Some other Source Follower Configurations

Hi Patrick,

Thanks for the reply.

.... If using EI, it is better to mount the transformer away from the amplifier to minimise magnetic disturbances.

Either a toroidal (if not too expensive) or a separate PSU case, then... No problem.

4. If you want to put a buffer upfront, you might as well add a proper line stage with gain instead.
But this depends on what headphone impedance you intend to drive with.

The reason for thinking about a headphone amp is an upcoming sabbatical leave, which will leave me without music while away unless I do something about it... (No way can I ship all my stereo and speakers...) So my idea was to build a headphone amp and buy a good (OK, self-indulgent ;)) pair of headphones when I get up to the US, since they cost a fortune down here. I've been looking at Grado or AKG, which appear to mostly have impedance around 32R, although I'm open to other suggestions. I presume when you said a gain stage "depends" on impedance that you meant a higher impedance than this would need one (say 300-600R or something) Right??

If I need gain juma's BF862 preamp (gain of 2) springs to mind, since I've made two and found them to be excellent. On the other hand it might be simpler not to bother, and just take one of them with me. Unless, of course, you meant something with more gain than that?

Cheers

Nigel
 
DAO is perfect for AKG701 or 702.
No need for voltage gain if you have a source with 2Vrms output, IMHO.
Gain of 2 does not bring much if you have 600R phones and you still want a volume control somewhere.

Whether you need a buffer (e.g. Curl Follower, otherwise known as B1) or not depends on your source (if it can drive a load of say 10k or less).


Happy building,
Patrick
 
... I first came across in Borbely's article, where the name Curl was mentioned....

Yes, it's the Borbely's article: "JFETS: The New Frontiers, part 2"
That article mentions Mr. John Curl's JC-2 module, where he used complementary, push-pull JFET buffer (fig. 16b) but not the JFET CCS loaded SE buffer (fig. 15c) we are talking about. It's treated as generic, classic circuit...
 
Disabled Account
Joined 2006
Nigel, if your volume control output impedance is not too large (and hence does not cut high frequencies when feeding the LU1014 gate capacitance), then you might not need the buffer. Just the volume control and the DAO stage.

1.225 V RMS (which most sources can supply) will get you 50 mW into 30 Ohm headphones. So if using modern high sensitivity 'phones you will likely get in excess of 110 dB. Most folks seem to listen in the range of 70 - 80 dB, so you will have plenty in reserve.
 
Crossfeed for headphone amps

Six or seven years ago I was playing with DIY headphone amps, based on Chu Moy's "headwize" website. It's a controversial subject, but my various fiddlings and listenings convinced me that headphone amps with crossfeed are better than those without for extended listening. Some of the folks from headwize went off and formed head fi about the time I turned attention to other topics, and that has since become huge. There is a lot of information on headphones, portable headphone amps, tube headphone amps, buffers, power supplies, you name it, at these sites, and if I were going to be without my stereos for a sabbatical year I would want to get the latest word and DIY accordingly.
 
Recordings vary and the amount of "required" amount of Xfeed will also vary to your taste, from none at all to quite a bit (very precise, yes!) - there are a number of ccts around but are fairly consistent about the freq points and attenuation and Xfeed levels.

As bd mentioned above, both "headwize" and head-fi" contain a lot of useful information, and links to some of the newer developments - suprising.

From my #81 post, you can see that I'm slowly playing around with the input stages to the unit - the amp itself is Patrick's standard Taylored design (280mA), and I just replaced the zener with a "string of leds" and tried different caps - I have simplified the power supply to basic Shottky diodes, R-C-R-C smoothing into a beefed up Salas Shunt Reg (Version 1) that replaced the previous Cmultiplier method - this, too, required the output delay/dc detect cct as the +/- rails don't come up together, unfortunately.

I have concentrated on a "tilt" control (HeadWize) plus a version of Rane tone controls but have a long way to go yet - Xfeed is a mixture of the J Meier and Ohmon ccts, adjustable and the type of caps here make a bit difference - I use styrene generally, but many other readily available good caps - fancy bits are not always best choice here, IMO.

