Scaled-down Bridged Zen for Headphones

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've seen the discussion on a bridged Zen. What about a scaled-down version for driving 32-ohm headphones? It should be straightforward to calculate the required power levels and scale the resistor values accordingly. Using a bridged version is essentially the equivalent of each amp driving 16 ohms.
It could be driven from the Zen balanced-out preamp (was that the Bride of Zen, I think?).
Why go to the extra expense of a bridged zen ?The transformer and heat sink cost alone would be way high for a headphone amp. Just add a resistor voltage divider to the output for headphones,mabe say 100 ohm/1 watt from each output to the corresponding leg of the phone jack and 10 ohm/1 watt from each of those legs to the ground lug of the jack.Check the HeadWize Library FAQ section for the specifics

Oops, right, the common ground. So that's what a brain fart feels like....

Okay, just the scaling-down part. Take the Headwize Szekeres buffer, make it common-source to get some gain, use a current source instead of a resistor,...suddenly it's a Zen. I don't want to just build a big amp and chop down its output; that really would be a big waste of power supply, heatsink, etc.
So I'll restate the question: has anyone tried scaling down the Zen to lower wattage / higher load impedance?
How much power are you interested in putting out? 1W?

I've never done this, but here are some ideas:

The starting point would be the Zen Revisited amp. You should be able to cut the idle current (increase R1), lower the gain (increase R11), and be done with it. Lowering the supply voltage (25V) should lower the heat output still more. You should also be able to cut the output capacitance in half, but this isn't necessary.

Going still further, cutting down on the input filter capacitance (5000uF), and going to a larger inductor (4mH/18ga) would probably decrease ripple. Going to a double-PI filter (3 caps and 2 inductors) would decrease the riple even more. Use PSUD2 ('') to check the ripple at the required load.

My understanding is that the distortion should drop with the lower gain.

[Edited by thoth on 09-15-2001 at 11:04 AM]
The Zen would be a fine choice, but what about the SOZ? A small version would be a cinch, and it's already scalable.
Amps are stable into higher loads, you need not worry about a 32 ohm load on a nominal 8 ohm output. Many designs actually give better performance into higher loads.

Yes, I have thought about the SOZ too. Something about me has a problem with a resistor instead of a current source, but in the interests of starting with simplicity maybe I should give it a whirl. I've seen some reports that the resistor version "sounds better". Hmmmm.
You could just use one of the outputs. If balance is important, a dummy 32-ohm resistor on the other output would be easy (remember, we're only driving maybe 50 mW into the phones).
Since the SOZ drives its load differentially (equivalent of 4 ohms per side), I think I could scale all the resistors by a factor of 8. This should reduce the power output by the same amount. If I go for the 5W version voltages, it would mean power of 0.625 watts - should be plenty for headphones.
The 'problem' with the SOZ seems to be that that it's a balanced amp. Stereo headphones have a TRS (Tip-Ring-Sleave) connector. The tip and ring supply the signal to the headphone drivers. The sleave is the common (not necessarilly ground).

I think that if the power supplies for the two amps are separate and isolated (amp grounds not connected together or to earth ground), and the undriven input is connected to amp ground, then the outputs associated with the grounded inputs could safely be connected together to form the common.

There is a potential safety issue here, in that the common is usually at ground potential. This headphone amplifier will have a floating common. It might be possible to connect this floating common to earth ground, but I 'm not sure.

Will this work?
Sit tight...since the issue has come up so many times, I'm whipping up a SOZ w/current sources, just so folks can have a starting point for discussion. I've been somewhat distracted by events this past week, but will try to post soon. I have one current source running already, and etched the board for the second one this afternoon. I'll drill and populate it tonight when I get home from work. The amp will then be subjected to various tests and insults in the process of working up Certain Variations which may be of interest.
Note that I, personally, feel that current sources are inferior to resistors for a number of reasons, but they are sometimes necessary. In the case of the SOZ, it should prove useful for dropping the heat dissipation. In your case, that shouldn't be all that big a problem, as you're not looking for a really high-powered version. Feel free to use either resistors or current sources.
Your point is well taken--sorry, hadn't thought about the common problem--I plead distraction. However, as Thoth notes, there is a difference between ground and common. There is no reason that the common (sleeve) of the headphones need be at ground potential. A well-matched pair of SOZ circuits (it might be necessary to match all four output devices) will have about the same DC offset, but in fact I'm not sure that it would really matter, anyway, as long as you've floated the amp circuit (if need be, the inputs could be capacitor coupled and floated using a voltage divider between the positive and negative rails--something folks might very well be doing anyway if they're using asymmetrical rails and current sources). Hooking, say, the negative outputs together will bring the two circuits into alignment voltage-wise, effectively creating a virtual ground some random number of volts above 'real' ground.

The more I think about this, the more I like it. While I'm interested in your modified SOZ, I think I would start with resistors. The power, heat sinking, etc. is not as critical a problem if everything is scaled down for headphones.
The essence of the SOZ, and the reason why we probably won't see a commercial offering of it, is that almost all of the power drawn by the circuit is wasted. Don't input circuits typically use this differential circuit because of its inherent linearity? It wouldn't be used for output circuits because of so much wasted power. In a headphone circuit, who cares? I can drive 20 watts into the circuit but only transfer 50 mW to the phones, so what.
Would it be true to say that the SOZ wastes more power but is more linear than the Zen?
I think you would need electrical isolation somewhere to allow you to connect differential outputs from two channels together. Both differential outputs change, even if one of the differential inputs is kept at ground. The two circuits would have to be totally isolated except at that one point. And if you do this, where does power line safety ground go?
I'm not worried about only using one of the differential outputs and driving a load referenced to ground.We're stealing such a small fraction of the current in the circuit anyway. I would have to use an output capacitor, but that doesn't concern me.
****..... I've been busy all day simulating a scaled SonOfZen and BalancedLineZen version for Headphones when it hit me.... the commom ground or the headphone makes it impossoble to use this circuit unless you are use capacitor decoupling (yes..or transformer..oke)..

just use the headwize class A buffer from headwize...easy and good sounding.. I've build two..realy sound oke
Yes, differentials drop a fair percentage of power, but for that matter, all front ends are class A. They're *all* wasting (proportionately) a lot of power for what they yield into load. The thing is...they're only using a watt or two, so who cares? Differentials are inherently a distortion-busting kind of topology, so at least in theory the SOZ should be better than the Zen. Add in the fact that you're knocking out caps, etc. and it should be better still.
As I said above, my feeling is that resistors are better than real world current sources. I'm just fiddling with the current source SOZ in order to lay out a game plan for others here who are interested in conserving heat. It isn't necessary for the experiments I have in mind. With winter coming on, I won't mind 'free' heat, anyway.
I can see no reason why the output of a SOZ need be referenced to ground. It's not necessary for signal purposes, whether the amp is driven single-ended or balanced. I don't see it as a safety issue, either. We're only talking about 10-15V, max. which is not dangerous. (Car batteries are 13.6V and you can touch both terminals without harm. Just don't do it when you're sweaty or you'll get a tingle. I routinely test 9V batteries by touching them to my tongue. Once you get used to the feeling, you can accurately estimate the life left in the battery by the tingle.)
Assuming I understood tschrama's posts correctly...I'm still not convinced that the common is a deal-killer for headphones. What matters is the *relative* signal change seen by each headphone driver. It doesn't matter if our common (aka floating ground) is bouncing like mad. If that's all that's tied between the two channels, it becomes a reference point for each circuit/driver, and they will only respond to a change that is valid for that channel.
I'd be more than happy to test this, but I've only got the one SOZ channel and it's currently disabled while I'm putting in the current sources. Time (and lack of heatsinks--I'm using air-cooled for the time being, rather than hooking it into the water-cooled system) is my limiting factor, here. I won't be happy until I see for myself whether this will or will not work in the real world, but it'll be a bit longer before I have two channels running.
If nothing else, Paul, even if the SOZ doesn't work out, you can always do Zens or one of these other circuits people are mentioning. For that matter, there are some OTL tube circuits at They seem to have rather an obsession with headphone circuits over there, and once you get rid of the output transformer, you've gotten rid of an awful lot of the stuff people complain about in tube circuits.

Hey GRollins.... The thing is that a Brigded driver like the SOZ have two advantages: (and a few more)

1] Less distortion due to cancelation
2] Direct couple is possible because the whole headphone is not at ground.

1] can only take place if each speaker two leads to drive it in bridge-mode.. There areonly three leads in the headphones so this can't happen.. unless heavy left-right channel separation is no issue (eg. mono!)

2] this is still possible with headphones but you needs a very very low impedance (again channel crosstalk! ) precise reference voltage and very close matched mosfets....

I think it's very unlucky that headphones have only three leads...yesterday I dedigned a brigde-mode headphone amp that simulates great.. really great.. but I realize I have to modify my Grado's to use it .... Maybe I will ???

I think the Only way you can use a bridged amp for headphones is if the two amplifier channels are totally electrically isolated, up to the point where they connect to the common of the headphones. This would also be the chassis and AC power line ground. Note this isolation also applies to the inputs of the amp, and their grounds, so you would need at least one transformer (caps won't do it) if you're driving them from a common source.
Not worth it, IMO.
SOZ calcs

I've done some calculations, pretty basic Ohm's law stuff, and I can't seem to duplicate the graphs in the SOZ article, at least at the 5W output level. Starting with the power per resistor, I calculate the current and then the transistor dissipation. All is well except at the 5W point, where I calculate a much lower dissipation for the transistors than the graph shows. I noticed the curves have a bit of a knee below the 10W point. Any ideas? Maybe tschrama in his simulation can fill me in...
I'm trying to understand the circuit to the point where I can scale it to drive a 32-ohm single-ended load at much lower power levels.

Hee Paul...
I have been simulating the SOZ circuit for hours and hours and hours..whole weekends trying to understand everything.. have simulated some headphone versions too (less than 0.01% THD, 0-200KHz, at 5 Watts dissipated maybe worth a headwize article, but not yet Build!!!!!!) .. it seems to me that Pass has optimized the SOZ pretty much .. But I would never build a SOZ whitout ConstantCurrentSources!!! it saves dissipated watts and performes better... anyways
I use the circuitmaker demo... try it , I love it
If you mail me any precise questions I will simulated me another headache...;-)

Good idea, I think I'll try the simulation. Sometime when I have hours and hours to spend...I'm not surprised to hear that the Pass circuit is pretty well optimized.

In the meantime, if you'd like to share your headphone version, which I assume is scaled down for lower power and higher load impedance, I'd like to look at it (if topology is the same, just the parts values will do). I happen to have some MOSFETs that I'm itching to do something with. I can burn a few watts.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.