making Digial Amp -- TA2020-20 --


Japanese version here.


We can make a digital amplifier easily using a power amplifier IC, "TA2020-20".
I wanted a nice one, I tried.
I love reggae musics, so I designed an amp that would sing powerful and deep which would match reggae.

Kamaden, a japanese electronic component company, produces a kit of TA2020-20.
When you get it, we will be able to make a nice digital amp very easily.
But I presaged that it wouldn't satisfy me.
I would want to modify it after complication, so I tried to make a TA2020 amp without a kit.

I knew next to nothing about electronics.
I asked Google as "TA2020", "TA2020 capacitor", and many other keywords and I borrowed ancestor's wisdoms.


a part of amp

parts model number price
1 digital amp IC TA2020-20(Tripath) 1,890yen
2 schottky diode * 8 1S4 41V 1A 25yen *8
3 resistor(10ohm) * 2 1/2W metallic platy resistor 1% (NIKKO) 42yen *2
4 resistor(20kohm) * 4 1/2W metallic platy resistor 1% (NIKKO) 42yen *4
5 resistor(8.2kohm) 1/2W metal-film resistor 1% 31yen
6 capacitor(0.01uF) * 2 MKT1826 metalized polyester filmy capacitor(ERO) 126yen *2
7 capacitor(0.1uF) * 8 MMD2E104N metalized polyester filmy capacitor (NITTSUKO) 42yen *8
8 capacitor(0.22uF * 2 MKT1826 metalized polyester filmy capacitor(ERO) 210yen *2
9 capacitor(0.33uF * 4 BG-PK electrolytic capacitor (BlackGate) 31yen *4
10 capacitor(1uF) * 2 MMD2E105N metalized polyester filmy capacitor (NITTSUKO) 42yen *2
11 capacitor(2.2uF) * 2 BG-C 50WV suppressed noise electrolytic capacitor for coupling (BlackGate) 94yen *2
12 capacitor(3300uF) * 4 KMG 105 Celsius grade (Japan chem-con) 100yen(4 included)
13 reactance(10uH) * 4 TA2020KIT-SP coil protected from magnetism 840yen(4 included)
14 universal board one side glass epoxidated board 72 * 47mm 60yen
15 wireing cable speaker cable subordinated from SONY mini MD CD system DHC-MD333 -
1: TA2020-20
TA2020-20 image
A digital power amp IC made by Tripath. It enables to improve electric efficiency up to 85%(at a speaker resistor value 6ohm). It will be good for the environment. Its output power is 20W+20W.
2: 1S4
1S4 image
This is a schottky diode for commutation. It endures 40V and 1A. VF = 0.4 - 0.5V.
I used these diodes in order to protect TA2020 from overshoot and undershoot, suppressing output voltage under VDD and upper GND. I selected high VF schottky diode.
3: NIKKOM 10ohm
NIKKOM image
This resistor consists of Zobel network.
I selected platy resistances for this part. Platy resistor sings between a metal-film resistor and a carbon one do.
4: NIKKOM 20kohm
NIKKOM image
These resistors determine the gain of a return circuit.
This part is important for sounding because these are setted in series.
5: metal-film resistor 8.2kohm
metal-film resistor 8.2kohm
This is used for a voltage bias.
I regret that I've forgot to purchase NIKKOM for this part.
6: MKT1826 0.01uF
MKT1826 image
A metalized polyester film capacitor for audio by ERO.
Differencial Output Capacitor. I guess this will act to reduce high frequency noise from or into a speaker, but I don't know sufficiently about this.
7: MMD2E104N 0.1uF
MMD2E104N 0.1uF image
These are decoupling capacitors.
It needed 8 altogether, so I chose a cheap film capacitor. Capacitors which have good quality in high frequency band will be better for the decoupling use.
8: MKT1826 0.22uF
MKT1826 image
These consist of Zobel network.
9: BG-PK 0.33uF
BG-PK 0.33uF image
Output capacitor.
We should adjust this value to the speaker impedance.
4ohm: 0.47uF
6ohm: 0.33uF
8ohm: 0.22uF
I love deep sound, so I chose BlackGate.
10: MMD2E105N 1uF
MMD2E105N 1uF image
This may be decoupling capacitor in output of a charge pump.
I don't have enough knowledge about this. Why is charge pump needed in this part?
11: BG-C 2.2uF
BG-C 2.2uF image
These are coupling capacitors.
This part is also important for sounding. I appointed BlackGate. I love BlackGate.
12: KMG 3300uF
KMG 3300uF image
I got these costing only 100yen, almost 1$. How cheap!!
These reduce ripple noise from switching AC power supply. Low impedance capacitors are desirable.
13: coil 10uH
coil 10uH image
These are protected from magnetism. These are packed in a TA2020KIT-SP kit.
I wanted to make coils by myself but I didn't have any measure tool which could monitor inductance.
14: universal board
universal board image
comparatively small
15: line
speaker cable image
These are speaker cables which accompanied with SONY mini MD system DHC-MD333. I guess 18 core lines.

