Author Topic: EEVblog #995 - Power Supply Sequencing  (Read 6265 times)

0 Members and 1 Guest are viewing this topic.

Offline EEVblog

  • Administrator
  • *****
  • Posts: 27951
  • Country: au
    • EEVblog
EEVblog #995 - Power Supply Sequencing
« on: May 21, 2017, 09:43:06 am »
Dave shows how to do power supply MOSFET rail switching and demonstrates and discusses why supply sequencing might be needed.

 
The following users thanked this post: SeanB

Offline Numex106

  • Newbie
  • Posts: 1
  • Country: ca
Re: EEVblog #995 - Power Supply Sequencing
« Reply #1 on: May 21, 2017, 02:08:33 pm »
At 17:00 Dave meant to say "ground it to turn it off" instead of "ground it to turn it on", correct? because N-type, you have to bring gate high?
 

Offline mtdoc

  • Super Contributor
  • ***
  • Posts: 3574
  • Country: us
Re: EEVblog #995 - Power Supply Sequencing
« Reply #2 on: May 21, 2017, 03:48:30 pm »
Nice video Dave.  Would love to see more like this.  Short, on point "how and why" videos on practical topics.

For this particular issue, I'd never really considered that the timing of different power rails could be so critical.  For those of us TEA members who have several old school PD or HP PSUs without soft output buttons, a simple power strip might do for simultaneous power up I suppose.  Nevetheless, very good to work through the mosfet solution.  :-+
 

Offline Barny

  • Frequent Contributor
  • **
  • Posts: 271
  • Country: at
  • I'm from Austria, not Australia ;)
Re: EEVblog #995 - Power Supply Sequencing
« Reply #3 on: May 21, 2017, 04:56:17 pm »
Thanks for the interesting video.
I never thought of that issue.


Wouldn't it be better to connect the gates to the -5V rail instead with a switch to ground?
With this the -5V rail witches on all other rails without bounceing.
 

Offline G0MJW

  • Contributor
  • Posts: 40
  • Country: gb
  • Mike
    • G0MJW
Re: EEVblog #995 - Power Supply Sequencing
« Reply #4 on: May 21, 2017, 06:05:02 pm »
I asked the same question in the comments. Its not quite so simple as you need to make sure the PSU in its off state doesn't ground the rail and also it doesn't object to having +15V on it via the gate resistor. This is easily done with a small switching transistor.

I am wondering if Dave has found the faulty component on the PSU? My money is on one of the diodes or maybe one of the switching transistors. You only see disconnected mains input connector faults on 1st April.

Mike
Mike
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 5279
  • Country: de
Re: EEVblog #995 - Power Supply Sequencing
« Reply #5 on: May 21, 2017, 06:09:55 pm »
Directly connecting the gates to the -5 V would not work: when turned off the -5 V is still at about fround level and this is enough to turn on at least the 15 V and +5 V and with a low level FET also the 3.3 V.

However the idea is not that bad. With an NPN transistor in base configuration (base at GND and let say 1 K to the - 5 V), this should work. The transistor would be off when the -5 V is off and output (collector = gates of the P MOSFETS) would be at about -0.6 V if the -5 V are on.
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 9356
  • Country: us
Re: EEVblog #995 - Power Supply Sequencing
« Reply #6 on: May 21, 2017, 06:11:56 pm »
I was surprised no mention of using the remote programming interface that better power supplies have and sending commands to turn on the outputs in the right sequence?
I'm not an EE--what am I doing here?
 

Offline G0MJW

  • Contributor
  • Posts: 40
  • Country: gb
  • Mike
    • G0MJW
Re: EEVblog #995 - Power Supply Sequencing
« Reply #7 on: May 21, 2017, 07:01:55 pm »
If it would work or not depends on the PSU and what happens when it is off. Some PSUs will be open circuit, some won't, some may explode. However another transistor would certainly solve that.

Mike
Mike
 

Offline blundar

  • Contributor
  • Posts: 5
Re: EEVblog #995 - Power Supply Sequencing
« Reply #8 on: May 22, 2017, 02:17:14 am »
One comment...  About 15:45 in the video where Dave's going over the multiple P-FET switch.

