Paralleling the two stereo outputs of a TA2024 power amp to form one mono output

[microbug]

I bought a TA2024 amp from eBay for my boombox project. At the time I didn't realise that it can't be bridged, so I am now looking to put the outputs in parallel. I only have one speaker which is rated at 30W nominal so this should be fine.

I found this thread on diyaudio.com which seemed to refer to putting multiple chips in parallel. Ideally I would like just to use the board I have. It also says that in the LM3886 datasheet there is an outline of a "parallel mode" but I can't find one. I'm about to make a Farnell order soon so I will need to know what power resistors to order if this is possible.

Thanks!

[Holograph]

The size of the sharing (output) resistors necessary will depend largely on the output power (up to 30W I guess, in your case) and the accuracy and precision of the amp. The lower the voltage differential across channels, the lower your sharing resistor could be. What you need could range from, say, 0.05 ohms to 0.50 ohms. The larger the sharing resistors, the less current between channels that will be induced by the voltage differential, but obviously at the cost of a lower efficiency (and damping factor - but that's a different can of worms - some people think damping factor is important and some do not).

[microbug]

I assume the voltage differential is simply a matter of the amp's accuracy - the left and right channels are combined using two 10k resistors and although the mono input currently only goes to the right channel, it's easy enough to hook it up to both. I'm not too familiar with audio – what should I be looking for in the datasheet (here)? From the other forum that I linked to I'm estimating that around 0.2 ohms should be about right for the sharing resistors.

Thanks for the info; as I said I don't have much experience with audio so extra info is always helpful. I looked at the datasheet to see the block diagram and I see what you mean!

Thanks!

EDIT: Yes, the output power is about 30W although in practice maybe a bit less; the TA2024 is rated for 10% THD+N at 15W (x2). I'm no audio nut but that doesn't sound too good.

[Holograph]

0.2 ohms would likely do the job easily and even 0.1 could possibly be fine. Personally I might play around with the chip a bit and measure the differences in gain/offset voltages under varying circumstances. That's something I tend to do since I don't do volume production of anything. So I like to verify how my specific chip works, just make sure there's not a large output voltage differential, and use that information when calculating the sharing resistor values.

Basically you just want the maximum power transfer between amps/channels due to that differential to be minimal compared to both the power handling of the chip(s) (so you don't burn them up or run them into protection such that they shut down) and also the typical output power (mainly for efficiency).

[edavid]

I'm not sure if the other people posting realize that this is a class D (class T, whatever) amp.

Tripath is so vague about the way their chips work, that it's hard to say what will happen.  You may just have to try it and see if it blows up.

It also may not work well without some additional filter inductance, since cheap eBay boards usually skimp on that.

Have you tried hooking up just one channel to see if it's loud enough?

[Holograph]

I'm not sure if the other people posting realize that this is a class D (class T, whatever) amp.

Tripath is so vague about the way their chips work, that it's hard to say what will happen.  You may just have to try it and see if it blows up.

It also may not work well without some additional filter inductance, since cheap eBay boards usually skimp on that.

Have you tried hooking up just one channel to see if it's loud enough?

I do realize that it's Tripath; I've got some experience with it but far from a ton. I usually deal with simpler Class AB stuff yes.

It is possible that the output filter will need to be modified, but I'm going to say that is unlikely. It should be fine and comparable to running 16 ohm speakers on a single amp. Most people don't run 8 ohm speakers but it's rare to find an amp that won't. Tripath IS more sensitive to output impedance than some other Class D amps, though. Personally I prefer Philips/Hypex UcD.

Edit: Now, if you also mean that it COULD be modified INSTEAD of using a resistor, that I am not really sure of. I suppose you could probably increase it somewhat, but if that's a better option, I am not sure offhand.

EDIT: Yes, the output power is about 30W although in practice maybe a bit less; the TA2024 is rated for 10% THD+N at 15W (x2). I'm no audio nut but that doesn't sound too good.

I don't know what the purpose of this project is, but if it is something in which you desire good quality, you want to aim for definitely under 1% THD, and ideally more like 0.1%. High-end audio gear is often much lower than even 0.1% (sometimes by another order of magnitude!), but it's not THAT important at those levels. 10%, though, is absolutely terrible. Tripath chips are not going to compete with high-end Class AB amps on THD, in my opinion, though they are plenty good when the circuit is properly designed. They can be prone to more offset voltage than some other amps, though.

