Thoughts on audiophile -isms

The debate between the subjectivist and objectivist camps of the audiophile community is probably about as old as the community itself. (Too) much has been written which there’s no need to repeat here, but if you are new to the game there’s an excellent post on this topic by Schiit audio co-founder Jason Stoddard here.

Incidentally, if you are the kind of diy’er that sometimes consider turning your hobby into a (sideline) business, then I urge you to read through all of Jason’s “blog-book” posts on head-fi as there is lots of useful information there. Anyway, that was a bit of a sidetrack…:)

(Also, feel free to skip the rest of this post if you’re the kind of audio diy’er that genuinely only really cares about the music (I’ve heard they exist although I don’t recall ever meeting one😀 ))

My own position on subjectivism vs. objectivism should hopefully be possible to read in my posts, but very briefly: I have too much of an engineering mind to believe that measurements are pointless, but I have also had plenty of “audiophile” experiences to suggest that measurements don’t necessarily always tell the whole story. Whether this is then complete placebo and wishful thinking on my part I don’t know, but in any case it’s all part of the experience for me:-)

My main point (and the only thing I want to add to the debate) is that you can be a little of both – but not at the same time. You have to make it clear whether there is objective reasoning behind you decisions and comments and when it is based on subjective opinions – however valid they may seem.

Case in point (and what actually prompted me thinking about this in the first place) – the number of questions I get on whether a specific amplifier needs an input buffer:

In a small number of borderline cases the answer is clear – the buffer is needed to prevent impedance mismatches that cause a significant (and objectively observable) change in the frequency response and loss of output drive capability. However, in most “normal” situations there is objectively no need (with emphasis) to include a buffer – but that does not mean that there isn’t a subjective change if you add it anyway and which might be preferable to you.

That to me is the essence of this debate (and about as far into it as I want to venture) – and hopefully I’ll be able to make it clear enough in my writings when I am being subjectivist and when I am being objectivist.

Anyway, in other (possibly more relevant) news: I’ve just placed my first new PCB order in months😀 Nothing really ground-breaking I’m afraid, but still something to look forward to in a few weeks:-)

Surrounded – again!

This is an old project that I have resurrected now as I would like to get my surround-sound setup back into working order (not that I expect I’ll be using it that much, but still…)

It’s a 2+3 channel ICEpower ASX-based setup with 125ASXs in BTL-mode at the front and 50ASXBTLs for center and rear. The 2-channel amplifier very nearly identical to my previous 125ASX-based amplifier but it does have three USPs compared to that build:

  • Transformer-coupled (balanced) inputs using Lundahl LL1527 transformers.
  • Two switchable inputs so it can be connected to both a stereo source and a surround-processor simultaneously.
  • ”Audiophile” form factor (i.e. around 44 cm. wide and much larger than is really necessary😀 )

The 3-channel amplifier also has Lundahls at the input but no input switching (for obvious reasons).

Many upmarket manufacturers use transformers on the inputs of ICEpower-based amps and Lundahl in Sweden make some of the best ones around. The LL1527 isn’t usually employed as an input transformer, but if I’m reading the specs correctly it’s actually fairly well-suited to the lowish input impedance of the ICEpower modules so it should work well. The alternative (which would also fit on my boards) is the LL1540 which is a purpose-built high impedance input transformer. And well, if all else fails the way that these are mounted would mean that I could probably develop an active circuit instead:-) (differential opamp-board anyone?)

Just like my as-yet-not-completed “Ring” amp project the front channel amp has switchable inputs so that it can be used in a combined stereo/surround setup. Switching after the respective volume controls make more sense to me, but of course I haven’t actually lived with it yet so let’s see if theory meets practice in this case😀 This switching is relay-based and uses the balanced selector modules I posted about earlier – yes, sometimes those piles of leftover prototype PCBs come in very handy😀.

There isn’t actually a lot missing – mostly cabling – before this is done, but I hate cabling so it might take a while to do it anyway😉

Third anniversary…

Yes, it’s that time of the year again – the third anniversary of this blog. Just like last time, I never imagined I’d still be here etc. etc.

