Class-T Audio Experiences

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Has anyone else attempted to build a class-T Tripath amplifier? I have a custom design (including boards which I implemented and tested and ran. The efficiency was unbelievably high when the amplifier was set to the correct idle frequency. The sound quality was good as well. The frequency response was flat over the entire audio band and above to the cutoff frequency of the output stage. Would anyone else like to comment on their positive and negative experience with Tripath amplifiers? Are they discontinuing the TA010X series or drivers?


Thank for the kind words regarding our products. I just wanted to pop in to let you know that we are not discontinuing the TA010x product line.

As you may know, the TA3020 product was developed to provide a more compact and cost-effective solution for applications requiring up to 300 Watts per channel (4 Ohm, single ended). The TA3020 basically consolidated the TA0102 and TA0103 market applications under one part. The TA0104 was retained as part of our active product line, as it provided higher power output than was achievable with the TA3020.

As you can see on our website, the TA0104 has been designated "end of life" as of October of 2002. However, we will soon announce another TA010x product which will have power output which starts where the TA3020 drops off and goes beyond what the TA0104 was capable of.

Specifications for the new part are not final, but we are disclosing the part to customers. A final data sheet should appear on our website "real soon now".

Thanks again for your interest,

Scott Bobo
Tripath Technology
...'Class T' amplifier chips be used to drive an extremly high voltage (say 2Kv) to direct drive an electrostatic speaker?
The modules can handle up to 100 volts. The output devices limits the module. With some creative thinking. The modules can handle higher voltages. However, the output devices needs special attention.

As with piezo elements, it is best to use professional audio grade transformers.
Tripath use what is probably the most interesting modulation scheme in the industry and were early to the game.

They are however behind in that they do not have a "direct from digital". They indicated several years ago that they were going to move in that direction, but alas this has not yet happened. It is possible that they will introduce products now that move them back towards the front runners -- I recommend a wait and see approach.

If you do move with Tripath, I suspect much gain can be had from improving the output filter.

The industry needs more competition in this area so I certainly hope Tripath gets moving with this critical piece of interface.


Anyone interested in building a power amp based on the Tripath TA0104 "evaluation board" can visit my website below to retrieve an article I wrote about the project. The article contains several tweaks and upgrades that bring the Tripath into the realm of high-end audio (but of course that's just my biased opinion):

Since I first made this article available two years ago, I have had nearly 100 inqueries from all over the world, with many going on to build their own. Most have written back that they were very satisfied with the results.

In addition, there was a thread earlier this year at DIYAudio by RJones (I think that was his name) outlining several additional tweaks and modifications, all of which are worth considering. A search on the Digital forum should enable you to find this.:)
A very brief follow-up:

John - Electro has got it basically correct. There are no MOSFET devices on the market which would allow us to directly drive an electrostatic loudspeaker. A good tranformer is your best friend.

Petter - Check our website from time to time this summer. :)

Chris - All of our evaluation boards are designed to show large commercial customers both how good our audio performance can be AND how inexpensive it can be to implement. Compromises are made on both sides, and we freely acknowledge that our eval boards are neither the highest-performance (although I think they're very good) nor the lowest cost. The evaluation boards are a sales tool, and so must meet the needs of the volume markets. Tripath is gratified that the DIY community has taken an interest in our evaluation boards and we are happy to serve their needs as best we can.

Scott Bobo
Tripath Technology

Thanks for the info - the issue here is that a transformer limits the bandwidth, and I can't seem to find a transformer at a reasonable cost that will accomodate the high voltages required. A full range (minus subwoofer) according to Wagner goes to as high as 6-8Kv, most step up transformers seem to be rated at 2Kv.

I'm thinking though of just using something like your amp and doing say several hundred volts - still requiring a transformer, but I could use less turns and/or higher bandwidth.

If I had a dual voltage setup where the output devices had a higher voltage rail, would your devices work? Would slew rate be an issue?
One of the issues which I had with the execution of the class-T design (TA0102A) was achieving a stable setting of the idle frequency using the feedback delay described in your Performance Enhancement Application Note. The resistor should be between 500 ohms and 1k however. The idle frequency was not in a normal range with minimized overshoot unless this resistor was a much higher value. The capacitor was 180pF for about 33 to 34V rail implementation. Is this an atypical result? Secondly, what is a typical cutoff frequency chosen for the second stage output filter. I implemented the second filter at very close the the first stage for the the best results utilizing a 2.54uH and 1uF capacitor (a later attempt used a 0.1uF). Also, when will the TA3020 be released for sale or is it available now?

John - There are a couple of possibilities for your ESL drive. Please contact me directly and we can discuss.

Slew rate is kind of an iffy spec where switching amplifiers are concerned. The step response of our driver products (TA3020, TA010x) is around 10kV/uSec. When used with typical output MOSFETs, the overall system is in the 3kV/usec range. If those seem like big numbers, remember that on/off time is critical to the basic operation of a switching amplifier. Comparing those sort of numbers to linear amplifier slew rates is not all that informative, but no, you will not be slew rate limited with our topology.

BeanZ - There are a number of things which will impact the value of Rnew (as referenced in the app note) when trying to settle in at 650kHz. MOSFET selection, load, rail voltage and circuit layout will all impact the final value of Rnew. The 500-1k range is an estimate and will cover 90% of the cases, but not all. As long as your final Rnew value puts you somewhere in the 600-700kHz range at idle (and at idle temperature), then you're fine.

Tuning the Zobel to 80kHz (our primary LC filter) is correct, as you noted. It's only function is to reduce the peaking in the primary LC filter under no-load conditions.

The TA3020 is in full production at this time.


Scott Bobo
Tripath Technology
Kay -

Tripath's present policy is to review, on request, any design for functionality. We realize that many linear amp designers may be encountering some of our design rules for the first time.

Send me the .pdf and I or someone on my team will review it. If the .pdf is unclear then we may ask to see the protel files (we can also accept PADS), but it is unlikely that we would need the gerbers. Please keep in mind that we can review only the circuitry included on our datasheet schematic and can review it only for functionality. We cannot comment on any other area of the circuitry, nor can we offer tuning advice. We see circuits from many customers every week, so we maintain a "clean room" policy with regard to design reviews.

Thanks for your interest in Tripath products.

Scott Bobo
Tripath Technology
Tango support

Hello blmn -

Unfortunately, we cannot offer any direct support for Tango or Tango-PCB. All of our designs are done in Protel/Pads, depending on the need.

There is a utility which allows us to output PCB data to a (large) ASCII file. If Tango supports the import of this type of file, then that may give you a starting point. The consensus here, though, is that these conversions are rarely more than 50% effective and that you're usually better off starting from scratch with a clean design.

On our website you can find the datasheet for the TA0104 Evaluation Board, and in that datasheet are many layout tips and the full schematic for the board. This would be a great place to start.


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