Working with the Max1674/1675 ICs (MM, MM-lite IC)

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Per the datasheet, the Max1675 and 1674 stepup ICs should be able to deliver a minimum of 80 and 160 mA at 3.3V respectively, at an input as low as 1.0V. Has anyone who has laid out a board using these ICs ever gotten this low Vin performance? As implemented in the MadMax and MadMax lite the output plummets to a much lower value for inputs below around 1.15V. The effect of this can be seen on all of the single 1 alkaline cell runtimes that various folks have done. This can be seen as the sharp change in slope of the runtime curves in the following examples: MM lite and MM Plus. Looking at the layout of the MM boards, it appears to me that Wayne has followed generally accepted guidelines for proper layout of a switching regulator. I remember a long time ago that Wayne complained about how quirky this IC was to work with [I don't remember if this was in a post or a private communication]. Using a MM lite board I varied the values of Cin, Cout, the inductor, and also deleted the forward diode [which is optional, BTW] and was unable to obtain any improvement in the low Vin performance. Has anyone managed to match the datasheet low Vin performance??
 

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,144
Location
New Jersey
Hi Doug S,

I have some bad news :-(

It looks theoretically impossible to guarantee an output of 3.3v at 160ma
using the Max1674/5/6 chips. The main reason for this is the spec on the
MOSFET switches, which is stated as a max of 0.6 ohms. Add to that some
series resistance in the inductor, some series resistance in the partially
depleted battery cell, and you've got a circuit that can put out at most
1.9v at 160ma with a 1v (before Rs) input source.

With say a 10uf very low esr input cap it gets a little better if we measure
1v at the input of the chip. Let me throw some numbers at ya...


The following theoretical values are obtained when measuring 1.000vdc at
the input to the chip.

With the MOSFET's at the 'typical' value we arent doing so bad:
using a 0.01 ohms inductor, at 210ma we get 3.3v output, nice.

As soon as we plug in the 'max' spec of the mosfets we run into problems:
at 160ma output even with a 0.01 ohms inductor we only get 2.6v output.
Had we used a 0.1 ohms inductor instead, we would get 2.2v output.

What this means, of course, is that if we dont get 3.3v output using a
low 0.01 ohms inductor when measuring 1.000v dc at the chip's input
then the MOSFET's 'on' resistance is above 'typical', but still within
spec. This means this chip cant be used for 160ma output at 3.3v or
some of them wont put out enough voltage and current. My guess is
that NONE of them will /ubbthreads/images/graemlins/smile.gif as i've had the same experiance with other
Max boost chips. They work, just not at as low of an input voltage
as we would like /ubbthreads/images/graemlins/smile.gif

If your inductor is 0.1 ohms i would suggest getting a 0.01 ohms or less
inductor. If your MOSFET switches happen to be lower (0.3 ohms) you'll
get a higher output.
If higher (0.6 ohms) it's not going to happen without
increasing the input voltage.


Take care,
Al
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Al, thanks for your input. I clearly misspoke when I referred to the minimums of 80 and 160 mA. I pulled these off of the maximum output gragh on the datasheet and apparently these are typicals. You are quite right about your analysis of the effect of considering the maximum Ron values on the datasheet. The phenomenon that is vexing me is something else, however, which is causing that discontinuity that is seen in the slope of the output vs Vin curve as shown in the posted links. I have test data for these IC's where their 3.3V output actually exceeds the datasheet maximum output graph for all voltages down to 1.2V or so but then abruptly plumments for lower Vin values.
 

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,144
Location
New Jersey
Hi again Doug,

I would have probably looked at that same graph, expecting
to see something i could 'use' rather than something i had
to go and calculate myself anyway. Gee, thanks Maxim for
the misleading datasheet /ubbthreads/images/graemlins/smile.gif Oh the other hand, when
have i found a data sheet that tells all (?).

Here's some more numbers for ya...

1.2v in; 170ma, 3.30v output
1.1v in; 145ma, 3.25v output
1.0v in; 110ma, 3.20v output

Here the output goes down with decreasing Vin, which is
what you are seeing, and it goes down quite fast.
Note, however, that we're still using 0.6 ohms for both
switches, which isnt necessarily the case as the spec
for 0.6 ohms is only valid for 2.0v inputs (data sheet).

