Tripath hotrodding

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello all,

Anyone try this stuff yet on the small tripath amps...

Has anyone bypassed the internal opamp with a better one (like opa627)? It looks possible and fairly easy.

How about trading the output filter network for ferrite beads (Fair-Rite # 2512067007Y3) like Texas Instruments does on the tpa3001? http://focus.ti.com/lit/ds/symlink/tpa3001d1.pdf
See page 14

Is there anything better than Os-Cons for power supply decoupling on these amps?
Check out part #s 08J0805 and 20C2913 at newark.com for the lower voltage tripath chipsets

Sorry if these things have already been discussed.

PSz.
 
PSz. said:
Hello all,

Anyone try this stuff yet on the small tripath amps...

Has anyone bypassed the internal opamp with a better one (like opa627)? It looks possible and fairly easy.


The opamp is not bypassable from what I can see, at least on the TA2024. The chip only provides the pins for changing the feedback level on the opamp.

PSz. said:

How about trading the output filter network for ferrite beads (Fair-Rite # 2512067007Y3) like Texas Instruments does on the tpa3001? http://focus.ti.com/lit/ds/symlink/tpa3001d1.pdf
See page 14


The ferrite beads in the TI chip are acting as a filter for the EMI radiated by the device. The inductors on the tripath serve a different purpose.

PSz. said:

Is there anything better than Os-Cons for power supply decoupling on these amps?
Check out part #s 08J0805 and 20C2913 at newark.com for the lower voltage tripath chipsets

Sorry if these things have already been discussed.

PSz.

Not sure about Oscons even being the best. Most capacitor afficiandos only believe in using oscons on digital devices.. I think any low-esr cap will work well, whether Panasonic FM's, Elna Cerafines, Blackgates, Nichicon HE/UHE, or whatever floats your boat. Mainly the exotics aren't the best choice due to their size.
 
The opamp is bypassable. Set up a unity gain device (buffer, opamp, tube…) and use pins OAOUT1 and OAOUT2 as inputs. Remove Ri and Rf from board.:devilr:

I know why we have been blessed with the output inductors, and I was just rambling out loud. I think that using the ferrite beads without the LC combo would stand a chance of sounding pretty good while maintaining a little EMI respectability.
Kind of like taking a 1930s vintage Ford car chopping its top off, building up its motor, bigger tires, etc... but leaving a muffler on it.

A hotrod Tripath.
:cool:

I was thinking that since the t-amps are digital devices that oscons would be best. But I suppose you are right, it really is subjective.:)

PSz.
 
PSz. said:
The opamp is bypassable. Set up a unity gain device (buffer, opamp, tube…) and use pins OAOUT1 and OAOUT2 as inputs. Remove Ri and Rf from board.:devilr:


I suppose you could do this, but I don't know if Tripath has any compensation circuitry in there or not. I also don't know how the internal opamp would behave being run wide-open. I suppose nobody will know unless someone tries it.

PSz. said:

I know why we have been blessed with the output inductors, and I was just rambling out loud. I think that using the ferrite beads without the LC combo would stand a chance of sounding pretty good while maintaining a little EMI respectability.
Kind of like taking a 1930s vintage Ford car chopping its top off, building up its motor, bigger tires, etc... but leaving a muffler on it.


I'm not sure if the amp would even function without the inductors. The purpose of the inductors is to demodulate the switching frequency into the audio waveform.. The RF Beads wouldn't even come close to providing the inductance needed to do that.

I'm guessing that you have to think of it as sort of a switch-mode powersupply.


PSz. said:

I was thinking that since the t-amps are digital devices that oscons would be best. But I suppose you are right, it really is subjective.:)

PSz.

Pretty much. I'm sure an oscon might work well, especially since the Tripath is a lower voltage device (or at least the TA2024 is). One of the main problems with using OScons, from my understanding, is that it's difficult to get them at voltages higher than 20-25v.

Mike
 
Input op amp

I have been thinking of bypassing the input op amp in the TA2024 and have a few questions.

