Sunday, October 5, 2008
DIY Class T amplifier - further development
I am no audio DIYer and have only a superficial understanding of audio electronics when I embarked on the DIY T-amp project. I have, however, built an Audio Note Kit 1 300B valves power amplifier in the past based on step-by-step instructions during which I learned how to use the soldering iron quite effectively, could differentiate between good joints from bad ones and also learned a couple of tricks. I have relied on online comments made by others who experimented with their T-amps before me and also acted on my gut feel in other areas to complete both my 2 units. I have owned and listened to numerous amplifiers in the past, yet I am still thoroughly charmed by the T-amp’s sonic qualities. Read about my listening to the DIY T-amp.
I learned a lot from the T-amp project and my findings and/or rationale of some of my actions are as follow:
(1) Input capacitors – The primary reason for their presence on the T-amp is to block DC current while allowing AC (Audio) signals to pass at the input line. I had initially thought they were only meant for signal filtering just like signals into individual loudspeaker drivers. The explanation by Blair Thompson, aka Justblair at the bottom of this article is a good read.
The input capacitors affect sonic qualities of the T-amp most as a change in these capacitors would make it sound very differently. Some folks went on to claim an improvement in resolution and soundstaging capability of their T-amps after using certain capacitors. In my case, I opt for moderately-priced metalised polyester film capacitors of Vishay Roederstein MKT1813 2.2uF 63V and Vishay Roederstein MKT1822 2.2uF 100V which are available at RS Components/Farnell in my 2 units of T-amp instead of exotic boutique ones. Of the two, I prefer the former more as its sound is much more relaxed, open-sounding, sweet and velvet-like. An article on capacitor test which include the Vishay MKT1822: http://www.humblehomemadehifi.com/Cap.html
We could try out any other capacitors between 1.0uF to 3.3uF so long as they could fit onto the T-amp boards. The use of capacitors with higher capacitance value though will lead to 'thud' sound during powering up.
(2) Power filtering/buffering electrolytic capacitors – Low ESR capacitors should be used here and many folks have reported a more dynamic, more punching solid bass to the sound of T-amp after a change to these capacitors. I opt for Panasonic FM series 470uF 25V ones.
I discovered that the ideal value of these capacitors on the T-amp should be 470uF 25V by accident after listening to my colleague’s unit on 27.09.2008 while we were troubleshooting a soldering-induced problem on his board. I had listened to and compared both our T-amps whereby mine has the 4 stocked capacitors replaced by Vishay MKT1813 2.2uF 63V axial type polyester film capacitors and Panasonic FM low ESR 1000uF 16V electrolytics whereas my colleague only changed the input capacitors to the same Vishay MKT1813 capacitors. Mine is also with a 24-steps 50K Dale resistors attenuator whereas his an Alpha 50K carbon potentiometer. PSU used is a 12V DC 2.5A SMPS, source a Sansui CD-X310 and transducers Q Acoustics 1010. I had reported at LYN forum that the difference in sound of our T-amps as very substantial and initially preferred the unit with Panasonic FM capacitors. The overall scale of music and tonal balance sound larger and more balanced respectively, the low frequencies also carry more weight which adds to the foundation of music making. I had concluded that the stocked buffer capacitors of 220uF 25V must be replaced, inferring that my choice of Panasonic FM 1000uF 16V is correct.
An afterthought to what I had reported earlier - upon further reflection, I found that Vishay MKT1813 capacitors when paired with the stocked buffer capacitors of 220uF 25V produce a livelier highs and midrange. Vocalists sound more airy and micro details are more easily picked up, the very feeling which evokes goose bumps in us (the very hallmark of TA2024 chip itself). With Panasonic FM capacitors on board, the image is bigger and music carries more weight but some of the micro details are overshadowed by the dominant bass weight and a shift in tonal balance from yin to yang which could be a tad tiring over long period of listening. Perhaps, an in-between value used here may be desired so as to strike a balance and this could be reason why finished T-amps usually come with low value capacitors in the buffer section. It was after my listening to the newer board with stocked 470uF 25V presumely general purpose capacitors in buffer section which confirmed that it has indeed more micro details than my well run-in old board and sufficient bass weight although a little flat in comparison to my old board, possibly attributed by the absence of discrete resistor-type attenuator.
