Need boost / step-up converter to 9v from 3 AAA batteries for Audio Project, high current, low dist

Thread Starter

65musclecar

Joined Sep 28, 2017
12
I'm building a small custom battery powered PCB for TPA3128D2. I only have space for 3 AAA batteries, but I need 9v. i.e - I need a DC-DC boost from 4.5v to 8v. System is stereo, has a 2W/8Ohm speaker and a 5W/4Ohm speaker so I will need 7W-8W (approx 1.8Amps). I'm Using 32dB gain. A 9v NiMH battery works but it's too big. 3AAA just barely fit. Don't want to use lithium batteries for safety reasons. Duty cycle is 25%-50% so short run time is ok.

I've tried an off the shelf DC-DC boost converter board MEZD41503A-B, but it cuts out and distorts at higher volume. I've been advised I need to add a capacitor to the voltage input fix this problem. I know it's currently producing 12v, so my first task is modding it to produce 8v or finding a different boost converter.

Any advice on what DC-DC boost converter to use or where to start would be greatly appreciated. Also any help on how to modify the MEZD41503A-B to eliminate the distortion and get the higher current needed would be awesome.

I've heard Web Bench software can help but not sure if the designs produced will achieve good quality audio. Based on WebBench recommendation I also just ordered http://www.ti.com/tool/TPS61089EVM-742 which will run at 9v volts. It's supposed to arrive today so I'll give that a shot too!

Thanks for any advice!!
 

Audioguru

Joined Dec 20, 2007
11,251
A 9V Ni-MH battery or two AAA cells cannot provide the 1A needed for the boost converter to provide 8V to the amplifier to produce 3.5W per channel into 8 ohms or be overloaded with 4 ohms.
 

Thread Starter

65musclecar

Joined Sep 28, 2017
12
A 9V Ni-MH battery or two AAA cells cannot provide the 1A needed for the boost converter to provide 8V to the amplifier to produce 3.5W per channel into 8 ohms or be overloaded with 4 ohms.
Thanks for the feedback Audioguru. The datasheet for the AAA battery I'm using lists the discharge characteristics for 2Amps which makes me think think it is an acceptable use case despite the short battery life.

http://www.fdk.com/battery/nimh_e/tech_info/HR-4U.pdf

I've read about AAA being used for camera flashes that require even more current, albeit for a short burst.

My system works great with a 9v NiMH battery (without the converter) so I believe the converter is the problem.

My plan is to use three AAA batteries (not two) with either the MEZD41503A-B (12.5v) or the TPS61089EVM-742 (9.5v) with the TPA3128D2 Dev board at 32dB gain. They both work at low audio input voltage, but as the audio input level increases past about 0.3 Vpk-pk (sine wave or 0.1 Vrms) the speaker output starts to get a weird phasing kind of sound and sometimes even drops out. Audio input of 0.3v pk-pk produces ~10v pk-pk output sine wave which is ~3Watt rms to the 4 ohm speaker and ~1.6W rms to the 8ohm speaker. I'm using Vrms = 0.3536 x Vpk-pk for a sine wave and Prms = Vrms x I put another way, Vrms = ( Vpk-pk / 2 ) / sqrt(2)

It's very possible I'm missing something (which is why I'm posting here ha ha!). If the AAA can produce 2Amps of current as shown on datasheet, then In theory 3 AAA should be able to produce 9Watts P = VI or P = 4.5v x 2amps , P = 9Watts.

In practice its not working...yet :)

My biggest questions is where to start with modding the TPS61089EVM-742 to fix this problem. I think capacitors will help buffer the voltage, but not sure how to connect them to the dev board and what values to try. Also whether to put on input, output or both.

Thanks!
 

ebp

Joined Feb 8, 2018
2,332
If a boost converter is operated in continuous inductor current mode (that is, the current in the inductor does not drop to zero each switching cycle), which is almost invariably the case at higher current, there is a "right half plane zero" in the transfer function. There is no way to directly compensate for a RHPZ, so it must instead be worked around. The workaround is to make the response of the system slow. This means that with a highly dynamic load it will have difficulty "keeping up" and there will be voltage overshoots and undershoots. An audio amplifier is a fairly dynamic load, depending on the signal frequency and the amplitude.

