TPA3255: Current source

Hi,

There are a few concerns here,

1. Audio amplifier is designed for audio application, which means it may have some problem to support a signal below 20Hz.

2. The current information feedback is a good idea. I understand the PA will be used as a block-box and provide simply a gain. You may search the app note, slaa788a, it may help you.

However, to avoid stability issue and anything else, is it possible for you to move the sum node of the loop before the input signal generator? I mean, suppose you have a DAC to generate the input signal for the PA, I would recommend to feed the feedback before the DAC. It would give us more flexibility.

Besides, a simulation on TINA or other tools may help a lot to verify your design.

Thanks!

Regards,

Sam

Thank you Sam,

Exactly we would like to use the TPA3255 as a back box that provides the power gain for our current loop. Your concerns with using this amp in a non-audio application echo my own. 

I will look at the post filter feedback note you reference (slaa788a).

At this stage it is possible to move the sum of the node of the loop before a signal generator (or any place that would be best). We have talked about reading the output signal from the current sensing amplifier with an ADC and then using that information to control a DAC to generate the PA input signal. We are concerned that we would not be able to do this (read the output and control the input) fast enough and a delay in the loop could cause stability issues and/or distortion on the output. Right now we have a MCU that runs at ~120MHz. If I get a prototype setup I can give it a shot, I agree that it would lend to a lot more flexibility.

I would like to be able to supply a DC (or near DC) signal to the inputs, but I'm concerned with not AC coupling the input signal. Reading around the E2E forum I believe it is possible to get away with not AC coupling the input signal as long as it's biased with AVDD/2. I'm just hesitant to try this with my eval board as it's the only one I have.

Good idea, I will also look into simulating this.

Is there any other part you would recommend using if I did not use the TPA3255? I would be looking at similar specs as the TPA3255 where we need high power (~10A through a ~2ohm load), low noise, and cooling constraints limit us to high efficiency designs such as a class-D amp.

Thanks!

Colen

Hi Colen,

1. Understand and agree your concern on the long delay of the control loop. If we can do pure analog, it would be perfect. However, for a 50kHz signal, I think a digital loop should also be good.

2. Yes, most audio amplifier has internal bias on the input, and need AC coupling capacitors. DC input is an fatal error in audio application that it can cause the speaker physically damaged. 

3. If DC signals are indeed needed, you may also take a look at TAS5614/34. They are class-d power stage, which takes PCM signal as input. PCM signals can comes from your MCU. 

Thanks!

Regards,

Sam

Thank you Sam, I will do some testing and if I need to I will look into the TAS5634 as a fallback plan.

Hi Sam,

I was able to perform some testing with a TPA3255EVM module that I have, and I've had some success but I'm running into a fault condition that is hanging me up.

First I tested if the module would work with a DC input, so I removed the DC blocking caps (C17, C28, C55, and C63) on the EVM PCB and soldered  wires directly to the pads that connect to INA, INB, INC, and IND. I used the wipers from two 10K POTs that were each connected between 7V and ground as my DC input sources. With the unit in x2 BTL mode, I had INA and INC connected to one DC input and INB and INC connected to a different DC input. So far so good, with no load (and later with a load) the amplifier output would track the DC input. e.g. I found that with a 40V source for PVDD, a 2.5V input would give about a 90% positive duty cycle at PWMA which gave ~36V at OUTA, and a ~5.2V input would give about a 10% positive duty cycle at PWMA which gave ~4V at OUTA. I noticed nearly identical behavior between DC input and DC output for the other channels as well. Great!

I then connected a load (1mH 2.2ohms) between OUTA and OUTB. I took the INA input and inverted it about AVDD/2 (~3.875v) and fed it into INB, so as I made OUTA increase in voltage,OUTB would decrease by the same amount. After that I implemented the current control by using the INA253 amplifier and feeding the signal back around to an integrating amplifier that summed my input voltage with the INA253 output. The integrating amplifier output was then connected the INA/B input. This also seemed to work as the current output would track an arbitrary input signal. Just what I was looking for!

Now to my problem, I can only get the amplifier to output about 9A maximum before the FAULTz LED turns ON (the CLIP OTW LED remains OFF) and the outputs turn OFF. I'm thinking this error is something other than an overload condition as I measured the current through the load with a DMM and only can get around 9A out when I should be able to get around 17A (I did not touch the 22K resistor attached to the OC_ADJ pin).

Furthermore I removed the current loop (I went back to just using the two 10K POTs for the inputs and removed the INA253 current sensor and the integrating amp on the input) so I just had what I started with where I input a DC voltage and get a DC voltage out. Here I'm still getting the same unexpected (to me) FAULTz trip.

Some test results that show the issue I'm seeing:

PVDD = 40V, RL = 1mH 2.2ohm coil.I put a 2.5V input on INA and INB (giving ~90% +duty on PWMA and PWMB,  OUTA and OUTB are at ~36V), I then increase INA to ~2.85V (giving an Iout  of approx. -1.7A, PWMA is at ~80% +duty and OUTA is ~32.2V ) and the FAULtz LED turns ON (CLIP_OTWz LED remains OFF).  I see the same behavior when I keep INA at 2.5V and increase INB to ~2.85V.

PVDD = 40V, RL = 1mH 2.2ohm coil. I put 5.2V on INA and INB (giving ~10% duty on PWMA/B and ~4V on OUTA/B). I then decrease INA to ~3V which makes Iout go form 0 to ~9A (PWMA goes from 10% to ~80% duty), and then the FAULTz LED turns ON and the OUTPUT turns off.

PVDD = 20V, RL = 1mH 2.2ohm coil. I put 3.85V on INA and INB (giving ~50% +duty on PWMA/B and ~ 10V on OUTA/B)  I then decrease INA to about 3.4V where Iout goes from 0A to ~2.65A (PWMA goes from 50% to ~80% + duty) before FAULTz LED turns ON and the outputs turn OFF. 

Any thoughts on what's causing the FAULTz trip?  Looking around on the e2e forum it seems like it's a similar problem as this question: "TPA3255: OC shutdown always at only 9A" :

https://e2e.ti.com/support/audio/f/6/t/821615?tisearch=e2e-sitesearch&keymatch=tpa3255%209A

Sorry for the long post, I'm hoping it's something obvious that I'm missing, I'd really like to be able to output higher currents (ideally around 12A but 10A at the least). 

Thanks!

-Colen

Hi Colen,

As the device has internal bias on the input pins, I'm not sure if removing the AC coupling cap and feed DC signals directly to the input can cause some problem.

Do you think it's a good idea to try with AC signals like 100Hz, with the AC coupling caps on board and keep your rest circuit active? It could be a way to isolate the problem from a DC input condition. 

Thanks!

Regards,

Sam