Class D vs Class AB

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
You are missing my point:
My personal opinion is that if you want to go the DIY route in this "power range" you are better off with the cheaper class A/B TDA7297, maybe 2 of these in parallel so you have one for each channel and better control with low impedance loads, use a 12v battery for PSU, and get a cmoy to use as a pre amp stage. You get a nice setup that sounds like a lot like more expensive stuff but on a toaster budget.

The TPA3116 is good, but in my opinion the TDA7297 is better, and much cheaper to build for one that really wants to DIY stuff. Not to mention easier to solder for a beginner that wants to make his/her own pcb.

Most of the TPA3116/3118 boards endorsed by people on this forum are, just as you say at least in part "Finished products from Asia."

If you really want to DIY it is still easier to go the class A or A/B route. Most people can solder a few through hole and maybe even some of the smallest smd components with a bit of practice and a keen eye, but to diy uBGA components and similar can be a bit expensive to get wrong.
 
Last edited:
TDA7297, maybe 2 of these in parallel so you have one for each channel and better control with low impedance loads, use a 12v battery for PSU, and get a cmoy to use as a pre amp stage. You get a nice setup that sounds like a lot like more expensive stuff but on a toaster budget.

KaffiMann, this sounds like a good idea. are there any issues with "load balancing" both sides of the TDA7297 ?
From memory there was some talk of the "geunine / retail" chips sounding better than the ones that come with the "bay" amplifiers.
 
Regarding ebay vs diy with parts from mouser/digikey/similar
I am not certain if the sound quality is due to the chips (possible 2nd/3rd grade sorting) or the often sub-par PCB's with tracks evaporating when in contact with solder iron, or low grade other parts on the boards. But I've had good stuff from Ebay, and absolute crap.

Due to low parts count needed you could also do a point to point setup with original TDA7297 and just forget about a PCB alltogether. There is really no need to parallel both outputs from one chip in normal use, but it has been done. With one wire to tie outputs, and another wire to tie inputs. The TDA7297 is internally limited to 2A, so it may have some merit pending on your intended application.

If nothing else, paralleling should at least improve crosstalk.
 
Last edited:
My personal opinion is that if you want to go the DIY route in this "power range" you are better off with the cheaper class A/B TDA7297, maybe 2 of these in parallel so you have one for each channel and better control with low impedance loads

I agree...this is a wonderful chip. I've never found a schematic to wire these in parallel. Is it as easy as running to same signal to pins 4 and 12, and running both pins 1 and 15 to driver + and both pins 2 and 14 to driver - ?

Any issues into an 8 ohm load, as I believe each output will "see" 1/4 of the driver load.

Thanks, Mike
 
No issues, you got your Ohm's Law wrong. Paralleling amplifiers doubles the current output, so you can now drive half the impedance. Bridging doubles the voltage output, and you will require double the impedance to remain within allowed current limits of the output devices.

Each channel of the TDA7297 is already made up of two amplifiers bridged internally, this is why you only have an 8 ohm rating. With both channels in each chip running in parallel you will be able to use a 4 ohm loudspeaker or two 8 ohm in parallel, but then you would be better off just using one output channel for each speaker, and only paralleling the inputs.

If you use parallel configuration, you will of course now need two chips for a stereo amplifier.
 
I am not sure it is that simple.

Class D: The manufacturers often describe in the datasheet that the outputs can be connected (paralleled) and then offer half the individual switch impedance. The output transistors are FETs driven into saturation and, as long as the two channels can be operated synchronously, the many small FETs of one output transistor can be connected in parallel with the many small FETs of the other. This is very convenient for driving lower impedance loads.

Class AB: The output transistors are kept in the linear range and may be bipolar types. The manufacturers very often do not mention paralleling. If we recall that the output of (higher power) class AB amplifiers often use more output transistors in parallel, we also know that emitter resistors are used to compensate for differences of each output transistor such that they are likely to share the load-current when operated by a common driver stage. If we just connect two class AB outputs (for paralleling), they may not share the load current because they are not designed for load-sharing and are operated from different driver stages. One output may start sourcing the other output and the individual current limiters may act on one output in disharmony with the other output. Therefore, do not a priori assume that you can just connect class AB amplifier outputs. The solution to the TDA7297 is to use a TDA7377 or STA540 (will accept 22V supply) instead as they have the double current capacity (4A) by design.
 
Last edited:
Nothing is ever simple if you start looking into the finer details.
"Simplicity" takes time and experience to perfect.

Not gonna argue about the chips and such, you got it right.
But I was merely pointing out that if we are talking low cost entry into DIY; an extreme low-cost alternative, that can actually sound really good. ca 2$ for two TDA7297 kits on ebay, including small cooling fin that is adequate, can very well be paralleled if your heart so desires.

