[panomaniac]

Hello everyone.
 This is my 1st post there, though I've been reading for awhile.

So, you think that your Sonic Impact Tripath amp sounds bass shy?  You're right, it is!  See the graph.
Think it lacks power and really needs very efficient speakers? It does! See why.

 All this information along with graphs and measurements over on my site.
Not news to you, of course, but you might like to see it all gathered together and measured.  Plenty of information on this fun little amp.

Some of you will already have seen the found the site via other audio forums, but there is new information there as of today.  Check it out at:

http://www.michael.mardis.com/sonic/

All comments, questions, suggstions and arguments welcome.  

[Panelhead]

MM,
  I am traveling, cannot access your site. Maybe it is the hotel supplied DSL connection.

                                        George

[gary]

Just copy and paste it into the address bar, then it'll work.

Gary

[panomaniac]

MM,
  I am traveling, cannot access your site.

I wasn't you, George. The link got corrupted in my post. Fixed now.
http://www.michael.mardis.com/sonic/

Thanks for pointing it out.

[JeffB]

panomaniac, thanks for the link, some fantastic information there.

[JeffB]

I had asked a question in another thread that I think your page answers.  I'll restate the information and see if you have any input.

I tried out a test tone CD with my Sonic Impact.  When playing frequencies below 60Hz there would be a volume point where the signal sounded ok, but then just barely notching the volume up, say 3 db, I would get horrible distortion.  Totally unlistenable.  The volume this occurred at was not very loud, probably under 80db.  The lower the frequency, the lower the volume before massive distortion kicked in.  At 40Hz, I bet the volume was around 70db where massive distortion kicked in.  I am waiting for a 12V SLA to arrive via UPS before taking more accurate measurements.  By measurements, I just mean an SPL meter and my hearing.  I was beginning to think that my Sonic Impact had a defect.  I also like Heavy Metal which appears to be a type of music the Sonic Impact will not handle very well.  Note that tests were done on a cheap speaker with a single 4" driver, with unknown sensitiviy, 6 ohms. Power handling of the speaker is listed as 5 watts.

Would a high pass cross over to remove the low frequency information help any with the clipping?  I am considering a Behringer DCX2496 which I think might provide a digital cross-over.

Is the clipping related to the 2024 chip itself or other circuit elements of the Sonic Impact?  Would the Clari-T perform better in this respect?

[_scotty_]

panomaniac,

  Excellent information and not too suprising. This chip has very limited heatsinking and current throughput capability. I wonder if the bass is rolled off to allow the amp to put out even 0.125watt  before clipping.  It wouldn't suprise me if this was the case.  By all acounts people think  it sounds good .  I had hoped that due to the low output it might be a little lower in ringing than a higher powered version of the Tripath technology.  I suspect it has inadequate local decoupling adjacent to the output devices and a power supply with too high an impedance at the switching frequencies this amp operates at. The storage caps are probably crap and need to be replaced with caps that have a low impedance at high frequencies.  This would do nothing to solve the bass roll off problem however.  Have you calculated the what effect if any that the size of the input  coupling cap has on the bass response.  I was thinking that if the 2040 was intended to power computer
multimedia speakers, augmented by a powered woofer its perfomance charateristics make a lot of sense.  The nearfield listening enviroment  requires very little power  and the apparently built in rolloff  below 200Hz
would limit the tiny drivers excursion and distortion below 200Hz.
   Thank again for posting and taking the time to measure the performance
parameters of this device .  Have you considered measuring the $99 dollar three channel Teac .  It has a little higher power rating and a good price.
   Scotty

[tianguis]

Panomaniac:
        Thanks for the new info. In a word or two or three: re-cap, re-cap, re-cap. The stock T-Amp is gritty, nasty and bass shy. Properly broken in, it loses the gritty nastiness. Caps give it bass and dynamics. Vinnie's Clari-T uses 10 X the 380 uf on the power rail. Mine uses a bit less, but each one I build I'm going bigger. Of course, less-efficient speaks contribute to your measurements. Larger caps will deliver the instantaneous energy in order for the speaks to move some air quickly, which is the only way to obtain any real bass with the SI (or almost any other amp).
       Which speaks were you using for these latest measurements?
        Keep up the good work!

Regards,
Larry Welsh

[panomaniac]

I tried out a test tone CD with my Sonic Impact.  When playing frequencies below 60Hz there would be a volume point where the signal sounded ok, but then just barely notching the volume up, say 3 db, I would get horrible distortion.  Totally unlistenable. .

I don't know what would be causing this, unless the clipping was coming from input overloading.  It seems that the input cap on the Sonic amp is too small, creating a passive hi-pass filter that reaches up into tthe audible zone.  You have to push the volume WAY up to get the ower frequencies to overload, as they are attenuated at the input.

I did not test clipping at frequencies below 200 Hz, but will, and will post the results here and on my site.  Thanks for bringing this to my attention.

Have you considered measuring the $99 dollar three channel Teac . It has a little higher power rating and a good price.

I have thought about it. Apex also has a couple of cheap HT in a Box that use Tripath chips.  Might go that route.

To "recap" Larry. Recapping will be #1 mod.  The input cap needs to go, too small and causes bass loss noted.  Bige better power caps on the board will be added too. My PSU has a 8500 uF cap across the output after the regulator.  Biggest I've read about.  (I come from the "Massive Passive" school of PSU design).

I will give me subject impressions once the mods are done.   Will see if it changes the measurements.