case, power supply, cable

_ parts index number(manufacuture) price
1 switching AC adapter(12V, 5A) STD-1205(GO FORWARD) 1,700yen
2 DC jack MJ-10 (MaruSin) 30yen
3 variable resistor(10kohm 2 accompanied A model) RK27112(Alps) 1,890yen
4 toggle switch MS-500K-B (miyama) 147yen
5 heat sink 17P23L25 105yen
6 LED(12V) CTL701R (sansei) 168yen
7 knob WT-37 262yen
8 RCA plug *2 RJ-2008AT 120yen *2
9 speaker terminal *2 BP-226G a set of black and red 451yen *2
10 line speaker cable 30core line 2m(DAIEI) 188yen
11 ferrite core a type of 5mm(TDK) 100yen
12 case UC9-5-12DD (Takachi) 1,610yen
1: switching power supply
switching power supply image
TA2020 needs up to 20W+20W power. It is recommended a high power one. I chose a 12V 5A switching AC adaptor. If we can set a power supply part outside of case, an amp can avoid noises which will be emitted by a switching power supply.
2: DC jack
DC jack image
Anything goes.
3: variable resistor
variable resistor
I chose a Alps's VR.
It is expensive, but it is worth to pay. I had to cut a shaft by a saw because its nose was so long.
I'm satisfied with the feel of clicking.
4: switch
switch image
General toggle switch.
It will confuse me if it has many terminals, so I chose one which has only 2 terminals.
5: heat sink
heat sink image
TA2020 doesn't become feverish. But if there is a enough space in the case, it'll be good to set.
6: LED
LED image
It contains resistor 12kohm.
I used this for the use of pilot lamp.
7: knob
knob image
This is made with aluminium, plated by gold
8: RCA plug
RCA plug image
I felt relieved finding a tag "hi-grade plug".
But it costed very low.
9: speaker terminal
speaker terminal
It costed so much because of its good shape.
10: cable
cable image
I used this between a amp circuit and a speaker terminal.
11: ferrite core
ferrite core image
I aimed at reducing high frequency noise to use this.
12: case
case image
This looks so cool. TAKACHI produces many aluminium cases.

listmarkmaking a circuit of amp

I made it almost the same as the circuit appeared in the TA2020-20 datasheet. (datasheet by Tripath)
I changed the value of output capacitor, 0.22uF to 0.33uF in order to match the speaker impedance (6ohm).
In addition to this I connected MUTE and SLEEP to GND.

I placed parts on a universal board.
TA2020 parts on a universal board

It looks like buildings of Manhattan.

I couldn't set TA2020-20 legs into the holes of a universal board, so I bent legs using pliers.

wiring.....(it took almost three hours)

front shape
TA2020 amp front shape

back shape
TA2020 amp back shape

side shape
TA2020 amp side shape

bottom shape
TA2020 amp bottom shape

I confused by complicated cables.
It was very hard because TA2020-20 has 32 pins and they ware concentrated in a narrow space.

I regretted that I should have selected a more large universal board.

listmarkprocessing case

TAKACHI case image

I made a woody case firstly, but I failed. It was difficult to varnish and it looked like a junor high school student's work.

So I decided to make a case using aluminium.
I had never processed aluminium.

processing aluminium image

I marked using pencil and drilled holes using hand-spinning drill.
Handy drill is very useful.
If you need to drill many holes, you should use a automatic drill.

processing aluminium image2

Then next step was expanding these holes in order to match each parts.
"Chassis reamer" helped me to accomplish that.
I bought it around $10 at a DIY shop near my house.
The left one of the upper photo is "Chassis reamer".

Chassis reamer image

It was easy expanding.
It took about 2 minutes to expand one hole.

