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

[pandapops]

Wondered if there was much difference in this circuit between polystyrene and siver-mica, you're probably the only person who's heard both.

[Deleted]

Oct 5, 2011 22:13:08 GMT pandapops said:

Wondered if there was much difference in this circuit between polystyrene and siver-mica, you're probably the only person who's heard both.

I am in 100% agreement with what Mike is saying.
You may notice small differences between silver mica and polystyrene in higher values, and when used in areas such as the main amplification stage of an amplifier ( VAS stage) where typical values are more likely to be 68pF, or more commonly 100pF.
I would never normally use a ceramic type capacitor at that location either. However 22pF is a VERY low value and any perceived differences between types of capacitors of that value would be vanishingly small.
Alex

[pandapops]

Replacing C8/16( as per schematic) with polystyrene 22pf was not at all a small change, it was the classic case of veil removal, I then did the other 22pfs and got some improvement and some harshness and sibilance as a trade-off.

[Deleted]

Oct 6, 2011 2:51:54 GMT pandapops said:

Replacing C8/16( as per schematic) with polystyrene 22pf was not at all a small change, it was the classic case of veil removal, I then did the other 22pfs and got some improvement and some harshness and sibilance as a trade-off.

I was referring to the differences between mica and polystyrene, not with ceramic.As I said , I would not have used ceramic, as they are quite non linear.
I wouldn't regard added sibilance and harshness a good tradeoff.

[pandapops]

It became tiring quickly. Added tantalum resistors on inputs and that calmed things down, will add some more of those to the signal path and then will consider swapping out the second batch of polystyrenes.
Will try Toshs for the driver stage TIPs.

Is there a guide for converting the voltage regulator to CCS? this should tidy the highs also.

[pandapops]

So for a 22pf its a choice of:
Ceramic - nah
Polystyrene - over done these a bit
Mica - audiophile choice but wouldn't these sound like the 'styrene
Polyester - like the green one people removed from the Jaycay, is this a default winner?

Tantalums in and they are great, I can see why Audio Note use them.

[leo]

I tend to be cautious with polystyrene,In my experience some of them are quite inductive and can make the high's sound glassy in some filter circuits.
NPO/COG ceramics should be better tbh in such a low value, problem is they can be microphonic. Mica's tend to be high voltage so physically bigger, if not too big should be decent.

BTW thats a lovely build Mike Wonder how much that would go for if it was a commercially ready built job in some swanky hi-fi shop

[pandapops]

Glassy is a very good description. Is it likely that the caps are causing the problem rather then revealing it?

This may be a provocative question for some but would the Panda be more natural with bipolar rather than JFET input section? I'm not in the least bit qualified to design an alternative and wouldn't just stick bit in, just asking.

[Deleted]

Glassy is a very good description. Is it likely that the caps are causing the problem rather then revealing it?

The only way to answer that to your own satisfaction would be to try silver mica at the same locations.
IMHO, bipolar input is more neutral sounding, although many like the added warmth that FETs give.A major advantage of bipolar is that they are more readily available as a very closely matched pair.
It is much harder to do this with FETs. Then of course, there is the availabilty problem with many manufacturers having discontinued discrete FET production.
Alex

[PinkFloyd]

Would cost mega money ready built....

[leo]

I totally agree, just shows the high quality available though after putting a lot of time an effort into a custom build. There's other much more expensive ready built stuff which is quite frankly shit in comparison.

[leo]

Oct 6, 2011 19:44:28 GMT pandapops said:

Glassy is a very good description. Is it likely that the caps are causing the problem rather then revealing it?

This may be a provocative question for some but would the Panda be more natural with bipolar rather than JFET input section? I'm not in the least bit qualified to design an alternative and wouldn't just stick bit in, just asking.

Its possible, some polystyrenes are much worse than others. Apart from a few mods most the caps in the Panda I have here are pretty much stock, it certainly never sounds overly bright, infact more warm if anything.

[PinkFloyd]

By the way.. the pot has been grounded

[rob100]

Mike, the amp is superb. It looks even better in the flesh than it does in the pics (well it does to me, the misses still needs convincing!) and sounds fantastic! I've posted a mini review in the headphone amp section.

I don't think my soldering skills will ever be up to the standard of building one of these myself, but I'm intrigued by how you can tune the sound and what difference the byassing and silvered mica caps make to the SQ and how this compares with stock?

[artur]

hi everyone.

I'm building a panda as well but i have a little problem.

during the build i accidentally swapped the B649 and the D669 after the firs power up i found that mistake and resoldert the board.

so far so Good

but now i have a new problem one channel is working really fine and i managed to trim the voltage. but the channel is giving me some trouble.

first of all it burns one of the 2.2 resistor so i started tho measure around in that channel i found that the voltage at the zener near the backside of the alps is 25 /25volts on both sides looks like there is something wrong.

could it be that i grilled something by installing the d669 on the wrong place ?

maybe someone could help me find my problem.

Thanks Artur

[pandapops]

Those 2.2 or 3ohm resisors are pretty big so can deal with quite a lot of power, I'd have thought they'd survive a small disaster, which suggests that something is drawing way more current than it should. A transistor can be tested by treating the legs as diodes, so look at the 649 and 669, also its worth testing the tip41/42s on the bad channel as these can die quite easily.

Most multimeters have a diode tester, these links should help
www.kpsec.freeuk.com/components/tran.htm
www.elexp.com/t_test.htm

Also check the 1K and 1.5K resistors near the 649/669. My guess would be that you've killed those 2 transistors so the zener becomes irrelevant, there are much more knowledgeable guys around here than me, so my advice: Don't power the amp up again before doing some basic tests and taking advice on the results from people that know, Mike, Solderdude, SandyK, Will and probably a bunch of other around here would help.

[gommer]

Hi Mike,

Did you ever get as far as comparing your different Panda's WRT cap rolling? I'm not asking what sounds best, but more like which decoupling actions have an affect for the better.

I might sound lazy, but I'm not. I simply don't have the time to experiment much and would like to immediately concentrate on what's tried and tested by others. Shameless, i know

Cheers, Marc

[Deleted]

Would it be possible to compile some sort of guide to the different mods one might perform to the Panda?
I know it sounds great in the standard configuration, but a lot of you guys have tweaked it, and some even did so in several steps in pursuit of the ultimate Panda. I know this will sound lazy, but a guide to the separate mods and what their effect is would be absolutely great for those of us lacking the knowhow (or balls) to start from scratch.
Something like "input caps" what does it do and what types to use, bypassing components etc.
I might build a second Panda soon and since I am in no hurry now, it would be fun to try and go for the "ultimate build".

[pandapops]

Okay what I did, what it cost and the results:

Bypass larger electrolytic caps with 100pf polystyrene - free as from my spares. Big improvement in detail, top end and also bass clarity. NB Zener voltage regulators are known for benefitting from bypassing the zener, which in this circuit is the same as bypassing the 100uf.

Didn't use protection circuit - they cannot help but be detrimental for day to day use.

Used a seperate PSU for each channel - 2 50VA transformers, rectifier, additional cap per rail with a very small value resistor per cap, housed in its own box. Cost about £40. Impact was amazing, I'd not heard a dual mono headphone amp before and now would never go back. In general its best to get either shielded transformers or move them from the signal with this circuit, had the same experience with another discrete design but not with bog standard IC opamp to buffer type amps.

Fitted Shinkoh resistors on signal path, 5 per channel - about £30. More detail, removal of some edge and more defined bass, bass definition was my major complaint with stock.

Polystyrene caps instead of ceramic, I think 4 a channel - about £10. Big detail improvement and some harshness, which was corrected with Shinkohs and Toshiba transistors.

Replaced driver stage Tip41/42 with Toshibas 2SA1930/2SC5171 - cost a couple of pounds, bought 5 pairs and used 2 for £9. More detail, in voice or string, precussion whatever, definitely gets something that TIP41/42 miss. Downside might be some echo or something becomes audible at time with some sounds I think this is just picking up what the source is giving, however it may mean a change in filter cap is needed.

Have not fitted an input cap or anything on output.

Current sound is beyond what I'd thought my CD player and headphones could manage. I've had more detail with a highly modified DAC, similar when trying Grados, but never liked the sound that much.

Further improvements - the voltage regs for the low power bit should have room for improvement, not sure how to do this without getting complex/expensive, so not for now.

Additional transformers for those low power sections and stepped attenuator would round things off. Amp is already the best bit in the chain, so may do this someday.

A headphone amp with 4 transformer, stepped attenuator, boutique parts and super regulators would cost thousands to have built, could be yours for half the cost of an A1.

[infundibolo]

Nov 18, 2011 2:51:18 GMT pandapops said:

Used a seperate PSU for each channel - 2 50VA transformers, rectifier, additional cap per rail with a very small value resistor per cap, housed in its own box. Cost about £40. Impact was amazing, I'd not heard a dual mono headphone amp before and now would never go back. In general its best to get either shielded transformers or move them from the signal with this circuit, had the same experience with another discrete design but not with bog standard IC opamp to buffer type amps.

This seems a lot interesting mod, but not an easy one (at leas for me). Any further detail on this mod will be very appreciated.

[Deleted]

Jan 17, 2012 15:16:22 GMT infundibolo said:

Nov 18, 2011 2:51:18 GMT pandapops said:

Used a seperate PSU for each channel - 2 50VA transformers, rectifier, additional cap per rail with a very small value resistor per cap, housed in its own box. Cost about £40. Impact was amazing, I'd not heard a dual mono headphone amp before and now would never go back. In general its best to get either shielded transformers or move them from the signal with this circuit, had the same experience with another discrete design but not with bog standard IC opamp to buffer type amps.

This seems a lot interesting mod, but not an easy one (at leas for me). Any further detail on this mod will be very appreciated.

COMMENT.
2 x 30VA transformers are more than adequate here.
Alex

[pandapops]

Briefly what I did:
1. Buy ABS box, or fancy pricey box if you prefer, and secure the transformers inside with shared switch and fuse. To get power into that box I just drilled a hole in the plastic and put the cable through and clamped it either side of the hole( fine for me, but not gonna meet industry standards). Cable had a particularly thick outer layer and won't be damaged by the case, a more normal method would be to use an IEC inlet.

Used 50VA transformers as bigger transformer means lower impedance, could go bigger still or smaller as Alex says.

2. Basic PSU circuits: 4 diode rectifier after each transformer giving a +/- voltage and taking a ground from the centre tap. Then filter caps on + to ground and - to ground, each followed by a 2ohm resistor and another cap. What caps are used and if you add a resistor is open to infinite variation, spares on hand, budget etc, arguments about a best result would go on for ever. At this stage check voltages are about right, remember that the low power section of the Panda has a voltage regulators anyway, so a couple of volts extra isn't a big deal.

My construction method is to glue the caps together, do the soldering that can be done outside of the box, glue that into the box and then finish off the soldering. Glue used was Bison impact adhesive from Rapidonline, which takes a day to dry. This is not industry standard but works fine and when dry the glue is strong enough to lift the box by the caps, most people would just use veroboard and then used standoffs.

3 Getting power to amp. Power entry points are the big blue resistors near the main big PSU caps: R9, R23 and R41 and 52, these are snipped at the end so the original power supply is circuit is broken and wired to the new PSU box.

From the schematic you can see that the juice flows through the resistors, is filtered by the big caps C7 and C15 and C25 and C33 and then goes to the output transistors and onto the rest of the circuit.

Those resistors are the ones labelled 2.2 on the PCB( mine were actually 3ohm), on the left in the second image down www.mcfrede.dk/projects/panda_headamp/panda_headamp_01.shtml

Obviously the leads from PSU to Panda can be anything from basic hookup wire to shielded cable with some sort of plug so that the whole thing can be easily dismantled.

[trolololguy]

Hi!
Trouble with a headphone amplifier. May only sound in left channel (sounds good But right channel is almost dead, you hear ytters bit of the song running in the right but as I said almost dead.
Have adjusted the "DC offset" on both channels get it down to 3mV. And more than 1.5 k resistors so it is 1.3 V for both channels.
Have checked it right resistance in the right place, same with transistors, everything seems to sit at the right place. Transformer I'm using is a 2x20V 0.5 A on both.
Here are also pictures of the amplifier.



So how do I do to troubleshoot it really? Have access to an oscilloscope and multimeter.

[Deleted]

It's difficult to be sure in the picture BUT the soldering on "TIP41" transistor on the right channel looks like it might be worth investigating for a short.
Otherwise you need to wait for someone familiar with the build.

[Deleted]

measure DC voltage on:

Emitter Q3 to ground (should be around +16V)
Emitter Q12 to ground (should be around -16V)
Emitter Q22 to ground (should be around +16V)
Emitter Q28 to ground (should be around -16V)

voltage over DZ2 (should be around 3.6V)
voltage over DZ4 (should be around 3.6V)

voltage on cathode DZ2 (measured to ground, should be around 1.8V)
voltage on cathode DZ4 (measured to ground, should be around 1.8V)