Must of the published specs were taken into 8 ohm resistors.  I did the listening on my Vifa/Dayton 2 ways at home and my big Dayton/ScanSpeak 2 ways at work.  Nothing fancy.

[panomaniac]

I did test clipping at low frequencies tonight.

FWIW
There seems to be no difference is clipping level from 1kHz on down to 20Hz.   Input level had to be raised at lower frequencies to acheive clipping because of the input low end roll off.

So I don't know what would be causing your low frequency problem, if you are using pure tones. Are you?

[JeffB]

Yes, they were pure tones.

[panomaniac]

Yes, they were pure tones.

 
Well i don't know, then.  I 'll do some listening to see if I can reproduce what you heard.  Maybe the clipping just sounds worse at low frequencies.

[commited]

To "recap" Larry. Recapping will be #1 mod.  The input cap needs to go, too small and causes bass loss noted.  Bige better power caps on the board will be added too. My PSU has a 8500 uF cap across the output after the regulator.  Biggest I've read about.  (I come from the "Massive Passive" school of PSU design).

I will give me subject impressions once the mods are done.   Will see if it  ...

Can anyone explain in more detail exactly where the capacitors should go and recommended sizes etc?

Any links to any sites that explain this?

[JeffB]

I have bookmarked the following two links for Sonic Impact mods:
http://rblank.home.texas.net/Digital%20Amp/
http://members.tripod.com/talkgarden/sonic_impact_5066.htm

[Occam]

.......
{Would someone who is comfortable with it please check my math?}

Edit - Chairguy,
I did check the math again and pulled the post,

[TheChairGuy]

{That would certainly not be me}

 

[panomaniac]

Can anyone explain in more detail exactly where the capacitors should go and recommended sizes etc?

The input caps are C2/C4. These caps block DC from the chip to the input.  They also act as a hi-pass filter causing the bass roll off this amp has. (I measured to be sure).  
They will be found on the top and bottom of the PCB.  Don't know yet what size to replace them with.  If you go off PCB with the replacements (easiest) then use at least a 2.2uF or larger.  There is 5VDC on the downstream side, so be carefull to orient it properly if you use an electrolytic.

The power decoupling cap is the "large" electrolytic nestled between the output chokes (the 4 black towers). This can be replaced with a bigger, better cap that will still fit.  Panasonic FM series 680uF 16V has been suggested over on the diyaudio forum.  

I will soon have pix showing were these caps are.

Good luck.

[Adamay]

If that's the main source of the bass roll-off, seems to be a key issue.  Anybody know the answer to this?  Thanks.

[commited]

I will soon have pix showing were these caps are. ..

Excellent

I think i'll wait until someone with more experience does it first  

[Gary P]

Hi All,

After reading about these little amps they just looked like too much fun. Got the first one a few days ago. Not bad in stock form but like everyone has said, lots of room for improvement.

The amp is now dissasembled for inspection and circuit analysis. Like pointed out in many places the bass response is lacking. The main culprit causing the weak bass is the input capacitors that are too small. In the Tripath datasheet they recommend 2.2uf. The parts in the Sonic Impact measure .33uf.

Here is a schematic of the input circuit the Sonic Impact.


As this input circuit is driving the inverting input of an opamp the input impedance is low. The impedance of an inverting opamp circuit is the value of the input resistor, in this case R1 and R2 at 20K. These are in  parallel with R01 and R02 bringing the input impedance down to 6.67K.
Add the 50K pot and the impedance drops down to 5.8K. This is getting kind of low for some sources. Not a problem for the portable CD players that it is designed for as they are setup to drive headphones directly.

The path I think I'm going to take is to try and see how good the performance can be while keeping the constraints of the stock case.

1. Input caps must go. Just for grins they will be replaced with Black Gate BGN 4.7uf caps. The ceramic caps will be removed and the caps soldered to the SMT pads. I don't think the vias are large enough to get the cap leads into.

2. R02 and R02 will be removed to ease the load on the source components.

3. The power supply cap will be replaced with a larger low ESR cap (choice of what kind has not been made yet). It looks like if the 4 inductors are raised up the leads can be bent sideways enough to get a larger cap in the space available.

4. Board surgery- I don't like the way the 12 volt supply is routed. The measurements that panomaniac has made shows that the supply is very noisy with high frequency switching noise. This noisy supply is routed all the way around the board including up to the pot/switch and directly under the input circuitry. My plan is to add a small SPST switch to the rear panel to handle the power switching. The 12 volt supply will then be fed into the circuit at the power supply cap. The run that loops around the board will be cut and soldered to ground instead. This will keep the supply noise away from the input circuitry and provide a touch more ground plane.

5. more board surgery. A couple of different ways to go here. The input signal runs along the edge of the board and is not shielded. Option 1 would be to cut the runs on the board and run coax from the inputs to the pot then use more coax back to the board. The other choice would be to wrap some grounded copper tape around the edge of the board where the input signals run providing some shielding and use coax from the existing connectors to the pot.

6. More capacitance to help things sound good while running from the AA batteries. There is just enough room to add a pair of 16x35mm capacitors, one beside each battery compartment. I’ve found caps up to 6800uf @16 volts in this size. Don’t know if maximum size or lower ESR is more important here as the cap on the main board will be a low ESR type.

7. The value of R4 and R5 will be reduced to 20K. This will reduce the gain of the amplifier. In my system the amps are too sensitive. The transformer volume controll is almost at minimum to get the level I like to listen at.

All these ideas will have to wait until the next units get here. Just ordered a pair of amps from Partsexpress. Due date is Mar. 1st.

Gary