PANDA Headphone amp kit, build thread, upgrades & tweaks.

[dicky]

real pioneering

;D ;D ;D

Not so much pushing back the frontiers of science as stumbling across the borders of incompetence

Thanks for your (and Mike's) help - although I don't have a clue what I'm doing I'm still very interested to see what effect these 'improvements' to the Panda have.

[Deleted]

Not so much pushing back the frontiers of science as stumbling across the borders of incompetence

If you have a keyboard, try this:

D GC B GE AD

To boldly listen in the kaazi.

[dicky]

I don't, I'm afraid.

[dicky]

I'll be using 0.5W resistors in parallel with the 1k5's. Can I also use 0.5W resistors when experimenting with R6/R17/R34/R50? Is there any benefit in using 1W resistors - heat dissipation, extended life?

[Deleted]

there is virtually no current there.
There is from the PS to the zener but almost nothing (a few uA) from the zener to the base of the transistor.
So 0.4W is already heavy overkill.

[dicky]

Thanks Solderdude 0.5W it is then.

I'll make some final voltage measurements now and put an order together. Should get the components for the weekend.

[Deleted]

A w1000?I loved your numerical reply!!! Talk about exactness? It just tickled the old rabbit tail.Frans, the output resistance seems to be a balancing act. I have played around with your rather nifty device and there are differences in the balance with different resistance but there is a point where the output of some amps simply runs out of steam and it becomes lifeless.So it seems that the output impedance is truly a balancing act between tone/impact. I kind of see that low output amps can be problematic if the output impedance is low and you're after the 'so called' flat response.It's really hard to reach an ultimate conclusion since it's subtle and depends very much on what you're listening to.

The wood looked a bit like the purple of the AD700

The resistors we were talking about is not in the output path (this time) but in the powersupply part of the Panda and if this value is changed it also changes the internal resistance of the regulator in the Panda which might or might not be audible.

the output resistance optimum is a matter of HP/amp and personal preference and might not be very audible to most, but knowing what to look for I can hear substantial differences with some cans, less with others (as explained in the impedance explanation pdf).
Small changes in output resistance are inaudible to me b.t.w., significant changes are.

[PinkFloyd]

Dicky,

I'd forgotten what I'd done to "Panda 4" (I still haven't fired it up yet!) so hoovered the dust off it today for a look:





Must fire her up one of these days for a listen

[PinkFloyd]

A few more pandas here:

s157.photobucket.com/albums/t69/pinkfloyd2007_bucket/Panda/

s157.photobucket.com/albums/t69/pinkfloyd2007_bucket/panda2/

s157.photobucket.com/albums/t69/pinkfloyd2007_bucket/Panda%203/

s157.photobucket.com/albums/t69/pinkfloyd2007_bucket/Panda%204/

s157.photobucket.com/albums/t69/pinkfloyd2007_bucket/ians%20panda%20amp/

s157.photobucket.com/albums/t69/pinkfloyd2007_bucket/panda%20three/

s157.photobucket.com/albums/t69/pinkfloyd2007_bucket/panda%20technical/

[dicky]

Panda 4 has evolved some extra components since I last saw it - the white caps.

[PinkFloyd]

Feb 15, 2011 22:30:26 GMT dicky said:

Panda 4 has evolved some extra components since I last saw it - the white caps.

If you PM me your address now I will send you four in the morning.... they're Chinese polyprops, very good quality too!

[dicky]

Thanks Mike, that's very kind of you - you have PM.

Is there a 'rule of thumb' for decoupling capacitors? At the moment I have 4700uF with 470nF (factor of 10000) polyprop and 22pF (another factor of 21000) polystyrene connected across it.

[joethearachnid]

Gosh, I feel quite dirty after looking all those Panda pictures. It's not every day that you get to see four separate amplifiers undressed, each one more sexy than the previous one.

I've been having some doubts about the Panda over the past few days. The bass seemed a bit overbloated and the treble was shouty. Either I was going insane or it's suddenly got through that phase, as I'm once again listening in the palaces of Olympus.

-JoetheArachnid

[dicky]

Hi Joe, so what do you think changed? I often find I have to be in the right mood for certain listening - otherwise it grates on my nerves and I just can't enjoy it.

[jonclancy]

Where are those polyprops used, Mike? Rails?

[PinkFloyd]

Feb 15, 2011 22:53:55 GMT dicky said:

Thanks Mike, that's very kind of you - you have PM.

Is there a 'rule of thumb' for decoupling capacitors? At the moment I have 4700uF with 470nF (factor of 10000) polyprop and 22pF (another factor of 21000) polystyrene connected across it.

Well..... It's a very individual thing Dicky... with a 1,000uF cap I normally go for anything between 100nF and 1uF... a 3,330uF anything between 300nF and 3uF.

I'm sure Frans will tell you, there should be measurable benefits to the high frequency impedance even using bypass caps. Your typical 3,300uF electrolytic will start bottoming out (become inductive) @ around 6kHZ and 100uF electrolytics @ around 30kHz.... your film cap will remain capacitative way up to to the 200kHz region (0.47uF) or 600kHz (0.1uF) or way up into the megahertz zone with even smaller caps.

Quite how this is perceived by the "human ear" is a different story altogeher which is why it is always a good idea to experiment with bypass caps (values and types) and sometimes even beneficial to use a multicap.... the 4700uF on my Panda have been bypassed with polystyrene, polyester and polypropylene, you could even add a silvered Mica if you so desire.... it's all about tuning to your personal "taste".

[joethearachnid]

Feb 15, 2011 23:03:56 GMT dicky said:

Hi Joe, so what do you think changed? I often find I have to be in the right mood for certain listening - otherwise it grates on my nerves and I just can't enjoy it.

I honestly can't say. Somehow the bass quantity seemed too much and was making the sound as a whole muddy and unfocused. I'd imagine that a good quantity (if not all) of it is in my head and that it's just my brain messing with me. Perhaps it's an operating temperature thing (possible), perhaps it's to do with cables (doubtful).

I'm just glad that I can enjoy listening to it again.

Mike - which 3.5mm socket did you use on the back of Ian's Panda? I'm considering putting in a pre-out for possible orthodynamic duties and that seems like the easiest way to wire one. I can't find any quality-looking ones that seem to fit the bill.

-JoetheArachnid

[PinkFloyd]

Feb 15, 2011 23:24:28 GMT joethearachnid said:

Feb 15, 2011 23:03:56 GMT dicky said:

Hi Joe, so what do you think changed? I often find I have to be in the right mood for certain listening - otherwise it grates on my nerves and I just can't enjoy it.

I honestly can't say. Somehow the bass quantity seemed too much and was making the sound as a whole muddy and unfocused. I'd imagine that a good quantity (if not all) of it is in my head and that it's just my brain messing with me. Perhaps it's an operating temperature thing (possible), perhaps it's to do with cables (doubtful).

I'm just glad that I can enjoy listening to it again.

Mike - which 3.5mm socket did you use on the back of Ian's Panda? I'm considering putting in a pre-out for possible orthodynamic duties and that seems like the easiest way to wire one. I can't find any quality-looking ones that seem to fit the bill.

-JoetheArachnid

Hi Joe,

It was a "lundberg".... I think I bought 5 which means I should still have "4" somewhere.... I will have a hunt around tomorrow.

Mike.

PS: what I have found is that when the sound starts to "sound" like shite it's best to take a few days away from it, come back with a fresh pair of ears after you've had a rest.

Most of the time, it's not the equipment that needs "tweaking" it's the listener... go outdoors, sniff the roses and get some daylight on your face.... come back in a few days later and have a listen again

[PinkFloyd]

.......................... having said that, the amp could well be sounding like shite! Thing is, did it sound OK before? Is this bloaty bass a recent thing or has it always been that way?

[Deleted]

I'm sure Frans will tell you, there should be measurable benefits to the high frequency impedance even using bypass caps. Your typical 3,300uF electrolytic will start bottoming out (become inductive) @ around 6kHZ and 100uF electrolytics @ around 30kHz.... your film cap will remain capacitative way up to to the 200kHz region (0.47uF) or 600kHz (0.1uF) or way up into the megahertz zone with even smaller caps.

3300uF will be bottoming out @ around 50kHz-100kHz (depending on make/type), 10uF electrolytic @ around 1MHz, 1uF ceramic @ 10MHz, 100nF @ 20MHz, 10nF@ 50MHz, 1nF @ 150MHz and 100pF @ 500MHz.

Since the Panda is class-A and therefore has no 'audiocurrent' on top of the idle current drawn from the power suplly (the beauty of class A) and there are no switching components nor superfast opamps nor components near oscillation 10MHz bottoming ESR should be more then sufficient and frequencies above this is not of importance anymore..

Now here is the kicker... the Panda designers already thought of this and in parallel to the 3300 they mounted 220uF and 100nF caps on the PCB in the power lines assuring low ESR across the line.
The ESR is already 'guaranteed' low to at least 10MHz.

Furthermore decoupling the big capacitors some physical distance away from the parts that need the decoupling (output devices and regs) needs to be as close as possible to these parts as the inductance of the 'traces' from the big caps to the circuit defeats the purpose of mounting them further away.
For this reason decoupling caps are always mounted as close as possible to the critical parts.

What they (Panda designers) didn't do (nor need ?) was to decouple their own regulator output with 100nF or 10uF (from Emitter Q3, Q12, Q22, Q28 to ground)
Seeing how they decoupled the PCB itself already it is likely they already tried such a thing.
Also it being class-A helps in this case.

In class AB (when used as a 'power' headphone amplifier) it is totally another matter as the audio load currents are superimposed on the PS current.
optimizing the zenerdiode current as mentioned helps in this case.
Also very fast opamps need their local decoupling and as soon as something digital is inserted in the circuit the rules change dramatically but is not the scope of this explanation as it would become a really big article with loads of piccies/graphs.

I hope this clarifies something about decoupling of power lines.

disclaimer:
I know in audioland there are different theories based on their observations but this explanation is based on how electronics behave.

[PinkFloyd]

As I said Frans..... all down to personal "taste".... this is one of those audiophool areas where the "ears" are the judge. Whether it be placebo or actual fact, I don't know, but bypassing electrolytics does make a difference to the SQ, whether it be an "improvement" or not, it definitely can (and does) tune the sonics.

I believe in the "house sound" of capacitors (as do many others) and it is possible to "tune" this house sound by the addition of different brands of film caps.... basically, we no longer have a 4700uF "nichicon" cap (with the Nichicon house sound) we now have a 4701uF "multicap" comprising of 4700uF electrolytic, 33pF polystyrene, 100nF polypropylene and 470nF polyester.... this, in effect, has "tuned" that electrolytic into a totally different sound from the Nichicon "house" sound.

I know that Leo (for example) doesn't like the Nichicon house sound (I'm not overly keen on it either) and this is one way of tweaking the house sound to your own preferences without physically replacing the cap for an other brand (other house sound)..... as I say, there are no rules here.... it is ALL down to personal taste and pretty much a "suck it and hear" type exercise.... fine tuning a cap to your own "house sound" takes a lot of trial and error but once you get it right you will be a firm believer

[jonclancy]

Like bypassing a polyprop with a teflon etc...

I will have to revisit this concept.

[dicky]

Well..... It's a very individual thing Dicky... with a 1,000uF cap I normally go for anything between 100nF and 1uF... a 3,330uF anything between 300nF and 3uF.

I'm sure Frans will tell you, there should be measurable benefits to the high frequency impedance even using bypass caps. Your typical 3,300uF electrolytic will start bottoming out (become inductive) @ around 6kHZ and 100uF electrolytics @ around 30kHz.... your film cap will remain capacitative way up to to the 200kHz region (0.47uF) or 600kHz (0.1uF) or way up into the megahertz zone with even smaller caps.

Quite how this is perceived by the "human ear" is a different story altogeher which is why it is always a good idea to experiment with bypass caps (values and types) and sometimes even beneficial to use a multicap.... the 4700uF on my Panda have been bypassed with polystyrene, polyester and polypropylene, you could even add a silvered Mica if you so desire.... it's all about tuning to your personal "taste".

Thanks Mike. I'll do the mods and see what effect they have. I have a few other components I can swap if there is a significant change - I have no idea what to expect.

[dicky]

If you PM me your address now I will send you four in the morning.... they're Chinese polyprops, very good quality too!

Hi Mike, polyprops arrived safely. Many thanks.

I'll fit them over the weekend, along with Solderdudes' zener resistor mod. It'll be nice to actually listen to this thing instead of sniffing solder fumes!

[dicky]

3300uF will be bottoming out @ around 50kHz-100kHz (depending on make/type), 10uF electrolytic @ around 1MHz, 1uF ceramic @ 10MHz, 100nF @ 20MHz, 10nF@ 50MHz, 1nF @ 150MHz and 100pF @ 500MHz.

Since the Panda is class-A and therefore has no 'audiocurrent' on top of the idle current drawn from the power suplly (the beauty of class A) and there are no switching components nor superfast opamps nor components near oscillation 10MHz bottoming ESR should be more then sufficient and frequencies above this is not of importance anymore..

Now here is the kicker... the Panda designers already thought of this and in parallel to the 3300 they mounted 220uF and 100nF caps on the PCB in the power lines assuring low ESR across the line.
The ESR is already 'guaranteed' low to at least 10MHz.

Furthermore decoupling the big capacitors some physical distance away from the parts that need the decoupling (output devices and regs) needs to be as close as possible to these parts as the inductance of the 'traces' from the big caps to the circuit defeats the purpose of mounting them further away.
For this reason decoupling caps are always mounted as close as possible to the critical parts.

What they (Panda designers) didn't do (nor need ?) was to decouple their own regulator output with 100nF or 10uF (from Emitter Q3, Q12, Q22, Q28 to ground)
Seeing how they decoupled the PCB itself already it is likely they already tried such a thing.
Also it being class-A helps in this case.

In class AB (when used as a 'power' headphone amplifier) it is totally another matter as the audio load currents are superimposed on the PS current.
optimizing the zenerdiode current as mentioned helps in this case.
Also very fast opamps need their local decoupling and as soon as something digital is inserted in the circuit the rules change dramatically but is not the scope of this explanation as it would become a really big article with loads of piccies/graphs.

I hope this clarifies something about decoupling of power lines.

disclaimer:
I know in audioland there are different theories based on their observations but this explanation is based on how electronics behave.

;D

Blimey, OK so these mods won't improve the tech spec but may change the flavour of the sound. Fair enough - still worth having a play.

One more question about the 1k power supply resistors; you said they can have any value - even a short circuit. It seems a little counter-intuitive, but I can simply short the resistor on the back of the board?!