John,
why do you insist on bridgecoupling? Previously I though you
had balanced input, but I see now that is not the case.
It will certainly give a cancellation effect, especially for the
active followers, but is it worth it in the case of diamond
buffers? I am not saying you shouldn't do, it may still give
some improvement, but the diamond buffer already has a
lot of the cancellation, although only from complementary
BJT types. You would save a $$$-prices opamp per channel,
considering your choice of type. Just trying to make you think
twice. I don't mind overkill, and might try it myself some
day, perhaps, but think about it.
Edit: It is probably just that you have omitted stuff that is
not relevant to the topology to simplify the schematics, but don't
forget the protection diodes in the diamond buffers, especially
those from input to output. Jung seem to consider them
important when the buffer is preceeded by an op amp and
within the closed loop. I also think the capacitors closing the
feedback loop already from the op amp output at HF should
not be omitted.
why do you insist on bridgecoupling? Previously I though you
had balanced input, but I see now that is not the case.
It will certainly give a cancellation effect, especially for the
active followers, but is it worth it in the case of diamond
buffers? I am not saying you shouldn't do, it may still give
some improvement, but the diamond buffer already has a
lot of the cancellation, although only from complementary
BJT types. You would save a $$$-prices opamp per channel,
considering your choice of type. Just trying to make you think
twice. I don't mind overkill, and might try it myself some
day, perhaps, but think about it.
Edit: It is probably just that you have omitted stuff that is
not relevant to the topology to simplify the schematics, but don't
forget the protection diodes in the diamond buffers, especially
those from input to output. Jung seem to consider them
important when the buffer is preceeded by an op amp and
within the closed loop. I also think the capacitors closing the
feedback loop already from the op amp output at HF should
not be omitted.
Christer said:
I'd like to try something that is a bit different. If nothing else, so that I have something interesting to discuss. I've got op amps. Yes, this is overkill. I hope not at the expense of good sound.
And I have plans to protect the headphones from DC (~+/-50 mVdc across the outputs trip). Also, plan to bias the op amps into class A with 2mA current sources (not drawn).
I wished I could use balanced inputs. However, I also need to include a volume control. So, I'd like to avoid an attenuator with four poles.
Thanks for your cautionary notes Christer!
JF
johnferrier said:
By all means, do it. I just wanted you to think about it so you
know why you are doing it. I hardly think it will deteriorate
the sound.
It should be an improvement, I guess, at least in theory and
measurements.
I don't want to confuse you further but since you are prepared
to take the complexity of bridging, have you considered using
CFB op amps and using Nelsons X scheme?
I haven't had time to analyse that special cross-feedback
you use, so maybe it is similar to the X, though.
I am currently working on something similar as you, so it is
interesting for me to.
However, I will use single diamond buffers and no
bridging. On the other hand I will have sockets for two sets
of op amps and a switch for A/B comparison between them, as
well as some other swithces and jumpers for various options.
That is meant more as a testbed than a finished project.
peranders said:
And what solution would you propose? Running the buffers
at a lower voltage than the op amp is a no no, and it seems
most agree that op amps perform better at high supply
voltages. The bridging doubles the possible voltage swing,
of course, but even the non-bridged variant vould give
close to +/-18V, which is probably also too much for the
headphones. Thus, in either case one has to either be
careful with the volume control or include some overvoltage
protection circuitry, as far as I can see.
Has already been asked and answered.
Actually my old headphones has an ordinary 1/4" plug, but
the cable is a twin coaxial which splits close to the headphones,
one coaxial going to each side of the phones. These could
easily be converted to work in bridge mode if I wanted to, and
I guess there are many haedphones that are done that way.
I'm aware that 20Vrms would be extremely loud with all but very ineffecient headphones. The datasheets I read (in the "swamp") indicate that circuits operate better at higher voltages. If nothing else, at normal listening levels, clipping will not be a problem. (Perhaps, though not likely, I will use the ineffecient AKG K1000.)
Likely using Sennheisers that easily modify to work with differential circuitry. (Currently, have Sennheiser Ministar MS100. Broken head strap for years, with intermitent contacts--may use conductive epoxy to fix. The the MS100s are the one thing I know of that I *can not* find something about on the internet. Kind of an enigma.) Will use the Ministars to test my amplifier, after testing with 300 ohm 1/4W resistors.
Unsure of the Nelson X, though it seems like these circuits are close. Prefer output transistors rather than MOSFETs. Also, not aware of people raving about the X. Also, prefer VFB to CFB. If my circuits are not Nelson X, then I'm trying something a little different.
Originally, I thought of using switches and relays to easily A/B circuitry, but decided that I wanted to avoid a never ending design. I'd like to avoid trying this and trying that. Also, doubt that my ears are that sensitive.
JF
Likely using Sennheisers that easily modify to work with differential circuitry. (Currently, have Sennheiser Ministar MS100. Broken head strap for years, with intermitent contacts--may use conductive epoxy to fix. The the MS100s are the one thing I know of that I *can not* find something about on the internet. Kind of an enigma.) Will use the Ministars to test my amplifier, after testing with 300 ohm 1/4W resistors.
