Ok, I went to the local electronic surplus store. Found four 10uH toroids (blue/green type, material 52) about 3/4" diamater with AWG 17 magnet wire 13 turns. Not the type 2 (low permeability) as most people recommended in class D forums. But they are NOS and the store asked for $1/piece so I have nothing to loose and took them home.
Initially I didn't know the inductance, but after some googling, I found the manufacturer's datasheet and it is the 10uH low loss type. I swapped out the on board small inductors (22uH) with these set. The sound did improve and the bass seems improved as well. I went ahead and shorted the power line protection diodes (3 of them) to make sure current from power supply will reach the board without the voltage drop from the diode. I also bypass the 2 on-board tank caps with 2 MKT1813 2.2uF caps.
After all these, the Sure TK2050 amp has good bass. Vocals have more warmth. The bass department seems more realistic and warm. Piano sounds full and sweet. I have the volume cranked up loud (when wife is not around) and have no noticeable distortion. I have been testing this board with my Tannoys 611 MKII towers. I don't know if my toroids have saturated or not but now the amp does sound good. I was liking more my Sonic Impact TA2024 more but now this TK2050 is also good.
I remember I read somewhere in this forum that the small factory inductors produce more triangular waveform compare to the toroids or aircore. I don't have an osc-scope or sig analyser but assuming this is true, the total area under a sine wave is more than the total area of a triangular wave with the same peak amplitude by maths. Therefore, my change of the tiny inductors to toroids change the output waveform to more of "true" (sine) wave. Therefore, the output maintain the "muscles" in the original signal source. The factory inductors "cut" short the area of output waveforms by making them more triangular like.
I now am a believer that the implementation and component selection is critical to class D amps.
I'm not sure if more powerful TA chips will fulfill your needs. I thought the same when I bought the Sure TK2050 amp but have to say disappointed without mods. I think it all comes down to implementation and component selections.
I have post pics in my other thread about my TK2050/TA2024 if you are interested:-
http://www.diyaudio.com/forums/clas...pace-left-inside-enclosure-2.html#post3295459
http://www.diyaudio.com/forums/clas...pace-left-inside-enclosure-2.html#post3295459
For those who purchased the Indeed TA2021 recently, or the ones who might purchase, let it burn for a while.
After I first turned it on, it was left burning in continuously for over 300 hours, with quick auditions every day. The first 30 hours were pretty disappointing, then began to settle it's character around 70 hours, and fully set between 180 and 200 hours. Actually, I laughed when I read Hammet's comment regarding Mr. Cheung's 200 hour recommendation, for I have reached the same conclusion. After 200 hours, changes were subtle to negligible.
I have yet to open it (absence of proper hexagonal screw) to see if it matches the description, specially the output inductors. But I am not even sure whether I want to mod it or not. Perhaps a simple bypass on tank caps, but unsure about changing the input caps. Based on the results with WIMA MKP10 on other boards, in this case a times expensive condenser will be needed for great improvement.
I have tested the Sure TK2050, YJ TA2020, Evaluation boards (similar to Arjen's) TA2024, and now the TA2021, and the Indeed is the best sounding from the get go.
My favorite tripath sonic signature is of the TA2020 chip, and fortunately the TA2021 offers a similar sound.
After I first turned it on, it was left burning in continuously for over 300 hours, with quick auditions every day. The first 30 hours were pretty disappointing, then began to settle it's character around 70 hours, and fully set between 180 and 200 hours. Actually, I laughed when I read Hammet's comment regarding Mr. Cheung's 200 hour recommendation, for I have reached the same conclusion. After 200 hours, changes were subtle to negligible.
I have yet to open it (absence of proper hexagonal screw) to see if it matches the description, specially the output inductors. But I am not even sure whether I want to mod it or not. Perhaps a simple bypass on tank caps, but unsure about changing the input caps. Based on the results with WIMA MKP10 on other boards, in this case a times expensive condenser will be needed for great improvement.
