Total noob with power supply noise

[Yosefu]

I built a portable class D amp with a TPA3118 and a preamp. It's powered by 8 AA nimh rechargeable batteries, but I added a power jack so I could recharge them without having to take them out. I know it's not the best way, but my plan was to recharge them with a 12v adapter.


Now, the amp works great on batteries, but gets awfully noisy when I connect the 12v adapter. I thought the voltage difference between the batteries and the adapter might be the culprit, so I took out the batteries: the noise persists. Then I tried my laptop's adapter: still the same, if not noisier.

So I figured this must be a common problem for amps, and there must be a generic way of filtering out this noise. But as a noob, I have no idea.

 

[rayma]

Heavy LC or RC filtering in the amplifier.

 

[Yosefu]

like a high pass filter cutting everything below 65hz? In the signal path (for instance, after the preamp) or in the power section?

 

[rayma]

Use power supply filtering, with a series L or R, and a shunt C, for each supply leg.
The filter corners would have to be well below 10Hz to do much good.

Series voltage regulators could also reduce the noise, if properly implemented.
I'd still filter the input voltage to the regulators, though.

 

[jjasniew]

You didnt mention what your line level audio "souce" is, when the noise occurs. Another plugged-in to AC device perhaps? If so, could be a ground loop that only happens when the "charger" PSU is connected.

 

[Yosefu]

You didnt mention what your line level audio "souce" is, when the noise occurs. Another plugged-in to AC device perhaps? If so, could be a ground loop that only happens when the "charger" PSU is connected.

Not sure I understand - what would be causing the ground loop?

Use power supply filtering, with a series L or R, and a shunt C, for each supply leg.
The filter corners would have to be well below 10Hz to do much good.

Thanks! So then something like a second order low pass filter? What values would you recommend? I was thinking about going for two 25 ohm resistors, so that way I could charge the batteries at C/10 rate.

 

[Vovk Z]

Try to add small LC-filter at charging adaptor input: something like 10uH (+-50%) inductor (with rated current good enough) and 100-1000 uF Low-ESR (or solid polymer) electrolyte cap.

 

[Vovk Z]

I was thinking about going for two 25 ohm resistors, so that way I could charge the batteries at C/10 rate.

- That can be fine too. RC filter, with C = 100-1000 uF (the larger the better). Or you may use RLC filter. R and L in series. Almost any small L can do - from 3.3 uH to 100 uH.

 

[Rick PA Stadel]

I'm worried about the long-term health of your NiMH cells. A generic 12V adapter is not a proper charging method.

The 25 ohm resistors won't give C/10 charge rate, even with the amp turned off. At the top of their charge curve, 8 NiMH cells in series will measure about 11.4 volts. 0.6V across 25 ohms produces a 24mA current -- about C/100. For 50 ohms, half that.

All of the 'modern' battery technologies are somewhat particular about charging specifications. Gone are the lead-acid car battery days when you could just set the voltage to 14.5V and pound in whatever current was available.

Noise is only one problem with charging while operating. The batteries prefer constant current charging(*), while the amplifier's current demands vary with the music. While it is possible to design a circuit that properly adjusts the current demands on the adapter, it is not a trivial undertaking.

The batteries will last a lot longer if you just turn the amp off to charge them.

Cheers

* during the initial 80%, 'bulk' interval

 

[mchambin]

I understand you need a 240mA constant current supply, that can go down to 0.6V.
This is easy to make with two transistors and two resistors.
Have a look at Rod Eliott ESP site for explanations about CCS.
Current Sources, Sinks and Mirrors in Audio

 

[Mark Johnson]

None of the constant current sources on Rod Elliott's page operate correctly with only 0.6V across the CCS circuit. To achieve that you'd need to heavily modify the circuit in Figure 5.2.1, plus you'd need to carefully choose just the right active components. Plenty of plausible looking parts will poop out and give a non-constant current, well above 0.6V across the CCS circuit.

 

[mchambin]

Right, assuming Silicium transistors.
A way is to use germanium transistors for their lower Vbe. About 0.3V.
Using a Germanium transistor it should be ok down to
Vbe + Vsat = 0.3V +0.1V = 0.4V

 

[Rick PA Stadel]

Of the 3 issues the OP is facing, the 0.6V minimum seems to be the easier to overcome -- a 14 or 19V adapter solves it in a snap.

The noise problem would probably readily surrender to Mark Johnson's recent contribution -- thanks for piping in here, MJ -- my 1/4-a$$ed *search* for it came up dry, though. Maybe he'll direct us to it.

It may be a smaller deal here than it has been on some of my projects, but I think the circuit to add the operating amp's current drain to the desired battery charge current, just might be beyond the OP's skill set. (Sorry if I'm wrong, but the bit about a 'couple 25 ohm resistors for C/10' strikes me as a pretty strong hint ..)

Also wondering how long battery operation (per charge) is expected.

Regards

 

[Mark Johnson]

Here is a constant current source made with silicon devices (BC337, LM358) whose output current remains a constant 2mA, all the way down to 0.15 volts.

_

 

[kimschips]

Can you also show schematic and maybe some measurements please?

 

[Yosefu]

Rayma's suggestion's worked pretty well in reducing the noise to a bearable level. Thanks!


Yes, I know this is very crude way to recharge batteries, but I couldn't bother with anything more complicated at this time. I've seen some ICs for this job, maybe next time I'll try them.

It may be a smaller deal here than it has been on some of my projects, but I think the circuit to add the operating amp's current drain to the desired battery charge current, just might be beyond the OP's skill set. (Sorry if I'm wrong, but the bit about a 'couple 25 ohm resistors for C/10' strikes me as a pretty strong hint ..)

Well, the title says "Total noob with power supply noise"...