It's been my experience that at very low input voltages
the gates dont get driven properly, which raises the
Ron resistance even higher. Since at 2.0v input it's
spec'd at 0.6 ohms max, i wouldnt doubt it if it's
double that at 1.0v input. The high switch resistance
wont allow the inductor to reach the required max
current flow in order to boost the output AND provide
a high enough level of current flow. This is most
pronounced with low input voltage because Vin/Ron doesnt
get up high enough, and Ron is also increasing as Vin
goes down, so there's that second effect too.

The way to tell for sure would be to measure it
with a scope. Looking at the voltage across one
of the switches while looking at the peak current
would tell the story in a heartbeat /ubbthreads/images/graemlins/smile.gif


Take care,
Al
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
[ QUOTE ]
MrAl said:
Hi again Doug,

I would have probably looked at that same graph, expecting
to see something i could 'use' rather than something i had
to go and calculate myself anyway. Gee, thanks Maxim for
the misleading datasheet /ubbthreads/images/graemlins/smile.gif Oh the other hand, when
have i found a data sheet that tells all (?).

Here's some more numbers for ya...

1.2v in; 170ma, 3.30v output
1.1v in; 145ma, 3.25v output
1.0v in; 110ma, 3.20v output

Here the output goes down with decreasing Vin, which is
what you are seeing, and it goes down quite fast.
Note, however, that we're still using 0.6 ohms for both
switches, which isnt necessarily the case as the spec
for 0.6 ohms is only valid for 2.0v inputs (data sheet).

It's been my experience that at very low input voltages
the gates dont get driven properly, which raises the
Ron resistance even higher. Since at 2.0v input it's
spec'd at 0.6 ohms max, i wouldnt doubt it if it's
double that at 1.0v input. The high switch resistance
wont allow the inductor to reach the required max
current flow in order to boost the output AND provide
a high enough level of current flow. This is most
pronounced with low input voltage because Vin/Ron doesnt
get up high enough, and Ron is also increasing as Vin
goes down, so there's that second effect too.

The way to tell for sure would be to measure it
with a scope. Looking at the voltage across one
of the switches while looking at the peak current
would tell the story in a heartbeat /ubbthreads/images/graemlins/smile.gif


Take care,
Al

[/ QUOTE ]

Al, where did those data in your I/O table come from? Are they some test data of yours or some more calculated values? Your comments on rising Ron with decreasing gate drive are noted but I don't believe apply to this situation. In this case, the gate drive levels are determined by the Vout since the gate drive supply is bootstrapped from the output for the bottom Mosfet [N-ch] and the top P-ch drive is inherently determined by the output voltage since the gate is driven to ground to turnon.
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
[ QUOTE ]
MrAl said:
Hi again Doug,

Here's some more numbers for ya...

1.2v in; 170ma, 3.30v output
1.1v in; 145ma, 3.25v output
1.0v in; 110ma, 3.20v output



[/ QUOTE ]
Al, for comparison, here is some actual test data for a Max1675 set for nominal 3.3Vout driving a Lux load:
1.30V in; 144ma, 3.16V out
1.25V in; 137ma, 3.16V out
1.20V in; 88ma, 3.06V out
1.15V in; 32ma, 2.88V out
1.10V in; 9ma, 2.72V out
 

Arjuna34

Newly Enlightened
Joined
Nov 20, 2004
Messages
10
Location
Virginia, USA
Re: Working with the Max1674/1675 ICs (MM, MM-lite

I've had the same experience with the MAX1795 boost switcher. Those "typical" graphs aren't so typical around a volt or lower. Moving up to the next size switcher (MAX1796 in my case) seemed to help - the FET resistance must be lower. Things still fall apart (for that series anyway) around a volt anyway, though.

Arjuna34
 

NewBie

*Retired*
Joined
Feb 18, 2004
Messages
4,944
Location
Oregon- United States of America
Have you tried the TPS61020 or this one?

http://www.onsemi.com/pub/Collateral/NCP1422-D.PDF

There is also this one, http://www.onsemi.com/site/products/summary/0,4450,NCP1450A,00.html?tax=149

See also, http://www.linear.com/pc/productDetail.do?navId=H0,C1,C1003,C1042,C1031,C1061,P2403

See also towards the end, output disconnect circuit, http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1042,C1031,C1061,P1250,D2794

See also, towards the end, single cell 0.9V, http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1042,C1031,C1061,P1897,D2673

As I recall, the lowest resistance switch I have seen in the low volts parts is 0.1 ohms.

