Load disconnect

[Andy_123]

May be a simple question, but I want to confirm:

I am using TPS61040 to get 16V from two AA batteries.
TPS does not provide complete load disconnect when disabled, so I am using circuit shown in the datasheet and it works (see below).
Max application current does not exceed 30 mA.
Question: can I replace PNP transistor with a P-channel MOSFET (IRLML6401TRPBF)?
If so what R3 value should I use?
I tested it with 100Ohm and it works, but I am not sure if this is correct value and a higher value (like 10K) should be used.

 

[ronsimpson]

IRLML6401TRPBF what a interesting part.
The maximum D to S voltage is -12V. You are switching 18V. Normally not good but:
>I think this might be OK because:
>>The MOSFET switches when Vin is about 2 volts less then the voltage on C2. (Vin=3V, V-C2=5V, Vout=5V/0V
The maximum G to S voltage is 8V. If Vin instantly went to 0 volts you will have 18V.
>When operating you will have 15V G to S.
>There are many MOSFETs that have a 20V G-S rating.

The value of R3; I don't think it matters much. I probably would have used 200k just because it was there. There is no reason to switch fast ( 10 ohms). I would have used 10k just to pick a value. (or use 160k because you are using one and that will reduce the different parts you need to get)

 

[Andy_123]

Here is a full circuit that I have today.
I have feedback resistor divider before MOSFET.

My main goal is to reduce number of different components that this device uses.
I don't have 160K, I am using 1M/80K pair instead
I have few 3K and 10K resistors already used, so I will try one of them.

IRLML6401TRPBF already used in reverse polarity protection and that is why I want to use it here.
Vin can go as low as 2.4V with NiMH batteries

 

[ronsimpson]

The first circuit, (not the second) the feedback of from the output of the MOSFET. This way any voltage loss (D-S) will be removed. If the RDS is low then the drop is small and it does not matter.

Measure the G-S voltage on you MOSFET and it will be too much.
>Us the 80k resistor and a 4 to 8 volt Zener to save the gate of the MOSFET.
OR
>R21=80k and add a G-S resistor of 80k. This will 2:1 reduce the G-D voltage.

 

[MrAl]

Hi,

Not that it is bad, but is there any specific reason why you want to use a mosfet instead of bipolar?

You should be aware that a 0.2v drop with 18v output is about 1 percent. That means if you use batteries you'll use 101 batteries instead of just 100 over the course of time.

But there are other bipolars too that drop even less voltage than that. Zetex for example makes bipolar transistors with extremely low voltage drop which would increase the efficiency even more and you would not have to modify the already existing circuit.

You can do it either way though, it's up to you.

 

[Andy_123]

I am not sure where 0.2V drop came from, this MOSFET has very low resistance (0.05ohm) and with typical load less that 20mA drop will be less than 1mV.

Output voltage value is not critical and does not have to be exact 18V, it can be as low as 14V for load to operate correctly and it works.
Device has occasional use, few minutes per week max. So battery life is not critical at all.
From what I know with previous designs with a mechanical power switch that batteries die because of age first.

My goal is to have a reliable device and reduce number of tapes used because of SMD machine limitation.
Adding another PNP transistor is not an option, IRLML6401TRPBF is already used in polarity protection, but I may use IRLML6402TRPBF instead, with higher GS voltage.
Also I like the idea of adding another 80K to G-S,
Will it be better is I use R21 3K and Rgs 10K to create 4:1 divider?

 

[ronsimpson]

The theory is that the Gate used 0 current, so 0 energy.
The Base uses current. Depends on the gain of the transistor. If the transistor is set to gain of 100 then you are back to 1% loss. (100:101)
If you want low E-C voltage drop you will need to drive the base hard.

If the output was 5V then the loss do to base current is not much. But with the output of 18V the base current X voltage = much power.

 

[ronsimpson]

My goal is to have a reliable device and reduce number of tapes used because of SMD machine limitation.
Adding another PNP transistor is not an option, IRLML6401TRPBF is already used in polarity protection, but I may use IRLML6402TRPBF instead, with higher GS voltage.
Also I like the idea of adding another 80K to G-S,
Will it be better is I use R21 3K and Rgs 10K to create 4:1 divider?

