Author Topic: Voltage Regulator Bypass Circuit  (Read 3131 times)

0 Members and 1 Guest are viewing this topic.

Offline LittleRainTopic starter

  • Contributor
  • Posts: 47
  • Country: ca
Voltage Regulator Bypass Circuit
« on: January 25, 2020, 05:33:06 am »
I am building a device which runs on USB or Bluetooth with a battery.
My power source goes to a 3.3v LDO regulator, and I want to bypass the regulator just before the regulator starts to dropout.

The plan was to use a voltage detector, the TPS3780, to detect when the battery drops below 3.4v, which will drive a mosfet to bypass the regulator.
The TPS3780 has 2 outputs, so I was going to use the second to put my MCU in a reset state when it drops below 3v.

Problem is my board is getting more and more expensive, and the 2-3$ IC's keep building up.

So I was thinking, what if I just used a 3.4v zener diode connected to the output the battery.
The zener would be attached to the positive input of a comparator, and a voltage divider from the battery on the other input.
When the regulator/battery drops below the zener it will stop flowing, toggling the comparator.
And then I would drive the mosfet with the output from the comparator.

Here is my schematic, only 1 would be connected of course.
I'm also going to switch out the comparator for an open-drain instead, and if you were wondering, the resistors at the TPS3779 value is "?" because I haven't calculated them yet.
Would my zener/comparator circuit work? And is there anything I should take into consideration?
914266-0
« Last Edit: January 25, 2020, 05:37:17 am by LittleRain »
 

Offline bob91343

  • Super Contributor
  • ***
  • Posts: 2675
  • Country: us
Re: Voltage Regulator Bypass Circuit
« Reply #1 on: January 25, 2020, 05:52:26 am »
A 3.4 V zener is a nonstandard value.  Even if you select from a lot of 3.3 V units, they are terrible regulators, with impedance too high for regulation, not to speak of a somewhat negative temperature coefficient.  You will be disappointed.
 

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 9336
  • Country: fi
Re: Voltage Regulator Bypass Circuit
« Reply #2 on: January 25, 2020, 01:26:55 pm »
Just pick an LDO with low-enough drop-out, i.e., one which internally does exactly what you are doing. FET based LDOs have everything needed internally, you are duplicating the functionality. Sort for drop-out voltage on parametric search.

Using a lower RDS(on) MOSFET to bypass it only gains you some 0.1V, maybe 0.2V extra compared to a properly selected regulator with very low drop-out. That extra 0.1 or 0.2V would gain you relatively little battery life; your solution have extra parts with extra quiescent current that may eat the benefit.
 
The following users thanked this post: SeanB, wraper

Offline MasterT

  • Frequent Contributor
  • **
  • !
  • Posts: 851
  • Country: ca
Re: Voltage Regulator Bypass Circuit
« Reply #3 on: January 25, 2020, 03:14:03 pm »
I see AP7215 states 250mV max drop-out at 100mA. There is an  APPLICATION NOTE 1853 from Maxim-integrated that :
Quote
a linear-regulator circuit whose dropout at 100mA load current is only 10mV.
 
The following users thanked this post: exe

Offline LittleRainTopic starter

  • Contributor
  • Posts: 47
  • Country: ca
Re: Voltage Regulator Bypass Circuit
« Reply #4 on: January 25, 2020, 09:47:47 pm »
I see AP7215 states 250mV max drop-out at 100mA. There is an  APPLICATION NOTE 1853 from Maxim-integrated that :
Quote
a linear-regulator circuit whose dropout at 100mA load current is only 10mV.
Just pick an LDO with low-enough drop-out, i.e., one which internally does exactly what you are doing.
Problem is I want the battery to be able to go down to 3v, and I'll also be using a CPLD and DC motors, so I'll be using close to the full 500mA from the USB port.
I really want to get as much battery life as possible.

Should I just switch to a 2.5v regulator, and make sure all the 3.3v components can run at 2.5v, and switch them if they cant?
Id have to check and see if the MCU can run at the clock speed I need from 2.5v first though.
The CPLD is the Max II EPM240T100C5, which can run off of 3.3 V/2.5 V or 1.8 V, pretty sure it doesn't matter as well, I'll be running it at 50mhz instead of its full 100/150mhz.

Or should I use the TPS3780 circuit for bypassing?

