Referee:
Hugh and Nico say the diode string sounds better. John remains firm that he should not be able to hear the difference. Others have not commented on the "sound" of the diode vs. Vbe multiplier, and testing has not specifically been done to show whether or not this difference can be heard.
1: large transistors such as the BD139 or japanese types are easily mounted.
a. It is said that for temperature stability the device only needs to be near the heatsink.
2: The fundamental Vbe multiplier can have significant V-I difference from a diode string depending on the Hfe of the transistor, and also with regard to temperature.
a. A true diode curve is best approximated by using a high-gain transistor, for temperature as well. A TO92 package is suitable for this, but high gain can be had in larger packages as well (John uses the 2SA1381).
3: A diode string has the smallest phase error, though surely this is insignificant.
Gaetan, I'm afraid I don't have much experience simulating with Opamps, since I've never bothered to get models for them to work. The most I could do is make an intrinsic model made of behavioral sources. But I don't see this as necessary. It is not advisable to try to evaluate sonic quality from the simulator. If there is something wrong with your circuit, someone would probably have pointed it out already (you could also research schematics for well-known tone controls/equalizers).
- keantoken
Hugh and Nico say the diode string sounds better. John remains firm that he should not be able to hear the difference. Others have not commented on the "sound" of the diode vs. Vbe multiplier, and testing has not specifically been done to show whether or not this difference can be heard.
1: large transistors such as the BD139 or japanese types are easily mounted.
a. It is said that for temperature stability the device only needs to be near the heatsink.
2: The fundamental Vbe multiplier can have significant V-I difference from a diode string depending on the Hfe of the transistor, and also with regard to temperature.
a. A true diode curve is best approximated by using a high-gain transistor, for temperature as well. A TO92 package is suitable for this, but high gain can be had in larger packages as well (John uses the 2SA1381).
3: A diode string has the smallest phase error, though surely this is insignificant.
Gaetan, I'm afraid I don't have much experience simulating with Opamps, since I've never bothered to get models for them to work. The most I could do is make an intrinsic model made of behavioral sources. But I don't see this as necessary. It is not advisable to try to evaluate sonic quality from the simulator. If there is something wrong with your circuit, someone would probably have pointed it out already (you could also research schematics for well-known tone controls/equalizers).
- keantoken
Sheldon,
Thanks, you are right, geezers it is. This is not a word we ever use in Oz, my bad, as you Yankees colorfully say... And you are right when you say we compensate for our deteriorating hearing. The brain is a marvellous thing!
Nico, I like the idea of a tone control too, but I'm worried about the IC, not just because it's an IC, but also for the needed bipolar power supply. I think the simple board should have a crossfeed circuit at the front end, implemented before being passed to C1, the input cap, but that the tone control should be the snap-off feature.
T in AZ, thank you for your power supply file. However, there are two cascaded regulators, and unsurprisingly the CMC is not in circuit - no models, I assume. By golly it has a very tight regulation, unresolvable ripple as far as I can make out!
I really would prefer the diode string. It sounds more refined, and I believe Anthony's (aka KT) simulations show us why. The output stage of this tiny headphone amp runs low current, and uses huge 10R emitter resistors. It is uncritical and would never thermally run away, as with 40mA quiescent the drop across the emitter resistors is 400mV, a higher proportion of the BE voltage of the outputs vis a vis conventional TO-3P outputs (typically 30mV). The load is much higher impedance, too, and cap coupled. I feel mounting the diodes close to the outputs, but not even touching them, would be just fine.
Folks, it's coming along, and looking good. Would someone like to try searching for a suitable CMC, that is, a mains/2 x 6Vac 5VA pcb mount trafo, nice and cheap, from Digikey or Mouser? Something Chinese and widely available would be ideal. Once we have the pinout, we can move ahead with board layout.
Cheers,
Hugh
Thanks, you are right, geezers it is. This is not a word we ever use in Oz, my bad, as you Yankees colorfully say... And you are right when you say we compensate for our deteriorating hearing. The brain is a marvellous thing!
Nico, I like the idea of a tone control too, but I'm worried about the IC, not just because it's an IC, but also for the needed bipolar power supply. I think the simple board should have a crossfeed circuit at the front end, implemented before being passed to C1, the input cap, but that the tone control should be the snap-off feature.