Some of the newer commercial headamps are worth a look, too.

my 2 cents, regards
 
Hi Guys,

Thanks for all the replies.

Nigel, if your volume control output impedance is not too large (and hence does not cut high frequencies when feeding the LU1014 gate capacitance), then you might not need the buffer. Just the volume control and the DAO stage.

Actually, it seems that the heatsinks will make the chassis a good deal bigger than the basic DAO circuit needs, so there should be plenty of space inside - I will probably do it first with no buffer, and pop one in later if it seems to need it.

Six or seven years ago I was playing with DIY headphone amps, based on Chu Moy's "headwize" website. It's a controversial subject, but my various fiddlings and listenings convinced me that headphone amps with crossfeed are better than those without for extended listening.

Yes, I've been looking through the sites you mention, and I'm fascinated by the crossfeed idea. I'm not usually a big headphone user, but since there's no way I can ship speakers I'm trying to learn up.

...if I were going to be without my stereos for a sabbatical year I would want to get the latest word and DIY accordingly.

Amen.

From my #81 post, you can see that I'm slowly playing around with the input stages to the unit - the amp itself is Patrick's standard Taylored design (280mA), and I just replaced the zener with a "string of leds"

Are you using the J511? I'm having trouble finding where to get them. (Just pricing things out for the moment)

I have simplified the power supply to basic Shottky diodes, R-C-R-C smoothing into a beefed up Salas Shunt Reg (Version 1) that replaced the previous Cmultiplier method - this, too, required the output delay/dc detect cct as the +/- rails don't come up together, unfortunately.

I'm thinking of either the Salas reg (by "first" you mean without the "R6 mod" that's discussed in the thread?) or the zen v3 supply that Steenoe used in the DAO. What delay/dc detect circuit are you using?

...I have concentrated on a "tilt" control (HeadWize) plus a version of Rane tone controls but have a long way to go yet - Xfeed is a mixture of the J Meier and Ohmon ccts...

OK, I have some reading to do....

The other issue with planning the amp is its size and weight. Taking a family of six abroad for six or seven months means an incredible amount of luggage, so with flight weight restrictions I may need to make something more portable and leave the DAO for when I come back. Not really clear yet, but I can almost hear my wife now ("You're taking WHAT??") The headwize and head-fi sites have loads of info, but does anyone have a recommendation for something lighter (but maybe not as small as an Altoids tin)?

Thanks for the help

Nigel
 
j511, Nigel? Not sure, not so common anymore

I just used a jfet + resistor - the resistor is there to set the current thru the leds and not really reqd for a zener - the simpler way works okay, so ...

Not sure about the shunt reg details but it was pretty complete (if that helps!) I did go a bit nuts with the caps and used a few of those big rushian k73-16 foils caps on it - ended up running about 400mA thru CCS - got nice and HOT! The headamp cct is dependent on p/supply "sound", so it's easy to 'tweak' sound for different 'phones.

The protection/delay was a Rod Elliot thing - plenty kits around - used oversized relays, double contacts, wipers, etc - use them in power amps -

Tone controls - a few years ago, Pete Millet did a rather interesting design called the HA4 with input filters - well worth a look.

For travel, might be worth looking at some of the hybrid headamps and 'tweaking' them - as usual, most of their power supplies suffer from size constraints.

Ah, 'morning Patrick, this little design of yours is going to last forever! the B1 buffer in front seems to soften the sound a bit, and the JC seems 'not there'.
 
Hi everyone,

Still thinking about how and if to build this, or do a different headphone amp. I'm ploughing through stuff here and on the other sites, but no clear ideas yet, except that sooner or later I'm going to do a DAO, whether before travelling or after.

As I mentioned above, my problem with doing a DAO to take on sabbatical is that it may be too big and heavy. Is anyone with a complete amp willing to measure and weigh it for me? (James? I'd love to see photos, too...)There may be things I can do to reduce weight, like choosing trafo and heatsinks with care, but it would help to see what other people have done. Also, juma posted a version above in post #96 with lower rail voltages and MJE243 in place of the IRFP240 - did anyone build it? (juma?) This would presumably have a slightly lower bias current (??) and therefore I could get away with smaller (and therefore lighter) heatsinks...

