[Vinnie R.]

3. The ultrasonic SPL into high impedance headphones will be twice the figures shown, or about 6db@53kHz. If you're listening to music at about 100db, you may be exposed to 106db of ultrasonics.
.

Hi Eric,

The simple solution is to use 5 watt 4-ohm dummy load resistors in parallel with the headphone outputs.  I did this with a Clari-T-Amp connected to AKG K-1000s and all I was presented with was beautiful music with a very sweet and clean souding top end.  Yes, this rolls off the higher frequencies of the audible range (which may or may not be noticed by listeners), but it does provide "peace of mind"  

Also, like you said, not all headphone drivers (or loudspeaker drivers) can reproduce such ultrasonic frequencies.  

Regards,

[eweitzman]

The simple solution is to use 5 watt 4-ohm dummy load resistors in parallel with the headphone outputs.

This is one of four possible solutions. It has a couple of downsides: most of the power is dumped into the resistors requiring the amp to be run "louder" (which includes running closer to the distortion curve's knee), and resistor noise will be relected in the signal reaching the phones.
If you're going to use this solution, I'd recommend using a 7 ohm resistor instead of 4 ohms. When paralleled with typical phones, the load on the amp will be close to 6 ohms, giving the flattest possible response at 20kHz with the Tripath reference circuit's filters.

Also, like you said, not all headphone drivers (or loudspeaker drivers) can reproduce such ultrasonic frequencies.

I don't want to rehash all the arguments again here. The driver may block the signal (harmless), reproduce it (possibly bad for your ears), or turn it into heat (bad for the phones). Like the HD-650s, it may block some (above 39kHz) but pass some (below 39kHz). One should know exactly how their phones will behave before using them with any class D amp.

- Eric

[Vinnie R.]

This is one of four possible solutions. It has a couple of downsides: most of the power is dumped into the resistors requiring the amp to be run "louder" (which includes running closer to the distortion curve's knee), and resistor noise will be relected in the signal reaching the phones.
If you're going to use this solution, I'd recommend using a 7 ohm resistor instead of 4 ohms. When paralleled with typical phones, the load on the amp will be close to 6 ohms, giving the flattest possible response at 20kHz with the Tripath reference circuit's filters.

- Eric

Hi Eric,

Yes, power is dumped into the resistors, but I believe this is well worth it, as the TA2024 likes to see a 4 to 8 ohm load.  I couldn't get the AKG K-1000's past 10 to 11am on the Clari-T-Amp volume because it was too loud for me.  In the range that is not dangerous to your hearing, you will not be near the distortion curve's knee.  With the headphones I mentioned, your ear drums will distort and be in pain before you have to be concerned with the amp or headphones clipping  

I agree with your recommendation for a 7-ohm load in parallel with the 120-ohm K-1000 headphones.  This gives you a 6.61 ohm load, which is pretty close.  I just happened to only have 4-ohms at the time...

Regards,

[eweitzman]

Yes, power is dumped into the resistors, but I believe this is well worth it, as the TA2024 likes to see a 4 to 8 ohm load.

This solution (or another) solution is pretty much necessary because of the ultrasonics.

I couldn't get the AKG K-1000's past 10 to 11am on the Clari-T-Amp volume because it was too loud for me.  In the range that is not dangerous to your hearing, you will not be near the distortion curve's knee.  With the headphones I mentioned, your ear drums will distort and be in pain before you have to be concerned with the amp or headphones clipping.

It will depend on the sensitivity of the phones. If a particular phone needs 500 mW on the peaks to get realistic dynamics and you're dumping most of the power into the resistors, the THD might be moving into the high range of the device.

I heard a bunch of phones at a meet on Saturday. I was surprised how insensitive the HD-650's are compared to most of the other phones that were there. I was pleasantly surprised that they sounded so good.

[superdow]

Looks like 6 moons will be reviewing the clari-T

http://www.positive-feedback.com/previews.htm

Vinnie, what's the time frame for the review. Good Luck!

[Wind Chaser]

Looks like 6 moons will be reviewing the clari-T

http://www.positive-feedback.com/previews.htm

Vinnie, what's the time frame for the review. Good Luck!

This should be very interesting; hopefully TNT, Enjoy The Music and others will do the same.  Heck I'd even by a copy of Stereophile if they'd do a review  on the Clari T!

[Bingo]

Hi everyone,
Not to change the subject too much. Would there be any advantage, whatsoever, in using only one channel of the SI...for one transducer?

[Panelhead]

Hi everyone,
Not to change the subject too much. Would there be any advantage, whatsoever, in using only one channel of the SI...for one transducer?

  Do not know, yet. Working on the boards for this. Using the two channels with the lowest dc offset.
  Hope that the use of a SI board per channel with give even more slam and dynamics. Terminating the unused channel with a 8 ohm resistor and 0.15 cap on the board. Grounding the unused input.
  Finally, did like some others and soldered a pair of 120 ufd Nichicon UHE caps right to the pins, bypassing the 0.10 ufd SMD cap. Also installed a 680 ufd cap as the onboard filter cap.
  If I get feeling brave, may try soldering the input power lead to the chip also. This is one trick the Clari T has on most of the diy 5066 jobs. But not quite that brave tonight.


                            George

[Mal P]

Hi guys,

This amp has made me very curious... what is the general distortion pattern of this amplifer? Is it even-ordered? If so, is that why it sounds "tube-like" or is it because it lacks the cross-over distortion and other such artefacts of transistor amps?

My speakers are 88dB/watt. I currently run an 8 wat 300B-based SET amp on them, which produce 6 volts RMS into 8 ohms and 4 volts RMS into 4 ohms. The volume level is fine, but that's because the amp clips gracefully. Will the Clari-T (or Sonic Impact) clip gracefully? I.e. not damage my speakers?

Thanks,
Mal

[Sean]

Panelhead said "Do not know, yet. Working on the boards for this. Using the two channels with the lowest dc offset.
Hope that the use of a SI board per channel with give even more slam and dynamics. Terminating the unused channel with a 8 ohm resistor and 0.15 cap on the board. Grounding the unused input.
Finally, did like some others and soldered a pair of 120 ufd Nichicon UHE caps right to the pins, bypassing the 0.10 ufd SMD cap. Also installed a 680 ufd cap as the onboard filter cap.
If I get feeling brave, may try soldering the input power lead to the chip also. This is one trick the Clari T has on most of the diy 5066 jobs. But not quite that brave tonight."

I would NOT terminate the output of the unused channel at 8 ohms.  If going that route, you want to use a VERY high value non-inductive resistor.  This reduces the thermal and current demands of the unused channel of amplifier circuitry.  Since the amplifier can't deliver much power into such a high impedance, the use of a physically smaller, lower wattage resistor is also possible.  This is cheaper, more effective and easier to work with i.e. a win / win situation in every aspect.    

If i were messing with one of these and wanted to go the "mono-block" route, i would take the above approach a step further and disable the input of the unused channel at the chip and shunt it to ground using a reasonably low value resistor i.e. 50 - 100 ohms.  You now have no potential for crosstalk between the channels, no potential for the channel being used to modulate the unused channel ( or vice-versa ) with signal or RFI / EMI.  Even if there was signal leakage / cross-contamination, the amount of signal being conveyed between the two channels would be so low due to the impedances chosen as both input and output shunts that it would be negligible.

As far as the power supply wire routing goes, bare in mind that these are "dirty" amplifiers.  Altering the feedpath will not only alter impedances, but also change how the power supply related noise is distributed ( sprayed around ) inside the amp.  While this can be beneficial in terms of increased current demands with less potential voltage sag, try to use some type of a heavy gauge shielded cable.  As far as shielding goes, a high percentage heavy copper braided shield is more effective than less braiding and a foil shield.   In this regards, a 98% copper braid is more effective than a 60% braid with a 100% foil shield, so don't fall for "double shielding is better than single shielding".  The shield should be connected to ground as that will keep unwanted radiation to a minimum.  Otherwise, your power supply lead will act as an antenna and could actually degrade the sonics of the system.  This is a common problem with mods as people try to improve the performance in one area and end up creating yet another problem.  The end result is that the mod is beneficial, but possibly not to the extent that it could be if all the various design attributes were taken into consideration.

You might also want to install a larger value cap and a bypass cap at the point where the shielded power cable feeds into the chip also.  This will not only act as a reserve for the chip right where it draws power in, it will also act as a RFI filter for any type of noise that the chip tries to feed back into the supply lines.  This may further reduce the potential for RFI from being sprayed around inside of the amp, which would find it's way into the signal related circuitry.  

I've not studied this amp at all, but those are just basic precautions that should be taken with the overall perforamance of the product taken into consideration.  Not all circuits are the same though, so not all "mods" or "tweaks" will work in the same manner or with the same effectiveness.   Sean
>

[Panelhead]

Sean,
  There is a reason for using the 8 ohm terminating resistor. The output of the amp has a very high frquency carrier riding on it. There is a filter applied to the output to remove it. The filters recommended by tripath are for 4 and 8 ohm loads. Using a high impedance termination can allow some very high frequency hash.
  The input I have grounded. There is a lot inbetween the the input line and the Tripath 2024 chip, buffering resistor, shunt cap and series inductor. These are there to help keep radiated noise off the input signal.

                          George