T in AZ, thank you for your power supply file. However, there are two cascaded regulators, and unsurprisingly the CMC is not in circuit - no models, I assume. By golly it has a very tight regulation, unresolvable ripple as far as I can make out!
I really would prefer the diode string. It sounds more refined, and I believe Anthony's (aka KT) simulations show us why. The output stage of this tiny headphone amp runs low current, and uses huge 10R emitter resistors. It is uncritical and would never thermally run away, as with 40mA quiescent the drop across the emitter resistors is 400mV, a higher proportion of the BE voltage of the outputs vis a vis conventional TO-3P outputs (typically 30mV). The load is much higher impedance, too, and cap coupled. I feel mounting the diodes close to the outputs, but not even touching them, would be just fine.
Folks, it's coming along, and looking good. Would someone like to try searching for a suitable CMC, that is, a mains/2 x 6Vac 5VA pcb mount trafo, nice and cheap, from Digikey or Mouser? Something Chinese and widely available would be ideal. Once we have the pinout, we can move ahead with board layout.
Cheers,
Hugh
Last edited:
Like this?: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=MT2098-ND
It has 5V secondaries. 6V are harder to source.
Nico. Geyser is a noun that refers to an source of water or steam and water that shoots up due to geothermal activity. Geezers is slang. MS spell check isn't quite that sophisticated.
In the U.K.: A guy, a bloke, a person in general. The British equivalent of the American slang word "dude".
In the U.S.: An old man, particularly one who is either cranky or eccentric. Rather derogatory term.
Nico, get Firefox! 
One more thing, and I'll leave the diodes alone.
John and others, I'm sorry if I seemed rude when defending the diodes.
By placing a special capacitor, we can greatly increase the regulation of a Vbe multiplier at AC, one advantage the diodes don't have. Here is the plot and my simulation file.
This works quite well with the BD139, and will probably work even better with the MPSA18.
- keantoken
One more thing, and I'll leave the diodes alone.
John and others, I'm sorry if I seemed rude when defending the diodes.
By placing a special capacitor, we can greatly increase the regulation of a Vbe multiplier at AC, one advantage the diodes don't have. Here is the plot and my simulation file.
This works quite well with the BD139, and will probably work even better with the MPSA18.
- keantoken
Attachments
So the supply is down to 12V now, I suppose in lieu of the recent advice on headphone maximum levels? If so an adjustment of VAS bias is in order, and are we using diodes+resistor or just plain diodes? I must update my schematic.
Hugh, are you okay with my modified bootstrap? Since no one has complained so far I'll assume the change is solid with your approval.
- keantoken
Hugh, are you okay with my modified bootstrap? Since no one has complained so far I'll assume the change is solid with your approval.
- keantoken
Member
Joined 2009
Paid Member
In the UK we use 'old Codger' as the equivalent of old Geezer. 
Tone controls - actually, I'm afraid I do like this idea. But if we did implement it I'd be wanting to go the whole way and include 3-band EQ as this would also address the ear canal resonance issue.
How well do snap-off boards work ? I haven't seen this before and I'm a bit skeptical based on my own pcb shaping abilities. An alternative is to agree on a circuit but leave individual DIYers to add an adjunct pcb of their own making into the chasis ?
Here's an interesting article on the 3-bad EQ approach:
http://www.headwize.com/projects/equal_prj.htm
We could implement it with a quad packaged op-amp if such things exist with high quality op-amps. We might be able to fake the dual supply if current draw is relatively low.
I know this EQ / tone control thing may be getting us off-track from a simple design, and we may readily kill this option quite quickly - but as I'll be an Old Codger one day I'm in favour of a bit more discussion on this topic before we decide one way or the other.
Bias - I think consensus has steered back to diodes, but at least I stirred up the pot for a page or two and learned a bunch of things about this topic. I still like Vbe multipliers and now that Keantoken has given the old faithful BD139 a thumbs up in light of our better understanding of these things I'm not inclined to change my mind
- but if it sounds better with diodes then I'm all behind using 'em here (with resistor to allow adjustment)
Tone controls - actually, I'm afraid I do like this idea. But if we did implement it I'd be wanting to go the whole way and include 3-band EQ as this would also address the ear canal resonance issue.