I couldn't broaden the hole of the AC plug because it was so huge, so I used a metallic file to expand.

finished processing aluminium image

I finished processing a case.
It took about 2 hours.

It was easy pocessing Aluminium.

Next, I cut a shaft of variable resistor because it was so long.
cua a shaft of variable resistor image

It was also easy to cut.
I borrowed saw from my girlfriend.
Why did she have it...

listmarksetting parts to a case and wiring

set TA2020 into a case image

I disposed almost every parts in the case.

I gave up to set a heat sink and a ferrate core on account of lacking space.

For further imfomation, please see a simple circuite figure.

TA2020 wired image

I wired.

I closed the case.
TA2020 front shape

TA2020 back shape

Forgive me to praise myself.
It looks very cool!!
I like it.

Now did it sound really?

Now red LED was lighted.

red LED was lighted

Oh no.
It turned me down.

It was "Just a case".
I paid much time and money for it.

I checked the wiring again.
It was difficult reviewing because it was like a cobweb.

Luckily, I noticed that I dind't wire 2pin, 8pin and 30pin.
I corrected the mistake and tested again hoping it would sound.

Now I switched on again.

There was no sound.
"Just a expensive case"

I was almost about to throw it, but I tried to check again using a voltage tester.

VDD : 12V (OK)
VDD5V : 5V (OK)
INPUT : about 3V (I had no idea whether this was correct.)
OUTPUT : about 7V (I had no idea whether this was correct.)
DGND : about 3V Oh!! This might be unusual.

DGND should be 0V becase it was GND.
I connected DGND and AGND.

Then I heard something.

Oh!! Finally it sounded.
I was touched by its marvelous sound.

Tears ware about to swim.

listmarkthe sound of my digital amp

My audio system is poor, TA2020-20 may cry.

foobar2000 (foo_mpg123.dll) or TV
USB sound card (Sound Blaster Digital Music LX)
RCA cable 0.5m (AT5A64)
TA2020 digital amp, up-and-coming star
speaker cable 0.3m (DAIEI speaker cable 50 parallel core lines)
mini MD audio system (SONY DHC-MD333) 2way speaker

I used the MD audio system (DHC-MD333) for 9 years before I made this TA2020 amp.

I compared the sound of TA2020 and DHC-MD333.

my subjective impression

high pitch sound TA2020's sound is very clear. Tinny sound doesn't annoy me. My favorite sound of TA2020 is tinkle of a bell. They are cool and fabulous beyond description.
The sorrow point is a sound of a female voice.
They are a bit dry. I want more moist sound.
middle range sound I feel volcals stick out. They are vivid.
low pitch sound It's very powerful. It suits to reggae musics that I love. Especially a sound of drum are excellent.
whole mood It is very clear and lively. It is easier to tell each instrument's sounds apart. I thought the sound of a digital amp were something cold, but TA2020 has a nice taste.

listmarkimproving and modifying

changed a position of VR

It was difficult to adjast volume.
I couldn't lessen the volume enough.

So I canged a position of a variable resistor.
I tried to placed it in place of return resistances (20k ohm).

It is dangerous because gain may become infinite when a bad electrical connection of VR (gariohm) will happen.
In order to avoid this tragedy, I connected 2pin and 3pin of VR.
If gariohm is happened, the gain will be under 1/2.

Now I'm satisfied.

connecting GND to the case

connecting GND to the case

GND should be stable.
I heard it is good to connect GND to a post that is buried in a yard.
I can't do that because I live in a apartment.
Hoping improvement, I connected GND to the case of amp instead of a yard.
There was a little improvement.
It might be effective to avoid noises.

Result: I feel low pitch sound got stronger.

exchanging decoupling capacitors for MUSE FX 3300uF/16V * 3

MUSE FX 3300uF/16V MUSE FX 3300uF/16V

The time has come to say good bye to KMG 3300uF.
They worked well.
In order to satisfy me, I threw a relief.
It is MUSE FX 3300uF/16V.
I bought it for 170 yen at Yahoo Auction.

MUSE FX is a audio capacitor that sounds gently.

wired MUSE FX 3300uF/16V

Sorry for troublesome wireings.

The sound changed dramatically.
1: volume got increased (Does it get high efficiency?)
2: It got more deep.
3: It got more easier to tell vocals apart.