I have tested the Sure TK2050, YJ TA2020, Evaluation boards (similar to Arjen's) TA2024, and now the TA2021, and the Indeed is the best sounding from the get go.
My favorite tripath sonic signature is of the TA2020 chip, and fortunately the TA2021 offers a similar sound.
Most class D amps, yours too, have low-pass filters between the amp output and the speaker. Those are usually calculated for a resistive load of 4, 6 or 8 Ohm. If you connect speakers that do not come with the right impedance and an impedance linearisation, those filters will not work as designed.
First you have to find out for which impedance the output filter was designed. If that amp is another copy of the datasheet, it was designed for 4 Ohm. Any speaker that has a higher impedance will lead to a rising frequency response which explains your impression of harsh sounding treble.
If your speakers have a higher impedance, you need to adapt the output filter. E.g. if your speakers are 8 Ohm, then you need to double the values of the inductors and halve the values of the capacitors in the output filter.
When the filter is adapted to the nominal speaker impedance you need to linearise the speaker impedance at least above 5-10 kHz. That is one thing you should also do when you work with tube amps. You can skip that for solid state and chip amps. I.e. you need to add an RC network in parallel to the speaker terminals to counteract the impedance rise due to the speaker coil inductance Impedance Equalization (L-Pad) Circuit Designer / Calculator Help, and you need to add RLC networks in parallel to counteract impedance peaks due to driver resonant frequencies or due to the crossover components Strassacker: Speaker Building, Components.
It's certainly been a good 300+hours that the little amp is on, playing music in my workshop. It was plugged on the Tannoys and I must admit it did impove "alot". There's a little glassy shine on the treble, still, but no longer the harshness it first demonstrated. Bass goes low, but a bit hollow. All in all, it has a nice opennes and good imaging, and it's quite detailed.
I bought a pair of Celestion Ditton 25 yesterday ($30 at a flea market, pristine cond.) They're much less detailed than the Tannoys, but flaws are less apparent as well. So, it sounds pretty good. I'm even surprised at how good these Celestion do image since I remember having owned that same model of speakers in the 70's and was not so found of them. I liked my 66 better.
Thanks for your suggestions. At the price the Indeed sell, it will make, as is, a very nice little amp in my bedroom. I even plan to give one to a friend.
[QUOTE=pacificblue;3301045
If your speakers have a higher impedance, you need to adapt the output filter. E.g. if your speakers are 8 Ohm, then you need to double the values of the inductors and halve the values of the capacitors in the output filter.
When the filter is adapted to the nominal speaker impedance you need to linearise the speaker impedance at least above 5-10 kHz. That is one thing you should also do when you work with tube amps. You can skip that for solid state and chip amps. I.e. you need to add an RC network in parallel to the speaker terminals to counteract the impedance rise due to the speaker coil inductance
Hi,
I appreciate your suggestions but since I am a beginner it sounds too complicated for me. Will you please suggest some solution for the impedance adjustment in my case - I have two two 2021 each driving left and right bass/trebles speakers via active filters. The mids are managed by a LINN. Now - the suggested impedance of the amp is 4 ohms but my speakers are 6 ohms each.What kind of the sequence I have to work out there ?
Thanks in advance,
Ignat
If your speakers have a higher impedance, you need to adapt the output filter. E.g. if your speakers are 8 Ohm, then you need to double the values of the inductors and halve the values of the capacitors in the output filter.
When the filter is adapted to the nominal speaker impedance you need to linearise the speaker impedance at least above 5-10 kHz. That is one thing you should also do when you work with tube amps. You can skip that for solid state and chip amps. I.e. you need to add an RC network in parallel to the speaker terminals to counteract the impedance rise due to the speaker coil inductance
Hi,
I appreciate your suggestions but since I am a beginner it sounds too complicated for me. Will you please suggest some solution for the impedance adjustment in my case - I have two two 2021 each driving left and right bass/trebles speakers via active filters. The mids are managed by a LINN. Now - the suggested impedance of the amp is 4 ohms but my speakers are 6 ohms each.What kind of the sequence I have to work out there ?
