Finished Truepath

[bonalux]

Hi all,
I've just finished (more or less) my amp project based on the Truepath board, but can't say I'm 100% satified for the following reasons:

1) Even with DC offset set to 0v, the TP for me heats up too much (at idle), the heatsink (that it's sized for a 50+50W class AB amp) became noticeably warm compared to my amp1-b that is always freezing cold. The efficiency for me it's comparable to a class AB amp for the TP. Apart this it sounds good, but not tried yet on my reference system.

2)The toroids became very hot also, another sign to me for the power inefficiency of the system. After some minutes of running at very low volume i can't touch them!

I've got a faulty board? Or my digital multimeter it's disturbed by the switching signal and don't read correctly the DC offset (tomorrow i will try with the scope)? Experience from other users regarding the heat?

If there is not a solution i hope at least to be satisfied with the sound, i'll let you know.
Thank you very much.

Regards.

Gianluca

[V-bro]

I tested my amp on 2x35Vac (2x47Vdc) and 8 ohm speakers without heatsinks the FETs hardly warmed up. Now even after longer testing on different (also 4 ohm) speakers it never got more than warm. The toroids remained entirely cold during all my tests...

Maybe I never take it as far as you do, but it's likely that there's something else wrong there....

------X-max. can be several meters on any driver, too bad it can be done only once...------

[BddB]

Hi Gianluca,
This amp typically runs very cold so clearly you have some other problem.
Make sure that all polarized components (like diodes, capacitors) are correctly mounted.

Also make sure that the input-wires are far away from every "switching hotspot" like toroids, FETs and other output-components, to avoid switching feedback.
A self sustaining feedback consumes a considerable amount of energy and will surely shorten the amps lifetime.
My Truepath draws (included transformer losses) some ~6-8W muted and ~21-23W unmuted. (If you have the PSU1-SS, the mains filter will consume another ~12-14W)
Regards
/B

[thomaseliot]

Yes, my Truepath too runs cold, heatsink and toroids (my oversized ones are colder than my hand when touching). With hot toroids you'll get high distorsion, so better find and fix the problem.

Take a look here:

http://www.41hz.com/Forums/topic.asp...2&whichpage=15

maybe something wrong with mosfets.

[bonalux]

Thanks to all for the reply, i've checked all the components more than twice and all (apparently) it's ok, and the amps sounds good, even connected to a pair of cheap speakers (the output power is low even at max volume but it's connected to a cheap mp3 player powered by only one 1,2v cell), so i can exclude defective components. Before mounting the amp in the case i have had the overheating problem only on one channel and was solved by shielding the V5 cable, but now both are overheating, despite having shielded cables on V5 and on the inputs. I'm puzzled by this problem....

The input wires passes reasonabily far from output ones (6-8 cm in the nearest point), only the VPP-GND-VNN cables intersecate in one point with output speaker wires, can this be a problem?

Watching the waveform on the V5 line, there is some 200mV (if i remember correctly but it was fairly high) of confused signal at 2 or 3 microsecond period, i think obviously that it's related with switching frequency of the amp but, it's normal this? I've tried to filtering it but with no luck (a similar signal is present in my amp1-b but the amp works well). The V5 (and also VN10) are generated by an LM317 bypassed on the adj pin with 2u2 tantalum cap and 22uF e-cap + 0.1u ceramic disc cap on output and there is also a 10uH inductor in series as an added filter.

Tomorrow i will try to visualize signals on all the lines and to put some shielding material behind hotspots and wires.

I hope to got this gremlin because i like Tripath sound so much, as i like so much the idea of powerful amp that doesn't waste energy

Happy listening.

Gianluca

[BddB]

Hi Gianluca,

Truepath has (if default comp. values are used) an input sensitivity of 2V for full output.
A cheap mp3 will have an output voltage-swing that is only a fraction of that.

The +5V is the analogue power and must be rock steady. 200mV ripple seems to be a lot.
2 us corresponds to 0,5 MHz and [u]might</u> be the switching frequency of the amp.
Are you using the PSU1-REG?