(3) Volume potentiometer – I opt for a 24-steps 50K Dale discreet resistor type attenuator for use as volume potentiometer instead of carbon type like Alps RK27 Blue Velvet as my past experience with Audio Note Kit 1 power amplifier clearly demonstrates that the former gives a far more transparent sound, better tonal balance, 3 dimensional soundstage and a killer midrange (tone and body) to-die-for over the latter. I had fitted onto the said amplifier a surprisingly cheap 100K Alps Blue resistor type potentiometer (its construction similar to present day TKD resistor potentiometer) that did not last very long and was replaced with a 100K Alps RK27 carbon type potentiometer shortly thereafter. Consequently, I lost the body and liquidity of midrange that I cherish.
I am often fascinated by the pursuit of others for transparency and musicality through the use of discreet resistor type attenuators in passive pre-amplifiers. Some interesting sites: http://www.eastmarinedrive.com/contents/non-static/passive-content.htm; and http://www.mh-audio.nl/MySystem/hifi.asp.
The selection of resistance value for volume potentiometer is dependant upon the input impedance of T-amp and some folks have also reported much success with 20K ones. With higher resistance value, the entire range of the volume potentiometer would be playable without the T-amp sounding way too loud. For more information, see: http://www.goldpt.com/info.html.
(4) Surface mount devices (SMDs) - I am of opinion the SMDs on my T-amp boards are of inferior quality and value of some are not according to those specified in the Bill of Materials. Replacing these with better grade ones or discrete components should improve sound of the T-amp. An article on understanding and working with SMDs: http://www.geocities.com/vk3em/smtguide/SMT-GuideV1-3.PDF and a YouTube video: http://www.youtube.com/watch?v=3NN7UGWYmBY&eurl=http://www.curiousinventor.com/guides/Surface_Mount_Soldering/101
(5) Internal wires – I only use very cheap AWG-26 multi (7) strands copper wires which cost me RM0.30/metre for internal wiring of my 2 units of T-amp. I have braided the single run of wires for signal just like those from Kimberkable PBJ interconnects with no shielding so as to filter out stray RFI (Radio Frequency Interference) picked up with 'Capacitive Cancellation' phenomenon by the crossing of wires in the cable. I have personally heard the PBJ sounded so much better than its shielded version of KC1 and also the tone richness of SilverStreak that PBJ clearly could not match let alone the full silver KCAG but as I am only using a short length of less than 9" in total, I opt for copper instead of silver which I thought is more suited to the sonic character (see Summary below) of the T-amp. As for the wiring to speaker binding posts and 12V DC, I have twisted and used the same wires in double run instead of thicker wires.
(6) Light Emitting Diode (LED) – I installed a blue coloured LED each to both units of my T-amp, tapping from the 12V DC input which powers the board and wired it parallel. A LED needs a resistor connected to its anode (+ve) leg so as to limit the current passing through. From the calculation, a 1W 1K resister is adequate for 12V application but in reality, we may need a higher value as the LED is merely for pilot light and need not be overly bright. A self explanatory article: http://www.kpsec.freeuk.com/components/led.htm
(7) Switched mode power supplies (SMPS) - I have been using a RM33 12V DC 36W 2.5A computer-notebook-style SMPS all along. I had originally wanted to get another similar unit as I have 2 units of T-amp but my colleague convinced and bought me a Meanwell S-50-12 SMPS for RM55 on 11.11.2008. The difference in sonic quality is substantial and the extra amount of money spent is well-worth. Coincidentally, size (width and depth) of this SMPS is identical to casing of my main T-amp and looks very much the dedicated external PSU specially made for it.
The Meanwell S-50-12 SMPS has 4.2A on tap and its DC voltage could be increased to 14V maximum. Even at 12V setting, it sounded so much better than my other SMPS – the sound becomes fuller, more 3 dimensional and the amplifier behaves like a 50W instead of a mere 20W. It is less flat-sounding and less tiring over long period of listening with ample of reserves when demanded upon. A really excellent stuff.