This modifying bore good results.
This capacitor's role is important because almost every electrical charges are given from their savings.

modifying to adjust a change of my speaker

I made a bass reflex speaker. (making bass reflex speaker)
Speaker impedance was changed from 6 ohm into 8 ohm.
So I had to change the output capacitor value.

BG-PK 0.33uF (for 6ohm)
BG-PK 0.22uF (for 8ohm)

my moody audio systems

I'm now satisfied with my audio system.
I feel comfortable hearing musics from it.

adjusting DC offset

TA2020 has a problem of pop noise.
When I turn on and off, a pop noise goes bang.

I thought that the cause of this might be DC offset.

I found a article about DC offset in the datasheet of TA2024C.
It appears at 11 page.

a circuit of adjusting DC offset

I found a potentiometer was sold at very cheap price, I bought it.
But it was 5k ohm. It will be easy to adjast DC offset if the value of potentiometer is as high as 50k ohm.

910k ohm resistor 1/2W 5%RMA (RIKEN) 78yen *2
5k ohm potentiometer RJ-9W multi-rotation fine adjast VR 210yen *2
0.1uF capacitor BG-PK 50WV (BlackGate) 31yen *2

RIKEN's resistor is famous for good sound.
It helps to sound vividly.

potentiometer image
These are potentiometers.
They are so vivid.

I love blackgate
I love it.
BlackGate has privided me with marvelous deep sound.

set in case

I made the same circuite as the one which were appeared in the TA2024's datasheet.

I couldn't remove a offset voltage under follow.
Left: 70mV
Right: 6mV

70mV was so high that I couldn't remove POP noises.

I thought it would be able to adjast deviding +5V.
I must give up.

additional diary:
I tested DC offset without connecting TA2020 input and GND, so DC offset never approached 0V.
When I connected IN1 and IN2 to GND, the DC offset got close to 0V.

But a small POP noise sounds yet.
It is smaller than before, but I'm not satisfied.
The cause of it may not be DC offset.

Interestingly, the sound got more clear and vivid.
This is mysterious for me.
It seemed that DC offset should be 0V.

after this
blue potentiometer image
I found a 50k ohm blue potentiometer, I changed yellow to blue.
I bought them 80yen per one.
They are 3296W made by BOURNS.

Please give me a imformation of removing POP noise.

removing POP noise by delay relay circuit

I couldn't remove POP noise completely by adjasting DC offset.
I guess that it will be caused by the process behavior of the voltage changing while power on and off.
I tried a delay relay circuit to wait output until power supply was stabilized.

I refered following sites.
Transistor Switches
RC_circuit - wikipedia

This is a delay relay circuit using a transistor switching.

relay circuit avoiding POP noise
(If you click, the figure will expand.)

I designed a relay would be on 4.2 seconds after amp's power was on and it would be off when the power supply voltage was under 10V.

I calculated t, delay time, by following method.
time constant: C(R1+R2) = 1000uF * 10470ohm = 10.47s
The transistor will be on when the voltage between a base and a emitter get 0.6V.
R1, R2 and Tr will have 12V -10.1 = 1.9V altogether because it is descended by a zener diode.
0.6V accounts for 31% of this 1.9V.
Then read the time when the voltage gets dropped till 31% reffering the figure of "Capacitor voltage step-response" in site "RC_circuit - wikipedia".
In my case, it gets 31% when it gets 0.4 times as long as the time constant value.
So I got following.
t = 10.47 * 0.4 = 4.188s

Actually, it took 4 seconds to get relay on.
I don't have enough confidence in this calculation method, but it may be correct.
If there is mistake, please notice me.

Be careful about determining R1.
I designed that Ic, collector current, would be 60mA because the standard current value of relay was 41.7mA.
60mA would be enough for the relay.
R1 = (12V-0.6V-10.1V) / (60mA/300) = 6.5[kohm]
2SC1815 was GR grade, so its amplification would be 200-400 and the voltage between Base and Emitter was 0.6V.