If this is the problem, the only way of coping with it is to increase the capacitance on the output of the boost converter. I see the dev board has provision for a through-hole cap. The is nominally only 66 µF on the output without that cap. A ghosted 470 µF is shown on the schematic, and should likely be easily physically possible at the required voltage.

AAA cells are pretty tiny, but I used to use AAs in a photoflash that was about as powerful as portable flashes get. It would draw 5 or 6 amps from the cells. You must remember that the input of a switcher looks like a negative (in slope, not absolute value) resistance, so if the voltage goes down the current goes up which, with a battery, means the voltage goes down so ...
 

Audioguru

Joined Dec 20, 2007
11,251
You do not understand that the TPA3128D2 amplifier has bridged (BTL) outputs with an amplifier driving one speaker wire and another amplifier driving the other speaker wire with opposite phase that almost doubles the output voltage swing and almost doubles the current. With a 9.5V supply the output into a 4 ohm speaker is shown on a graph at about 10W at 1% distortion. Stereo needs a supply that delivers 20W plus
2W efficiency loss= 22W. Then the continuous current from the 9.5V output of the boost converter is 22W/9.5V= 2.3A.

I agree that the output of the boost converter needs a huge capacitor since peaks in the sounds use 1.414 times the current than for a continuous loud tone.

The boost converter has a 10% efficiency loss then the power from the battery is 22W + 2W= 24W. If three AAA Ni-MH cells are used then their voltage falls flat on their face at maybe 1V per cell for a total voltage of 3V. Then the continuous current from the battery at 1% distortion is 24W/3V= 8A which is impossible from the tiny battery. If you play very distorted acid rock then the current is 8A x 1.414= 11.3A.
 

Thread Starter

65musclecar

Joined Sep 28, 2017
12
Thanks again for the feedback AudioGuru. I will do some reading on BTL. I will say that my system works fine with a NiMH 9v battery, albeit with a short run time. I will do some measurements to see how much power the 9v NiMH is generating. My guess is I haven't accurately described my power requirements.
 

Thread Starter

65musclecar

Joined Sep 28, 2017
12
Thanks EBP. I added the 470µF cap on the output and it indeed improved performance. I also added 3 330µF caps on the input, which also improved performance, but it's still not where I'd like and a fair bit worse than my 9v NiMH battery. Getting closer though!!!

If a boost converter is operated in continuous inductor current mode (that is, the current in the inductor does not drop to zero each switching cycle), which is almost invariably the case at higher current, there is a "right half plane zero" in the transfer function. There is no way to directly compensate for a RHPZ, so it must instead be worked around. The workaround is to make the response of the system slow. This means that with a highly dynamic load it will have difficulty "keeping up" and there will be voltage overshoots and undershoots. An audio amplifier is a fairly dynamic load, depending on the signal frequency and the amplitude.

If this is the problem, the only way of coping with it is to increase the capacitance on the output of the boost converter. I see the dev board has provision for a through-hole cap. The is nominally only 66 µF on the output without that cap. A ghosted 470 µF is shown on the schematic, and should likely be easily physically possible at the required voltage.

AAA cells are pretty tiny, but I used to use AAs in a photoflash that was about as powerful as portable flashes get. It would draw 5 or 6 amps from the cells. You must remember that the input of a switcher looks like a negative (in slope, not absolute value) resistance, so if the voltage goes down the current goes up which, with a battery, means the voltage goes down so ...
If a boost converter is operated in continuous inductor current mode (that is, the current in the inductor does not drop to zero each switching cycle), which is almost invariably the case at higher current, there is a "right half plane zero" in the transfer function. There is no way to directly compensate for a RHPZ, so it must instead be worked around. The workaround is to make the response of the system slow. This means that with a highly dynamic load it will have difficulty "keeping up" and there will be voltage overshoots and undershoots. An audio amplifier is a fairly dynamic load, depending on the signal frequency and the amplitude.

