Headphone output tweaks

sthcoaster

Member
Paid Member
2016-03-03 9:46 pm
Washington DC
Greetings
Forgive me if I'm posting in completely the wrong place - only my second post.

In the domestic listening environment one must resort to headphones when music wants to be heard at volume. So to keep the peace, so to speak, and like many other music lovers I do stand in front of the stereo ensemble and crank my head to music only I can hear while the rest of the family is left unmolested say by Einsturzende Naubauten's Haus der Luge album at industrial volumes.

Question:
Are their specific things that I can do to the headphone output and circuit of an integrated amp? Or does it get the same as the usual analogue outputs? ie. Not having past through any other filters etc. Or are all the best tweaks on the path through the amplification.

I like minimalism hence a fondness for 80s & 90s Nad kit but also this extends to the numbers of kit laying around. So I don't necessarily want to add anything to my stack. BUT I might be tempted to have ago at fashioning a headphone gainclone?amp if folks could suggest their favourites. I imagine I should go search the forums on this one myself.

My amp is a Nad C350 (only mod is a Belden mains cable hard wired). Source until recently, was a NAD 541i, I've recently cycled out a Marantz 6000ose, before rediscovering the Rotel 965bx I bought off a mate years ago stashed away in storage learning about the respectful following it has. This was all of course prior to heading down the diyaudio rabbit hole when by cutting out the muting transistors I broke my music collection and I am still grieving the loss of murky homogenous sound that coloured my listening pleasure before.

I listen through a pair of completely stock Sennheiser HD25-1 - that's good enough for my budget... for now .... so any helpful suggestions beyond "spend more on cans".

Best from the sthcoast where I have got nothing better to do so apologies for the ramble
 
Consult the schematics for details, but generally the headphone outputs in integrated amps tend to be simple resistive tapoffs at the output. Very unexciting, extremely inefficient (bigger integrated amps tend to be burning 40-50 watts all the time), possibly noisy if the amp is, and it is to be hoped that your headphones have a somewhat constant impedance response or else the high output impedance will color the sound.

HD-25s are pretty easy to drive really (about 120 dB/V, 70 ohms), you can get ear-splitting volumes from these on a portable player. Actually they're about ideal for portable use. At home, however, I'd rather have something more cozy and with some soundstage to speak of...
 
Question:
Are their specific things that I can do to the headphone output and circuit of an integrated amp?

In my experience - ditch the usual resistive attenuator from the output and fit a step-down transformer.

You can wind the transformer yourself on (say) PQ26 or PQ32 ferrite cores. If you say what the impedance of your particular cans is and the rated power output of your amp I can calculate the number of turns and wire diameters you'll need. You might need to buy a winding machine (fairly cheap on Taobao) though as the number of turns for the primary can run into the thousands which tends to tax most people's patience.

Your patience be rewarded with improved dynamics as compared to a resistive attenuator.
 

sthcoaster

Member
Paid Member
2016-03-03 9:46 pm
Washington DC
Thanks abraxalito.
I won't pretend to know immediately how to do this but like a challenge and there is always plenty of help online - this is something I think I could handle ... At least winding A transformer. Do you then connect each end of the winding to where the resistive attenuator was? Will it matter which way around it goes?

If this solution is implemented what does it mean when using other cans? The output will be optimised for a particular impedance so what if you use something with higher/lower impedance? For example, if I do decide I "need" new headphones for at home listening I may/will need to redo this differently

Sennheiser HD25-1 headphones are rated 75ohm
NAD C350 Amp is rated 60w per channel for 8 ohm

I've read the spec sheet for this amp it seems to indicate line output for the headphones is 220ohms? Does this sound right? Am I reading the right bit of the data sheet? Sounds very high according to other reading I have been doing - according to which the attenuation should follow a 1/8 rule which would mean impedance should be 9ohm for these headphones?

I fancy I can wind this thing by hand - I'll be doing this instead of my crochet!;)
 
Yes, that's 220 ohms of output series resistor. 220 ohms / 2 W, to be precise (R211/212) - the schematic for this unit is available on HFE. (And that's not even a very high value. I've seen 330, 390, 470 and even 1k2||1k2 = 600 ohms.) A common tweak for a circuit like this would be installing parallel resistors of 10-ish ohms at the output to get output impedance down (so we are getting a loaded voltage divider), but here the headphone output is always connected and if you do the math, the poor little 2 watt resistors could just about withstand full speaker power output.