[microbug]

I am well aware that this is a Class-D (Class-T is marketing bullshit) amp.

I will order the appropriate power resistors for testing the amp (my speaker is 4 ohms). Offset voltage is rated between 50 and 150 mV.

I have hooked it up and while it's fairly loud, it would be nice to have a little more volume given that I specifically bought a 30W speaker. I will probably need to add some extra parts to make it up to Farnell's free postage (Add £3 to your order and get next postage worth £3.95 free) so I'll investigate some better inductors and replace the undoubtedly fake Nichicon caps (and the pile-of-crap "HuaHong") with genuine ones. I was considering doing this anyway,  given the likelihood that a board costing £6.28 incl. shipping wouldn't be using high-spec parts.

[retrolefty]

Not sure my audio amp theory is too rusty or not, but I don't think you will gain any increase in output power even paralleling the amp channels unless you also decrease the load impedance of the speaker load. The paralleling of the channel outputs doesn't increase the voltage swing, but could increase the total current drive available.

[Holograph]

it would be nice to have a little more volume given that I specifically bought a 30W speaker.

FYI, there is MUCH more to a speaker than just the size and power you run through it. One extremely important rating is the sensitivity or efficiency (they are not the exact same thing but they are extremely similar and directly proportional) which basically tells you how much sound to expect from a given power. But this can be measured in various types of boxes or can be free-air or on an open baffle mount, etc. Best to look at the frequency response graph and make sure it meets your expectations. Also keep in mind sensitivity is often specified as either at 1W/1m (measured 1 meter away with 1 watt running to the speaker) or 2.83V/1m (which is 1 watt at 8 ohms, but keep in mind your driver has varying impedance throughout its frequency range so it won't be exactly 1 watt and will vary with frequency). Depending on what you're doing with the data, it can be more convenient to look at 1W or 2.83V ratings. Crossover design, for example, will use the 2.83V/1m method. You, however, will probably want 1W/1m measurements.

The box design you use also makes a HUGE difference in frequency response, sensitivity, output power handling, AND sound quality.

There is also a thing in speaker design called Hoffman's Iron Law. (If you know the audio brand KLH (which is junk now but they used to be good and they discovered new technology), Hoffman was the H.) The simplest explanation is: enclosure volume, system efficiency, bass extension - you must give at least one of these things up.

I've built way more speakers (systems, not raw drivers) than amps, heh.

[Holograph]

Not sure my audio amp theory is too rusty or not, but I don't think you will gain any increase in output power even paralleling the amp channels unless you also decrease the load impedance of the speaker load. The paralleling of the channel outputs doesn't increase the voltage swing, but could increase the total current drive available.

Yes, this is correct, but it has other potential benefits like lower distortion and slightly better temperature characteristics inside the amp. I say "slightly" because both channels are in one chip so you won't gain that much uniformity or lose that much thermal resistance, but if it were separate chips it would affect that even more. Obviously the distortion matters far more, but temperature fluctuations can affect the voltage differential between channels somewhat, so it's good to have that under control even more so if you're paralleling amps.

Whether OP needs them to be parallel or not, I am not sure. Would need more info about the system as a whole, like which speaker it is in specific. But keep in mind, as discussed earlier in the thread, Tripath gives some of their power ratings at 10% THD, which is terrible and really an unacceptable way for them to rate power.

[edavid]

It is possible that the output filter will need to be modified, but I'm going to say that is unlikely.

Well, I'm just wondering if there will be shoot-through current from one output to the other.

[microbug]

Since you asked:

The speaker is a Pioneer TS-G1021i 10cm dual-cone speaker. Specs:
Maximum input power: 180 W
Nominal input power: 30 W
Impedance: 4 Ohms
Frequency response: 45 - 26,000 Hz
Sensitivity (1W/1m): 88 dB
Woofer size: 10 cm
Woofer material: IMPP cone
Cut-out Hole: 106,0 mm
Mounting Depth: 43,0 mm

Amplifier is this one.

Mounting in a large tic-tac box with the amp, Li-Poly battery and various other bits & bobs. See the pictures!

