going to have to tell me
about...

and I do like paranoid designs.....

Harvey

Harvey, you raised several very good points. I cannot say I disagree with anything you said.

Regards,
John

Harvey, you raised several very good points. I cannot say I disagree with anything you said.

Thank you. I prefer paranoid designs myself. However, I don't design
commercial products, I design stuff for myself. It does make a
difference. If I get it wrong, I have to fix it... and I have LOTS
of stuff to fix myself.

I also don't have to design to a price point, at least, not as much as
for a commercial product.

Harvey

When you design low cost hardware, you have to make certain decisions to get the cost down.

  1. As few components as possible.
  2. Limit the application. Only one application,
  3. Push as much cost outside, for example the power supply.
  4. Lowest cost components.
  5. Limit the features.
  6. Cut the profit.

Yes, there are several things I could have done different. Many of these no one has even identified. But if I had, you would not have bought it because it cost too much. After all hardware is supposed to be cheap. That is where the value is, in the price. Not the value…

Nobody asked how I took it from $89 to $49. They just bought them up and complained that it didn’t do all the things they wanted it to do for $49.

If anyone of you want to change the design, add more features, make it more robust, add more cost, increase the price, manufacture it and sell it, by all means, go ahead. I am sure there will b a few folks that value the hardware and recognize that value, and will pay for it.

But, I suspect the majority will complain that it is too expensive and will stay with the BBB and instead ask how to flash the latest image in the BBB and why does my my GPIO does not work…

When you design low cost hardware, you have to make certain decisions to get the cost down.

  1. As few components as possible.
  2. Limit the application. Only one application,
  3. Push as much cost outside, for example the power supply.
  4. Lowest cost components.
  5. Limit the features.
  6. Cut the profit.

Yes, there are several things I could have done different. Many of these no one has even identified. But if I had, you would not have bought it because it cost too much. After all hardware is supposed to be cheap. That is where the value is, in the price. Not the value…

Nobody asked how I took it from $89 to $49. They just bought them up and complained that it didn’t do all the things they wanted it to do for $49.

If anyone of you want to change the design, add more features, make it more robust, add more cost, increase the price, manufacture it and sell it, by all means, go ahead. I am sure there will b a few folks that value the hardware and recognize that value, and will pay for it.

But, I suspect the majority will complain that it is too expensive and will stay with the BBB and instead ask how to flash the latest image in the BBB and why does my my GPIO does not work…

Exactly, or close enough to what I was getting at. So Instead of me saying: “Pay no attention to John, as he tends to pontificate on others comments, and has no idea what the hell he is talking about.” Let me just say that I figured Gerald had the sole purpose of designing this board to work “good enough” while remaining within a specified price range. e.g. it works, and it’s cheap.

@Gerald

Actually, my buddy and I would have bought the white, if that’s all that was available. So, much to our surprise when the beaglebone black was announced . . . we immediately jumped on the pre-order list for two boards . . .

Would we have prefered you did one thing, or another slightly differently ? Sure ! But as it is, you saved us 50% right away, by doing what you did with the BBB versus the BBW. So . . . we nite our lip. Then implement what we have to on the side to make the BBB work for our own application. Too bad many youngsters would prefer to complain about what the board ISNT versus what the board IS.

So, we bit our lip *

Thanks for the words. But, I think my days are numbered here.

Gerald