[WORKLOG] 6 channel integrated 'stereo' amp with lots of features

first and foremost, BE NICE.
this is a sharing community, and i don't get paid nor do i seek acknowledgement from some virtual sifu by documenting this.

the initial test is the 10 channel inverted gainclone explained here. the amp is a preliminary testbed before moving on to bigger and more complex design. explains the wires hanging around the case.
multichannel gainclone with more than 2 channels are rarely documented on the net. i need a place to test.

the planned outcome is to build a stereo amp with 2 channel input (duh) and 6 channel output with a 3 way active crossover driving my relatively cheap 3 way speaker.

the planned power amplification section is to use the LM3886 in the best configuration possible, which is the Mauro Penasa Rev. C (essentially a multiloop gainclone in howland current pump arrangement).
the planned 3 way active crossover is using the Linkwitz-Riley 24db/octave active crossover in 3 way configuration.

a simple tube preamp (bypass-able) is also planned to add more flavor to the sound, modified from the Cavalli-Lovell-Pasculle hybrid tube headamp. the working voltage of this amp is rather safe (circa 75V DC) and parts are easy to source without the need for output transformer coupling. and they don't sound half bad.

also planning a HT bypass function (maybe a crude one at that) to be able to integrate the amp into my current speaker setup.

the poweramp schematic. lots of people are saying this beats some rather well known class A amp and comes close to the best of class A amps whilst running in class AB.

don't trust everything you read on the net, so build yourself one and try with your own ears. costs less than RM100 to build a case-less 1 working channel. beats the price of any good class A that you could buy.

yeah an electronics degree could probably help. that would cost you about RM30k. that class A amp does look enticing once you factor in the hidden cost.


populating the PCB board with components. the PCB is selfmade and self-designed. 3 channels per board. parts are measured to the ones that we could get locally. and there's a few modification here and there according to my personal preference (good thing about building your own amp).

and yes there's a test board made prior to this for brief listening test to test for possible upgrades.

upgrades made includes:
class A biasing using low-noise toshiba jfet (about 3-5mA for each channel) for the voltage gain stage (LM318).
regulated DC voltage for the dual-ended LM318 power supply using LM317 and LM337.
increased LM318 supply voltage to 18V after reading that it increases the LM318 slew rate and made it sound better. well the voltage is adjustable from 8V to 20V.

both board fully populated, and tested. DC offset per channel is below 20mV with most channel giving around 10mV while on load.

looks right at home inside the planned case.

ok now have to work on the active crossover PCB.
planning to integrate the tube section right inside the PCB while still measuring less than 15cm x 15cm with everything.
which could take a while seeing how my current progress not in a hurry anyway

the schematic that i've posted before is cropped. there's no power supply section and there's no speaker protection section. the design comes with an integrated speaker protection circuit and power on delay to protect the connected speaker if should there's something wrong somewhere.
and with the power on delay, there's no pop or clicks during turn on and turn off.
power on delay is about 2 seconds, power off delay is instatenous, and speaker protection kicks in at about 1.5V of offset.

the full schematic:


well i did say an electronics degree would probably help

PCB for the tube pre, active crossover and output equalization buffer are completed. developed using UV exposure.
board measuring 15cm x 15cm.


it looks small during drawing, but freaking huge once developed. i'm starting to have doubts whether such a large and complex circuit could sound better than a simple passive crossover inside the loudspeaker. most builder claim that an active crossover sounds better than a passive one, but looking how complex the circuit is (there's 12 dual channel opamp in there) i could only guess.

i'm a firm believer than a simple circuit would (almost) always produce better sound than a complex one. less parts would mean less colouration to the sound.
let's hope i'm proven wrong.

p/s: it would also be beneficial to have access to a PCB lab for such a large circuit. toner transfer method just won't cut it.

the preamp is taking shape pretty nicely. the sockets are for resistors which will tune the crossover point.

just a few more capacitors and it'll be complete. but think i'll stop for a while.

Hi Jat,
Very neat and tidy work ! Much better than mine !
Must be many hours of soldering.
Your pcb is very nice. Can you show some info about making the pcb ?
thanks

kp93300

preliminary test wiring.


all the channels are tested. the active crossover are working perfectly (minus 30 minutes of troubleshooting one of the high pass filter giving 4V of offset to other channels because i forgot to link one of the jumper wires).
and wasted a few hours trying to figure out why the tube pre is giving such a high humming/buzzing noise. i've built the same configuration a few times before this and all are buzz free so no idea why this is giving me such a headache.
the circuit are all correctly drew and fabricated. current to the plate are correct, bias voltage, plate voltage are all checked and double checked. must be the tube itself but since i have no other compatible tube to swap it with, i could only guess.

decided not to waste anymore time with the tube pre so i bypassed it for the time being. good thing i built the tube pre on a separate section of the pcb, making it totally bypass-able.
have to build the tube pre on a protoboard and check again. revisiting it later.


the preamp/active crossover section are completed (minus the tube section).
crossover point are (tried to) set at 140Hz and 2.2KHz as per manufacturer's passive crossover. just to get an idea how much improvement could be had moving from a passive crossover to an active one.

the grey capacitor used are Vishay MKP1837, 47nF. raved quite a lot of the net. compared against other capacitors here.
other better sounding caps are either too expensive or too big. this strikes a balance between the two and comes highly recomended.
tho i don't really like them as they sound too bright out of the box. maybe will improve with time but i usually don't have the patience to wait for that long.

but already bought more than a hundred of them so i'm forced to use them and wait for them to burn in.
the crossover point are set at 140Hz and 2.2Khz but i'm setting them at 143Hz and 2.195 KHz because i have lots of resistors for that point.

kp93300 wrote:Hi Jat,
Very neat and tidy work ! Much better than mine !
Must be many hours of soldering.
Your pcb is very nice. Can you show some info about making the pcb ?
thanks
kp93300

soldering is the easy part, troubleshooting is the hard part. easy mainly because i have these pair:

they do help a lot. tho out of reach of most diy enthusiast. once you try these you definitely won't go back to normal handheld iron and a sucker.

i'll snap pics of the pcb making process once i start working on my DAC.

jat wrote:

both board fully populated, and tested. DC offset per channel is below 20mV with most channel giving around 10mV while on load.

looks right at home inside the planned case.

ok now have to work on the active crossover PCB.
planning to integrate the tube section right inside the PCB while still measuring less than 15cm x 15cm with everything.
which could take a while seeing how my current progress not in a hurry anyway

Hi Jat, that's surely a lot of work. Take my hats off to you.
On a side note, I was wondering if you notice, perhaps your mains transformer is undersized for the job to power 10 channels?

cheers.

Wikin wrote:Hi Jat, that's surely a lot of work. Take my hats off to you.
On a side note, I was wondering if you notice, perhaps your mains transformer is undersized for the job to power 10 channels?

cheers.

that's a 500VA transformer. i've ran one of the channels through my dc power supply and played some music through it. it doesn't suck more than an amp of current even at ear bleeding volume. at moderate volume, one channel doesn't even reach 500mA of current draw.

500VA is good for 10A of continuous current supply. and the amp is only running 6 channel in which not all 6 would be running 100% load at any given time with the active crossover separating the load to 3 amps per channel.

in any case i will get another toroid rated at 1kVA just to test if there's any improvement to be had by oversizing the transformer. seen one selling at RM300+. this 500VA costs me RM200.

and a friend of mine is getting a Nuvotem-Talema 500VA toroid from rs components. that's a proper toroid with (hopefully) good quality control. unlike my cheap china made toroid. will be comparing if it makes any sonic benefit to splurge a bit more for branded toroids.

added:
i could maybe run a channel powered through my dc power supply and load the channel using a dummy 4 ohm resistor. and feed the channel with some 20Hz sine wave. that should load the amp at 100% and monitor the current draw.
i might hit more than 1A at that config. just to get a brief idea on the max current the amp would require.

jat wrote:
in any case i will get another toroid rated at 1kVA just to test if there's any improvement to be had by oversizing the transformer. seen one selling at RM300+. this 500VA costs me RM200.

Yes this will be a good exercise and experience. There are a lot of things on the bench top testing in static condition that somehow does no correlate to the sound produced. If static current, harmonic distortion measurements etc etc is 100% correlated to the sound then we won't have that 1980s solid state amplifier crisis and all amplifiers should sound the same when designed up to 0.001% THD. Not forgetting Cyrus 2 will not produce a dedicated PSX-R

cheers

Wikin wrote:
Yes this will be a good exercise and experience. There are a lot of things on the bench top testing in static condition that somehow does no correlate to the sound produced. If static current, harmonic distortion measurements etc etc is 100% correlated to the sound then we won't have that 1980s solid state amplifier crisis and all amplifiers should sound the same when designed up to 0.001% THD. Not forgetting Cyrus 2 will not produce a dedicated PSX-R

cheers

precisely. not everything are based on just numbers and calculation. tho i am someone that you would call a technical aficionado (my line of work require me to be so), i don't believe that everything could be measured or be put into a formula.

even a different power supply capacitor could make such a great difference to the final outcome that you hear on the speakers, yet i doubt a measurement device (any) could measure how that simple power cap changes the THD and produce the difference.

going technical or believing in measurement is a good thing. but once you reach a certain point, you start to throw all that textbook through the window. especially when you build some certain circuit exactly the same, yet with a component change they would sound totally different from one another (yet still measure exactly the same), you know that something is amiss somewhere.

or when something that measure worse than another circuit could sound sonically better to your ears.

tho i have to agree, better numbers will look good on paper (and on your report).

final picture of the current resident 6 channel stereo amp.

it's been giving good performance all these while. tomorrow will be the last day the 6 channel myref rev. c see the light of day. it will be stripped down and rebuilt in a bridged mono configuration, 6 channel active with 100w each into 8 ohm.

the active crossover will also be rebuilt in a 2nd order 6db/octave slope. goodbye myref, you've been kind. but you lack bass impact and i found something better.