How well do snap-off boards work ? I haven't seen this before and I'm a bit skeptical based on my own pcb shaping abilities. An alternative is to agree on a circuit but leave individual DIYers to add an adjunct pcb of their own making into the chasis ?
Here's an interesting article on the 3-bad EQ approach:
http://www.headwize.com/projects/equal_prj.htm
We could implement it with a quad packaged op-amp if such things exist with high quality op-amps. We might be able to fake the dual supply if current draw is relatively low.
I know this EQ / tone control thing may be getting us off-track from a simple design, and we may readily kill this option quite quickly - but as I'll be an Old Codger one day I'm in favour of a bit more discussion on this topic before we decide one way or the other.
Bias - I think consensus has steered back to diodes, but at least I stirred up the pot for a page or two and learned a bunch of things about this topic. I still like Vbe multipliers and now that Keantoken has given the old faithful BD139 a thumbs up in light of our better understanding of these things I'm not inclined to change my mind
- but if it sounds better with diodes then I'm all behind using 'em here (with resistor to allow adjustment)
Last edited:
Hello Hugh
Sorry, I've forgot that it was single rail amp. I did think about the op-amp tone control also for my LF amp who now replaced the Radford, so I've done it double rail.
Until few month ago I was using a Radford HD250 integrated amp, there is a tone control in that amp, I was needed to raise the treble to get the high freq. correctly, but there is also a possible high lost from my room acoustic.
I will use a tone control with the Haksa, so for old geezers like me who will want to use an ad-on tone control line amp, here is the schematic that I may use, it's a Crimson Electric tone control line amp, it's single rail. A 30 years old circuit but it still a fine circuit like my old Crimson power amp. Btw I did not simulated this tone control line amp, but it should work good as is.
KT, my Tina simulator are much better for discrete transistors circuit.
Dont worry guys, I will not hack the thread with tone control circuits.
Thank
Bye
Gaetan
Attachments
Last edited:
KT,
No problem on the bootstrap from the emitter of the upper device. But fb should still come from the output, I feel.
BG,
EQ? Que? I'm OK with a tone control, but full EQ, this thing will be bigger than a Black Eyed Pea concert!! Unless it's just two or three components, I'm not keen about this, I admit, if only because I, with Nico, will be negotiating the layout..... groan.......
Gaetan,
Thanks for the new tone control, I like it. Anything Radford gets my vote, Arthur Bailey did the SS amps for Radford, and they were very, very good. Your circuit could be preceded by the cross feed, too. I will work on this later today.
KT,
Where is this cap which improves the Vbe multiplier? I've got so many sims floating around in my files now that I'm reluctant to save yet another. Does it also improve phase shift over the diode string? I confess your compound graphs confused me somewhat.
Supply remains at 24V. There are advantages with a very high Vcc. I use 125V in a SS gain block in one of my preamps. The distortion is very low, and dynamics amazing. Rupert Neve uses more than 100V in his mixing desks, in fact.
Gareth,
Surely that was old tosser?
Hugh
No problem on the bootstrap from the emitter of the upper device. But fb should still come from the output, I feel.
BG,
EQ? Que? I'm OK with a tone control, but full EQ, this thing will be bigger than a Black Eyed Pea concert!! Unless it's just two or three components, I'm not keen about this, I admit, if only because I, with Nico, will be negotiating the layout..... groan.......
Gaetan,
Thanks for the new tone control, I like it. Anything Radford gets my vote, Arthur Bailey did the SS amps for Radford, and they were very, very good. Your circuit could be preceded by the cross feed, too. I will work on this later today.
KT,
Where is this cap which improves the Vbe multiplier? I've got so many sims floating around in my files now that I'm reluctant to save yet another. Does it also improve phase shift over the diode string? I confess your compound graphs confused me somewhat.
Supply remains at 24V. There are advantages with a very high Vcc. I use 125V in a SS gain block in one of my preamps. The distortion is very low, and dynamics amazing. Rupert Neve uses more than 100V in his mixing desks, in fact.
Gareth,
Surely that was old tosser?
Hugh
Member
Joined 2009
Paid Member
two or three components - that's a bit optimistic !
Well, I read somewhere that decent tone controls means an opamp at least and treble and bass is 2-band so one more band and, well, in for a penny in for a pound...
(check out that link)
Okay then, it's decided.