[Mark Hennessy]

Not sure my audio amp theory is too rusty or not, but I don't think you will gain any increase in output power even paralleling the amp channels unless you also decrease the load impedance of the speaker load. The paralleling of the channel outputs doesn't increase the voltage swing, but could increase the total current drive available.

Indeed, this is correct.

The TA2024 is happy to work into 4 ohms, so putting them in parallel - if that were even remotely a good idea - would gain nothing. Forget about it...

Putting a pair of LM3886s in parallel is another thing altogether. Very easy, and it works, especially if you then go on to bridge the combination (as in a bridged setup, each half "sees" half the impedance). I've done it myself. But there is no "voodoo" with these class B amps.

The TA2024 uses bridged outputs, so each hot and cold output would need to be connected together. Get that wrong and they fight. To be honest, as the switching scheme isn't discussed in the datasheet - so we don't know the phasing of the PWM relative to each half - you might end up with a fight at HF anyway.

BTW, the THD+N is much better than 10%. It's common for chip manufacturers to quote the power output at 10% THD, but below these powers it improves dramatically. There is a graph on page 1 of the datasheet.

Are you are aware that loudspeaker power ratings - unless quoted as "continuous average sine wave power" - are wild guesses, based on an optimistic view of the peak to mean ratio of typical programme material. So the 30 watt rating of your loudspeaker (at unspecified impedance and unspecified frequencies) is pretty meaningless. Feeding it with a 15 watt amp could be described as "sensible engineering".

Are you are also aware that the difference between 15 watts and 30 watts is only 3dB? The human hearing system can only just about hear a change of 1dB, providing you have a reference, and 3dB is actually a tiny change.

If you want to do something meaningful with a stereo IC, use the other half to drive a tweeter. Design a low-level crossover to split the HF and LF at line level. Hey presto - you have an active loudspeaker - very common in professional setups. The power loss in a passive crossover can easily exceed the 3dB you're worrying about if you properly equalise for "baffle-step" and other things...

[Holograph]

Okay, that's a decent enough speaker. I don't know if you noticed but I've modified my post explaining speakers a few times since I wrote it to add further info, should it interest you.

The sensitivity is not super great, but you won't get great sensitivity at that size of driver (driver = "raw speaker," as opposed to a complete speaker system), so that is to be expected. The polypropylene cone will be relatively weather resistant.

And yes, you will definitely want to parallel the amps since it's a 4 ohm speaker and not 8 ohm. For some reason I had the idea that you had said 8 ohm, but I reread your posts and apparently that's just something I made up in my own head 

Is the box ported, or are you sealing it up for mobile/outdoor usage? If you port it and you tune the port reasonably well, you should get a few extra decibels of sensitivity in the bass and lower midrange. Well, for most speakers, anyway. Some speakers are designed only for sealed usage and don't do well with a port (the Qts rating can be used to roughly judge whether to port a speaker or not), but car door speakers are not designed like that.

[Holograph]

BTW, the THD+N is much better than 10%. It's common for chip manufacturers to quote the power output at 10% THD, but below these powers it improves dramatically. There is a graph on page 1 of the datasheet.

I'm going to say you're wrong to assume that he'll simply be happy with half the wattage of his original idea. 3dB is not often considered a LOT, but it is not insignificant at all. He specifically said he was unhappy with it. Your idea of changing his expectations is not your decision; you should stick to answers to what HE wants (assuming it is possible, which it is), not what you want.

Plenty of people have done this with Tripath chips. It can work - he just has to be careful about more things. Is it the BEST solution? No. I'm not a huge fan of Tripath in general. But he already has it, and this has already been done by others as mentioned in his original link and in other places.

[Mark Hennessy]

Since you asked:

The speaker is a Pioneer TS-G1021i 10cm dual-cone speaker. Specs:
Maximum input power: 180 W
Nominal input power: 30 W
Impedance: 4 Ohms
Frequency response: 45 - 26,000 Hz

Like I said: meaningless.

How big do you think that voice coil is? 25mm at best? Give me a 30 watt amplifier, and I'll destroy that within an hour.

And a frequency response with no limits is also meaningless. It might well make noise at 45Hz, but it could be 20 or 30 dB down relative to 200Hz. If they'd said 45Hz to 26kHz +/-3dB, that might actually convey something.