But I didn't have 5kohm risistance, I adopted 10kohm which was in my stock.
It was a gamble.
10kohm would be lack for current supply of relay.

relay G5V-2 DC12V OMRON

two points of contact, two circuits relay G5V-2 DC12V 283yen OMRON

2SC1815 GR

2SC1815 GR 21yen

I set parts.

a delay circuit avoiding POP noise

I used low ESR capacitor for C.

power on: success
power off: failure

When I switched amp on, there was no POP noise.
I can say goodbye for POP noise when power on.
It would be OK that I adopted 10kohm resistor as R1.
But I can't say goodbye when power off.
POP noise goes bang about 0.5 seconds after and the relay gets off about 2 seconds after I've turned off.
I don't know that I can cut down the time shorter than 0.5 seconds by increasing the value of zener diode close to 11.4V.
I guess the reason of this will be that it takes longer time to drop power supply voltage than the time constant value of relay circuit because of huge capacity of amp's smoothing capacitors.
If you want to remove POP noise completely, it will need another relay circuit that will quickly respond to power switch when power off.

The other day I simulated this circuit using "LTspice/SWitcherCAD3".
It seemed to be quite ridiculous to simulate after completion, but I din't care any more.

simulate my delay circuit

I used a 9V zener diode model because I couldn't find 10.1V one.
Values of the other parts were the same as the actual circuit.
Vcc, the power supply voltage, was set as following.
0 - 0.1s: 0.6V - 12V
0.1 - 20s: 12V constant
20 - 22s: 12V - 0.6V
I set initial voltage as 0.6V because there were some charges in smoothing capacitors when I switched off.

simuration result

Relay will be on 4 seconds after Vcc gets 12V.
I(RL) will be maximum to 33mA, 10kohm has seemed to be so high.
I regret that I haven't simulated in advance.

stabilizing 5V

I had already stablized the 12V line, but I didn't provide any measure for 5V.

TA2020-20 has a own regulator inside of it and supplies 5V from 30pin.
2pin and 8pin recieve 5V from it.
2pin: V5D(digital 5V input)
8pin: A5D(analog 5V input)

2pin and 8pin are connected to some impotant parts in TA2020-20.
5V should be stabilized.

TA2020 a circuit around 5V

I inserted decoupling capacitors between 2pin and 3pin, also 8pin and 7pin.
Their values are recommended to be as high as some hundred uF.

The capacitors of 0.1uF roles to absorb high frequency noises.

This is C1, 22uF NICHICON MUSE ES 16V UES1C220MEM 47yen.
It should be higher value, but I didn't have such capacitor in my pocket.

C2 is OS-CON.
OS-CON is very famous.
100uF SANYO OS-CON 20V 20SA100 231yen
It offers unbelievable sound and I can't believe that it is electrolytic capacitor.

changes by this modifying
Low pitch sound got more powerful.
This modifying is very effective.
I felt that this modyfying evokes comfort and affluence.

after this, supplying 5V from a external switching regulator

It was convenient that a 5V switching regulator was left in my parts box.
I tried to supply 5V from it instead of a regulator in TA2020-20.
It was TA4805S(5V 1A) that I bought for $1.

supplying 5V from a external switching regulator
I placed laminated ceramic capacitors (0.1uF) between each terminals of the regulator.

I connected its output to 2pin and 8pin in exchange of 30pin(TA2020-20 VGEN).
30 pin should be none connect.

There wasn't enough space to set a heat sink, so I united it to the case instead of a heat sink.
Because TA4805S is molded package, it is unnecessary to take a measure for isolation.
I sticked it by double-faced tapes.

supplying 5V from a external switching regulator

I'm about to blowout by lacking in space.

change of sound by this modifying
It changed dramatically.
It got being sharpen and vivid as if I changed from cassette to CD.
It seemed to be good for audio device that volutage supply is given by independent regulator rather than built-in one.


I decided to put letterings because I couldn't distinguish ON from OFF.
I bought a lettering set for about $2.

lettering on-off

lettering each terminals

It was easy to letter.
My amp looks like as if it appeared in the market.


compared size with a lighter

My amp is very small. It's clear to compared with a lighter.
It has 20W+20W output, I can't believe that.
I made it for about $100 and It sounds so nice.
How hight cost and space perfomances!!

I heard that it is much more easier to make amp of TA2020 with the kit of TA2020.
But I don't regret making on a universal board because it costed lower and it appered my originality.

I'll be grad if you get interested in making audio.

This homepage allows you to link to your site.
It is unnecessary to contact me linking.

I hope the exitement of making audio will spread.

listmark If you have a question or comment, please use BBS.
If BBS doen't satisfy your needs, give me a mail. (replace # by @)

Notice!! I don't take any responsibility for the trouble that can occur for this homepage. Please do at your own risk.