The wiki of diyAudio.com (http://www.diyaudio.com/wiki/index.php?page=Class+D+Amplification) states that 3A – 5A regulated supply is typically aplenty as the T-amps only draw about 300mA with a musical signal, with many loudspeakers. Even when a battery is to be used, its recommended amperage-hour is 7Ah. I would think the T-amp does not only draw 300mA constantly with musical signal all the time as very loud and complex music passages with lots of bass will definitely draw more current. And with difficult load loudspeakers (their impedances dipping very low), the T-amp will require even more current. With the T-amp driving medium sensitivity loudspeakers of benign loads and music not being played to head banging levels, 4.2A should be more than sufficient.
In my opinion, SMPS is cheap, convenient and the Meanwell is with flexibility of switching to a slightly higher DC voltage. Apart from SMPS, there are the linear power supply and 12V DC 7Ah batteries. On sonic quality of using these different means to power up the T-amps, do check out comments by a reviewer who has used all three to power his T-amp: http://www.tnt-audio.com/ampli/amp-6n_e.html
Summary: The TA2024 chip is designed primarily by the now defunct Tripath as a multimedia amplifier, its main characteristics - ultra clean sounding and prominence to the highs and upper midrange. Bass is slightly lacking. Use of suitable valued low ESR electrolytic caps would restore the balance slightly (I found success with Panasonic FM 470uF 25V ones). Replacing SMDs with discrete components and/or using a full resistor attenuator will further 'cure' the ultra clean sound and render the T-amp slightly more 'dirty', colourful and musical. A few of us have heard a T-amp board with most of its SMDs replaced with discrete components, its sound as what I have ascribed. We also heard how the Vishay MKT1813 2.2uF 63V metalised axial caps made the T-amp even more musical, especially in the midrange. All harshness and hardness from the stocked input caps were totally eliminated.
What am I on now? I have joined a fellow forumer at LYN in performing a ‘mad’ modification of removing all the SMDs and replacing them with discreet components (completely ignoring the benefits of the former over the latter) in the hope of milking the very last bit of performance possible from the T-amp.
I appreciate Justblair’s mention of my review of the T-amp at Post 621 in a diyAudio forum on Sure Electronics board (http://www.diyaudio.com/forums/showthread.php?s=7baaed79463029223d2c822c871e5fc5&threadid=112395&perpage=25&pagenumber=25) and his website (http://www.justblair.co.uk/blog/12-audio/27-welcome-to-malaysia.html).
I have in the evening of 04.10.2008 installed a flag counter to provide me with an indication of number of visitors to my blogsite and must say I am deeply honoured to have close to 50 people from all over the world visiting it within the first 24 hours.
Justblair's explanation on input capacitors:
"... The best solution is to have no capacitors in a circuit altogether. However, the capacitors have a very useful property in the sense that they block DC signals while allowing AC signals (ie Audio) to pass. To understand how, look inside the capacitors. There are two layers of conductor, separated by a very thin layer of insulation. What happens when electrons are removed from one side than the other (ie a positive charge) is that electrons on the opposite layer are attracted to the positive charge, they flow in.
. Of course when enough have flowed in, they stop, and we have no current, ie the DC current has been blocked. Add some electrons to the now positive side, and the electrons on the other side will no longer be in balance and will flow away from the plates until equilibrium is now reached. Remove them again and the opposite occurs. What is happening now is that the AC signal occurs on the other side of the capacitors as a reaction to the AC signal on the source side, or to put it another way, AC signal has been allowed to pass.
. The fact that we get a filter this way is almost consequential. If we want the amplifier to feed a high range speaker we could use very small input capacitors to restrict not just the DC but also the slow charging AC (lower frequencies). Capacitors also filter low frequencies apart from DC current. It has to do with the charging and discharging times. That is where the resistor comes in. It reduces the current, meaning the capacitor takes a finite time to discharge and recharge. If the frequency of the signal is low enough, it changes aptitude (voltage) more slowly. The capacitor discharges quicker than the change in source signal, so the source signal does not pass through..."