If this is the problem, the only way of coping with it is to increase the capacitance on the output of the boost converter. I see the dev board has provision for a through-hole cap. The is nominally only 66 µF on the output without that cap. A ghosted 470 µF is shown on the schematic, and should likely be easily physically possible at the required voltage.

AAA cells are pretty tiny, but I used to use AAs in a photoflash that was about as powerful as portable flashes get. It would draw 5 or 6 amps from the cells. You must remember that the input of a switcher looks like a negative (in slope, not absolute value) resistance, so if the voltage goes down the current goes up which, with a battery, means the voltage goes down so ...
 

ArakelTheDragon

Joined Nov 18, 2016
1,264
A 9V Ni-MH battery or two AAA cells cannot provide the 1A needed for the boost converter to provide 8V to the amplifier to produce 3.5W per channel into 8 ohms or be overloaded with 4 ohms.
3 x 3.7VDC 18650 batteries will be better. They can be at 4000-5000mA/h and will provide the current you want for 12 hours approximately.

I bought these ones recently and they didn't drop after the first minute. I have yet to test them extensively. One deficiency is they look a mm or 2mm bigger than normal or maybe the box was 1 or 2mm smaller so it was hard to put them in, now I can't remove them :D.

CARPRIE 2PCS 3.7V 7800mAH for 5.24 USD
https://www.cfcbazar.com/products/carprie-2pcs-3-7v-7800mah-li-ion-rechargeable-18650-battery-for-flashlight-torch-emergency-lighting-portable-devices-power-tools/

Edit:
My giess is you have a normal "9VDC" battery which is not rechargable, they don't put the "mA/h" on them, so it should be about 2000-3000mA/h.
 

ebp

Joined Feb 8, 2018
2,332
You might try even more capacitance on the output of the booster if you have room. Some switchers get cranky if you put too much capacitance on the output, but I'm reasonably sure that dev board will be OK. Ordinary aluminum electrolytic caps don't need to be "derated" in terms of operating voltage, so you can use the lowest voltage rating that is at least equal to the booster's output voltage. That may allow using a physically smaller cap (e.g. 10 V rating vs. 16 V - I think there are some 8 V parts around, but I don't think it is a very common rating).

Usually with switchers you want to have the output capacitor very close to the switcher circuit and sort of "between" the switcher and the load. However, because there are good high-frequency ceramic capacitors in the right place on the circuit board, it isn't likely to be a problem if you need to use some short extensions for the leads of a big cap across the output to allow you to position it conveniently.

The fact that you get some improvement with extra capacitance across the input does say that the source impedance of the AAAs is an issue with the dynamic nature of your load.
 

Audioguru

Joined Dec 20, 2007
11,251
I bought these ones recently and they didn't drop after the first minute. I have yet to test them extensively. One deficiency is they look a mm or 2mm bigger than normal or maybe the box was 1 or 2mm smaller so it was hard to put them in, now I can't remove them :D.

CARPRIE 2PCS 3.7V 7800mAH for 5.24 USD
https://www.cfcbazar.com/products/carprie-2pcs-3-7v-7800mah-li-ion-rechargeable-18650-battery-for-flashlight-torch-emergency-lighting-portable-devices-power-tools/
Your no-name-brand batteries are too long to be 18650 which should be 18mm in diameter and 65mm long. Name-brand 18650 lithium batteries have half the faked capacity claimed by your no-name-brand batteries. Here are some more cheap Chinese fake batteries:
 

Attachments

Last edited:

Thread Starter

65musclecar

Joined Sep 28, 2017
12
I'm concerned about using Lithium Ion batteries because the use case is up against the skin and a lipo failure could be dangerous. It's possible my concern is not rational based on current lipo failure rates, but I don't want to find out the hard way either. (love to get some data on failure rates and safety because lipo would give more run time /smaller form factor / faster charge)
 
Top