To run lower, you can bootstrap the chip input voltage to the output voltage in many cases, and even enhance performance since it can drive the switch harder.

A year ago, when I did this, I was able to get them to run down to 0.2V input (had to be started), but not at your load power. In this case, the switch power had to be separate from chip power.
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
@Arjuna34: I see this is your first post. Welcome and thank you. I had for some reason overlooked the Max 1795/6/7 series of IC's. They look interesting. In your experience with them did they seem to get pretty close to 1V in before they deviated from their datasheet output capabilities? If so, they would at least be a bit better than the Max1674/5 IC's which start to go South around 1.2V.

@Newbie: I've got plenty of hands-on with the TPS61020. I believe we've exchanged a few PM's on that one. I've no experience with those two OnSemis that you suggest [BTW, all but your first link seem to be disfunctional]. Do you have any hands-on with either? If so, any comments? Well behaved? I couldn't make out what parts you were suggesting in the other links. I'd be interested in knowing if that low Vin part with the 0.1 ohm switch was single cell capable.
 

Arjuna34

Newly Enlightened
Joined
Nov 20, 2004
Messages
10
Location
Virginia, USA
Re: Working with the Max1674/1675 ICs (MM, MM-lite

Thanks for the welcome!
We ended up using the MAX1795 with two cells - the 1795 and 1796 weren't too good with a single AAA nicad (which was being boosting to 5.4V at about 80mA). The MAX1797 may have been OK, but was ruled out due to long lead times (this was about a year ago). I think I've got the eval kit for the MAX1797 laying around - I could fire it up if you're interested. However, it can't be a good sign that the MAX1797 datasheet guarantees startup at 0.85V, but the eval kit datasheet warns that the input must be >1.0V for guaranteed startup.

Arjuna34
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Re: Working with the Max1674/1675 ICs (MM, MM-lite

[ QUOTE ]
Arjuna34 said:
Thanks for the welcome!
We ended up using the MAX1795 with two cells - the 1795 and 1796 weren't too good with a single AAA nicad (which was being boosting to 5.4V at about 80mA). The MAX1797 may have been OK, but was ruled out due to long lead times (this was about a year ago). I think I've got the eval kit for the MAX1797 laying around - I could fire it up if you're interested. However, it can't be a good sign that the MAX1797 datasheet guarantees startup at 0.85V, but the eval kit datasheet warns that the input must be >1.0V for guaranteed startup.

Arjuna34

[/ QUOTE ]
Yes, if it's not too much trouble, I'd be interested in seeing some actual I/O performance data from that eval board in the 1.0 to 1.5V in range when set for a nominal 3.3Vout. From the datasheets these seem to be able to do better at the 3.3V setting than the 5V setting at low Vin on a Pout basis. BTW, from the look of the datasheets, the Max1674/5/6 series seem to have very similar internal architecture to the Max1795/6/7.
 

Arjuna34

Newly Enlightened
Joined
Nov 20, 2004
Messages
10
Location
Virginia, USA
Re: Working with the Max1674/1675 ICs (MM, MM-lite

What kind of load? I don't have any Luxeons, but I could use a resistive load, or series some diodes to get a Vf in the 3V ballpark, and measure the current.

Or, I could adjust the input voltage, and vary the load to find the peak current available at that input.

Arjuna34
 

NewBie

*Retired*
Joined
Feb 18, 2004
Messages
4,944
Location
Oregon- United States of America
Re: Working with the Max1674/1675 ICs (MM, MM-lite

I see that series has a 0.17 - 0.35 ohm N-ch FET and a 0.27-0.45 ohm P-ch FET when driving .33 Watts out off 2V Vbatt. Also notice it nees less than 10mA load during startup off 1.2V when putting out 3.3V, they mis-labeled the side of the chart as Efficiency. Sheet six has some interesting curves, max current out vs. input voltage.
 