I still very much enjoy writing here when time permits. I also have plenty of unfinished projects to write about, so even if we’re nearing 150 posts I hope there’ll be many more to come.

A “virtual” toast to that – and of course to all of you reading out there!:)


A cheap fantasy….

Yes, it’s a not a very good joke – sorry!:-)

Haven’t done one of these “cheap kits” for a while, but I am possibly getting a bit more picky. However, this particular board was around 20 USD for a kit (excl. tubes) and so the risk was manageable. It’s supposedly a clone of the famous Matisse “Fantasy” preamp which uses 5670 tubes. I wouldn’t know to be honest, but I have been intrigued by low(ish)-voltage tube circuits for a long time and so I took the plunge.

I’ve replaced some components (mainly the capacitors and the volume pot), but there was still plenty of kit parts left to salvage to make it worthwhile (tube sockets, semiconductors, heat sinks etc.) over buying just a blank PCB. The caps probably would have worked just fine, so that replacement was mainly cosmetic. One notably exception was the 80V rated electrolytic in a part of the circuit that sees rectified 55-60VAC – a little too close to the limit for my liking.

The tubes ‘ve bought from an Eastern European seller and they are 6n3p-E which is supposedly a long-life Russian version of the 5670. The circuit runs off 50-60VAC, so it’s what I would call a “medium-voltage” project. As usual, the instructions that came with the kit were poor (especially if you can’t read Chinese😉 ) but with a bit of care it wasn’t a problem to put together.

Sound quality? Well, as usual with these projects I’ve only really done basic bench testing for now and so the only thing I can really confirm is that it produces sound. One of the things I did notice though was that it seems to be a well-behaved circuit. By that I mean no big turn-on/turn-off transients, no excessive noise and no microphonics from the tubes. I might just have to splurge for a proper transformer and case for this anyway…

JLH Evo Update

All right, no more moaning about lack of build time (at least not for now😉 )

Managed to do a little work on the mechanics of my “JLH evo” concept allowing you to get an idea of what the end result will look like (when it’s eventually finished…).

I’m still figuring out a final chassis design so I may well leave it in this state for quite a while, but at least the mechanics seem to fit together as planned and there’s enough room to run the wires. The baseplate size is app. 170 x 230 mm per mono block.

The extra PCB is a CRC-regulator stage that I will add to reduce the noise and ripple from the switching PSU (a Mean Well EPP-150-27). The advantage here is that the ripple frequency from a switching power supply is very high (typically 65-100 kHz) so the attenuation is much, much greater than at the normal 100/120 Hz ripple from a linear supply – meaning in other words that it’s possible to get away with much smaller filtering caps than a usual class A amp.

This should therefore reduce the highish app. 240mV ripple that’s specified in the data sheet for the supply that I am using down to something much, much less. Any HF-noise on the output should be well attenuated as well. Not sure what effect it really has but that’s part of the experiment😀

April fools…

No, this isn’t really a joke as such. However, there’s something strangely appropriate about posting this today I think😀

It’s an “extended” version of my ZenHP amp, but I went a little overboard and added the gigantic polypropylene caps I purchased on my last trip to Japan as the output caps. As mentioned then, I’d want to test if using film capacitors on the output made any sonic difference compared to the electrolytics that are normally used. We shall see how that works out later on….

The downside of this (perceived) audiophile greatness (…) is that in order to make everything fit in a 2U/350mm enclosure I basically had to cheat a bit on the PSU. It’s either going to be an internal IRM20 switching type from Mean Well as shown or a very simple linear one that can be fed from an external transformer. Obviously having an amp as big as this requiring an external PSU is a bit stupid, but hey – it’s an experiment!😀

Also, since it’s an experiment I’m not going to order fancy front and rear panels for this amp yet. Once I’m through travelling for work in a couple of weeks I’ll have to do a bit of metalwork of my own instead. Not much else missing before it’s ready to play though, but with my current workload it might still take a while to do.

PS: I’ve you see any good audiophile April fools jokes online, feel free to post a link in the comments section.


Project files: A bal. driver with the THAT1646

Still busy at work, but being home for (most of) the weekend and I have time to dig in the back catalogue a bit – hope this is useful for someone:-)

What is it?
A simple SE/BAL line driver board using the THAT1646 line driver IC. This should make it compatible with the DRV134 from TI and the – now obsolete – SSM2142 from AD as well. The board can be used as a pure line driver to feed a balanced input or of course also to bridge two suitable power amps.
The THAT-chip is combined with an on-board opamp, partly to ensure that it is driven by a low-impedance source as per the datasheet recommendation, partly to increase the versatility as the opamp can provide more gain if required.

How big are the boards?
The board measures just 1.6” x 1.9” (app. 41 x 48 mm.) The boards can be placed side-by-side or stacked. I originally had this as a stereo board with two channels on the same board, but decided that the mono-version was probably more versatile overall. If you disagree feel free to let me know😀

What is the status of the boards?
The board is v1.0. I’ve built a single prototype and tested it (I needed one channel for a test setup) and it sounds fine as far as I can tell. No further sound impressions yet I am afraid.

Does it use any special/expensive/hard-to-find parts?
None. I am not actually aware of a source for the THAT-chip in Europe, but Mouser has them and that should work for most people I guess:-)

Anything else I need to know?
A few things:

  • Protection circuitry: I’ve omitted the protection circuitry described in the data sheet for the THAT1646, mainly because I only expect to use it in home applications where there is no risk of a phantom power supply being present. If you are using it with PA gear that (potentially) has a phantom power supply on the inputs then you might need to look into this.
  • Grounding: I have connected the ground pin of the output connector on the PCB to GND on the board, which is actually a no-no. Connecting all three pins to the XLR would (potentially) give you “pin 1-problem”.
    As I understand it, the proper way of wiring an XLR is therefore to only connect “hot” and “cold” from the PCB connector to the XLR out connector and then connect  PIN 1 on the XLR connector to the chassis ground via as short a wire as possible.
  • Preamp-mode: Given that the THAT1646 already has an opamp onboard to drive it, if you use 100k-220k input impedance (R1) it should be possible to put a 10k-20k log pot in front of the input capacitor and convert the board (well, two of them…) to a stand-alone preamp with SE in and Balanced out. I haven’t tested this, but I see a couple of potential applications here:)
  • Chip substitution: The DRV134 has the same pin connections as the THAT1646. The only thing I can see that makes them different is that the DRV134 data sheet specifies 1uF decoupling caps on the supply pins rather than the 100nF for the THAT1646.
    Note that the DRV134 also has a reputation (at least in some DIY-circles) for sounding pretty unspectacular. I have no personal experience to offer here, so try for yourself if you want😀

Download design files here

Edit 14th april 2016: Link has been updated to point to the correct file:-)

Related information:
As usual, RTFD! (= read the f’ing datasheets :D)

Note: Always read the “intro post” for additional important information about my designs.


Just read over on Innerfidelity that the founder of the Headwize site and namesake of the CMoy headphone amplifier, Chu Moy, has passed away recently.

As a relatively early member of Headwize, I can remember discovering a site with an active discussion forum and a lot of inspiration for a novice audio diy’er like myself. Apart from the direct contribution of the Headwize forum (including the projects gallery that I have mentioned several times in the past), Headwize of course also spawned and many other international and local headphone forums that still run today.

Headwize was also the birthplace of the first community headphone amplifier build projects (like the ones from Tangent and AMB) that have helped others get started in this new hobby. I am sure there are quite a few people out there who first experienced the elation of playing music through a self-built amp via one of these projects:)

Although I never knew Chu Moy personally and he was out of the “spotlight” for many years, I am very grateful for his contribution to the audio hobby and of course for being one of the reasons why I am in this at all:)

Full story by Tyll Hertsens

High voltage…

Yes I am still here, but it’s another busy period for me at work so updates to the blog are correspondingly few and far between. As usual when I don’t have a lot of time for diy I still somehow manage to start up new projects. Even with less than 48 hours at home in a weekend, there’s still time to do a little soldering to relax and unwind😀

Among the overdue projects I’ve managed to start up lately are some amplifiers for my Stax electrostatic headphones. This is actually more than a little overdue, because I haven’t had a Stax amplifier for nearly a year now and so the headphones I have aren’t getting any use which is a shame really.

The pile of half-assembled boards in the picture actually consists of the following designs, all by Kevin Gilmore:

– A pair of KGST tube amp boards and matching 350V PSU

– A mini-version of the KGSSHV amp and matching 400V PSU

– A version of the KGSSIC/“Carbon” amp and matching 400/450V PSU

Most of the boards were all acquired through various group-buys on the forums, but Kevin graciously keeps the Gerber files for all of his designs available for free download as well.

I’ve soldered more or less all the parts I have available, so still to do are:

1) Order remaining parts (working on that – since it’s also possible to do from hotels after work :D)

2) Figure out the mechanical stuff (mostly done, but still needs a bit of work – and some tools I don’t have regular access to)

3) Select and match a pile of semis (saving that one for a rainy day:-) )

4) Finish and test boards (as quickly as possible)

I’m not really used to high-voltage stuff so I am being extra careful with these boards. Just like when you move up in frequency, moving up in voltage means that things that were not previously issues suddenly become very important. Fortunately I have a working variac again (fixed after stupidly blowing a fuse in it a few weeks ago) which makes testing much easier – not to mention safer all round.

These aren’t the only Gilmore-designs I’m working at the moment by the way, but the rest involves much more pedestrian voltages😀


Project files: The Zen Headphone Amplifier

What is it?
The board files for my Zen Headphone Amplifier “remake” shown here.

How big are the boards?
The board measures 3.575” x 3.75” (app. 91 x 95 mm). This is obviously for a mono-channel.

What is the status of the boards?
The boards are version 1.0. The prototypes seem to work well and there wasn’t really anything that needed changing in my view.

Does it use any special/expensive/hard-to-find parts?
Not really.

Anything else I need to know?
A few things:

  • Heatsinks: The basic type for me here in Europe is the Fischer SK129, but there are many manufacturers of this profile. The board is designed for the heatsinks to be soldered in place with pins, but screw-mounted versions might work as well. You can use 1″/25mm heat sinks, but in that case I recommend to tune the bias a little lower. My prototype measured app. 240 mA of bias and the heat sinks seemed to stabilise at around 55C in free air, which probably is a bit too much when the board is cased. So, either turn the bias down a bit and/or use taller heat sinks if your case allows for it.
  • Adjustments: Space around R10 and R12 is quite tight, especially with heat sinks/output caps taller than 25mm. In order to easily be able to adjust bias and balance of the amp, my suggestion would be that you don’t trim the leads of the two resistors completely flush but leave enough of the resistor legs that you can connect crocodile clips to them on the underside.
  • Output capacitors: The recommended value is 2 x 470uF from the original schematic, but if you’re using low-impedance headphones I think you should consider 2 x 1000uF instead. This is one place where I think “audiophile” capacitors can’t hurt, so look for Nichicon Muse (KZ/KW, FG/FW, ES etc.), Elna Cerafine/Silmic capacitors or similar. Bypassing the electrolytics with small film capacitors is easily done on the underside of the board if you want to.
  • Transistors: The Q3 footprint on the board is for a BC550C, but the original ZTX450 from the schematic can be used as well if it’s turned 180 degrees. Remember also to match at least the two Q2 FETs between channels as described in the build article. If you buy 8-10 of the IRF610 FETs you should be able to get a couple of very tight matches and the leftovers can be used for the current source (Q1).

Download design files here

Related information:
See the original post for some more information and links to the build article for this design. After posting I actually also managed to find the original headwize article cached here – amazingly it seems that most of the headwize library has been kept intact there! :)

Note: Always read the “intro post” for additional important information about my designs.



Get every new post delivered to your Inbox.

Join 62 other followers