It's generally good form for EACH FET to always have its own gate resistor.  It can help prevent bounce and other such bad things happening.  Probably able to be ignored with small FETs but larger ones... Not so much.

In a similar light, adding some extra capacitance to the gate control line probably wouldn't be a bad idea, especially if any kind of mechanical actuator is being used, again to avoid bounce.  If you're trying to synchronize power supplies, bounce will defeat the purpose of the circuit.
 

Offline Poe

  • Regular Contributor
  • *
  • Posts: 201
Re: EEVblog #995 - Power Supply Sequencing
« Reply #9 on: May 22, 2017, 11:33:06 am »
If you're just turning them all on at the same time, just use an N-type on the low side?

 

Offline German_EE

  • Super Contributor
  • ***
  • Posts: 2069
  • Country: de
Re: EEVblog #995 - Power Supply Sequencing
« Reply #10 on: May 22, 2017, 07:02:34 pm »
Dave, I owe you a beer!   :-+ :-+

For months I've been chasing a weird issue in a homebrew amateur radio project where one of the chips keeps on failing (and by failing I mean going bang). Thanks to video #995 I now realize that the 13v8 input signal on one pin is being applied before the 15v supply rail is being turned on and the internal chip protection diodes are probably dying.

I'm now off to do some revised power supply sequencing.

EDIT: I now realize that this thread is also a power supply sequencing problem http://www.eevblog.com/forum/chat/tv-av-receiver-hdmi-turn-on-synchronization/
« Last Edit: May 22, 2017, 07:09:43 pm by German_EE »
Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks.

Warren Buffett
 

Online johnh

  • Regular Contributor
  • *
  • Posts: 125
  • Country: au
Re: EEVblog #995 - Power Supply Sequencing
« Reply #11 on: May 22, 2017, 07:35:32 pm »
I saw the effect of incorrect power sequencing a long time ago in the mid 80's

The magazine (shelf unit) for AXE Line cards had three different power supplies.  If they weren't not switched on in the correct order, there was chance the ring current relay would be activated and power applied to circuit continuously.  Saw quite a few card that were charred, from a nice light brown color to black.
 

Offline ggchab

  • Regular Contributor
  • *
  • Posts: 157
  • Country: be
Re: EEVblog #995 - Power Supply Sequencing
« Reply #12 on: May 22, 2017, 09:38:23 pm »
Very interesting. Thank you.
Why can't the negative supply be started with a similar circuit ? Or not necessary for this specific case ?
« Last Edit: May 22, 2017, 09:47:05 pm by ggchab »
 

Offline German_EE

  • Super Contributor
  • ***
  • Posts: 2069
  • Country: de
Re: EEVblog #995 - Power Supply Sequencing
« Reply #13 on: May 23, 2017, 01:02:03 am »
Switching a negative supply using a positive logic level enabling signal can be a difficult exercise. The challenge is to get the circuit to 'fail safe' and not get your negative supply blowing your logic circuits should something go short circuit.

I normally cheat and use either an opto-isolator or a solid state relay (which is just a heavier version of the same thing).
Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks.

Warren Buffett
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 5279
  • Country: de
Re: EEVblog #995 - Power Supply Sequencing
« Reply #14 on: May 23, 2017, 04:11:06 am »
If you're just turning them all on at the same time, just use an N-type on the low side?

For dave`s case, one could use the +15 V supply as the master and use it to turn on all the other voltages with N-channel mosfets: the 5 V and 3 V are easy, just connect the gate to the 15 V supply. The -15 V supply would need a PNP in base configuration, to limit the gate voltage, and make it turn on at something like a 3 or 5  V level. Carefully scaled / adjusted just resistors (and maybe a diode) might also work.
 

Online GeorgeOfTheJungle

  • Super Contributor
  • ***
  • Posts: 1531
  • Country: pl
Re: EEVblog #995 - Power Supply Sequencing
« Reply #15 on: May 23, 2017, 04:33:00 am »
If you're just turning them all on at the same time, just use an N-type on the low side?