Amplifying only to divide back down is stupid anyway. You'd be better served tapping off the signal at the pre-out using a splitter cable and making an external headphone buffer to connect there (maybe using the venerable JRC 4556A opamp at either unity or -1 gain on a split +/-9 V battery or +/-12 V mains supply, but you can pretty much do whatever you want here). Pre-amp gain seems to be only 5 dB and change, that would be ample for HD25-1s, and you may even get the volume control into some sort of sensible range that doesn't exhibit major channel imbalance.

If you've got an amp with an old-fashioned mechanical input selector switch, you can even use a recording output to drive a dedicated headphone amplifier with the big amp turned off.
 
Thanks abraxalito.
I won't pretend to know immediately how to do this but like a challenge and there is always plenty of help online - this is something I think I could handle ... At least winding A transformer. Do you then connect each end of the winding to where the resistive attenuator was? Will it matter which way around it goes?

You connect the primary winding between the speaker output terminals (you might find you also need a capacitor in case there's a significant DC offset) or from the 'hot' side of the attenuator to 0V. Yes it does make a difference if you connect them out of phase, its fairly easy to spot when listening - its happened to me as I rarely bother to verify polarity when I'm keen to hear how a new setup sounds.

If this solution is implemented what does it mean when using other cans? The output will be optimised for a particular impedance so what if you use something with higher/lower impedance? For example, if I do decide I "need" new headphones for at home listening I may/will need to redo this differently

You can do what I do and make several secondary windings. Then you can series/parallel them according to your cans' impedance. I've done four on some trafos - this gives 3 options of low, medium and high impedance.

Sennheiser HD25-1 headphones are rated 75ohm
NAD C350 Amp is rated 60w per channel for 8 ohm

Thanks I'll look into the design for this based on those figures.

I fancy I can wind this thing by hand - I'll be doing this instead of my crochet!;)

Great to find such a gung-ho attitude here on DIYA :)

@richie00boy - no formal measurements no but I am often testing my trafos driven from my signal generator (50ohm source impedance) and into my AC voltmeter (with no load). The bandwidth comfortably exceeds the audio band by at least an order of magnitude - is the limitation only when the output's heavily loaded? I've noticed no subjective HF roll-off when using any of my trafos.
 
OK then next time I am checking a trafo I'll do a loaded test, with 50ohms and see what happens.

LF-wise there's definitely a roll-off as all trafos are high-pass filters. The roll-off is chosen - there's a trade-off involved as in all engineering. I tend to design trafos to handle full power only down to 40Hz and haven't noticed any subjective bass degradation. Rather the bass sounds deeper with the trafo in circuit than without.
 
After doing a few calculations with smaller core sizes, I reckon the smallest one to suit your application is a PQ32/30. I've chosen this because any smaller and the primary wire becomes rather thin, too easy to break it and also its more costly.

With this size core, your secondary - which you wind on first - is six layers of 0.33mm magnet wire. Wind it without any gaps for maximum neatness - the number of turns should turn out to be about 300. If you keep it neatly packed you won't need to bother counting.

The primary uses 0.15mm diameter wire - as much as you can fit on. This will be around 2000 turns by my estimation.

If you want to handle higher impedance cans too then your primary can be made in two parts, each 300 turns but thinner wire, say 0.23mm. If you do this you will need to count turns accurately in order to be able to parallel the windings for lower impedance cans.
 

sthcoaster

Member
Paid Member
2016-03-03 9:46 pm
Washington DC
Yes, that's 220 ohms of output series resistor. 220 ohms / 2 W, to be precise (R211/212) - the schematic for this unit is available on HFE.

Yes I already have this from Hifi ENgine but the scan is incomplete - only has the schematic not the PCB layouts. One day the schematic will be enough for me but for now I also like to be able to see on paper where the components are so I can more easily understand what I am looking for/at when I crack open the unit. Can anyone point me in the direction of a complete/better scan of the service manual for NAD C350? Otherwise I will just work with what I have and if I need to will buy pdf as I can see these are available (maybe it is the same scan?)