The TDA7377 is actually cheaper to DIY from scratch than paralleling the smaller brother, when buying from mouser or similar, and on Ebay you pay 4$, so double price for one single board, for the cheapest TDA7377 kit but that is not including cooling fins, so either you must have some leftovers, or buy some bits of aluminum. The PCB seems to be of slightly better quality than the cheapest TDA7297 boards though, so it may be well worth the extra 2$.

The very cheapest TPA3116 I could find is also ca 2 USD for 1 board:
TPA3116D2 Dual Channel Mini Digital Amplifier Board D Class DC 12V-24V 2x50W | eBay
I will admit that I have no idea what is needed to "Optimalize" this, as I have not heard that particular board and it may very well sound fantastic out of the box, despite my scepticism based on previous experiences.
 
Last edited:
I would consider spending a little more and get a more complete implementation:
XH M189 TPA3116 2*50W digital amplifier board 2X50W TPA3116D2 two channel stereo amplifier board-in Replacement Parts from Consumer Electronics on Aliexpress.com | Alibaba Group
I have tried TDA7297 as well and it is amazing what you get for next to nothing. An ideal first project. The TPA3116D2 may be a next step. For 3.5EUR you can do some experiments and if you damage it, you can afford it. Very good sound for the modest price.
When you want something good, go for TPA325x but be careful with experiments as it may be 50EUR that smoke.

My impression is that many forget the power supply may cost the same or even more than the amplifier. TDA7297 may run from a 12V car battery. For TPA3116D2 use a laptop charger with a capacitor. A TPA325x deserves a good power supply that is not cheap.

One slight concern I have noticed with the TPA325x is that the input coupling capacitors are 10uF and therefore often not implemented with foil capacitors. Is there a way around that?
 
I thought this was a class d vs ab discussion.
You seem more interested in the argument than the topic. Which is fine, we are all different.

If you're suddenly on the tpa32xx family, you should compare them to some quality solid state circuits, or chip amps like lm3886 and similar. All of the different options have their strengths and weaknesses. I am a fan of the lm3875/lm3886/lm4780 family, but there is a myriad of equal and better solutions if you look at stuff like the various derivatives of the circlotron, Elliott's p3a, the list is endless.

Of course the psu is important! Why do you think many of the low power solutions rely on voltages like 12, 24, 36 and 48v? Because it's easier to implement them with battery banks, in case you need to design a solution that must be powered off-grid. But today we are moving away from this, and we are also littered with power supplies that work perfectly fine, power supplies are cheap now, and can often be had for nothing if the equipment it was intended for has been delegated to the trash bin. A little knowledge of power supplies and how to improve them goes a long way.

I do not see foil cap issue, do not have the opportunity to look into it right now.
 
Might be some of the difference between subjective opinion on performance difference favors the EVM module is because of the 10uF cap, a bit high have not seen any mention of it any other places than documents regarding the EVM module. Do not know what value the various manufacturers use for that component, but it's probably much lower.

I still do not understand the problem, if you do not wish to get some random smd cap, then why not something like the Wima MKS2B051001N00JS, it is 10uF and 5% tolerance for not a lot of money.
I have some Jantzen 33uF laying about that could be used, but that's a bit excessive methinks, want to save them for another 1st order xo project.

Edit:
Cheap Kemet PP cap with 5% tolerance and through hole is double price of that polyester Wima.
 
Last edited:
Are they (class D) showing any advantage in crossover distortion and bias stabilization ?
I sort of like that warm, somewhat tube like sound of the TDA7294/93 for music amps but you need to gang them to get an easy 300 watts or so. I haven't tried them in that power range. The elephant weight power 50/60hz transformer needed for that keeps it from vibrating off the top of the speaker cabinet.
The class d amps can be so light you push a button and they slide around. Very annoying.
 
I think the biggest difference with class AB and D is that many decent quality off the shelf class AB amps (but far from all of them), are specified with rated power at 0.1 or 0.01% THD. If you expect the same quality sound from a class D amp rated at the same power but 10% THD you are in for a disappointment.
The topologies are different, they behave differently, but try to achieve the same thing, both have pro's and cons. What you lack in crossover distortion in most class d amps you might expect to get back at some other point of the wave. If you expect 90watts from a TPA3255 you are in for a treat, if you expect 250watts from the same TPA3255 there will be distortion.

I have not heard the TDA7294/93.

