Parallel TDA2052 vs Parallel LM3886 for subwoofer

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am thinking of parallelling 4 TDA2052 to drive an 8" sub kinda driver.
The driver has a 2 ohm nominal impedance so each chip actually 'sees' 8ohm.
The expected max power is 100 watts.
Given that ST mentions output of 28watts (@0.01 %distortion) with 25 volts supply into 8 Ohm load, total power of over 100watts is available.
http://www.st.com/stonline/products/literature/ds/1585.pdf


The following are the reasons to choose TDA2052 :-
a) I can then use a longer, slimmer (less depth) heatsink and distribute the chips along the length. This way the amps can be mounted outside, on the rear of sub cabinets without protruding much into the room.
b) 2052 is cheaper than 3886.
c) Slimmer heatsinks are cheaper than the ones required for LM3886 contributing to the cost difference even more.

What are the other pros and cons of not using proven chips like 3886 for this application?
Specifically, is it sonically inferior to a parallel LM3886.

Thanks in advance,
Goldy
 
a) With the same output power and the same load, you will need similar heatsink sizes. Sharing the heat across more ICs helps a bit, but the TDA is smaller, therefore can dissipate less heat per IC. You will end up with more or less the same heatsink for two LMs as for four TDAs.

The TDA2052 is sonically inferior to the LM3886, but in a subwoofer application that makes little difference. The human ear is not so sensitive in that frequency range, so the price could indeed be the decisive factor.
 
a) With the same output power and the same load, you will need similar heatsink sizes. Sharing the heat across more ICs helps a bit.

Yes, This is one of the benefits.


but the TDA is smaller, therefore can dissipate less heat per IC. You will end up with more or less the same heatsink for two LMs as for four TDAs.

Yes, this is another benefit. With 2 LM3886 chips (for 100 watts), I guess a deep heatsink, cubish shaped one, would be required.
With 4 TDAs, each one disspates less (25 watts each for same 100 watts total output) and a slimmer heatsink would do the job. Slimmer heatsinks are easier to find and cost less for the same weight.


The TDA2052 is sonically inferior to the LM3886, but in a subwoofer application that makes little difference. The human ear is not so sensitive in that frequency range, so the price could indeed be the decisive factor.

I will cross over to subs at about 200hz-250hz, use the system at moderate volume and have total of four 8" drivers to do the job.
Will the sonic compromise be noticable?
Thanks
 
Yes, this is another benefit. With 2 LM3886 chips (for 100 watts), I guess a deep heatsink, cubish shaped one, would be required.
With 4 TDAs, each one disspates less (25 watts each for same 100 watts total output) and a slimmer heatsink would do the job. Slimmer heatsinks are easier to find and cost less for the same weight.

It is actually the other way round. The heat dissipation is the same, but the TDA, being smaller cannot get rid of the heat as good as the LM. You will need pretty much the same heatsink for both configurations.

With four woofers it could be better to use one amplifier per woofer or one per two woofers. That way you don't need to match the feedback resistors to 0,1% and you can save the load sharing resistors, too.

Will the sonic compromise be noticable?
Thanks

Depends on the quality of the woofer. 200-250 Hz is quite high for a subwoofer. It will already reproduce human voices, which could reveal the shortcomings of the TDAs.
And you will be able to locate sounds coming from the subwoofer instead of from the main speakers. To avoid any locatability, the crossing frequency should be 80 Hz or less and the filter should be steep with 18 dB/octave or more. If the subwoofer is placed between the main speakers you can go up with the frequency to 120 Hz, even a bit more, if the woofer is at the exact center between the main speakers.
With frequencies as low as that, the difference between the ICs won't matter so much.
 
Moderator
Joined 2002
Paid Member
Two TDA 7294s will do the same job, and cost a lot less than the 3886. They sound acceptable, and are (were) readily available in your city at not very outrageous prices.

PB is right, the heatsink will be exactly the same for both. I use a locally available cheap extrusion (~Rs 100/6inches) for chip amps, and 6" is pretty much enough for a LM3886 parallel pair or a LM4780 running off 30V (which is about right for a driver that looks like 4 ohms). Higher voltages will mean larger heatsinks and the possibility of failure.

If you are paralleling 8 ohm drivers, it may make more sense to run individual drivers in series/parallel off single chips, you will eliminate the waste heat in the output resistor, as well as additional heat from the balancing currents. A TDA 7294 or LM3886 will run a 4 ohm load of two 8" 8 ohm drivers just fine.

If you are *not* planning an El-Pipe-O, I would just stick the money into a single large subwoofer driver. The Lab12 is available from the Chennai distributor, and is a pretty capable driver even for home use.