Unsure of the Nelson X, though it seems like these circuits are close. Prefer output transistors rather than MOSFETs. Also, not aware of people raving about the X. Also, prefer VFB to CFB. If my circuits are not Nelson X, then I'm trying something a little different.
Originally, I thought of using switches and relays to easily A/B circuitry, but decided that I wanted to avoid a never ending design. I'd like to avoid trying this and trying that. Also, doubt that my ears are that sensitive.
JF
John,
The X concept does not rely on MOSFETs or anything. It is a
more general topology, but does assume CFB amps. However,
I had no intention trying to talk you into doing an X amp instead.
I just thought there might be some similarities, so I just wanted
to give you a pointer in that direction, and then leave it to
you whether to follow it or not.
As for switches, it is rather the other way around for me. I
planned a buffer + discrete diamond buffer combo. Then I
thought that an op amp doesn't take that much space
comapred to a discrete buffer, so why not have two sets of
them and and A/B facility? Then I started thinking about other
options too. I admit my design were about to turn into another
everything-but-the-kitcheen-sink design, but I think it is doable.
What worries me most is long non-straight feedback paths
because of the switches.
Well, you have double buffers, I have double op amps.
The X concept does not rely on MOSFETs or anything. It is a
more general topology, but does assume CFB amps. However,
I had no intention trying to talk you into doing an X amp instead.
I just thought there might be some similarities, so I just wanted
to give you a pointer in that direction, and then leave it to
you whether to follow it or not.
As for switches, it is rather the other way around for me. I
planned a buffer + discrete diamond buffer combo. Then I
thought that an op amp doesn't take that much space
comapred to a discrete buffer, so why not have two sets of
them and and A/B facility? Then I started thinking about other
options too. I admit my design were about to turn into another
everything-but-the-kitcheen-sink design, but I think it is doable.
What worries me most is long non-straight feedback paths
because of the switches.
Well, you have double buffers, I have double op amps.
Why not single pole switches to control good DPDT relays that non-energize in the prefered state? SPST switch is cheaper. Relay is more expensive, but you can put it right next to your circuit. Alternatively, you could use header and jumpers. Regardless, I agree that having a quickly way to switch A and B is preferable (rather than powering down and changing parts). As I wrote, my ears are not super sensitive, so I would need all the aid I could get to have a good A/B comparison....
I note that you started another thread. I'll read through it later today to see if I can contribute.
Thanks.
JF
I note that you started another thread. I'll read through it later today to see if I can contribute.
Thanks.
JF
I have thought about relays too, and maybe that would be
better, but I might still need quite many relays and I happen
to have some switches that are suitable. I will use jumpers for
certain options that I will not use for A/B comparison, like
changing the gain. However, although a rather trivial and
straightforward design, it did turn out to be trickier than I
expected. For instance, if you take the feedback from after
the buffer and have two op amps, but only one of them
drives the buffer, what to do about the other one? It would
get the feedback for the wrong amplifier! That might cause
strange effects that might carry over to the active op amp
via the rails, perhaps. There are several ways to solve this,
but they all seem to need rather complex switching networks.
I have settled on using separate feedback networks for each
op amp, the active one closed around the buffer, the other one
closed around its op amp only so it will at least be well behaved.
Thus, I need at least 3 poles per channel. I have some 8-pole
switches lying around and plan to use one of them, but I have
to basically start from this switch when doing the layout.
Anyway, this is not meant to be a final amplifier for permanent
use, but a kind of testbed for experiments.
better, but I might still need quite many relays and I happen
to have some switches that are suitable. I will use jumpers for
certain options that I will not use for A/B comparison, like
changing the gain. However, although a rather trivial and
straightforward design, it did turn out to be trickier than I
expected. For instance, if you take the feedback from after
the buffer and have two op amps, but only one of them
drives the buffer, what to do about the other one? It would
get the feedback for the wrong amplifier! That might cause
strange effects that might carry over to the active op amp
via the rails, perhaps. There are several ways to solve this,
but they all seem to need rather complex switching networks.
I have settled on using separate feedback networks for each
op amp, the active one closed around the buffer, the other one
closed around its op amp only so it will at least be well behaved.
Thus, I need at least 3 poles per channel. I have some 8-pole
switches lying around and plan to use one of them, but I have
to basically start from this switch when doing the layout.
Anyway, this is not meant to be a final amplifier for permanent
use, but a kind of testbed for experiments.
> I'm using a cross coupled
That topology has a very bad name in "Pro audio". It is prone to act funny.
It does allow either output to be grounded, and preserves balance as much as possible, but that is not at all necessary for driving headphones.
I think it also has a low input impedance, which varies as you apply shorts to the output.
Just use noninverting + inverting modes, or a whatever with an inverter for the other side.
> it may have just the right amount of sophistication.
I'm not sure that sophistication has much to do with sound quality. (Yes, it can be fun.)
That topology has a very bad name in "Pro audio". It is prone to act funny.
It does allow either output to be grounded, and preserves balance as much as possible, but that is not at all necessary for driving headphones.
I think it also has a low input impedance, which varies as you apply shorts to the output.
Just use noninverting + inverting modes, or a whatever with an inverter for the other side.
> it may have just the right amount of sophistication.
I'm not sure that sophistication has much to do with sound quality. (Yes, it can be fun.)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.