pocket radio, ham-radio transceiver.

It has to connect standard microphone with speaker which couldn't be powered through differential pair of wires.

enter image description here

Minimal specs

  • single ended output (must have!)
  • 4.5-14V supply voltage
  • at least 1W on 4Ω load of power when powered from 5V
  • analog input
  • should be capable withstand shorts & hanging output

Ideally class D amp with low quiescent current and stand by mode.

Design problem

I found many great amplifiers but all of them are BTL. I spent several hours on many ICs vendors and had no luck. All modern power efficient amps are BTL.

Initially, tried to use TS4962 in single-ended configuration but it died right away. Also, doesn't tolerate shorts and opens. I killed several ICs and stopped my experiments. TBH, it is amazing IC and when all connections are soldered as BTL provides really good performance.

After I tried to use LM386 but was not happy by its internal noise level.

Then I tried SSM2211 (in a SE configuration) which did work well but seemed having less gain. When I tried to make it more sensitive I ran into stability problem.

TDA2003 seems like ideal amplifier but it is very outdated, consumes too much current, makes too much heat.

Disclaimer: I understand rules, this is not website for recommendations but I did my homework there and I am stuck therefore I am looking for help of the community.

  • \$\begingroup\$ That's a product recommendation question. You'll notice that the TDA2003 is something pretty different in specs than the (ancient) LM386, but you know the companies that produce these – go and check their website. It's as easy as that. They have customizable tables of amplifier ICs with supply voltage, speaker impedance, power, … \$\endgroup\$ – Marcus Müller Dec 28 '19 at 22:21
  • \$\begingroup\$ What does this mean " 1+ W RMS 4Ω @ 5V"? you need it to have more than 1W? \$\endgroup\$ – jDAQ Dec 28 '19 at 22:22
  • \$\begingroup\$ Also, for a pocket radio you would be best served with a class C or D, considering energy consumption. \$\endgroup\$ – jDAQ Dec 28 '19 at 22:23
  • \$\begingroup\$ How about an I2S interface 72W bridge Amp (4x28W) TDA7802 is a single chip quad bridge amplifier in advanced BCD technology integrating: a full D/A converter, digital input for direct connection to I 2 S (or TDM) and powerful MOSFET output stages. $16 digikey.ca/products/en/integrated-circuits-ics/… \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 28 '19 at 22:56
  • \$\begingroup\$ @MarcusMüller, I checked many websites but had no luck. They are all BTL. It seems like nobody makes single ended devices. \$\endgroup\$ – zoonman Dec 29 '19 at 2:04

With rail to rail output, at 5V supply, we get +/- 2.5V peaks which means a non-BTL amp will only do 0.78W "RMS" into 4 ohms. So what you're asking isn't strictly possible without a voltage booster, but well, I'll assume 0.78W should be enough.

The type of jack connector you have tends to short when inserted or extracted, so you need a short circuit proof amplifier. TS4962 datasheet mentions it is not protected against shorts, so it wouldn't be suitable anyway.

I guess filter-free class D amps have to be BTL. If it was not BTL but had the speaker connected to GND instead then the speaker wires and voice coil would have a common mode of half the unfiltered output square wave, which would cause lots of emissions.

So you'll need a LC filter.

But you could get a class-D amp that is short circuit tolerant, like PAM8302 which is available on cheap pre-soldered modules. This one is filter-free, but nothing stops you from putting a filter on the output anyway.

If you put a capacitor in series with the speaker to block DC, I don't see why it wouldn't work with only one output connected.

If it still smokes with one output unconnected, you could try putting just the LC filter on the unconnected output, and maybe add a dummy load, as high resistor value as possible, between the outputs...

| improve this answer | |

An easy way out would be converting your BTL output to a single-ended output; that should actually be pretty benign at these power levels; use a transformer whose primary side is your output, and whose secondary side's one end is tied to output ground, and the other is your single-ended output signal.

You can also trade current capability for voltage, should your wiring / chassis favor that, by using a non-1:1 winding. But, since 4Ω is the design goal for many integrated amplifiers, 1:1 seems a sensible winding ratio.

Since this seems to be a probably relatively low-frequency radio application: do make sure that you have good output filtering for your class-D amplifier (you'll need class D, otherwise the 1W from 5V into 4Ω will be hard) or you'd risk EMI into bands of interest. It probably wouldn't hurt to figure out how to make sure the amplifier is disconnected when the microphone is used.

I was about to say "try the TPA3137D2", but after I had some trouble with the quality of the datasheet, I'd say maybe stick with a different vendor. Maxim, for example, or analog devices, or ST, or NXP...

| improve this answer | |
  • \$\begingroup\$ Do you mean like an old school lamp radio transformer? \$\endgroup\$ – zoonman Dec 30 '19 at 5:27
  • \$\begingroup\$ I don't know what a lamp radio transformer is, but I was thinking of a normal signal transformer. At the low frequencies we're working at (audio--> lower than 40 kHz), a lot of power transformers, e.g. for flyback SMPSes would work, too. \$\endgroup\$ – Marcus Müller Dec 30 '19 at 10:50
  • \$\begingroup\$ ah, is a "lamp radio" maybe a "tube radio"? In that case, yes. But: you don't need the high winding ratio that these needed; 1:1 or 1:2 would totally work. \$\endgroup\$ – Marcus Müller Dec 30 '19 at 10:51
  • \$\begingroup\$ Yes, tube radio. Transformer feels like an overkill for such application but it seems like I have no choice. \$\endgroup\$ – zoonman Dec 30 '19 at 15:28
  • \$\begingroup\$ well, you simply win roughly a factor of 2 in output power through BTL, so you either live with half the output power and go single-ended, or you implement some form of balun. \$\endgroup\$ – Marcus Müller Dec 30 '19 at 15:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.