200Hz is too high for a sub. Even at 150Hz with the sub below my feet, it does not integrate correctly, but it's a gaming system so it's fine. I would think a maximum of 100Hz is where a sub should go to, maybe even 80Hz if the mains can take it.
 
Thanks for your inputs.

Depends on the quality of the woofer. 200-250 Hz is quite high for a subwoofer.

The woofers go quite well above 700hz. I plan to use plenty of them (2 per side, totalling four) for subwoofer duty.
They are very good designs, Xmax of 11 mm. Power handling of 100 watts RMS.


It will already reproduce human voices, which could reveal the shortcomings of the TDAs. Does'nt fourth order crossover help to get 250 Hz crossover.

I have a line of 2" mids that cannot go lower than 200hz-250hz. I really need to do a test to figure out the exact xover frequency.


And you will be able to locate sounds coming from the subwoofer instead of from the main speakers.

I have the woofers right next to front towers. I saw maggies being crossed to a woofer close to the ground at about 450Hz.

With frequencies as low as that, the difference between the ICs won't matter so much.

Do you mean to say, If I cross about 200hz, then TDA must not be used. If the difference is audible then I prefer LM3886.


Two TDA 7294s will do the same job, and cost a lot less than the 3886. They sound acceptable, and are (were) readily available in your city at not very outrageous prices.

My woofers are 2 ohms each. Though TDA 7294 is readily available as an assembled kit but cannot be paralleled:(


The Lab12 is available from the Chennai distributor.

I already have the woofers in my hand.

Even at 150Hz with the sub below my feet, it does not integrate correctly.
I have two boxes, each placed next to the fronts. Each box has two 8" woofers.

As mentioned above , I saw maggies being crossed to a woofer close to the ground at about 450Hz.
I added another woofer (on each side) just to negate the ground reflection problem.
 
Moderator
Joined 2002
Paid Member
Paralleling requires a few ballast resistors. It doesn't matter which amp it is, though for non-matched gains you will have some extra current being drawn at idle and the chips will run warmer than usual.

If you have 2 ohm woofers and 4 of them, I would run them as sets of 2 in series per channel, place the woofers near/under the mids and face them forward. This way you have a 4 ohm bass driver with a single TDA 7294/3886/xxxx. The other TDA option you are considering will also work this way.
 
Paralleling requires a few ballast resistors. It doesn't matter which amp it is, though for non-matched gains you will have some extra current being drawn at idle and the chips will run warmer than usual.
That is true for bipolar outputs, but MOSFET outputs self balance. You do need to synchronize the output drive signals, of course.

Here's a hybrid that would work well for your application:
http://focus.ti.com/docs/prod/folders/print/tas5630.html
You'll have to run it in single channel BTL for a 2 ohm load.
 
Moderator
Joined 2002
Paid Member
^^Thanks for that :) - I forgot the TDA 7294 uses a MOSFET construction too so maybe a ballast will not be required, but I suppose no harm could come of it...

That TI chip looks interesting. I hope goldy has the right resources to get a PCB made for it. I certainly don't know anybody in the whole country who would do a one-off for this chip :)
 
I have a line of 2" mids that cannot go lower than 200hz-250hz. I really need to do a test to figure out the exact xover frequency.

A line array? Can be tricky to integrate with a subwoofer, due to the different energy distribution.


Do you mean to say, If I cross about 200hz, then TDA must not be used. If the difference is audible then I prefer LM3886.

There will be an audible difference. The question you want answered is, if it matters. That depends on the speakers you employ, the music you listen to and your taste. Nobody can take that decision for you. The LM3886 and TDA729x are better than the TDA2052 in every aspect. You can try the TDA2052 and see, if it fulfills your expectations, and replace it, if it doesn't or you can use the LM3886 or TDA729x to begin with.

The class-D amplifier star882 proposed is a totally different thing. It is efficient, powerful and still small, but you should not even consider to build something like it without an extensive park of measuring equipment, beginning with an oscilloscope and some profound experience.


My woofers are 2 ohms each. Though TDA 7294 is readily available as an assembled kit but cannot be paralleled:(
Yes, it can be paralleled. This is a PCB with four parallel TDA7293. The only difference to the TDA7294 is a higher rail voltage rating.
 
I hope goldy has the right resources to get a PCB made for it.
Class D is too complicated. Not an option for me.


A line array? Can be tricky to integrate with a subwoofer, due to the different energy distribution.
My subs (actually 2 woofers per side) are right next to the mains. Is it still a problem. Several Martin logins are crossed ~ 450 with a woofer at the bottom ?
Also I have all the drivers with me so not much can change now :(


You can try the TDA2052 and see, if it fulfills your expectations, and replace it.
Clear so, TDA 2052 is out. What about LM1875? I am looking at several chips each sharing the power dissipation, rather than big chips with huge dissipation (for the same total).