Well, thanks for any help.

Cheers

Nigel
 
Quote]

If size and weight is important, you should not do Class A.

The circuit weighs nothing, as always.
But heat sinks and power supplies do.

Light weight, compact solution means Class D or Class AB to me.


Patrick

Agreed, for sure, but in this case I'm just trying to figure out *how much* it'll weigh. I don't need anything portable-and-battery-powered, but I can't take something that weighs as much as my F5, for instance. (At least, not if my wife catches me...) My best guess so far is that the heatsinks will have to be about half the size of the heatsinks I used on the F5 and the trafo will be *much* lighter and smaller. In terms of size, if I can make it no larger than a shoebox that's probably OK. (My guess is that's doable, if not a little smaller.) If I can get it in around 3 kilos that's probably OK, if it's 10 kilos that's probably not...

PS Class A means great sound. So you have to set priorities.

Again, I understand and agree - this is why I'm trying to find out what other people's experiences have been like.

Cheers

Nigel
 
...I use batteries, so no reference for you...

I've only just remembered you mentioned batteries in an earlier post. You mean you are using batteries on the DAO with Taylor? How? At what rail voltages? I'd have thought it would run through them pretty quickly, but perhaps I have misunderstood something (again...)

This may be a convenient alternative for me - just ship the amp without any batteries in it, and buy rechargeables online to be delivered to where I'm going... In fact, with a little thought and care maybe it could be built with two power options- either batteries or a PSU in a separate case. Leaving the PSU at home would certainly help with weight...

Cheers

Nigel
 
Hi Patrick,

I use 24V truck batteries. No option for you.

Patrick

I should have guessed it would be something like this because of the power draw... You're quite right - no option for me for any number of reasons...

I've been researching importing toroidal trafos, which will be pricier than an EI, but (it appears) a good deal lighter. (And better, anyway...) Mouser has what appear to be nice ones from Triad Magnetics with parallel 18V secondaries, at 600g for 50VA and 1kg for 100VA. Seems to me that either of these would do the job if I run 20V rails, or a little less, so the 50VA one seems a better choice, to save a little weight without compromising the sound too much. (At all?)

I'm still considering doing the version juma posted in #96, although maybe with the 16V rails Patrick suggested. Would there be any disadvantage?

Cheers, and thanks for any input.

Nigel
 
The cascode device in the DAO circuit is an integral part that contributes to the triode characteristics of the LU1014.
By changing that device to say a BJT, the dynamic Vds of the 1014 will be different, and so will its triode characteristics.

I am by no means suggesting that it is better or worse, because I have not tried.
But I would suggest you try to measure the behaviour of that transistor group first.
Maybe you can change a few component values to get it to work.

Or maybe juma has already tried ??


Patrick
 
I made the circuit I proposed in post #96 and results are very pleasing, it sounds better with my HD555 (Z=50 Ohm) than any other popular unity gain headamp. It's that good that I didn't feel the need to compare it to original DAO circuit.

About triode characteristic of the JFET there are at least two aspects of it:
1. triode transfer characteristic (Id dependance from -Vgs) is intrinsic. It's there whatever you do.
2. triode output characteristic (Id dependance from Vds). It normally doesn't exist - output characteristic of JFET is pehthode by nature. The way to change its' character to triode is to induce feedback from drain to gate. Of course, LU1014 is known for its' a small triode region that we aim for: around -1V of Vgs and Vds less than 5V. That small region is also intrinsic - not dependant on cascoding device.

Since in DAO we have constant Vds (by means of tight cascoding - cascode's reference is tied to JFET's Source) there can not be any influence of that kind; so using a BJT or MOSFET as cascoding device doesn't make any difference to JFET's or resulting device's (complex sum of JFET and cascoding device) behavior.
There is an arangement with such an effect - Mr. Pass implemented it in F3, and it's very different from DAO - a modulated cascode, where resulting device exhibits aforementioned properties.
 
Last edited: