9 Channel 3E Audio Amp(s) project

[activewire]

My new amp modules just arrived! I will soon (after a 2 week trip) start building these into my home theater system - an Onkyo RZ50 in 5.2.4 configuration. This build is actually 2 separate but identically looking amps:

- a chassis containing (3) mono amp modules powered by Mornsun LOF750-20B48 (this is still in transit) that will drive my L/C/R front sound stage
- a second chassis containing (3) stereo amp modules again powered by a LOF750-20B48 to drive my rear surround pair and 4 atmos height speakers

These modules are the Rev B Jan 2024 with the extra 12 volt header, Nichicon caps, 1656 op amps and I must say the build quality is simply outstanding. Hard to believe the power to size ratio of these Class D modules, with sound quality measurements surpassing my 80 pound McIntosh MC7300 beast.

Because my AVR (like every AVR costing under $5K) has only RCA outputs, I must perform surgical mods on each module - solder on an SMD OpAmp, remove 2 resistors, and bridge one jumper. That will be so much fun ... times 9. Yet this still seems like a better solution (SQ and power wise) than RCA-to-XLR adapter cables, but feel free anyone who can talk me out of it.

Also in this pic is an older 4-channel 3E amp module from a different project and I've been extremely happy with! Will keep you posted on my progress, as more parts arrive and I get going on the build.

 

[Guddu]

My new amp modules just arrived! I will soon (after a 2 week trip) start building these into my home theater system - an Onkyo RZ50 in 5.4.2 configuration. This build is actually 2 separate but identically looking amps:

- a chassis containing (3) mono amp modules powered by Mornsun LOF750-20B48 (this is still in transit) that will drive my L/C/R front sound stage
- a second chassis containing (3) stereo amp modules again powered by a LOF750-20B48 to drive my rear surround pair and 4 atmos height speakers

These modules are the Rev B Jan 2024 with the extra 12 volt header, Nichicon caps, 1656 op amps and I must say the build quality is simply outstanding. Hard to believe the power to size ratio of these Class D modules, with sound quality measurements surpassing my 80 pound McIntosh MC7300 beast.

Because my AVR (like every AVR costing under $5K) has only RCA outputs, I must perform surgical mods on each module - solder on an SMD OpAmp, remove 2 resistors, and bridge one jumper. That will be so much fun ... times 9. Yet this still seems like a better solution (SQ and power wise) than RCA-to-XLR adapter cables, but feel free anyone who can talk me out of it.

Also in this pic is an older 4-channel 3E amp module from a different project and I've been extremely happy with! Will keep you posted on my progress, as more parts arrive and I get going on the build.
View attachment 364352

All the best.

A few pictures as you progress would be great addition here!

 

[Sokel]

This cap seems like is put at the opposite:

And by that I would check all components.

 

[MCH]

My new amp modules just arrived! I will soon (after a 2 week trip) start building these into my home theater system - an Onkyo RZ50 in 5.4.2 configuration. This build is actually 2 separate but identically looking amps:

- a chassis containing (3) mono amp modules powered by Mornsun LOF750-20B48 (this is still in transit) that will drive my L/C/R front sound stage
- a second chassis containing (3) stereo amp modules again powered by a LOF750-20B48 to drive my rear surround pair and 4 atmos height speakers

These modules are the Rev B Jan 2024 with the extra 12 volt header, Nichicon caps, 1656 op amps and I must say the build quality is simply outstanding. Hard to believe the power to size ratio of these Class D modules, with sound quality measurements surpassing my 80 pound McIntosh MC7300 beast.

Because my AVR (like every AVR costing under $5K) has only RCA outputs, I must perform surgical mods on each module - solder on an SMD OpAmp, remove 2 resistors, and bridge one jumper. That will be so much fun ... times 9. Yet this still seems like a better solution (SQ and power wise) than RCA-to-XLR adapter cables, but feel free anyone who can talk me out of it.

Also in this pic is an older 4-channel 3E amp module from a different project and I've been extremely happy with! Will keep you posted on my progress, as more parts arrive and I get going on the build.
View attachment 364352

Cool project, please keep us posted.

I recommend you to read this:

https://www.ti.com/lit/an/slaa787/slaa787.pdf?ts=1713359668373&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FTPA3255

I believe the 3e modules have all necessary pins exposed, what is excellent.

 

[activewire]

Cool project, please keep us posted.

I recommend you to read this:

https://www.ti.com/lit/an/slaa787/slaa787.pdf?ts=1713359668373&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FTPA3255

I believe the 3e modules have all necessary pins exposed, what is excellent.

Indeed the pins ARE exposed for master/slave configuration. Yet I’ve read (in multiple places) that this is generally NOT necessary (perhaps even ill advised?) unless you fall into an exotic use-case of having AM radio receiver parts within the same chassis? My use-case is home theater with blu-ray and AppleTV source, occasional 2 channel audio with Sonos source (also my amp chassis will NOT contain any other components beyond 3 amp modules and power supply).

I will do some testing both ways, although not sure I have the knowledge or equipment to properly evaluate this condition.

Here is one example Post on topic
https://audiosciencereview.com/foru...o-v3-mono-amplifier-review.53474/post-1945437

 

[activewire]

This cap seems like is put at the opposite:

View attachment 364410

And by that I would check all components.

Oh wow great catch! You just saved me from an 8 channel / 5.2.3 atmos theater. While Ken Griffey Jr may have mainstreamed a backwards cap, that’s no bueno for circuit boards.

I will check everything when I get back to my home country. Paging @3eaudio please advise.

 

[activewire]

My progress is slow. Here I have de-soldered

And then re-oriented the capacitor thanks again @Sokel

My power supplies have arrived and the dimension is quite small, nearly the same as amp boards. I am playing around with different layouts and now considering perhaps a single chassis for this whole 9 channel build. The white base in this picture is 17" x 14" which is about same size as a typical audio component so what do you guys think should I go one or two chassis?

 

[DVDdoug]

Because my AVR (like every AVR costing under $5K) has only RCA outputs, I must perform surgical mods on each module - solder on an SMD OpAmp, remove 2 resistors, and bridge one jumper. That will be so much fun ... times 9. Yet this still seems like a better solution (SQ and power wise) than RCA-to-XLR adapter cables, but feel free anyone who can talk me out of it.

I'd go with XLR adapters. There are potential benefits to a balanced input even with an unbalanced signal, (It's trickier going the other way.)

 

[3eaudio]

Hi @activewire great project on multi-channel amp.

1. for Master/Slave setting, if you don't have AM device around then it is NOT necessary to do this as involve many wire routing. and also think about old device or other amp(hypex?ICE) how can they set Master/Slave in multi-amp case, my experience is that the simple the better for diy.
2. for RCA input, the best implementation is having a modify in the amp to adopt a single-end to balance convertor, but this is assume you have the capability to do that and want to get best performance with RCA input. Without a modification still can feed single-end input but won't hit the maximum output power as one of the amp output bridge not scale fully balance.So if you don't need full power of the amp and your RCA Gain is good then you don't need to modify the amp.

 

[PlasticDoc]

.

It is a significant issue to sell amplifier boards with electrolytic capacitors mounted in reverse polarity. This is poor quality control.

It is even more problematic to keep quiet about it after the client shows the photography here, and not properly acknowledge the issue and send a new board to the client.

This will hurt 3eAudio credibility.

 

[activewire]

It is a significant issue to sell amplifier boards with electrolytic capacitors mounted in reverse polarity. This is poor quality control.

It is even more problematic to keep quiet about it after the client shows the photography here, and not properly acknowledge the issue and send a new board to the client.

This will hurt 3eAudio credibility.I fixed the board myself in 10 minutes effort.

You saw my post about having fixed the board, not a big deal.

 

[activewire]

I am now about halfway done with the board alterations for single ended RCA inputs. Here are some pics of my workstation and a magnified view of one of the channels with OpAmp added. Hey @3eaudio Is there a decisive way for me to test these alterations?

As expected this is quite a pain and I would gladly have paid an extra $20 per board to not be doing this. @3eaudio please consider the next board Rev just include the OpAmp and have a jumper to select balanced vs SE, very few people are brave like me to solder SMD components.

 

[rsqt]

Nice project.
Do you know if the LM317 regulator is for the voltage of the opamps?

 

[activewire]

Update: I have the amplifier up and running, although at the moment this is inside a "Frankenstein" temporary case (an old Denon AVR) while I have been looking (without success) for the "perfect" case.

Internal wiring: despite my best efforts to layout and route cables following all audio design best practices with regard to cable crossings, this is a 9 channel amp inside a single case so that means there WILL be cables crossed. As such, I have decided to shield all (most) of my cable runs (signal inputs, 48V amp power, speaker line outs) by wrapping them with copper foil. Yes of course all cables are twisted/braided!

Then (at one side only) each shield gets drained directly into the case (which itself is grounded at a single location) by soldering a tail to the copper foil. This photo shows the signal wire on one of the stereo modules, where I use just one drain wire to span both the left and right channels.

The final result gets outer braid cover and shrink wrap at each ends

With 2 power supplies, 3 mono amp modules, 3 stereo modules this was a LOT of custom cable fabrication! I also made several errors which caused rework for many of these!

My next challenge was getting 3 pairs (positive/negative) of leads from each power supply which were not really designed for this especially in a such tight case.

My solution was to fabricate dual vertical bus bars from copper like so

For safety (avoiding exposed power bus) I fabricated the white shields from PVC pipe using dremel and shaped by heat gun.

 

[activewire]

Major System Components (I will provide links to each part and price after I finalize the design this is still evolving)

• 2 power supplies 750 watts each (one direct to 110V mains; other one is relay controlled)
• 3 MONO amp modules powered from PowerSupply#1 config as 1 Master + 2 Slaves (independent from Stereo bank)
• 3 STEREO amp modules powered from PowerSupply#2 config as 1 Master + 2 Slaves (independent from Mono bank)
• 12V trigger input: electrically isolated with PC817 photocoupler
• Front panel controls: clone of LSA Warp 1 design with power switch (5 volt, latching, with ring illumination blue); yellow standby LED; red fault LED; red clip LED; red temperature LED
• Rear panel: 9 speakON outputs; 9 RCA inputs (yes I know XLR>RCA but this is home theater and I cannot afford a $6K AVR with XLR outs!)
• MCU: my working prototype uses RasberryPiZero and a simple Python program, meanwhile I am dusting off my C language development skills and learning the PIC16 architecture, targeting either a PIC16F18456 or (slightly smaller) PIC16F18446, am now tinkering with Curiosity Nano Board
• Amp status monitor circuit: instead of needing 12 digital inputs on MCU (to supervise 6 amp modules each having a Clip and Fault) I have used a 7425 (Dual 4-Input NOR Gates) which combines all amps into a single Clip detect and a second one for Fault detect, thereby demanding only 2 digital inputs on MCU
• Air temperature sensor: probe mounted inside case
• 5V fan connected to PowerSupply#1 (cooling available at PlugIn state)
• 12V fan connected to SlaveAmp#3 (cooling available at TriggerOn state)

Operational States

• System Off: plug out and/or rear IEC rocker switch off; all systems and lights off
• Plug In: mains 110V directly to PowerSupply#1 in standby (5V on / 48V off); MCU boots and active monitoring loops for temp control+front power switch; all panel lights off
• Switch On: yellow Standby light on; MCU active monitoring loops for temp control+front power switch+12V trigger
• Trigger On: yellow Standby light off; blue power light flashing; relay mains 110V on to PowerSupply#2 standby (5V on / 48V off); PowerSupply#1 full on (48V+5V); PowerSupply#2 full on (48V+5V); soft amp start sequence MasterAmp#1→ MasterAmp#4 → SlaveAmp#2 → SlaveAmp#5 → SlaveAmp#3 → SlaveAmp#6; MCU active monitoring loops for temp control+front power switch+12V trigger+amp faults+amp clipping; blue power light solid (cease flashing)
• Trigger Off: blue power light flashing; soft amp shutdown sequence SlaveAmp#6→ SlaveAmp#3 → SlaveAmp#5 → SlaveAmp#2 → MasterAmp#4 → MasterAmp#1; relay mains 110V off to PowerSupply#2; PowerSupply#1 standby (5V on / 48V off); transition lights/MCU into SwitchOn state
• Switch Off @ TriggerOn: Perform TriggerOff sequence; transition lights/MCU into PlugIn state
• Switch Off @ TriggerOff: transition lights/MCU into PlugIn state

MCU Control Loops

• Temp Control ideally the fans would never run (ie, not create ANY noise although these are super quiet fans) so the MCU measures air temp with probe from ADC input; temp<XXX both fans off; temp>XXX switch 5V fan on, front temp LED slow flash; temp > YYY switch 12V fan on (continue 5V fan on) front temp LED rapid flash; temp>ZZZ perform high temperature shutdown with front temp LED solid and fault solid red
• Front Power Switch Power switch is low voltage (5V), so MCU monitors switch position On/Off and then software performs the requested function
• 12V Trigger Trigger is optically isolated to avoid bad things like ground loops. At first I tried just powering ALL components on at once but found this produces a 7.5 amp inrush current. That discovery is what lead me to an MCU in the first place. A carefully orchestrated soft start (single component at a time, with slight delays between) now has this down to a worst case inrush of 1.5 amps
• Amp Clipping The 7425 combines all 6 amp modules into a single Clip signal to MCU which turns on the front panel red light, any clipping would typically be short duration and does not require any further action from MCU just signal LED as clip occurs
• Amp Fault The 7425 combines all 6 amp modules into a single Fault signal to MCU which turns on the front panel red light and performs a soft shutdown, perhaps instead software could perform an automatic reset of the amps, but for now I am just curious will I ever get a Fault (and what conditions)

 

[richardm]

This is inspirational. I've just won a non-working 16-channel zone amp on eBay. If it's not an easy fix I plan to gut it and use the chassis for a project similar your own (only far less ambitious). The plan is to throw a 7.1 USB "sound card" in there and some Wondom modules along with a Meanwell SMPS.

 

[activewire]

Shots below of my working amp in Frankenstein case, some pieces not shown (fans & prototype raspberry pi controller ... I am working on an PIC16 based MCU). What I want is a quality aluminum case 17" wide (standard width) and 12" deep by 5" tall but I just cannot find ANYTHING out there (even in the $150-$200 price range) in this approx size which does not also bring multiple compromises. This may sound ridiculous but I am contemplating custom building my case from aluminum stock with laser cut panels.

 

[activewire]

Being a fairly inexperienced amp "designer" I have lots of questions so I would like to measure several variations of the amp: master/slave vs all masters .... with/without shielded drains connected to chassis ... etc. Therefore I am purchasing the necessary equipment to take distortion measurements. The ADC has arrived but I need to purchase a DAC and some dummy resistor loads. Anything else I will need? I already have a digital oscilloscope (not sure if I need that). Lots more to do, lots more to learn.

 

[activewire]

My MCU design, still working on this in breadboard form

UPDATE: old diagram is posted below see revised design on Page 2 of thread

 

[mcdn]

My MCU design, still working on this in breadboard form

Looks good! A couple of recommendations for robustness:

• Add a 1k resistor before TRIGGER+, and a 1n4148 or similar after TRIGGER-. This will limit the current to 12mV and protect against reverse voltages across the optocoupler (it only has a 6V reverse voltage limit)
• It is best to lower the voltage across the power relay after switching it on, this will greatly increase its life and reduce power consumption. A passive circuit like this can be used (use the net at Node2 to drive the relay) or use a transistor controlled by a second pin on the MCU
• Consider using transistors to control the LEDs, rather than drawing current from the MCU pins directly
• Consider adding a diode in front of pin 1 to protect against reverse voltage from VCC

Hope that's helpful!