NewBie

*Retired*
Joined
Feb 18, 2004
Messages
4,944
Location
Oregon- United States of America
Re: Working with the Max1674/1675 ICs (MM, MM-lite

[ QUOTE ]
NewBie said:
I see that series has a 0.17 - 0.35 ohm N-ch FET and a 0.27-0.45 ohm P-ch FET when driving .33 Watts out off 2V Vbatt. Also notice it nees less than 10mA load during startup off 1.2V when putting out 3.3V, they mis-labeled the side of the chart as Efficiency. Sheet six has some interesting curves, max current out vs. input voltage.

[/ QUOTE ]

BTW, I know you are sharp Doug, just a gentile reminder that the cell voltage will drop under load...
 

dat2zip

Flashlight Enthusiast
Joined
Jan 5, 2002
Messages
3,420
Location
Bay Area
Re: Working with the Max1674/1675 ICs (MM, MM-lite

I never looked to closely at running waveforms on the 1674/1675 under 1.5V. I did notice it did run way down to well below 0.5V. If you consider a dim glow acceptable.

Maxim specifications aren't always up to snuff and most of their ICs have a specific customer requirement or solve a customers requirements. Whether we can use it for something else is never quite fully understood since Maxim IC folks tend to scratch their heads as well as to how their parts actually work.

I've got the TPS61020 working downto 1V and with decent outputs. I'd have to go dig up my initial tests results, but, offhand I recall 180mA to the Luxeon flat CC from 1.5V to 1V.

If you need more info, send me email or PM me.
 

rdshores

Enlightened
Joined
May 3, 2004
Messages
400
Location
Vermont
[ QUOTE ]
NewBie said:
Have you tried the TPS61020 or this one?

http://www.onsemi.com/pub/Collateral/NCP1422-D.PDF

There is also this one, http://www.onsemi.com/site/products/summary/0,4450,NCP1450A,00.html?tax=149

See also, http://www.linear.com/pc/productDetail.do?navId=H0,C1,C1003,C1042,C1031,C1061,P2403

See also towards the end, output disconnect circuit, http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1042,C1031,C1061,P1250,D2794

See also, towards the end, single cell 0.9V, http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1042,C1031,C1061,P1897,D2673

As I recall, the lowest resistance switch I have seen in the low volts parts is 0.1 ohms.

To run lower, you can bootstrap the chip input voltage to the output voltage in many cases, and even enhance performance since it can drive the switch harder.

A year ago, when I did this, I was able to get them to run down to 0.2V input (had to be started), but not at your load power. In this case, the switch power had to be separate from chip power.

[/ QUOTE ]

CORRECTED LINKS:

Have you tried the TPS61020 or this one?
HERE

There is also this one,
HERE

See also,
HERE

See also towards the end, output disconnect circuit,
HERE

See also, towards the end, single cell 0.9V,
HERE

As I recall, the lowest resistance switch I have seen in the low volts parts is 0.1 ohms.

To run lower, you can bootstrap the chip input voltage to the output voltage in many cases, and even enhance performance since it can drive the switch harder.

A year ago, when I did this, I was able to get them to run down to 0.2V input (had to be started), but not at your load power. In this case, the switch power had to be separate from chip power.
 

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,144
Location
New Jersey
Hi Doug S,

[ QUOTE ]
Doug S said:
[ QUOTE ]
MrAl said:
Hi again Doug,

Here's some more numbers for ya...

1.2v in; 170ma, 3.30v output
1.1v in; 145ma, 3.25v output
1.0v in; 110ma, 3.20v output



[/ QUOTE ]
Al, for comparison, here is some actual test data for a Max1675 set for nominal 3.3Vout driving a Lux load:
1.30V in; 144ma, 3.16V out
1.25V in; 137ma, 3.16V out
1.20V in; 88ma, 3.06V out
1.15V in; 32ma, 2.88V out
1.10V in; 9ma, 2.72V out

[/ QUOTE ]


Wow, your numbers are much lower than the theoretical
ones i posted, and i was using 0.6 ohm switches too.

I couldnt find the bootstrapping notes on the data sheet,
but even with bootstrapping the numbers on the spec sheet
are with 2v input, so below that all bets are still off.
There are probably too many things that could go wrong
to try to guess what it is... rise/fall times could
increase, etc., and we dont even know for sure if
the feedback circuit functions normally at very low
voltage (ie is their control law adequate?). Also, if
something else causes the Vout
to fall then the P MOSFET gate voltage causes the
voltage to fall even more.
Of course if the inductor has high resistance (0.1 ohm
for example) that could be the whole problem. Did you
check it? Note that even some of the inductors shown
on the data sheet wont be adequate. I'd check this
first.