I wonder what would be the best thing to do with the internal op amps – leave the inputs floating, short the inputs together, tie them to ground through a resistor – what are your thoughts?

I notice that some of the Tripath products, for example the TA041, have the + input to the input op amp biased at 2.5V. I wonder why this is? Does the “Processing and Modulation” circuitry run at 5 volts and would like to see the analog signal at half of that. Of course the 2.5V is reflected back to the – input and to the input, thus the need for a capacitor on the input. But then the TA2024 data sheet shows a capacitor in its input and that chip is not shown biased at 2.5V. I wonder what is going on there?
 
Bypassing the opamp sounded like THE next step with these t-amps, but I just spoke with a tech rep at tripath (very helpful, didn't catch his name) and he told me that while the data sheets show a single opamp for the input stage there are actually two. The second converts the single ended signal to differential. I was told that all of their chips (single ended and diff output models) do this. Further, he advised against driving the OAOUT pins as we would be driving the output of the first opamp.

I am still intrigued by the idea of bypassing the first input stage, but I am probably not the one who could pull it off.

My thoughts had been to cut the analog 5 volt input and shut down the opamp. But with TWO in there...I dont know.

Any ideas?

PSz.:smash:
 
Hmmm, a second opamp? That complicates things because now you really don’t know what is going on in there. The combination of the two may work together to provide a desired result and bypassing one would upset something or other. That also screws up another idea that I had which was to force the opamp into class-A operation; but now who knows what that would do? Did the rep say anything about a 2.5V bias?
 
Re: Input op amp

David Davenport said:
I have been thinking of bypassing the input op amp in the TA2024 and have a few questions.

I wonder what would be the best thing to do with the internal op amps – leave the inputs floating, short the inputs together, tie them to ground through a resistor – what are your thoughts?

I notice that some of the Tripath products, for example the TA041, have the + input to the input op amp biased at 2.5V. I wonder why this is? Does the “Processing and Modulation” circuitry run at 5 volts and would like to see the analog signal at half of that. Of course the 2.5V is reflected back to the – input and to the input, thus the need for a capacitor on the input. But then the TA2024 data sheet shows a capacitor in its input and that chip is not shown biased at 2.5V. I wonder what is going on there?

The input is biased at 2.5 volts. This is needed to do a push pull output with a single rail supply. The output stage is bridged, neither side is grounded.
A transformer could do the same thing, but would cost more than a Tripath chip.
I continue to be amazed how a circuit with so many drawbacks and sonic no no's can sound. Guess there is still a lot to be learned about how to amplify voltages into real world loads.

George
 
The input is biased at 2.5 volts. This is needed to do a push pull output with a single rail supply. The output stage is bridged, neither side is grounded.

I suspected as much.

A transformer could do the same thing, but would cost more than a Tripath chip.

Sure, but then a transformer costs more than a tube. :)

I would not hesitate to use a transformer with these chips if I thought it would pay off with great sound.

Dave
 
Ok,
next crazy idea. How about using a tube amp output transformer INSTEAD of the standard output filter, and using the natural frequency roll off provided by the transformer for filtering. Has anyone tried it?

Oh, I just had another one (crazy idea)! How about just using a cap across the output for a 6dB/octave rolloff and do away with the inductors. We would still have filtering in place, just lower order.

Sorry guys, I have an uncommon, unexplainable urge to be rid of the inductors!

regards,
PSz.
 
PSz. said:
Ok,
next crazy idea. How about using a tube amp output transformer INSTEAD of the standard output filter, and using the natural frequency roll off provided by the transformer for filtering. Has anyone tried it?

regards,
PSz.


I believe some of the very early didgital amplifiers, called class D, in the '60 used regular output transformers just as you suggested.

I guess that the modern approach is less costly.

Dave
 
So, if some of the early ones used transformers, then it should work! If anyone wants to donate some trafos I will tear into my demo board and give a full report! :cool:

As far as I know the rolloff of transformers is first or second order (I don't know much about 'em), so it seems that my other crazy idea might work too! That is unless the reason why early ClassD didn't sound good was due to the filtering! Then again, ClassT is spread spectrum so it might do better.

any thoughts?

PSz.
 
Oh, I just had another one (crazy idea)! How about just using a cap across the output for a 6dB/octave rolloff and do away with the inductors. We would still have filtering in place, just lower order.

I was in the process of trying exactly that today, but after I took out the inductors and bridged the connections I managed to accidentally fry the chip when it was powered up before I could listen to how it sounded :cannotbe: .
 
David Davenport said:



I believe some of the very early didgital amplifiers, called class D, in the '60 used regular output transformers just as you suggested.


Dave


A bit of trivia: Who invented the digital amplifier? Extra points for when.

In his article “Updating Pulse Modulation” in issue one/1987 of Audio Amateur magazine, Norman Crowhurst wrote “In 1966 I was issued a patent for a form of pulse code modulation that promised to revolutionize audio...”

That was part one of a three-part series. Part two, “Developing High Quality Pulse Width Modulation” was published in the two/1987 issue of AA and I can’t find where the third part was published. I only surmise a third part because Crowhurst referred to it at the end of the second part. These articles were an update to original articles “The Two State Amplifier” written in the July and August 1965 issues of Radio-Electronics magazine. I have only the July part of that series.

In the February 1966 issue of Electronics World magazine, Don Lancaster wrote an article “Amplification Using Switching Techniques.”

Any good technical library should have the 1965 issues of Radio-Electronics and the 1966 issues of Electronic World. You could get the 1987 AA articles directly from Audio Amateur Publications who now publish audioXpress. They also should be able to tell you what happened to the third installment of that series.

Dave
 
I know this isn't anything as radical as you guys but how about as a first step to experimentation platform:

Has anybody tried to bypass output circuit of t-amp by removing on-board inductors and connecting air-cored inductors to output diode and then using own output circuit with caps to ground and zobel network if necessary?

Signal after output inductors no longer contains high freq waveforms, so tight circuit with smd parts not necessary?

would allow experimenting with different quality & value caps as well as eliminating zobel.

John
 
A few comments:

A capacitor is like a short circuit at high frequencies, and inductor is an high impedance device at these frequencies.
So, removing the inductor will fry the chip...

Using a Zobel is about the same as connecting the speaker directly to the chip without any filter.

A transformer with a Zobel may do the trick - but where is the advantage? The inductor will still be there...

Crowhurst:

I remember his original article (yes, I was alive by then), and the AA one.
I understand he passed away before writing part 3.
But it is a very interesting reading - if I recall correctly, he was more interested in power efficiency than quality (the transistors of the time were quite slow).
 
I got a different impression about hooking up with a first order (or no filter) from Tripath app note #4, bottom of page 3. They suggest that it is possible to hook up a speaker without any filter. AND that the quality of sound is indistinguishable from an amp with an ideal filter. This is what lead me to my crazy thoughts on the output filter...

Can you help me understand why the chip would blow up without the inductance? As opposed to a first order filter? Wouldn't a zobel that cut starting at say 50k hz. just cause the resister to get hot? And maybe the chip would actually require a heatsink...

:confused:

regards,
PSz.:)
 
PSz. said:

Can you help me understand why the chip would blow up without the inductance? As opposed to a first order filter? Wouldn't a zobel that cut starting at say 50k hz. just cause the resister to get hot? And maybe the chip would actually require a heatsink...

:confused:

regards,
PSz.:)

Why does one needs a filter in class D amps?
Because one may blow the tweeter, radiate lots of electro magnetic interference, etc. But indeed, it will operate without the filter (the speaker itself will act like a filter).

A first order filter using only an capacitor can blow the chip, as I've posted. A capacitor with say 8 ohms impedance at 20kHz will have 0.8 ohms at 200kHz and 0.2 ohms at 800k - about the switching frequency, so almost like a short circuit to the chip.

Using a Zobel or no filter will have almost the same effect, since a Zobel is not a filter - it's an impedance compensation circuit (from a given frequency up the impedance is constant, the resistor value)..
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.