Using a lower RDS(on) MOSFET to bypass it only gains you some 0.1V, maybe 0.2V extra compared to a properly selected regulator with very low drop-out. That extra 0.1 or 0.2V would gain you relatively little battery life; your solution have extra parts with extra quiescent current that may eat the benefit.

Here is an image of simulated battery life using the TPS3780 to bypass the regulator, you do get a bit more life using it.(I didn't take the pic, its from here)

Here is the image without.


A 3.4 V zener is a nonstandard value.  Even if you select from a lot of 3.3 V units, they are terrible regulators, with impedance too high for regulation, not to speak of a somewhat negative temperature coefficient.  You will be disappointed.
I was wondering about that, I know they have 3.3v but haven't checked about 3.4v.
While writing this I just checked, there is this zener, 3.425v for $0.27 CAD, which I could probably find cheaper.
The point of the zener(D3) is not to regulate though, just run a comparator(U17), which will run a mosfet(Q13).
 

Offline MasterT

  • Frequent Contributor
  • **
  • !
  • Posts: 851
  • Country: ca
Re: Voltage Regulator Bypass Circuit
« Reply #5 on: January 25, 2020, 10:01:25 pm »
Forget about zener, it's incompatible with word "efficiency". What you need is step-up/step-down 3.3V regulator, they make it in a single chip specifically for this case. TPS63060/TPS63061
 

Offline LittleRainTopic starter

  • Contributor
  • Posts: 47
  • Country: ca
Re: Voltage Regulator Bypass Circuit
« Reply #6 on: January 26, 2020, 06:24:57 am »
Forget about zener, it's incompatible with word "efficiency". What you need is step-up/step-down 3.3V regulator, they make it in a single chip specifically for this case. TPS63060/TPS63061


Would driving a mosfet/comparator with a zener really be that inefficient?
I am going to throw that idea out, but just wondering.

Big thing about a boost converter is cost, and ESD, that IC is $3.32, vs $0.753 for the LDO. The market I'm trying to enter is pretty competitive(gamepad), I really need to keep cost down as much as I can.
Not only that, but a friend of mine I'm working with, he is using a step-down for his device, and he has failed ESR testing 2 times now, which I really don't have the money for.
I can't remember if he was using a step-down before, but I know he switched regulators for a new design of his device, and all of a sudden its causing all these noise problems. He is using the LM2735.

My initial plan actually was a step-down regulator, but the thought of my friends ESD problems kind of scared me out of it.

And what do you think about scrapping the bypassing altogether and just going to down a 2.5v LDO instead? I mentioned the idea in my previous reply.


So I had written this reply earlier but had to go before I could post it.
When I got home my friend ended up calling me, and he suggested this.

He said I could use a charge pump, which would boost the voltage to 5v, then the input voltage could be anywhere from 3-5v.
Then he suggested to use a battery charger with a built in LDO, that would output to 3.3v.
I really feel like he mentioned another IC, but we had a long conversation directly after and I can't remember. I had it typed out, but I accidentally closed it.  :palm:
He said this would be a cheap option, and minimize the amount of IC's I need, and he didn't mention anything about efficiency.
Another thing he mentioned was if the charge pump doesn't work the way I need it, to use a boost converter, but then mentioned his ESD issues again.
He was under the impression I only planned on small production runs, and said I could probably skip FCC certification, but my end goal is large production.

Do you have any thoughts on his advice?



Also, thanks to everyone who replied so far.
 

Offline MasterT

  • Frequent Contributor
  • **
  • !
  • Posts: 851
  • Country: ca
Re: Voltage Regulator Bypass Circuit
« Reply #7 on: January 26, 2020, 02:51:16 pm »
I don't think charge pump is a good solution, for 500mA it 'd be both - costly & inefficient.
I already advised two option:
1. ultra-low saturation BJT with LDO, it may squize battery down to 1-2V;
2. step-up/step-down in single chip. I was referring to TI as an example, probably there is low cost IC somewhere - google. If one chip is costly, than option with two IC is still on the  table.
 Don't know anything about EMI testing, never did.
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20360
  • Country: gb
  • 0999
Re: Voltage Regulator Bypass Circuit
« Reply #8 on: January 26, 2020, 02:54:34 pm »
Zener diodes require a considerable current to regulate properly and they're not even very well regulated.