I need 100watts @ 2 ohms or 200 watts @ 1 ohm. Am I out of options given my requirements?
 
Moderator
Joined 2002
Paid Member
At such low impedances chips are not the way to go. They simply can't handle it - even paralleled, I wouldn't put my faith into the weak output stages. If you are not trying to wire the woofers in series (which would solve a lot of problems with a 4 ohm impedance), Class D is the only thing that will work out economically for you.

200 watts at 1 ohm is a lot of current. Good luck finding chips that can do it. And specially tinny ones like the LM1875 - that will fry at 3 amperes, so you will need 6 of them at least, maybe seven.

Again, consider wiring your '2 ohm' (are these DVC woofers with two 4 ohm coils?) drivers in a manner that you can do with the meager current supplies of a chipamp, if you wish to continue pursuing the chipamp method.

I had two 3886 chips driving a full range 4 ohm load, and it was no easy time. With only a LF load and even if you parallel 4 chips (3 chips are not 2 ohm stable), the 3886 will be under quite a bit of stress. And that is about the highest current you can get from a package that size.

You can find suitable Class D and discrete modules on a few members' pages, and there are the pricey Hypex modules as well. I don't know if this is going to work out.

Good Luck.
 
Why did you buy 2ohm drivers?

4ohm and 8ohm are the standard driver impedances.
8ohm suits both discrete and chip amplifiers.
4ohm suits discrete amplifiers.

If these are dual voice coil drivers, then wire the voice coils in series.
Now wire two drivers in series. What is the resulting impedance for the pair of drivers?
 
Again, consider wiring your '2 ohm'.

The reason to not wire in series is that somewhere I read that drivers in series degrade sonic performace. I feel I have a good set of drivers and need to make the best compromise I can, so I thought of using each with a set of chipamps adapted for the 2ohm impedance. Is it the correct way, given the drawback of series wiring?

I had two 3886 chips driving a full range 4 ohm load, and it was no easy time. With only a LF load and even if you parallel 4 chips the 3886 will be under quite a bit of stress.

A single 3886 can provide the necessary current for 50 watts rms into 4 ohm, now when we reduce the impedance to 2 ohms and double the power then we need to double the current too (for 100watts this time). Another parellel amp can provide the same. Each 3886 sees only 4 ohm impedance and provides the same current as if it had a 4 ohm load, so whats the problem? I am assuming good heatsinks and sufficient power supply and caps.

Why did you buy 2ohm drivers?

The driver were custom-made with 2 ohm. 2 ohm was chosen to keep Le and Qe down.
 
Moderator
Joined 2002
Paid Member
A single 3886 can provide the necessary current for 50 watts rms into 4 ohm

In practice, Spike operates much sooner than one would think. I've heard audible clipping when running 4 ohm speakers off a single 3886, and that was full range. It took two of them to achieve stable operation at maximum output (Dynaudio speakers). In subwoofer duty usually a larger amount of power is required to cope with transients, so you will be operating near the chip's limits at all times, and exceed them fairly often. Note that it takes 3x3886s to be fully 3 ohm stable, I wouldn't go any less.

Drivers in series do not degrade performance but they do not add up output the way parallel drivers do. In your case, it makes perfect sense to do so as your impedance does not bode well for chipamps (though is fine for some high-power, high-current discrete designs).

I think everyone here is trying to give you two options:

1. Don't use chipamps

OR

2. Give the chipamps something they can work with. 2 ohms is not.
 
The reason to not wire in series is that somewhere I read that drivers in series degrade sonic performance.


A single 3886 can provide the necessary current for 50 watts rms into 4 ohm,


The driver were custom-made with 2 ohm. 2 ohm was chosen to keep Le and Qe down.
series connected drivers do not have a performance disadvantage.

A 3886 can supply 50W into 4r0 (pure resistance).
A 3886 can supply ~50W into a moderate reactance 4ohm speaker.
A 3886 cannot supply 50W into a moderate to severe reactance 4ohm speaker.

Who determined that Le and Qe should be kept at those low levels and who chose 2ohm as the only way to meet that specification?
 
Drivers in series ..................... do not add up output the way parallel drivers do.

1. Don't use chipamps

OR

2. Give the chipamps something they can work with. 2 ohms is not.
Drivers in series do add up when the wavelength is greater than the spacing of the drivers times two. You get +3dB increase in efficiency when two drivers are driven together that are located near each other.
You get a further 3dB when each speaker is driven to the same reference power. i.e one driver with 50W is 3dB quieter than two well spaced speakers each with 50W and 6dB quieter than two close spaced speakers each with 50W.

I agree on advice 1. and 2. I would go further and say that a 4ohm speaker is not a good load for a chipamp that is to be driven hard. 8ohm is much better.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.