diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Class D (http://www.diyaudio.com/forums/class-d/)
-   -   Tripath TA2024/20 Bridge/Parallel Config (http://www.diyaudio.com/forums/class-d/67160-tripath-ta2024-20-bridge-parallel-config.html)

john65b 2nd November 2005 04:16 PM

Tripath TA2024/20 Bridge/Parallel Config
 
I know the topic of Bridging the Tripath TA2020 & TA2024 has been covered ad nauseum here and elsewhere (can't be done), but there is no talk of Paralleling them. The chips are already Bridged and cannot be rebridged, understood. But Paralleling them should be possible to get almost double the power...also reducing the "work" required by each module (not that heat is an issue with these cool chipsets)

I can try Paralleling two TA2020 boards, as I have a few I can afford to mess with, but wondering if anyone has already attempted it?

I know the gainclone chips - LM3886, LM3875 - can be bridged(they are not already Bridged like the Tripaths), or paralleled, or both to double and quadruple the output power.

Any thoughts before I destroy two perfectly fine Tripath boards?

BWRX 2nd November 2005 06:23 PM

Interesting idea, but you wouldn't really get double the power unless you halved the load impedance. Paralleling them would keep the same output voltage swing but would be doubling (theoretically) the current capability. It would certainly help with some 4 ohm nominal drivers or even lower impedances.

Then there's the issue of where to parallel them: before or after the output filters? Since these amps don't seem to have a set modulation frequency, it would seem that you'd have to parallel them after the output filters. If you could somehow get them to synchronize their modulation frequencies with each other you could theoretically parallel the outputs before the filters and just use one output filter. Also, it might be helpful to add low impedance (maybe 1 ohm) non-inductive resistors in series with all outputs (both sides of each channel because the outputs are already bridged) to help ensure current sharing and to "cope" with the different DC offsets (each channel will surely have a different DC offset).

john65b 2nd November 2005 07:17 PM

Yep...
 
On a parallel arangement, the Output should not double (maybe 1.75x or so), but bridging should be very close to doubling (which we all know we can't do with the Tripaths)

The LM3886 Parallel schematic recommends a 3W .1 ohm resistor on each output before tieing them so the output channels don't fight each other.

Think I'll fry my boards if I try?

BWRX 3rd November 2005 02:21 AM

Re: Yep...
 
Quote:

Originally posted by john65b
The LM3886 Parallel schematic recommends a 3W .1 ohm resistor on each output before tieing them so the output channels don't fight each other.

Think I'll fry my boards if I try?

I actually meant to say .1 ohm in my above post :o but you could probably go a little higher (maybe even .5 ohm) if you had to. 2W would be a good safe rating for these power levels, but you could use lower rated ones. Twisting a bunch of Don't forget that you'd need 4 of these resistors, one for each side of the channel outputs.

I don't think you'll fry the boards if you try because the amps are fairly robust, but you never know. Regardless, definitely use a "disposable" speaker for testing! What load impedance would you use? 4 ohm nominal speaker? 4 ohm power resistor? If you do decide to go ahead with this please let us know the outcome! :cool:

john65b 3rd November 2005 03:25 AM

4 Ohm speaker
 
Yes, I will be using a 4 ohm speaker that I made a few years back. I will test with a 4 ohm resistor and check DC offset first, but I do not expect anything major.

Was going to try tonight, but got in a bit too late...tomorrow for sure. I'll keep you posted

Pano 3rd November 2005 08:31 AM

Brian's right about the chips being robust. I've thought several times that I killed one, but no - it lived. Even with 12 volts going into the right speaker wires. :xeye: Still alive!

Let us kow how it sounds. They should be combined post filter, as they are not synchronized (as mentioned above).

But what would be gained in a 4 ohm system? Lower output impedance is all, really, yeah? More current. Let us know if it works.

BWRX 3rd November 2005 03:26 PM

Quote:

Originally posted by panomaniac
I've thought several times that I killed one, but no - it lived. Even with 12 volts going into the right speaker wires. :xeye: Still alive!
I thought I might have been the only one to do this!!! I too was amazed the little guy fired right up after reconnecting the power wires properly :spin:

john65b 4th November 2005 03:29 AM

Done
 
2 Attachment(s)
Well, got all the soldering done and ran the TA202's in parallel. It didn't blow up. It did sound a bit louder...20 wpc to perhaps a 35 wpc??

The boards I have have six independent channels and I paralleled four of the six...see pic below. The board also has two mini-pots to control gain on each channel. I noticed if one channel gain was lowered at high volumes, that would take that pair (the TA2020 chip) into fault mode, while the other 2 channels remained operating, and needed to push the mute button to reset the fault. Must have been a reverse voltage trip from the overpowering of the other 2 channels.

Anyway, it sounded just OK. Seems that the sound wasn't as precise as regular two channels straight up. But this could have been my power supply gasping for more amps - as I noticed the individual chanel LEDs would flicker on deep bass notes. Also could have been the .1 ohm 5 watt wirewound resistor (all I could find) on the four + outputs. The negative outputs were tied without resistors, as I don't think it was required. Maybe they are needed....

Anyway, if anything this is encouraging. I still beleive this can be done corrctly to increase the power of these chips by 50% - 75%.

Anyone else want to try and post results? I don't think it will harm your amp. If the TA2024 faults out, you should be able to just repower the unit up to reset the fault...

I have 10 1 ohm- 1 Watt resistors I can try on all positive and negative channel outs too. Maybe tomorrow...

I will repost the results.

Later

BWRX 4th November 2005 04:23 AM

Re: Done
 
Quote:

Originally posted by john65b
The negative outputs were tied without resistors, as I don't think it was required. Maybe they are needed....
They are needed because of the differential outputs :att'n: I tried to explain that in my posts above but guess I didn't make it as clear as I thought!


Quote:

Originally posted by john65b
Anyone else want to try and post results?
I would love to try this, but I don't have two spare amps to test with or the appropriate resistors.
Thanks for sharing what you did so far! I bet it will sound better with equal resistors on ALL outputs - positive and negative.

john65b 4th November 2005 05:27 AM

Will re-try
 
BWRX --> You said four resistors in your previous post, and from the LM3886 datahseet outlining the parallel mode, you only need to put the four resistors on the four positive outputs and the neg outputs go to ground resistorless ...

I will try again with all eight leads resistored. Looks like I will have to do the 1 ohm 1 watt since I am 2 resistors short of the 8 required at .5 ohm 5 watts......

Think the wire wound resistors vs the Carbon types would lead to a major sound difference? After all, carbon types are used in the amps and the wire wounds are used in the crossovers, so tthey are both audio worthy...

I hope it sounds quite a bit better, as I was disappointed in the test sound quality...

More tomorrow...


All times are GMT. The time now is 10:02 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2