Don't let me put you off - I'm sure the thing will make a nice noise (I presume you've got the T-S parameters and have designed a box for it with the correct materials and tuning frequency?), but worrying about getting another 3dB from the amplifier is utterly immaterial. Not to mention impossible from a 12 volt battery without a step-up DC-DC converter. To get 30 watts into 4 ohms (nominal, as impedance varies enormously with frequency), you need 11 volts RMS. That's 31 volts peak to peak. See where I'm going?

[microbug]

Thanks for the big response – knowing this forum I wouldn't expect any less but nevertheless it still took me a few minutes to read all the responses. I did know that the THD+N was better than 10% at lower power: 0.1% @ 11W with a 4 ohm speaker.

It now seems that my options are either to:
– Put the two outputs in parallel and drive the speaker with both to get a fairly small gain in volume;
– Buy a tweeter (or two) and use the second output to drive it (them) separately. Having just looked briefly on the same website I bought my current speaker from, they appear to be rather expensive; my budget is max £15, maybe stretchable to £20. Any better sources? Also, how much would a tweeter improve the sound quality – if it would make a comparatively big difference I'll go with it.

I am going to take a stab in the dark and guess that 'porting' is drilling holes to allow air to escape so the speaker can function properly. If that is the case, I was not planning on doing it but then again I didn't know about it. I will be using this outdoors but only in mild weather and I wouldn't leave it out in the rain. How does one tune a hole (or am I barking up the wrong tree?).

EDIT: having just seen Mark's latest post, I'll go with the tweeter. And no, the box is just a plastic thing I saved for some project a while back, and I found a use for it – how much of a difference will the box make? Yes, I see where you're going!

EDIT 2: the lack of information / data sheet regarding the speaker is also annoying but having looked at similar ones when selecting it, it is fairly typical.

[fcb]

As a general rule, by the way, you can only bridge dual-supply amps. If an amp doesn't use dual supplies, it cannot be bridged. Or, well, it wouldn't be worth bridging. You could do other tricks like use transformers but those are too big and heavy and can contribute distortion etc.

Wrong.  Most bridge amplifiers use single rails - that's one of the beauties of bridge configuration.

But back onto the TA2024 parallel issue.
You won't get any increase in power (theoretically) as you aren't increasing the voltage swing across the speaker - you'd only benefit if the TA2024 can't supply the current required for the voltage swing (not likely to be a problem as you are using 8ohm speakers).

I'd also be surprised if the Class D (because that's what the are) enjoyed being paralleled, feedback design being non-trivial.  Try it and have a go, but you will probably need larger ballast resistors than suggested.

[Holograph]

Like I said: meaningless.

How big do you think that voice coil is? 25mm at best? Give me a 30 watt amplifier, and I'll destroy that within an hour.

And a frequency response with no limits is also meaningless. It might well make noise at 45Hz, but it could be 20 or 30 dB down relative to 200Hz. If they'd said 45Hz to 26kHz +/-3dB, that might actually convey something.

Don't let me put you off - I'm sure the thing will make a nice noise (I presume you've got the T-S parameters and have designed a box for it with the correct materials and tuning frequency?), but worrying about getting another 3dB from the amplifier is utterly immaterial. Not to mention impossible from a 12 volt battery without a step-up DC-DC converter. To get 30 watts into 4 ohms (nominal, as impedance varies enormously with frequency), you need 11 volts RMS. That's 31 volts peak to peak. See where I'm going?

It's certainly not a good list of specs - car speakers are often not listed with full T/S parameters, unfortunately. In fact, almost none are given, typically. But meaningless? Not totally.

The voice coil size matters and the main thing that affects is the highpass filter used (I do recommend a highpass filter, though it's technically not 100% necessary if you're careful enough). As a side note, if you do port the box, you MUST use a highpass filter, and it must be at the port tuning frequency or higher (with beefy speakers you can sometimes go under the port tuning frequency by a little but not here).

As for the 26kHz, I'll give that one a laugh. It might make sound there but it's going to be rollercoaster in the midrange and up since it's an inexpensive coaxial speaker.