I chose 80k because you have them, and the current in (80+80=160k) is not much. I think 3k+3k is too much power loss.

With the 80+80 the gate turn on voltage will effectively double. If the MOSFET needs 2V to turn on then it will now take 4 volts. Vin=3V + 4V so the output will be near 7v to turn on. (note the gate turn on voltage is not the same in every part) I think 4:1 is too much. Maybe 3:1. Something for you to think about.

 

[Andy_123]

First, I need to correct myself, output voltage is 16.75V, not 18V.

It does not make a big difference without a divider.
It will exceed 8V Vgs while running and this is not good. (it will be about 14V)

With 2:1 divider with max battery voltage of 3.2V, this will create G-S voltage of about -6.8V to turn MOSFET on.
With battery voltage going to Zero it will be about 8.3V for a very short time.

I also have 560K available - will it be better than 80K?

 

[Andy_123]

I tested divider 80K/80K with IRLML6401TRPBF and it looks like it works just fine.
IRLML6402TRPBF may be even better with 12V Vgs, but I don't have it.

But I am not sure if I should keep this or go back to PNP transistor?
Will 80K work with PNP instead of 200K shown?

Also will something like 2N2906 work instead of BC857C?

 

[ronsimpson]

I don't have time to look. but:
A mosfet turns on with voltage. You have plenty of voltage. (if not too much)
The transistor is a long complicated story. You need to know if the part will be on HARD at all temperatures and all parts. If a typical transistor work, does not mean a worst case part will work.
The 2N2906A has lower gain then the BC857C.
>From data sheet of BC857C:
>>Gain is very much a function of temperature. Gain at -55C compared to 25C goes up 2:1, and 25C to 150C goes up 2:1 gain.
>>The gain at Vce=5V, Ic=2mA, Gain is some where from 125 to 475. You want to be running at 20mA and 0.5 volts or less, so the gain will be much lower.

 

[Andy_123]

I will stay with MOSFET and divider, I think it works just fine with 80.6K+80.6K divider
I tested 3 devices, all work just fine.
Not sure how I missed 8V Vgs limitation in my original design.
I have few devices that use just 100 Ohm and they are working for a few months already, so looks like this MOSFET can handle way more than 8V.
Now I have to replace resistors there just in case.

Thank you for your help!

 

[ronsimpson]

I have few devices that use just 100 Ohm and they are working for a few months already, so looks like this MOSFET can handle way more than 8V.
Now I have to replace resistors there just in case.

The manufacture states that the part will survive a G-S voltage of + or - 8 volts. Beyond that.......
You could have a batch of parts that will survive 2x but the next batch might blow at 1.1X.
In production you can count on someone switching to a different part.

 

[Andy_123]

All devices made so far used MOSFET from the same tape and I was just stating the fact that so far they survived 2x.
But I will not take a chance and replace 100 ohm resistor with 80.6K + adding G-S resistor.

I agree that I must stay within specifications to have a reliable device.
Production will have resistor divider to limit the voltage below 8V and most likely IRLML6402TRPBF that has 12V G-S voltage

 

[MrAl]

I am not sure where 0.2V drop came from, this MOSFET has very low resistance (0.05ohm) and with typical load less that 20mA drop will be less than 1mV.

Output voltage value is not critical and does not have to be exact 18V, it can be as low as 14V for load to operate correctly and it works.
Device has occasional use, few minutes per week max. So battery life is not critical at all.
From what I know with previous designs with a mechanical power switch that batteries die because of age first.

My goal is to have a reliable device and reduce number of tapes used because of SMD machine limitation.
Adding another PNP transistor is not an option, IRLML6401TRPBF is already used in polarity protection, but I may use IRLML6402TRPBF instead, with higher GS voltage.
Also I like the idea of adding another 80K to G-S,
Will it be better is I use R21 3K and Rgs 10K to create 4:1 divider?

Hi,

The 0.2v is the typical voltage drop from collector to emitter of a common transistor. That's used to estimate the efficiency, although it could be even lower than that. You also want as little base current as possible.
The MOSFET will be good too.

 

[Andy_123]

This is what I got so far. I wonder if it would be better to use a higher R21/R32 value?
I have 560K and 1M available as well.