Thanks in advance,
Ignat
Step 1) Take an impedance measurement of the speakers, in your case only of the tweeters.
Step 2) Linearise the impedance with the help of a capacitor and a resistor.
Step 3) Take an impedance measurement fo the tweeters again. Determine the average impedance above 5 kHz after the linearisation.
Step 4) Check which impedance the amp's output filter was designed for. The standard implementation fo the TA2021 is for 4 Ohm loads. If it coincides with the impedance you measured in step 3 with less than 20-25 % of deviation your work is done. Otherwise there is
Step 5) Replace all capacitors and inductors in the amp's output filter.
If the impedance is higher than the filter was designed for, increase the inductor values and decrease the capacitor values by the same ratio and vice versa. E.g. double the speaker impedance means double the inductor values and halve the capacitor values. A quarter speaker impedance means quarter the inductor values and four times the capacitor values.
All the above is easier than it sounds. Describing it all in detail would however be too much off topic writing for this thread. So the biggest part of the work is to research how everything is done. E.g. steps 1) and 2) you will find with the search function in the loudspeaker section. Step 4) means to use a crossover calculator from the web or do a spice simulation with the components you find in the amp and different loads.
Step 2) Linearise the impedance with the help of a capacitor and a resistor.
Step 3) Take an impedance measurement fo the tweeters again. Determine the average impedance above 5 kHz after the linearisation.
Step 4) Check which impedance the amp's output filter was designed for. The standard implementation fo the TA2021 is for 4 Ohm loads. If it coincides with the impedance you measured in step 3 with less than 20-25 % of deviation your work is done. Otherwise there is
Step 5) Replace all capacitors and inductors in the amp's output filter.
If the impedance is higher than the filter was designed for, increase the inductor values and decrease the capacitor values by the same ratio and vice versa. E.g. double the speaker impedance means double the inductor values and halve the capacitor values. A quarter speaker impedance means quarter the inductor values and four times the capacitor values.
All the above is easier than it sounds. Describing it all in detail would however be too much off topic writing for this thread. So the biggest part of the work is to research how everything is done. E.g. steps 1) and 2) you will find with the search function in the loudspeaker section. Step 4) means to use a crossover calculator from the web or do a spice simulation with the components you find in the amp and different loads.
I have bought three SA-S3 SMSL TA2021B amps from ebay.
All of them have an unreal sound on highs (comparing to my Gainclones).
I strongly believe this is because of the cheap input caps (MKT).
Even after burn- in you will still have a cheap MKT cap at the input of your amp effecting mostly the high frequencies.
So these caps must be replaced either way. The problem is to find a decent 2,2 uf MKP cap to fit in that small case.
Pacificblue's suggestion to adjust the low pass filter of the amp is my next suggestion as well.
All of them have an unreal sound on highs (comparing to my Gainclones).
I strongly believe this is because of the cheap input caps (MKT).
Even after burn- in you will still have a cheap MKT cap at the input of your amp effecting mostly the high frequencies.
So these caps must be replaced either way. The problem is to find a decent 2,2 uf MKP cap to fit in that small case.
Pacificblue's suggestion to adjust the low pass filter of the amp is my next suggestion as well.
Poor treble imho is due to the lack of power. Get yourself a TDA7498 100W Sure Class-D amp with a 24-34 VDC supply and you'll stop complaining about the treble. Also, a good class-d amp can show you all the drawbacks in your system. I mean - poor speakres, source and anything else. In my system 2021 played great, the power was not sufficient, though.
The problem is to find a decent 2,2 uf MKP cap to fit in that small case.
I have installed these and they sound better than the original ones - Vishay ERO MKP1839 2.2uF 400V. Still I do not like the bass section and will try SOLEN there, also the toroids suggested in the previous posts.
WRONG!!! Poor treble has nothing what so ever to do with power capability. What it is is either a too low switching frequency requiring a output filter that attenuates too much of the high audio band in order to not send out tweeter frying carrier residuals, or too much lowpass filtering of the audio input.
