Discrete Opamp Vs Monolythic!

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Which file is which?

  • File "A" is Discrete and "B" is Monolythic

    Votes: 0 0.0%
  • File "A" is Monolithic and "B" is Discrete

    Votes: 7 100.0%

  • Total voters
    7
  • Poll closed .
As a kid/teenager in 80's I've built shitloads of "big" opamps - 60, 120W etc poweramps, and just for fun, some smaller 2-5 watters (that would probably qualify as some sort of homebrew DOA's even if I didn't have a clue about what is a DOA then... )

I still remember vividly the difference(s) in sound of different flavors of these circuits - discretes vs. other variants of discretes, and even more pronounced, discretes vs. "chipamps" of that era (TBA, TCA and TDA chips). These differences intrigued me and as a result, I built - shitloads of diff. circuits and tweaked away as mad.

And I'm hearing the SAME underlying tonal characteristics here. Not much has changed.

I experimented a lot back then (even as schoolwork, homework etc) with different configurations and it came down to one characteristic:

a "resistor-loaded" circuit (including bootstrapped) will sound sweeter (or phatter) compared to a CCS or current mirror loaded circuit. CCS-loaded amp can sound "wider", but not "deeper" (i.e. the sound character gets closer to chipamps).

Most optimal sound character in those "big" opamps for me was a "quasi-ccs" (resistor/zener combo in the LTP, resistor-loaded collector) and CCS-loaded VAS. Some (look @ diyaudio forums) prefer bootstraped VAS (cap/resistor), but CCS can swing closer to rails if beefed enough ..

I remember liking the "sound" of BD139/140 in VAS, BD243/244 (?) and BDX18 in output stage (and of course 2N3055 and 3054 for the cheap ones). BC182 usually found its way into LTP in most of my undertakings.

I remember once I got some unknown japanese TO220 transistors which I swapped for BD139/40 in one of my "working specimens". It was like "what the f** happened?" Totally bigger, deeper and phatter sound that didn't last for long because I messed up with a human error resulting in smoke/sparks.

These "topological" differences really do change the tonal character. And as I said, I'm hearing the same underlying tonal character in these mp3's.
 
tv said:
...I remember once I got some unknown japanese TO220 transistors which I swapped for BD139/40 in one of my "working specimens". It was like "what the f** happened?" Totally bigger, deeper and phatter sound that didn't last for long because I messed up with a human error resulting in smoke/sparks...

interesting! I had the same thing happen to me when recently testing an opamp I made. Sounding AMAZING charged with energy, very very nice sounding. One of those things you could instantly tell it was better sounding than all the other  ones. But it didn't last very long either! I had a diode backwards so it fried the output resistors. But It showed me the more current you can shovel in the output stage, the better these seem to sound.
 
abechap024 said:
... the more current you can shovel in the output stage, the better these seem to sound.
That's the idea behind the Class-A operation mode. This can get very hot and expensive. Been there, will never return (with poweramps).

What I described was what happened in a low-bias AB-class "opamp" poweramp with just a VAS transistor change.

But, I also discovered that you can get the phat "Class-A-like" vibe if you run more current through your VAS, but it must of course have beefier devices. With very modestly biased AB output stage, as described.

In 90's, I also designed an odd-looking amp upon this principle (very funny VAS and output stage), that went into "beta-testing" phase with a friend of mine. But, there was a very serious, randomly occuring problem with components (it occasionally fried all the output devices and drivers, but IIRC only the same 3 of 6 identical channels, so wtf??) and it was decided not to develop or even market it further.

But yeah, if I look upon it this way, I've built a ton of "fat" DOAs in my life, haha.

One thing that struck me was when (some time ago) I read a thread on diyaudio where the designer of AKSA, the succesful DIY-kit amp, spoke about his testing and listening experience when voicing his amps - very similar findings I described in my earlier post (and yeah, I had much "cheaper" components at my disposal then).


In other words, try running more current thru the Class-A VAS and less current thru your output stage if you want more fat but less sparks and smoke.

You can also try to look upon this as a small Class-A amp (i.e your VAS) wrapped inside a big bad current-dumping Class-AB output stage. Or not ..

Ah, memories..
 
Just took a listen on my laptop speakers. The discrete sample jumped out of the speakers, while the monolithic sample sounded a little cloudy. Going to listen on my monitors (soon I hope). Very cool test!
 
But it was pretty obvious for anybody that has heard a difference between a chip-sound and discrete sound. Not saying that chip-sound is inherently worse, just different character that may or may not suit a certain style.

That aside, in the meantime I redesigned the melcor opamp now with the "dumping" output stage - and some other tweaks that would make it a more contemporary, perhaps faster circuit, with lower distortion - but with similar "harmonics" footprint (similar looking falloff when viewed in FFT, but a few dB lower). It simulates well and from my previous experience with LTspice, it "should" work right when built.

If you have some melcor PCB leftovers, you can probably "deadbug" it from there. In fact the redesign reduces component count and unifies some comp.values. In other words, I think it will offer a good performance with the "frugal league" components (BC550, BD139 et al). The design as-is sets the output BJT pair quiescent current at approx 25mA, so it should stay in class-A for 600ohm loads. But that would imho require some heatsinking. Iq can be pre-set with the choice of different degeneration resistors, though. For small-footprint amps like DOAs perhaps it would be better to set it to cca. 10mA, but that woud be best done "live", with the actual devices that would be used in the final build.

If you are interested, I'll open a new thread in "drawing board". The schematic will be free for DIY use, I will think about what compensations regarding any other commercial exploitation.
 
I need still to take some care of square-wave response (symmetry), but I think I'm 95% there. I'll probably try to convert the 2520 as well, because this one seems to be the "peoples darling".

Sometime over the weekend, perhaps. Will you beta-test?
 
Only 2 interested so far?

Not that I would "seek attention and/or fame" here, but I'm not really eager to polish and publish schematics (and description) for circuits that wouldn't be "consumed" by DIYers here. Especially since these DOAs are circuits that are somewhat known to be "in demand" and have potential to generate ROI for random unknown vendors.

In other words, I'm not eager to do free R&D for random idea seekers that would commercially exploit "my efforts" without even letting me know about it (and there are a multitude of ways to generate profit with DOAs).


Anybody else of "resident DIYers" interested? Raise your hands.
 
Hands up, baby hands up, eh?

I'm almost done with simulation of two different melcor circuits. The core circuit will most probably be very tolerant of active device variations (it simulates well with every possible model of BC54x, 55x, 56x I have).

The actual builders will have to geat their hands dirty with the output section, though. No simulation can replace that. This means tailoring two resistors to match your actual output devices (BD139,140 - use at least the 10 suffix, 16 is recommended, i.e. BD139-16) and setting the quiescent current to cca 25mA. If you plan to build several opamps, my findings were that once measured, it will be OK within a batch of devices. When you buy new batch or from a different manufacturer, repeat the procedure...

Beware - these are funky class-A circuits, which means there is always danger of thermal runaway! Based on my experience with similar circuits running at similar voltages (30V-something), the output devices will have to be somehow heatsinked.

The good news is that the component cost per one opamp will be under 2EURO, based on farnell prices.

I'll try to post a basic perfboard layout as well.
 
I'm almost done, I think it's going to be in "drawing board" tomorrow.

Schematics, writeup, FFT and THD comparison (at 12Vpp output into 600ohm - 14.77dBu). Two different versions, near-identical twins so to speak.

Frugal to the core, including perfboard layout (you will need one anyway if you plan to build more of these - to correctly setup the output stage and perform some necessary testing, meaning socketing some components and using a multimeter).
 
Back
Top