Scrap the voltage regulator bypass idea. Only use the AP7215, to power the parts of the circuit which don't need much power and require a regulated power supply i.e. CPLD, ADC and MCU etc. The higher current stuff such as motors and any LEDs can be powered from an unregulated power supply, thus saving voltage drop and power dissipation. If you want the motor speed to be independent of supply voltage, then PWM the motor, monitor the unregulated rail, using a potential divider and adjust the duty cycle to compensate.
 

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 9336
  • Country: fi
Re: Voltage Regulator Bypass Circuit
« Reply #9 on: January 26, 2020, 03:49:01 pm »
Don't regulate the DC motor supply - problem solved.

With lower current, very low drop-outs below 100mV are easily available.
 

Offline LittleRainTopic starter

  • Contributor
  • Posts: 47
  • Country: ca
Re: Voltage Regulator Bypass Circuit
« Reply #10 on: January 26, 2020, 11:20:23 pm »
Thanks guys, I appreciate your replies.

I don't think charge pump is a good solution, for 500mA it 'd be both - costly & inefficient.
I already advised two option:
1. ultra-low saturation BJT with LDO, it may squize battery down to 1-2V;
2. step-up/step-down in single chip. I was referring to TI as an example, probably there is low cost IC somewhere - google. If one chip is costly, than option with two IC is still on the  table.
 Don't know anything about EMI testing, never did.

I am using a lipo, it can't go below 3v. It has a range from 3.7 down to 3v. And I the only way to get the full 3v with an LDO is to bypass it once voltage reaches around 3.4 volts.
And I don't understand what you mean by using BJT with LDO.

If I can find a solution where I don't have to worry about EMI, I will go with step-up/step-down. I looked for metal shields, but didn't see any on digikey or mouser, but maybe I was just searching for the wrong thing.

Thanks

Edit: I found them, RF shield is what you need to search. Problem with those though, is they take up a lot of space, which I don't have.
Smallest one is also $1.50 on digikey, atleast for 1 unit.

Zener diodes require a considerable current to regulate properly and they're not even very well regulated.

Right, I hear you. I guess 10mA is more than I thought.
I've never worked with batteries where I had to be power conscious before.

Scrap the voltage regulator bypass idea. Only use the AP7215, to power the parts of the circuit which don't need much power and require a regulated power supply i.e. CPLD, ADC and MCU etc. The higher current stuff such as motors and any LEDs can be powered from an unregulated power supply, thus saving voltage drop and power dissipation.
I can't do that, my USB port will also some times be plugged into a 3.3v source, and if it is, I need to be able to bypass the regulator, or as MasterT said, use a Step-up/step-down regulator, which I'm really trying to avoid.
Yes I know all USB ports are 5v, but I am designing it so it can be plugged into something else, which is a must.

I could do the bypassing with the TPS3780, but its just going to add another $2 to my board.
Actually, I just realized I can use it for under voltage protection as well, at least I think I can.
It will allow me to drive a p-channel mosfet to shut off the circuit if it goes under 3v, and will pay for itself because I can take off the IC that handles that.

If you want the motor speed to be independent of supply voltage, then PWM the motor, monitor the unregulated rail, using a potential divider and adjust the duty cycle to compensate.

Good idea, but I actually do have Motors and LEDs setup with an N-channel for PWM.  ;)
I just took them off of the regulator, that I wasn't doing, I was trying to avoid the variable output, but I think most products on the market have that problem anyways.
So with PWM to avoid variable output, I would have the change the duty cycle depending on voltage, correct?

Thanks.

Don't regulate the DC motor supply - problem solved.

With lower current, very low drop-outs below 100mV are easily available.

Yes I will stop regulating it.

This is in my reply to Zero, its why I still need to bypass the regulator, I can't guarantee when the USB plug is in a 3.3v source, that it will be higher than 100mV.
I can't do that, my USB port will occasionally be plugged into a 3.3v source, and if it is, I need to be able to bypass the regulator, or as MasterT said, use a Step-up/step-down regulator, which I'm really trying to avoid.
Yes I know all USB ports are 5v, but I am designing it so it can be plugged into something else, which is a must.


The only way I can not bypass the regulator, is if I make my MCU/CPLD run at 2.5v, which may be an option, I need to look into it further to make sure all the features I need can run at 2.5v, or if I use a step-up/step-down.
I know some MCU's cannot do USB at 2.7v, but I read that you may be able to use a level-shifter, and I already have level-shifter with more channels than I need. That still doesn't guarantee anything, but its a start.
Right now I'm really just trying to find the best cost/efficiency ratio, for bypassing the LDO.