Not a good idea! (all rails on + ground disconnected => high probability of magic smoke)
« Last Edit: May 23, 2017, 04:35:26 am by GeorgeOfTheJungle »
int main (void) { while (1) fork(); }
 

Offline SilverSolder

  • Frequent Contributor
  • **
  • Posts: 265
  • Country: 00
Re: EEVblog #995 - Power Supply Sequencing
« Reply #16 on: May 23, 2017, 09:11:17 am »
Too simple to just use a 3 or 4 pole switch?
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 8442
  • Country: au
Re: EEVblog #995 - Power Supply Sequencing
« Reply #17 on: May 23, 2017, 11:04:41 am »
Where's the fun in that?

Besides - that might be an issue if there is need to pass some serious current from one or more supplies.
 

Offline bagnoz

  • Contributor
  • Posts: 7
  • Country: de
Re: EEVblog #995 - Power Supply Sequencing
« Reply #18 on: May 23, 2017, 08:18:25 pm »
In my intuition one would want the "ordering" of the voltages to be correct during the complete time of the startup transition.
So when switching, I would recommend to first power the 15.7 V then 5.2 V then 3.3 V so that there is never a reverse voltage between the different supplies.

I would assume that gives the best chance of doing right by the circuit.
Does that make sense?
How would a single negative supply fit into that model? My model says it would not matter at all, but that sounds wrong  ^-^

A simple power supply would probably do something similar in that it would derive the lower voltages from the higher voltages.
If I remember correctly the ATX standard requires such a behavior too.

I think one always has to sequence them, because there is no such thing as "at the same time" ;)
 

Offline bktemp

  • Super Contributor
  • ***
  • Posts: 1623
  • Country: de
Re: EEVblog #995 - Power Supply Sequencing
« Reply #19 on: May 23, 2017, 08:27:32 pm »
In my intuition one would want the "ordering" of the voltages to be correct during the complete time of the startup transition.
So when switching, I would recommend to first power the 15.7 V then 5.2 V then 3.3 V so that there is never a reverse voltage between the different supplies.
If there are diodes from each lower to the next higher voltage, then it makes sense.
But it totally depends on the circuit. If the analogue section (+/-15V) is driving ADCs in the digital section (probably +5V), then it may not like it if the input signal is present before its power supply voltage.

I also remember some Epson display controller ics, that need the lower core voltage first before the IO voltage, otherwise they will get quite hot:
"When applying supply voltages to the S1D13513, Core VDD must be applied to the chip before, or simultaneously with H VDD, or damage to the chip may result."

If you don't know any specific requirements, switching everything on at the same time is probably the best solution.
 
The following users thanked this post: bagnoz

Online Fungus

  • Super Contributor
  • ***
  • Posts: 9278
  • Country: 00
Re: EEVblog #995 - Power Supply Sequencing
« Reply #20 on: May 23, 2017, 08:51:51 pm »
I think one always has to sequence them, because there is no such thing as "at the same time" ;)

Even if the power supply powers on all buses perfectly there can still be rise-time delays on the PCB because of decoupling capacitors, etc.

eg. You might have a ton of decoupling on the 5V rails.
 

Offline tom66

  • Super Contributor
  • ***
  • Posts: 3374
  • Country: gb
  • Electronic Engineer & Hobbyist
Re: EEVblog #995 - Power Supply Sequencing
« Reply #21 on: May 24, 2017, 08:44:42 am »
Old plasma TVs used to have an issue with the data buffers that were bonded to the bottom of the screen (like in Dave's Panasonic teardown video).  If the high voltage, 70V or so data voltage, is applied after the 5V logic voltage the chips can be destroyed resulting in vertical lines.  Fine if the power supply works fine but if an engineer is fiddling with it and plugs it in after the set is off, the 70V on the caps (not discharging, due to lack of load) without the low voltage rail is instantaneously enough to fry the ICs.
 

Offline mikerj

  • Super Contributor
  • ***
  • Posts: 1752
  • Country: gb
Re: EEVblog #995 - Power Supply Sequencing
« Reply #22 on: May 24, 2017, 09:57:28 pm »
If the high voltage, 70V or so data voltage, is applied after the 5V logic voltage the chips can be destroyed resulting in vertical lines.

Do you mean 'before', i.e. if the 70v is applied before the 5v logic rail is powered, the buffers will die?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf