Time to DIY like it's 2009! (hobby has changed, man)

There is Discrete dynamics power amp design listed here (comes as fully populated PCB with heatsink attached):
FS: Discrete Dynamics Class A/B Power Amp modules - diyAudio
Another option is PeeCeebee amp:
PeeCeeBee - diyAudio

 

We can still DIY in 2021 with the best of them. As far as tubes are concerned there are very good players on the market for iron and materials have generally come down in price quite a bit. The beauty of DIY is expressing one's self and taking time to craft something cool.

Here's one no one tried before, a 300B driving 300B setup for current hungry planars. Doubles as a nice space heater and needs 200W

 

Yeah, there is definitely still room to create something incredible.

I think what I really lament is the barrier to entry, and what that means for the future. That insane-in-a-good-way 330B^2 is surely amazing, but the knowledge and resources needed to do that are incredibly high. Where are hobbyists learning the skills to do that, if there isn't a good community around the cheap-and-cheerful designs, where they can learn and hone their craft? It would be really sad if the only people who could feasibly DIY are experienced electrical engineers.

 

You used to be able to buy a cmoy kit for <$20 and really make a mess of things using papa's old soldering iron with the rusted tip.

Looks like JDS still sells a kit (sans tin) for $42, but by that point it's still a cmoy and I'm spending a money on batteries, and I'm pretty sure I can find a FiiO e10k for just a couple bucks more and that's both dac and amp... So why bother?

If I were just getting into the hobby now and didn't have any knowhow, I wouldn't see much to encourage me to go down that path. The Bottlehead Crack might be the best "entry level" project for someone adventurous, but that's still going to be pricier than most beginners can afford.

 

Yep. There's no obvious gateway drug.

*

My projects are proceeding slowly. Life gets in the way, and still trying to get hold of Soekris dam1021 modules. But I did receive Mouser and Digikey orders today with the last parts for several things, so hopefully I can post some more nostalgia soon...

 

Attempting DIY at entry level is not so rewarding with companies offering good sounding amp/dacs at $100, used ones go for even cheaper pricing. But if one wants to build a high performing amp or a dac which can sit with stuff in kilobuck range, DIY still provides lot of value.

 

I feel like I should share the disaster that is my current work bench. This in contrast to the pristine condition of @TomB ultra clean workspace where he's assembling the walnut dacs. I'm more like a soldering troll with the remnants of my discarded projects scattered in boxes around the basement.

 

My workspace is currently loaded with all my 3d printing crap. It doesn't look nearly as cluttered as that at first glance, but there are few things quite as awful as sitting down to work, then discovering a previously unknown puddle of semi-dried resin.

 

due to wfh, i've been able to justify lots of diy projects and upgrades. some of which i have shared and more to come
thanks @tommytakis for helping me with this

 

I don't normally like posting my projects until they are complete, and this new build is the end point of my little journey...... but I am just so giddy with DIY excitement that I just had to share an in-progress pic!

 

I can share some excitement as well, half way done with the PSU of a tube amp. It's a heavy boy.

 

nice beginnings of a Soerkris DAC

 

Bloody hell, that's a LOT of chokes. Obscene in its magnificence. Or maybe magnificent in its obscenity.

Yep! It took me four months to snag a dam1021 rev7 board, but I think it will be totally worth the wait because it is a significant redesign that brings a lot of the improvements from the dac2541.

 

@m17xr2b my back hurts just looking at that!

 

All for that one little tube.

 

And it only has the heater connected! Glow baby, glow.

/s

 

In many ways, this post represents the end of my journey, even though I haven't completed many of the interim steps. Time is precious, and building this new project has taken less time and effort than the other upgrades I'm working on, so I'll come back to those.

Long and short of it... in my gear reassessment I decided I needed a new DAC, and I still wanted to DIY. But the value proposition of DIY DACs is pretty poor, with only one exception I could find - the Soekris R-2R ladder DAC modules, which are extremely compelling. I chose the dam1021 because it works entirely standalone (the 1121 really needs a controller) and I can make it very minimal with my own digital inputs (the 1941 is complex and really needs a display).

Historically, I like how Soekris developed this module. It started out as a really ambitious DIY product, community feedback was used to launch the commercial DACs, and the commercial experience has progressively improved the DIY offerings. I have the recently released rev7 board, which has a whole boatload of advancements implemented in the flagship 2541. Everyone wins.

I built the module into a full DAC with a JLSounds I2SoverUSB v.III input (pulled from my USB-SPDIF box; this DAC is replacing my γ2 at work). This is an XMOS based interface, with full galvanic isolation and reclocking, and it works wonderfully. Power supplies are DIYINHK LT3045-based modules; bipolar ±9VDC for the DAC board, and dual channel 3.3/5VDC for the isolated side of the USB interface, and isolators on the DAC board. Two single ended outputs tap the R-2R ladder directly, and the buffered output. Transformers are nice Talema units, but I probably need to replace them with a custom unit. They are hugely over-volting even under load, meaning the power supplies are generating way more heat than they should.

Just a side note before the pics, I love the tools available for planning builds these days. When I was considering a dual-board balanced build, I laid it out in Tinkercad to see if it would all fit in the case. Never mind that I cocked it up because I forgot the bulk of the USB cable connectors. So despite the enormous amount of extra space, I had to work to keep the defined AC, DC and signal zones.

Spoiler: Tinkercad Layout

Spoiler: Front

Spoiler: Back

Spoiler: Internal

The DIY value proposition here is very interesting. With everything considered, I probably spent about US$800, which is a smidge less than the commercial dac1421. What is lost in my build is the extra inputs (don't want them) and the headphone output (don't need it). But what is gained is aesthetics to create a matching DIY stack (I call it elegant brutalism), and better output options for me; I'll take the outputs directly from the R-2R network for my own headphone amp, and the buffered outputs can power my Audioengine A5.

I haven't done any critical listening yet, but early impressions feeding my Crack and HD650 are extremely positive. This will ultimately end up feeding my M3 and Focal Clear. All things considered, I think this is a clear win for DIY.

 

Next project is an amplifier that everyone here is surely familiar with, the venerable M3. It is often exclusively attributed to Ti Kan of AMB, but was a group effort with 'Morsel', with additional input from several other old-school DIY'ers. It's hard for me to believe now, but the design was first published 16 years ago. At the time it was a revelation; very smart and minimalistic, giving incredible power at an excellent price. IMO, it simply slaughtered many of its contemporary commercial and DIY equivalents. The gold-standard documentation made the project extremely accessible.

The core elements are simple, an OPAMP gain stage directly driving a MOSFET output stage biased deeply (>80 mA) in class A. Builders could roll OPAMPs for different sound flavors, and there was the option to install a bass-boost circuit (capacitor and variable resistor in the feedback loop) or variable gain (just the variable resistor). The real twist comes from the active ground channel; a third channel at unity gain, same topology as the L/R channels, buffering between the headphones and 'true' ground. This allows the use of a single rail power supply with a rail splitter to float the V+ and V- symmetrically around ground, and in-turn allowed the amp to be completely DC coupled.

Much has been said about the active ground channel over the years, including a lot of attention from everyone's favorite jackass NwAvGuy. Certainly the strategy does have plusses and minusses... In the plus column, the single rail power supply is cheaper and easier to implement, the headphone doesn't "see" the power supply, and signal ground doesn't "see" the headphone. In the minus column, the ground channel adds back complexity, increases output impedance and stereo crosstalk, and the ground channel has to source/sink current for both channels meaning the ground channel could drop out of class A before the amplifier channels (although imagining what crossover distortion looks like in an amplifier held at zero volts just destroys my brain). Frankly, all of this is entirely academic, because by all subjective and objective criteria the amp was a raging success. You'd never use an active ground today - you'd just go dual rail power supply and fully balanced topology and be done with it - but it worked.

I first started my build way back in 2008, my very first DIY audio project in fact! A fairly standard M3 with S11 power supply at 24V, AD8610 OPAMPs, variable gain, and Hammond cases. It was my home amp for a while, then my work amp for a while longer. In this time, it fended off challenges from a Corda Arietta, several Millett Hybrids, and a selection of other forgettable amps that never left any lasting impression. But the M3 has mostly been in storage for the last few years - my HD650 were the best dynamic cans I owned, and they sounded a bit better from my Crack (not surprising, it's one of those classic synergies). Re-assessing the M3 in the context of modern amps is a big part of why I started this thread.

What I found, is that with the general demise of DIY, the M3 just fell off the radar of many newcomers. But it experienced quite a renaissance in 2015 thanks to Bob Katz writing about his build on Inner Fidelity. I found an SBAF thread with some positive recollections. Some measurements from 2018 suggest the M3 can still bloody noses in one-on-one fights. A derivative of the M3's OPAMP/MOSFET circuit is being sold commercially as the RebelAmp, with generally favorable comments.

All things considered, it really seems this clever little bugger has stood the test of time. So I decided to give mine a spit-polish, and upgraded/configured as follows to try and compete with the big boys:

• Complete re-casing in a HiFi2000 galaxy case with FPE panels
• ADA4627-1B OPAMPs
• TKD 2CP2511 pot (10k)
• Removed the variable gain and set a fixed gain of 5x (+14 dB)
• Replaced all electrolytic capacitors with Panasonic FR and Nichicon UHE
• Replaced rectifiers with Cree zero-rec SiC diodes
• Replaced 1.5" heatsinks with 2", allowing higher bias (110 mA minimum) and lower temperatures
• S11 at 27V
• Custom SumR magnetically-shielded transformer, 30VA, 12+12VAC
• Mains-filtered and earth-choked IEC inlet
• The biggest shiniest knob I could manage; it's tacky, it's gaudy, I bloody love it

Spoiler: M3 original build, circa Feb 2009

Spoiler: M3 updated, front

(Yes, the knob is so shiny, I had to edit out my own reflection!)

Spoiler: M3 updated, back

Spoiler: M3 updated, internal

So, how does this hold up to the amps of today? I still honestly have no idea. @Armaegis was generous enough to offer me several loaners that I haven't been able to accept, and I missed a local meet that would have been very instructive. The HD650 still sounds a bit better from the Crack, and I have nothing else in the way of direct amp comparisons. Yet I can say that the Focal Clear from the M3 is head and shoulders above my HD650/Crack and Lambda/Exstata combinations. It sounds magnificent. It's clear to me that modern OPAMPs with global NFB don't have to sound dull or harsh, and the M3's implementation is stellar. And amongst modern design trickery, there is still something to be said for brute force class A bias to provide clean power (the MOSFETs here burn almost 9W of power to put less than 1 milliwatt into headphones). The noise floor is completely inaudible thanks to a great circuit and decent execution.

But what about the value proposition? As I've said throughout the thread, it's really not great. For only a little more money than just the M3 casework, I could have bought an entire Schiit Jotunheim. For the cost of the TKD pot alone I could have bought a Magni. To built my amp entirely from scratch would surely surpass US$700-800, at which point there is going to be some very serious competition.

So to TLDR it..... it pains me to say, but I would find it hard to recommend that anybody would build something like my amp today. But at the same time, there's no way I would ever get rid of this amp. The nostalgia is too strong, and it simply sounds too good. There will always be a place for this amp in my system.

 

@Beefy it's time for you to build a Pass HPA-1 clone.

I've got Gerbers if you want them....

 

The WHAMMY by Nelson Pass also has a similar topology with an opamp gain stage and MOSFET output. Could be a good comparison