Here is a write up on the general differences, you you want a short version summary chart, look after "One amplifier to rule them all" down below:
Audio Amplifier Classes (A, A/B, D, G, and H): What are the Differences? | Audioholics
 
Are they (class D) showing any advantage in crossover distortion and bias stabilization ?
I sort of like that warm, somewhat tube like sound of the TDA7294/93 for music amps but you need to gang them to get an easy 300 watts or so. I haven't tried them in that power range. The elephant weight power 50/60hz transformer needed for that keeps it from vibrating off the top of the speaker cabinet.
The class d amps can be so light you push a button and they slide around. Very annoying.

When I experimented a lot with class D it seemed that it tended to have pretty much trouble with crossover distortion. That was a long time ago, something like 30 years, when I noticed it particularly bad and converting a simple homemade class AB amp with just output biasing diodes gave a really bad crossover distortion sound. Possibly, the bipolar output transistors were getting stuck from saturation during switching on and off, and causing the magnification of the problem. Also I wasn't using feedback from the output if memory serves, and was using an injected triangle wave at the input stage that I was modulating with audio. I had learned about class D from a 1960s era copy of Electronics World magazine.
 
Last edited:
Might be some of the difference between subjective opinion on performance difference favors the EVM module is because of the 10uF cap, a bit high have not seen any mention of it any other places than documents regarding the EVM module. Do not know what value the various manufacturers use for that component, but it's probably much lower.

I still do not understand the problem, if you do not wish to get some random smd cap, then why not something like the Wima MKS2B051001N00JS, it is 10uF and 5% tolerance for not a lot of money.

Thank you for pointing me to the 10uF Wima capacitors. They are both decent in size and price.
I only looked in my storage and found some 2.2uF that were already pretty large, so, I concluded that 10uF would be too large to be used on a board.
I was wrong in that assumption.

In a later post (different thread it seems) you write that you called your wife in as second opinion on the TPA325x amplifier. That is a clever approach because wifes, mainly being concerned about that we do not leave our electronic stuff all over the place, normally do not have a prejudice on performance issues and see it much more from a layman's perspective. If she also agrees, in comparative test, that the TPA325x is better I am close to buy one myself. At least it will give me some basis for disturbing in the discussions about how much better it is than the other class D chips.

Thanks
 
Thank you for pointing me to the 10uF Wima capacitors. They are both decent in size and price.
I only looked in my storage and found some 2.2uF that were already pretty large, so, I concluded that 10uF would be too large to be used on a board.
I was wrong in that assumption. Thanks

I don't think I've ever seen electrolytic DC blocking caps. Is this common? I believe I read that if used, they should be 10 times the normal value, due to the wide tolerance ranges and the desire to keep the voltage across them low to extend their somewhat short "lifespan".

How hard is it to replace smt with through-the-hole components?

Mike
 
I don't think I've ever seen electrolytic DC blocking caps. Is this common?

How hard is it to replace smt with through-the-hole components?

Mike

Some 30 years ago it was very common to use electrolytic DC blocking caps both at outputs (2200uF-4700uF) and inputs. The electrolytic DC blocking capacitors absolutely posed limits to the performance of the amplifiers, in particular at the output.
Reliability is rather low, they are rather noisy, have important leakage currents etc. 10 times the value needed sounds a bit exaggerated to me - I would do twice.
There are special audio grade capacitors that perform reasonably well but not as good as foil capacitors.
Two examples with electrolytic input DC blocking capacitors:
MINI HIFI TPA3255 300W + 300W Class D audio Digital Amplifier Board DC50V -in Amplifier from Consumer Electronics on Aliexpress.com | Alibaba Group
(on each side of the RCA input connectors), and
Free Shipping TK2050 50W+50W Dual Channel Class T HIFI Stereo Audio Digital Amplifier Board-in Amplifier from Consumer Electronics on Aliexpress.com | Alibaba Group
(blue ones next to the three pin input connector)

Replacing SMD (ceramic) capacitors with through-hole foil capacitors is difficult because the SMD capacitors can be very small and the PCB pads holding the capacitors are very small as well. Therefore, making a soldering connection to a 0.5mm terminal wire of a through-hole capacitor when the PCB pad is only 2mm wide is difficult.
 
Last edited:
Looks like hash noise could form in a feedback amp trying to maintain zero volts. Some hunting about zero or very low output.

When I experimented a lot with class D it seemed that it tended to have pretty much trouble with crossover distortion. That was a long time ago, something like 30 years, when I noticed it particularly bad and converting a simple homemade class AB amp with just output biasing diodes gave a really bad crossover distortion sound. Possibly, the bipolar output transistors were getting stuck from saturation during switching on and off, and causing the magnification of the problem. Also I wasn't using feedback from the output if memory serves, and was using an injected triangle wave at the input stage that I was modulating with audio. I had learned about class D from a 1960s era copy of Electronics World magazine.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.