As i said in the other post, the way to tell for sure is
to look at the switch resistance with a scope. The other
things i would look for would be rise/fall times,
duty cycle (make sure it varies according to Vout
demands, not just increasing for decreasing Vout), and
input voltage ripple (to make sure the voltage is always
what is measured on a dc voltmeter) ...recall that if
the input plummets even 0.1 volt during a switch turnon
that could cause the problem too. What's the esr of
the input cap and how close is it to the chip?


Take care,
Al
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Re: Working with the Max1674/1675 ICs (MM, MM-lite

[ QUOTE ]
Arjuna34 said:
What kind of load? I don't have any Luxeons, but I could use a resistive load, or series some diodes to get a Vf in the 3V ballpark, and measure the current.

Or, I could adjust the input voltage, and vary the load to find the peak current available at that input.

Arjuna34

[/ QUOTE ]
Thanks. I'd say that a string of series diodes to the 3V ballpark would be the best match for my application. We want the load to pull the output down a small amount under the setpoint voltage to insure that the IC should be running flat out.
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Re: Working with the Max1674/1675 ICs (MM, MM-lite

@Al
The reference to bootstrapping is buried in the Pin description table for pin 8 on page 6 of the DS. You are quite right that the datasheet values for Ron that you are using for your calcs are at a condition of Vin=2.0, Vout=3.3, [and BTW, I=100mA so things might even be worse than your calcs by a hair]. The complete definition of the test conditions reads Vin=2.0V, Vout=3.3V....unless otherwise noted. I am focussing more on the Max output current vs Vin graphs on page 5 of the DS where clearly the "unless otherwise noted" applies w.r.t. Vin. It is this depicted performance that I am failing to see across all of the few samples I have tested.
Re inductor and input capacitor characteristics, my standard test protocol on benchtesting a converter board off of a bench supply is to add a 47uF Sanyo OS-CON cap soldered directly to the input [ESR in the low m-ohm range] to decouple supply leads w.r.t. inductance and resistance. The stock input cap on the MM board is 1uF and [I'm guessing here] probably X5R dielectric [that is what I would specify if it were my design]. Regarding closeness of the input cap to the IC, there is no direct connection between the input and the IC since the IC bootstraps off of the output. This *should* make the IC control immune to effects of input voltage sag during the switch on period. Since the IC control works off of the output voltage and the datasheet range for output voltage extends down to 2.0V, we should not expect internal control circuit problems related to available voltage when set for a nominal 3.3V even when the output is sagging down to even a bit below 3V. I am using the MM/MM-lite boards as the test fixture in this testing. I have substituted inductors with resistances down to under 30 m-ohm without achieving improvement in the low Vin performance. The data I provided above, however, was taken with the stock MM inductor which I measured at .148 ohms using a test current of .40A.

Other relevant test data not mentioned earlier. Even in an unloaded condition, the output voltage falls below nominal at input voltages below around 1.15V or so. The same is true for a CC sink load set at any value between 2.6mA and 92mA [this was the range of the adjustable CC sink that I was using].
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
[ QUOTE ]
MrAl said:
Hi Doug S,

[ QUOTE ]
Doug S said:
[ QUOTE ]
MrAl said:
Hi again Doug,

Here's some more numbers for ya...

1.2v in; 170ma, 3.30v output
1.1v in; 145ma, 3.25v output
1.0v in; 110ma, 3.20v output



[/ QUOTE ]
Al, for comparison, here is some actual test data for a Max1675 set for nominal 3.3Vout driving a Lux load:
1.30V in; 144ma, 3.16V out
1.25V in; 137ma, 3.16V out
1.20V in; 88ma, 3.06V out
1.15V in; 32ma, 2.88V out
1.10V in; 9ma, 2.72V out

[/ QUOTE ]


Wow, your numbers are much lower than the theoretical
ones i posted, and i was using 0.6 ohm switches too.

Al

[/ QUOTE ]

Note that my data are for the Max1675 which is the 0.5A CL part. Are your calcs for the Max1675 [0.5A CL] or Max1674 [1.0A CL] part?
 
Top