As a general rule, by the way, you can only bridge dual-supply amps. If an amp doesn't use dual supplies, it cannot be bridged. Or, well, it wouldn't be worth bridging. You could do other tricks like use transformers but those are too big and heavy and can contribute distortion etc.

Wrong.  Most bridge amplifiers use single rails - that's one of the beauties of bridge configuration.

Ah. Guess it's just been the specific amps I've used that have happened to be like that. But since I've used various amps from various brands I took it to MEAN that. My bad. Thinking about it from the electronics perspective - and not what I've actually seen advertised - what you say does make sense. I've removed the offending statement from my previous post.

[microbug]

I knew there was lots to consider but never realised quite how much!

– @fcb - my speaker is 4 ohm, not 8.
– When you refer to a high pass filter, do you mean for a hypothetical tweeter, for the mid/high range cone or for the whole speaker?
– I looked up porting; I'll have a closer look later but I don't know if I'll have enough room in the enclosure judging by the photos I saw (first glance, could be completely wrong).
– Although it may not be a problem in the short term, I will probably install some better components in the output filter of the amp since I don't completely trust that the phoney caps currently there won't explode in my face.

[Holograph]

I knew there was lots to consider but never realised quite how much!

– @fcb - my speaker is 4 ohm, not 8.
– When you refer to a high pass filter, do you mean for a hypothetical tweeter, for the mid/high range cone or for the whole speaker?
– I looked up porting; I'll have a closer look later but I don't know if I'll have enough room in the enclosure judging by the photos I saw (first glance, could be completely wrong).
– Although it may not be a problem in the short term, I will probably install some better components in the output filter of the amp since I don't completely trust that the phoney caps currently there won't explode in my face.

In this specific case, the highpass filter I am referring to may also be called a "subsonic" filter, though that term is more appropriate with subwoofers. It cuts out the stuff your speaker won't be able to play because of the speaker's natural frequency response (or in the cases of beefy subwoofers, the limits of our hearing - which is technically the more accurate definition of a subsonic filter).

In most audio, most of your power is going to be in the bass frequencies. There's a great graph on sound.westhost.com (which is an EXCELLENT site that I learned a LOT from several years back) that gives an estimation of this. http://sound.westhost.com/tweeters.htm -- first chart. The way you read it is that at any given frequency, find the power level in % in the graph, and that - on average - is how much of the signal's power is contained in that frequency and below. You can see that it's biased towards bass. That means we need to cut some of that bass out so that we can run enough power in the midrange and treble while not exceeding the speaker's power or excursion limits. This is the highpass filter. With that speaker you will probably want it around 80Hz but you can play with the value a bit depending on what types of material you listen to and stuff.

As a side note, that graph also explains why biamping speakers (running the woofer on one amp and the tweeter/mid on another) doesn't really add much power handling. Your limit is nearly always your woofer, so unless a speaker has multiple woofers that can be run with multiple amps, you aren't going to be able to gain much output capability.

With a ported speaker you typically get more efficiency (most speakers work better ported but not all) but you get a suspension unwinding effect below the port tuning frequency (you tune the port by its diameter and length but it varies per speaker and with box size). This means that it's very easy to hit the excursion limit (Xmax for linear excursion (it'll add way more distortion above that value) or Xmech for actual physical clearance) below that frequency, which can destroy the driver just like running too much power and melting the voice coil can.

I'd normally recommend software like WinISD or Unibox to calculate the box, though it's going to be difficult if your driver doesn't have Thiele-Small (T/S) parameters available. Hopefully Pioneer gives some information about choosing the right enclosure for it. But they might not, since most people are just going to put that in a car door and be done with it.

One thing I had meant to explain in an earlier post but forgot (Mark's reply after this one reminded me). It is more efficient to get your sound from surface area** (edit: fixed typo) than stroke. Meaning larger speakers are typically more sensitive. The reason is mainly because as you increase excursion/stroke capability, you need to add more tolerance in the gap so the coil doesn't rub up against it. But the more tolerance you add, the weaker the magnetic forces become, lowering sensitivity. Also, as power handling increases, voice coil size and weight increases, which hurts high-frequency reproduction (which is a major deal with a coaxial speaker like this). But if you go with a larger speaker with similar Mms (moving mass - another T/S param) and Cms (compliance of the meahcnical suspension) but larger Sd (surface displacement of the cone - yet another T/S param) you will get more sensitivity and thus more volume.