Please post some pictures of amp, waveforms etc., if possible.
/B

[audiorl24]

Hi all
I have a running truepath and i woud like to know the normal level on the output after filter, off the commuting signal.
my toroids are a little warm without signal on the inputs.
thanks

[bonalux]

Hi Benn, the mp3 player it's only for testing purpose, the input FB res are setted standard (47K & 22K). Actually i can't say for sure if the signal visualized on my scope is noise catched up by the ground cable of the probe or what...for V5 and VN10 i use a self built supply that uses LM317 with bypass caps on adj pin, so is the "state of the art" of (cheap & simple) linear regulation, but i've tried also a desk PSU with the same result on the waveform, so maybe the scope is acting as an antenna.

I've discovered that my digi multimeter, if placed to near the amp reads wrong! (as i was suspecting) I've checked the offset with an analog multimeter and it was off by near a volt, but putting it to zero didn't solved the problem of overheating.

This morning i've checked some waveforms and you can find them attached, F1 and F2 are referenced to AGND, others to GND, inputs are shorted. DC offset on input is ~2V. The strange thing is that on F9 and F10 there is 100Vpp of switching signal instead of the 10Vpp i was expecting!! What's up? Also on the output there is a 4Vpp constant sine at 1,5 uS on both channels.

What's thats pads (there are four of them on the board) circled in red on one of the last pictures?? It seems a resistor but it's not mentioned int the AI or BOM.

The last picture is my less than perfect cabling (not tied up for now), the blue caps are to short the inputs (i've mounted input caps on the volume board).

Take a look and please let me know if you can solve this puzzle, i' can't....
Thank you beforehand.

















Gianluca

[audiophool]

Thanks for posting the photos Gianluca! It's interesting to see what others are up to!

Alas, Tripath amps are only highly efficient at or towards rated power.

The current drawn from VPP and VNN supplies at idle (no input) should be about 120mA regardless VPP and VNN voltages. I just verified this by winding my supplies from unmute voltage all the way up to 60V. At +-50 this corresponds to ~6W per channel. (Class AB does not draw this sort of current at ide.) Virtually all of it is the result of MOSFET shoot-through. I have three working Truepaths and all six channels get pretty warm at idle.

Benn, what VPP/VNN current do your amps draw at idle? I wonder if there could be significant variation between individual MOSFETs.

&gt;for V5 and VN10 i use a self built supply that uses LM317

I only use 7805/7810. Not ideal (it's on my to-list) but it works OK. It will not be the cause of your concerns.

&gt;digi multimeter, if placed to near the amp reads wrong!

Interesting. I never tried this or noticed it. What setting did you have it on?

&gt;The strange thing is that on F9 and F10 there is 100Vpp of switching signal instead of the 10Vpp i was expecting!!

100V is correct and it's elegant when you look into it. The 10V on the gate of the MOSFET is with respect to the gate (VGS) not ground! In the case of the high side MOSFETS, the source is the VNN to VPP square wave that's fed to the coil. VBoot lithely floats on top of this square wave.

&gt; 4Vpp constant sine at 1,5 uS on both [output] channels.

This is the residual HF square wave that the filter hasn't removed. It's also what causes nasty interference with your radio, and apparently your multimeter. If you look closely you should find that they are exactly the same frequency, unless you are using the same value of C107 and C207

&gt; What's thats pads (there are four of them on the board) circled in red on one of the last pictures??

Benn??? I assumed that they are test points. Notice that they're also on the ground sides of the outputs.

&gt;The last picture is my less than perfect cabling

I hope that grey cable tied to the black V5 cable is not an output to your speakers!


In summary, everything about your amps look OK to me. If you are concerned about the temperature though, you could try increasing BBM.

[bonalux]

Thanks for your reply Tim,
my multimeter is an autorange 4+1/2 digit of standard quality, don't know if it was fooled by the EMI, apparently it is.

&gt;100V is correct and it's elegant when you look into it.

Ok my fault, glad that it's not a dead mosfet.

&gt;I hope that grey cable tied to the black V5 cable is not an output to your speakers!

No, that is the VN10 cable, speaker cables are the big gray UTP CAT5 cable on the bottom that turns on the rear of the heatsink. The maroon, yellow and blue are VNN-GND-VPP ones.

Increasing the BBM, as per Tripath instructions, lends to an increase of distortion at mid-high freq so i will eventually try it only for testing purposes.

I've checked the curent drawn only from the VDD and it's about 120 mA, so the puzzle became more tricky. Maybe i've to live with it.

Bye


Gianluca