TPS63020 Based Driver

[foulmist]

I coulda sworn I saw a zener in your design earlier. Now I am really curious... -.-

You might have seen some of my previous boosts not the TPS63020...
I use zener for OVP protection but the TPS63020 already has one built in (tested).

 

[rfhv]

I added the zener to mine for overvoltage protection. I know that the TPS63020 has built in OVP but I do not know where it limits. If the limit is above 6.3 V then you may destroy your output caps unless you are using 10 V caps. I went the safe route and added the 5.1 V zener to be sure I do not damage my output caps.

 

[foulmist]

I added the zener to mine for overvoltage protection. I know that the TPS63020 has built in OVP but I do not know where it limits. If the limit is above 6.3 V then you may destroy your output caps unless you are using 10 V caps. I went the safe route and added the 5.1 V zener to be sure I do not damage my output caps.

As far as I remember it was 6.8V so better use 10V caps

 

[HTV]

Can you publish oscillograms of driver’s work at 0.5A and 1A?

It isn’t a useless interest. The same scheme was published around year ago on our site. Our scheme has big pulsations. Difference is in one chip. You used ZXCT1009 instead ina138.

 

[foulmist]

Can you publish oscillograms of driver’s work at 0.5A and 1A?

It isn’t a useless interest. The same scheme was published around year ago on our site. Our scheme has big pulsations. Difference is in one chip. You used ZXCT1009 instead ina138.

I will make oscillograms when I finish it. I am currently waiting on some inductors. No I don't use Ina or ZXCT or anything other than the TPS63020 itself. I don't see why everyone adds current regulators when there is already a method in getting constant current out of voltage regulators and IT WORKS. No pulsations so far

 

[HTV]

No pulsations so far

Congratulations! :beer: We couldn’t solve this problem.

 

[benmwv]

I will make oscillograms when I finish it. I am currently waiting on some inductors. No I don't use Ina or ZXCT or anything other than the TPS63020 itself. I don't see why everyone adds current regulators when there is already a method in getting constant current out of voltage regulators and IT WORKS. No pulsations so far

So you can adjust with a pot, and so you have a common ground.

 

[foulmist]

So you can adjust with a pot, and so you have a common ground.

Well, yeah but I don't like the pot thing

 

[benmwv]

Well, yeah but I don't like the pot thing

True, fixed resistor is much easier to use.

It depends on the situation. If I know exactly what current I want, and can get it with a fixed resistor then they are preferable. But an adjustable driver is more versatile.

If only those tiny smd multiturn pots weren't so expensive, they would be much more precise.

 

[foulmist]

True, fixed resistor is much easier to use.

It depends on the situation. If I know exactly what current I want, and can get it with a fixed resistor then they are preferable. But an adjustable driver is more versatile.

If only those tiny smd multiturn pots weren't so expensive, they would be much more precise.

aah yeah I like the MULTITURN option! but single turn pots I hate

 

[rfhv]

My sensing scheme allows a common ground for everything. I use three wires in my design. I predict that using low side sensing you will not hit quite the maximum current out that my driver can hit. This is because the TPS has a feedback voltage of 0.5 V. If you use low side sensing and do not amplify the sense resistor voltage you will loose 0.5 V across the sense resistor. I picked high side sensing to squeeze out every last volt from this driver.

I switched to a 1.5 uH inductor and saw no difference in the performance of my driver.

I have the following scope shots to share. All scope shots are of output current into my test load. One division from the bottom is 0 A. Sensitivity is 500 mA per division. Coupling is DC. Probe is 10X.

Here is the driver's output with 3.68 V in at 2.79 A. Output current was 1.83 A at 4.61 V. Efficiency was 82%. Ripple current is about 200 mA. Ripple frequency is about 200 kHz.

For this next image the input was 3.85 V at 1.33A. The output was 1.03 A at 3.91 V. Efficiency is 79%. Ripple is the same as before, about 200 mA

Past 1.8 A, the output waveform gets distorted (probably due to the driver nearing its maximum duty cycle). Here is 3.63 V at 3.20 A in. Output is 1.94 A at 4.67 V. Efficiency is 78%. Increasing the input voltage to 4.2A causes this distortion to go away. Ripple is a little more, around 300 mA I suspect.

To summarize, I hit 1.8 A out with 3.7 V in with 82% efficiency.

Here is my test setup.

 

[Wolfman29]

Damnit Foulmist, share this revelation! It will bring about a whole new world of DIY switch converters

 

[foulmist]

Damnit Foulmist, share this revelation! It will bring about a whole new world of DIY switch converters

what revelation it's nothing new... I use low-side current sensing that's all

@rfhv

try putting 2x22uF ceramic caps on the output and as close as possible to the IC for better stability. If you can put a smaller cap very close the the IC and then the 2x22uF close to that cap.

 

[Wolfman29]

Ah. Well, I will need to read up on how current sensing works =p

 

[foulmist]

@rhfv

Check the Vin, Vout and GND traces if they are wide enough,

Try a different layout with the same components.

Try not to think of how small can you make it - do it a little bigger just to check if the layout is the problem

:beer:

 

[rfhv]

I suspect the ripple is a consequence of the TPS chip itself and the fact that I am in current mode rather than voltage mode. No value of capacitor can remove the ripple. If you can show that with low side sensing you have no ripple then I may redesign my circuit to have a much higher bandwidth feedback loop. I suspect it will be very hard to remove the ripple.

Either way, I don't expect the ripple current to be any problem. The frequency of the ripple is so high that the diode will only see the average current. 200 kHz is far too fast for the diode to see any thermal shock, etc.