[Mark Hennessy]

BTW, the THD+N is much better than 10%. It's common for chip manufacturers to quote the power output at 10% THD, but below these powers it improves dramatically. There is a graph on page 1 of the datasheet.

I'm going to say you're wrong to assume that he'll simply be happy with half the wattage of his original idea. 3dB is not often considered a LOT, but it is not insignificant at all. He specifically said he was unhappy with it. Your idea of changing his expectations is not your decision; you should stick to answers to what HE wants (assuming it is possible, which it is), not what you want.

It's not fair to accuse me of "changing his expectations" simply because I've pointed out the stark reality that his expectations are not achievable with the chip he has. According to the datasheet, 15 watts into 4 ohms is as good as it gets. Don't shoot the messenger.

And exactly how is 3dB a "LOT", especially without a reference? If it's not loud enough for him currently, he'll need rather more than 3dB, which is a barely noticeable change in reality. Hence my suggestion of a boost converter (and a different chip). Or a more efficient loudspeaker, if that's an option at this stage (probably not).

[microbug]

If I were trying to make a professional speaker system I'm sure I could do much better, but I'm making a Bluetooth boombox to listen to tunes in my room & the garden – it doesn't need to be studio quality and I don't have studio money. It's plenty loud as it is, I was only thinking that given that there's an unused output I might as well hook it up, but since it will barely make a difference I won't bother going down that route.

The TA2024 datasheet says it will take between 8.5V and 13.2V. My battery, being a 3S Li-Poly, will vary between 12.8V (fully charged), 11.1 (nominal) and 9V (depleted). Given the comments that paralleling will make little difference, I suppose it would make more of a difference to buy one of the many cheap adjustable boost converter boards from eBay and set it to 13V (and put in a low battery light - without the volume difference I wouldn't notice and depleting an Li-Poly too far can destroy the battery).

Given my budget I think I'll just work with what I have – I'll keep the amp & speaker, fix the rubbish output filter components on the amp and possibly buy a tweeter. I have space restrictions though; I just measured: 3cm max mounted surface diameter if the tweeter is on the front, 44.6mm if it's on the bottom (perpendicular to the ground if you lie down the box flat, but since I have ordered raised feet it will face the ground slightly. Will a tweeter be viable?

EDIT: Just realised I could have the tweeter on the top of the box. Max diameter would be about 48mm.

[Holograph]

BTW, the THD+N is much better than 10%. It's common for chip manufacturers to quote the power output at 10% THD, but below these powers it improves dramatically. There is a graph on page 1 of the datasheet.

I'm going to say you're wrong to assume that he'll simply be happy with half the wattage of his original idea. 3dB is not often considered a LOT, but it is not insignificant at all. He specifically said he was unhappy with it. Your idea of changing his expectations is not your decision; you should stick to answers to what HE wants (assuming it is possible, which it is), not what you want.

It's not fair to accuse me of "changing his expectations" simply because I've pointed out the stark reality that his expectations are not achievable with the chip he has. According to the datasheet, 15 watts into 4 ohms is as good as it gets. Don't shoot the messenger.

And exactly how is 3dB a "LOT", especially without a reference? If it's not loud enough for him currently, he'll need rather more than 3dB, which is a barely noticeable change in reality. Hence my suggestion of a boost converter (and a different chip). Or a more efficient loudspeaker, if that's an option at this stage (probably not).

I'm just saying he should try to use the whole chip before deciding, since he already has the thing. If he's REALLY not impressed with it then yes, a beefier amp or speaker is likely necessary. But I wouldn't say "Use only half your chip, or buy a better speaker." I don't think those are the only two valid options here.

I didn't call 3dB a lot - I said it was not a lot, but also not insignificant.

[microbug]

I'm going to guess that a boost converter putting the input voltage to the TA2024 at 13V will make more of a difference than paralleling the inputs since it will actually increase the voltage swings. If not, at least it will make the volume more stable while the battery is discharged.

I am not "REALLY" unimpressed with the speaker by any means, I just thought I might as well get the most out of the chip. However, it seems adding a tweeter would probably be a better option if that's my goal.