This cap is placed in parallel with the top resistor, the one that goes from C-B. It decreases the divider's impedance, so that the base current of the transistor causes minimal error.
I'm not sure I understand about the transformer then. How do we get 24V from a 5V trafo? Voltage multiplier???
Off, topic, but with respect to canal resonance:
I'll bet we could have 2 pots, one for resonance frequency (say to vary +-2KHz of "standard" resonance) and the other for volume of the resonance correction. Not sure how easy this is, but it would be perfect I think. I'm not suggesting we should include it in the project, but it would work better than a full-blown equalizer.
- keantoken
KT,
Thanks for the cap trick, how big, presumably as an impedance at 20Hz a small fraction of 1K5, the usual value of this resistor, so around 22uF, same as the bypass?
We need ANOTHER trafo for the power supply, the Tamuras selected by Sheldon are simply the common mode chokes!! Two 10V secondaries, or, a single 20Vac secondary, is all we need.
What is the usual frequency of canal resonance? Does it vary much from individual to individual? This is unexpected; hadn't realised it was an issue.....
Cheers,
Hugh
Thanks for the cap trick, how big, presumably as an impedance at 20Hz a small fraction of 1K5, the usual value of this resistor, so around 22uF, same as the bypass?
We need ANOTHER trafo for the power supply, the Tamuras selected by Sheldon are simply the common mode chokes!! Two 10V secondaries, or, a single 20Vac secondary, is all we need.
What is the usual frequency of canal resonance? Does it vary much from individual to individual? This is unexpected; hadn't realised it was an issue.....
Cheers,
Hugh
I heard on some other audio forum that canal resonance is around 3KHz. Google brought up this:
http://www.neurophys.wisc.edu/h&b/textbook/external_ear.html
But I think one could find better info by searching on headwize, etc.
Canal resonance does vary, with age as well.
As for the cap, I just did a simulation and for an equivalent response of -3db at 20Hz, the cap is 47uF. (but the usual bypass cap is smaller than this, I'm not sure how important bass response of the Vbe multiplier is).
- keantoken
http://www.neurophys.wisc.edu/h&b/textbook/external_ear.html
But I think one could find better info by searching on headwize, etc.
Canal resonance does vary, with age as well.
As for the cap, I just did a simulation and for an equivalent response of -3db at 20Hz, the cap is 47uF. (but the usual bypass cap is smaller than this, I'm not sure how important bass response of the Vbe multiplier is).
- keantoken
Oh, look what I found. It has the crossfeed and stuff, everything we need to know. Is it desirable in any way to have an inductive crossfeed? Hmmm...
http://www.headwize.com/tech/sshd_tech.htm
It doesn't seem to say anything specific about canal resonance though.
EDIT: This page has some more stuff:
http://www.headwize.com/projects/equal_prj.htm
So I've now heard figures of 5KHz and 3KHz for canal resonance.
- keantoken
http://www.headwize.com/tech/sshd_tech.htm
It doesn't seem to say anything specific about canal resonance though.
EDIT: This page has some more stuff:
http://www.headwize.com/projects/equal_prj.htm
So I've now heard figures of 5KHz and 3KHz for canal resonance.
- keantoken
Last edited:
Member
Joined 2009
Paid Member
That's the same link I posted earlier - perhaps there's a bit too much to digest with all these posts, I will step back for a page or so... and see how things settle out a bit.
Hello Hugh and keantoken
The Bauer's passive auralization crossfeed network circuit seem very interesting and relatively simple.
http://www.headwize.com/tech/sshd_tech.htm
Bye
Gaetan
The Bauer's passive auralization crossfeed network circuit seem very interesting and relatively simple.
http://www.headwize.com/tech/sshd_tech.htm
Bye
Gaetan
Attachments
Okay, here is the updated schematic. Doesn't yet include CF or muter, etc.
I've also attached the PSU noise as seen by a headphone impedance of 32 ohms, at 100db/mW. I couldn't match the datasheet curve for the 10u bypass, so treat this like "worst case" I guess. Looks good.
- keantoken
I've also attached the PSU noise as seen by a headphone impedance of 32 ohms, at 100db/mW. I couldn't match the datasheet curve for the 10u bypass, so treat this like "worst case" I guess. Looks good.
- keantoken
Attachments