I am going to work on both solutions that have been suggested as I think this will be a good way to learn - of course not implementing them at the same time.

I fancy that might even try to be clever to pick up the preamp inside the case of the unit rather than use a splitter cable and install the buffer inside rather than as a separate box externally - do you think it possible to pick up a line to power the buffer from the existing board? Or will this add "noise"?

I can see various voltages in the schematic but none of them are the 9v or 12v that sgrossklass indicates. Is the voltage specific or are these values chosen because they can be readily produced by a battery or small transformer that I may have laying around and can repurpose?

Headphone buffer? Do you mean something like these? High-Performance Quad Headphone Buffer
RJM Audio - Szekeres VE Headphone Buffer

If this is coming off the preamp does this mean the volume is set - so the main volume control will no longer control the volume from headphone output? I should add a volume control? I think I remember seeing somewhere that this is not so simple as adding a pot and introduces another set of issues?
 
There's a full SM for C352. Might be similar.

A +/- 18V supply would be available inside the amp but you may have to regulate that down to +/-12...15 V. You don't normally run a 4556A higher than that, even +/-9V may be advisable when trying to drive very low impedances like 32 ohms in order to avoid overheating. Plus, when picking up a supply inside the amp you always have to watch out for grounding issues. Power supply ground + shielded cable = potential ground loop.

An output transformer is easier to wire up, but so is a resistor divider. I don't think I've seen a transformer used for a headphone output since hollow-state era communications gear, they're just too nonideal. Might be a fun (if not necessarily trivial) exercise if you're into that stuff though.

The volume control is in front of the preamp, so of course it'll still operated if you tap off after the preamp.

In terms of buffer, I was basically thinking the second half of the popular O2 headphone amp.
 

sthcoaster

Member
Paid Member
2016-03-03 9:46 pm
Washington DC
A +/- 18V supply would be available inside the amp but you may have to regulate that down to +/-12...15 V.

Something like this after picking up the 18V± On both -ve and +ve? Has to be both rails right?

What is the V value on the caps? Should they be higher than the 18v going in? - I have a couple here already that might do?
47uF 25v
220uF 16v

The 4556 that I can easily get is the AV SSOP variety - could mount on a pin adapter if I were to use a breadboard. Though I am considering a point-to-point solder on this project! Nothing like making it harder for myself.
 

Attachments

  • image.gif
    image.gif
    3.9 KB · Views: 89
abraxalito, have you measured the frequency response of such a transformer? Its usually not easy to get good response at frequency extremes with a basic winding.

For the sake of completeness here I did the test on my latest transformer, using a 50ohm load (two 100R in parallel). Its a 3:1 step down with a 2R secondary resistance, 20R primary resistance and it was driven from my Siglent signal generator which has 50ohm output impedance.

There was indeed some HF roll-off. I set the 0dB reference at 1kHz and the loss was 0.5dB at 15kHz. I'd expect slightly less (can't say how much less) droop when driven from a normal amp which has lower output impedance.
 
Something like this after picking up the 18V± On both -ve and +ve? Has to be both rails right?
Yep. Use 78Mxx/79xx or others with good noise specs (study datasheets from actual mfr), or LM317/337 low-noise circuits (including Cadj). Beware of pinout differences.
What is the V value on the caps? Should they be higher than the 18v going in? - I have a couple here already that might do?
47uF 25v
220uF 16v
You could use those for 18 V in / 12 V out, it's not a super-generous voltage spec though. A 16 V cap is about the minimum I'd use on 12 V, a 25 V on 18 V (or 16 V on 9 V) sounds better already, but if you were concerned about really long life you'd go for about +100%, i.e. 35V/25V.
The 4556 that I can easily get is the AV SSOP variety
Which has the lowest power dissipation of the bunch (250 mW vs. 700 mW for the DIP). The O2 uses two DIPs in parallel and people have still managed to occasionally fry those driving 16 ohms on stock supplies (+/-11 V and change). Some creativity with cooling may prove a good idea there.

BTW, last time I checked one could still get stock of the AL in antiquated SIP format. 800 mW Pd, possibly not too bad a choice for a one-off project.