Added on 12.12.2008:
I have finally completed the modification of replacing the entire SMDs with discrete components on my third T-amp board on 04.12.2008 after acquiring an Antax 18W fine tip soldering iron. This T-amp board subsequently found its way into the casing of my first T-amp pictured above. The replacement discreet components and their respective costs (all from Farnell) used on my third T-amp board are as per below:-
Vishay Roederstein MKT1813 2.2uF 250V caps (2) - RM14.66 ;
Panasonic FC 470uF 25V electrolytic caps (2) - RM4.82 ;
Vishay BC 0.47uF 63V caps (6) - RM15.84 ;
Vishay BC 0.1uF 63V caps (6) - RM7.02 ;
Vishay BC 1000pF 400V caps (4) - RM2.68 ;
Vishay BC 100pF 50V caps (2) - RM2.58 ;
Panasonic Audio Grade 1uF 50V electrolytic caps (2) - RM1.90 ;
Vishay BC 1% Metal Film 20K ohm resistors (4) - RM1.68 ;
Vishay BC 1% Metal Film 10 ohm resistors (2) - RM0.84 ; and
Vishay BC 1% Metal Film 8.2K ohm resistor (1) - RM1.48.
If electrolytic caps are to be used (I myself use such Panasonic 1.0uF caps in C5 and C17), their polarity must be observed. A better way to mount the resistors on the T-amp board would be to let them stand vertically instead of being horizontal as in my case. I could hardly find enough space to mount them in R9 and R10 so much so that I need to remove the 10 ohm resistor after having soldered it properly and in the process accidentally lifted the soldering pad of R9 bottom and ended up having to tap its leg to the blotch of solder below C28 bottom. Please exercise care and do not solder and desolder many times as the soldering pads could come out quite easily. If the power buffer caps are to be replaced, the stock caps should to be removed prior to the soldering in of the discrete components.
This modification is rather 'mad' as I put in discrete components onto a T-amp board meant for SMDs. It took me 6 hours to complete my entire modification but I feel the outcome to the sound changes of the T-amp totally fantastic, no regrets. Most important of all, at the end of day we are using it to listen to our music. On its sound quality, after playing music for many hours now, it is beautiful and easily some 20% better in all areas (resolution, airiness, decay, expressiveness, micro details, timing, 3D soundstaging, dynamics and even dynamic shadings) in comparison to my 1st board. As I am using a discrete attenuator rather than a carbon type volume pot, I find the high frequencies extended yet not bright sounding to the extent of being unforgiving as reported by a LYN forumer. Total cost for replacement discrete components is RM53.50, almost the price of 1 T-amp board. However, to me, the mod is well worth every sen spent so for those who could handle the soldering iron reasonably well and are deliberating whether to perform the mod or otherwise, my advice would be to go for it and tell me later if I have been incorrect in my assessment. I would even go further to pronounce that my T-amp with full discrete components will take on and give many highend amplifiers a run for money. What it could not produce in quantity (huge wattage) is compensated by the sheer quality it oozes.
A week after I completed the modification, the sound of my T-amp with discrete components is very different from the time it was newly done. I have used bigger than usual capacitors (Vishay Roederstein MKT1813 2.2uF 250V in the input stage and Vishay BC 1000pF 400V in the output biasing stage) and a result of this is my T-amp sounds so much more smoother and fuller. Absolute speed is just a tad slower compared to my other units of T-amp with SMDs or even another LYN forumer’s unit where most SMDs were replaced with discrete components. I have not installed any additional PSU cap(s) so as to maintain the same level of 'micro details' and overall balance of sound. I thought its timing now is almost perfect. I also hear no noise nor distortion after replacing the SMDs with discrete components but I must admit that the super clean and almost clinical sound of the stock T-amp is now gone. Could whatever distortion induced be the 'even-order harmonic distortion' which are so pleasing to our ears? Yes, you got me correctly, I am inferring that my T-amp with full discrete components is sounding pretty close to a valve (tube) amplifier not only in its smoothness, warmth and micro dynamics but also with its timing gotten right, lots of rhythmic bounce and sufficient bass and bass weight we could possibly seek from an amplifier. Shocking, but the sound of my modded T-amp is that good and extremely transparent at such a low price without any use of fancy powercord, interconnects or speaker cables.
I am certainly not the only guy to have raved that the T-amp sounds pretty close to a valve amplifier as many others have actually indicated their preference of T-amps over their other exotic valve amplifiers. Enjoy tweaking and listening.