Thanks.
« Last Edit: January 26, 2020, 11:35:24 pm by LittleRain »
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20360
  • Country: gb
  • 0999
Re: Voltage Regulator Bypass Circuit
« Reply #11 on: January 27, 2020, 07:14:03 pm »
Scrap the voltage regulator bypass idea. Only use the AP7215, to power the parts of the circuit which don't need much power and require a regulated power supply i.e. CPLD, ADC and MCU etc. The higher current stuff such as motors and any LEDs can be powered from an unregulated power supply, thus saving voltage drop and power dissipation.
I can't do that, my USB port will also some times be plugged into a 3.3v source, and if it is, I need to be able to bypass the regulator, or as MasterT said, use a Step-up/step-down regulator, which I'm really trying to avoid.
Yes I know all USB ports are 5v, but I am designing it so it can be plugged into something else, which is a must.

I could do the bypassing with the TPS3780, but its just going to add another $2 to my board.
Actually, I just realized I can use it for under voltage protection as well, at least I think I can.
It will allow me to drive a p-channel mosfet to shut off the circuit if it goes under 3v, and will pay for itself because I can take off the IC that handles that.
Quote
Don't regulate the DC motor supply - problem solved.

With lower current, very low drop-outs below 100mV are easily available.

Yes I will stop regulating it.

This is in my reply to Zero, its why I still need to bypass the regulator, I can't guarantee when the USB plug is in a 3.3v source, that it will be higher than 100mV.
I can't do that, my USB port will occasionally be plugged into a 3.3v source, and if it is, I need to be able to bypass the regulator, or as MasterT said, use a Step-up/step-down regulator, which I'm really trying to avoid.
Yes I know all USB ports are 5v, but I am designing it so it can be plugged into something else, which is a must.


The only way I can not bypass the regulator, is if I make my MCU/CPLD run at 2.5v, which may be an option, I need to look into it further to make sure all the features I need can run at 2.5v, or if I use a step-up/step-down.
I know some MCU's cannot do USB at 2.7v, but I read that you may be able to use a level-shifter, and I already have level-shifter with more channels than I need. That still doesn't guarantee anything, but its a start.
Right now I'm really just trying to find the best cost/efficiency ratio, for bypassing the LDO.

Thanks.

How much current does the logic and control circuitry require?

I still don't see the need to bypass the AP7215. It only has a maximum dropout voltage of 250mV, even with a 100mA load, so if your USB voltage is 3.3V, the output voltage will drop to just over 3V, and most ICs designed for 3.3V operation can handle that. A MOSFET will have a similar voltage drop to a low dropout linear regulator, so adding one doesn't really gain you much. The extra cost would be better spent on a lower dropout regulator or a regulator designed for higher currents, which should give a lower dropout, at much lower currents, if it has a MOSFET output device.

The TPS7A05 has a lower dropout voltage, than the AP7215, even at 200mA. At much lower currents, it's even less.
http://www.ti.com/lit/ds/symlink/tps7a05.pdf

Then there are parts which use an N-channel pass device and either require an external higher bias voltage, such as the LP38841, or have a built-in charge pump, like the LTC3035.
http://www.ti.com/lit/ds/symlink/lp38841.pdf
https://www.analog.com/media/en/technical-documentation/data-sheets/3035f.pdf


Quote
If you want the motor speed to be independent of supply voltage, then PWM the motor, monitor the unregulated rail, using a potential divider and adjust the duty cycle to compensate.

Good idea, but I actually do have Motors and LEDs setup with an N-channel for PWM.  ;)
I just took them off of the regulator, that I wasn't doing, I was trying to avoid the variable output, but I think most products on the market have that problem anyways.
So with PWM to avoid variable output, I would have the change the duty cycle depending on voltage, correct?

Thanks.
Yes, the reduction in motor speed can be compensated for by increasing the duty cycle. Motor speed scales fairly linearly with voltage and duty cycle, so 5V PWM'd at 50% duty will give roughly the same speed as 2.5V steady DC, which makes it fairly simple to calculate. Using a low loss Schottky for the freewheeling diode also helps. LEDs can be driven off constant current sources which will keep their brightness constant.
« Last Edit: January 28, 2020, 08:47:39 am by Zero999 »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf