• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

Aspen Headphone Amp

Hi Sheldon,

The reasoning behind the need for a good power supply is that the amp is relatively simple and the power supply ripple rejection is only around 17 dB. In other words if there is ripple on the PSU of say 1V, the ripple would manifest itself as a signal at the output of 103 mV. This, on even a mediocre set of cans would be considered serious hum.

One wants to reduce the audible PSU hum during quiet passages to an inaudible level. Although this is different for different sensitivity levels and hearing threshold, one must regard what will be the worst case.

This is easily specified and someone may just want to do this practically, firstly it must be a youngster who's hearing threshold is still very good.

Dial up 100 Hz on your signal gen and listen to it on your headphones. Now turn the level down until you do not hear the hum. Subtract 10 dB from it for a safety margin (in case there is someone that has a hearing threshold five times better than you)

Write down this value and there you have specification number 1. Everything can be related to a practical application.



Nico
 
Last edited:
Ikoflexer,

you have a good argument. Any project starts with a functional specification which defines what, and all the what questions should be measurably verifiable in the product you are setting out to design else you have not achieved the objective.

If you have no specification then what happens here on this thread is shooting from the hip, or what I call design on the fly.

The second part is defining how and the how leads to a paper design which will define the topology of the schematic, this is how I am going to do it (of course this involves a lot of maths and knowing at least a little about electronics.)

Simulators came of age and this does make things a little easier, you can virtually cut and paste, guess many component values, use other existing designs and the like and get to a result.

The result you need to measure against what you specified and this will tell you if you seceded or failed at the attempt.

Finally you will practically build what you have simulated and what I would call hack to work - in the audio world this is called tweaking but the results are again verifiable against the original functional definition.

If you shoot from the hip you normally miss everything unless you are a gunslinger (weathered audio designer) who are able to design a product off the cuff, that needs many years of experience.

Nico
 
mlloyd,

This is a problem when you get older you tend the tweak to what you cannot hear so well anymore. Funny thing is we lost the bass and treble controls over the years because they are not fashionable in high end. Half of us would actually appreciate high end with controls so that you can compensate for your hearing loss.
 
Hello

I'm 53 and few week ago I haved a hearing test at the hospital, I did know that I have earing lost at high frequencies but I was surprise to see that test graphic.

At 2k 5db, 4k 45db, 10k 50db of earing lost.

So a treble boost are interesting, but it would complicate the amp by needing a preamp at the input, maby as an other part of that project.

Bye

Gaetan
 
Hi Sheldon,

The reasoning behind the need for a good power supply is that the amp is relatively simple and the power supply ripple rejection is only around 17 dB. In other words if there is ripple on the PSU of say 1V, the ripple would manifest itself as a signal at the output of 103 mV. This, on even a mediocre set of cans would be considered serious hum.

One wants to reduce the audible PSU hum during quiet passages to an inaudible level. Although this is different for different sensitivity levels and hearing threshold, one must regard what will be the worst case.

This is easily specified and someone may just want to do this practically, firstly it must be a youngster who's hearing threshold is still very good.

Dial up 100 Hz on your signal gen and listen to it on your headphones. Now turn the level down until you do not hear the hum. Subtract 10 dB from it for a safety margin (in case there is someone that has a hearing threshold five times better than you)

Write down this value and there you have specification number 1. Everything can be related to a practical application.
Nico

Sure, I understand the general goal. So let's do have some fun with numbers.

I assume Ikoflexers sims are for the active portion of the supply only, and do not include the filter after the rectifier (otherwise the ripple at the output would be something like 45V/600, or 60mV, and it's nowhere near that). As shown for Mihai's circuit, the ripple prior to the regulator (as simmed in PSUD) is about 42mV. At 76dB line rejection that would be reduced by a factor of about 600. Let's round those figures to 50mV/500. That means an output ripple of 0.1mV. Reduce that by a factor of 7 (17dB), and the ripple at the amp output should be 14uV. Assuming I haven't misplaced a decimal or some other silly error, is that adequate for the task at hand?

Sheldon
 
Last edited:
Hi Sheldon,

The reasoning behind the need for a good power supply is that the amp is relatively simple and the power supply ripple rejection is only around 17 dB. In other words if there is ripple on the PSU of say 1V, the ripple would manifest itself as a signal at the output of 103 mV. This, on even a mediocre set of cans would be considered serious hum.

Nico

Hello

I was thinking of a regulation at the input LTP transistors, it would boost the power supply ripple rejection, but I will wait the regulator from Hugh, Bigun, KT, Nico and John.

Bye

Gaetan
 
Last edited:
Hi Keantoken,

I actually do not want to joint the design team at all. What I would like to do here is play devils advocate and ask questions and give a clue here and there. I would like a good amplifier progress to a great amplifier. I want to provide the focus and the movement from this focus.

I assume nothing is good at the start, all DIY audio are monkeys, they will progress to human purely BY what we will learn and argue about here.

It is Gareth's task to lead the project and collect all useful data while discarding the rubbish in the project "scrap book" because I believe we are still wondering what the functional requirement is. You already see a solution to a problem that most don't know they have.

So procrastinate a little then when the problems arises you have a ready solution. If the problem does not manifest itself then obviously the solution is useless no matter what improvements could be had from it you will be labeled a terrorist and not a team player, it will not fly.


We have already seen this on three power supply offerings. I bet most don't even remember what these three pages earlier contain.

What Gareth should do is on the first thread of a new page summaries what the latest spec is, and we will not loose sight of the task at hand.

Gareth there has been three very important statements made and I say it is worth listening to it and noting it down. It is how older people enjoy music

THE PENSIONERS ARE DEAF, BUT THEY HAVE BUCKS, SO IF YOU WANT THEIR BUCKS SPENT WITH YOU, THE PRODUCT HAS TO ACCOMMODATE WHAT THEY NEED.

THINK ABOUT IT - IF THIS AMP IS OFFERED TO ME NOW, I WOULD PROBABLY SAY IT SOUNDS LIKE MUSIC PLAYED THROUGH A SOILED BABY NAPPY, THE SMELL ACTUALLY enhances THE CRAP I HEAR
 
Hello

I was thinking of a regulation at the input LTP transistors, it would boost the power supply ripple rejection, but I will wait the regulator from Hugh, Bigun, KT, Nico and John.

Bye

Gaetan

A good point. But, I'd prefer not to meddle with the LTP or anything else since the overall topology of the amplifier is important for achieving the AKSA sound.

Nico - I will give some thought to how we can incorporate a smelly nappy into this project :D
 
Iko,

Hell, no. Three reasons: I cannot sit on a prostate crusher for six straight hours; the bike might get wet, as most of yesterday it was 12C and raining along the way, and I had to take my wife, who loathes the motorcycle....... Besides, tyre costs over any distance are about half the cost of gasoline, and I restrict myself to about 2000 miles a year!

Sheldon, IKO

I appreciate your technical input and obvious desire for rigor, particularly as hum on a power amp (the AKSA) is scarcely noticeable when the same hum on cans would be unlistenable. The supply I am discussing at present has 80uV of simulated 120Hz ripple at full output. I think there is a balance here, I don't want this issue being too locked up in simulations, either, as I know from experience that building the product reveals many issues not well covered in simulations. Simulations are necessary, particularly with feedback amps, but hum issues should not be too difficult to solve, and the best course is to select the right topology at the outset, a point we are yet to reach.

Nico is right about the old geysers having no top end, and Gaetan's information confirms this. I won't embarrass myself with my hearing figures, just appalling. However, the processor behind this degraded hearing does to a degree compensate, though Nico's point about tone controls is well made. What is significant is his market angle; how do you think Focal speakers have been so successful?

My experience has been that SS amp power supplies benefit more from iron than from active devices. I use a simple CMC supply on my premium amp and the results are just stunning, far better than regulation. I would think this argument is transferable to cans. A judicious combination of iron (a trafo used as a CMC) and simple regulation should do the trick very nicely. It might even be practical to use a simple LM317 straight off the rectifier cap, THEN put the power through a CMC and another quality cap. I've found a little sag in the supply to be no bad thing; precision is not required, rather, it is speed, the absence of savage charge pulses in the final supply cap, and a low order harmonic spectrum.

Paul B, thank you for your suggestion about muting. Good idea. If we are to drive 32R cans, we should keep that first resistor low, so no more than 10R should do the trick. I would be happy to meet up with you to discuss this if have have the time.

KT, thank you for your masterful circuit. However, seven transistors plus a 431! Can you do it more simply? I would like no more than four. Simplicity is mandatory, performance is essential, of course.

Again, thanks to all here, great discussion,

Hugh
 
Disabled Account
Joined 2007
Nico is right on the money about establishing the specs early and working toward achieving them.
Leaving the basic amp design alone we see that it's PSRR is not great so it makes sense (to me at least) to design a power supply that will make up for this shortfall.
Many opinions, many egos clash and at this stage, we still don't have a firm PS plan. Simple or complex, series or shunt, it needs to be selected.
Will it make everyone happy? Nope! - but (here especially) that's the rule, not the exception.
:)
 
It might even be practical to use a simple LM317 straight off the rectifier cap, THEN put the power through a CMC and another quality cap. I've found a little sag in the supply to be no bad thing; precision is not required, rather, it is speed, the absence of savage charge pulses in the final supply cap, and a low order harmonic spectrum.

Hugh

Hello Hugh

Maby a simple LM317 straight off the rectifier cap for a basic amp and those who want more could use a better reg from KT.

I think that it is normal that somes of use will want to ad their own mods to the amp and they can even open a sub-thread for all those mods, and the basic amp would still the foundation of the project and the main thread.

Thank

Bye

Gaetan
 
Last edited:
Okay, what exactly did Hugh mean by "private"?

I thought we were going to do this by Email.

I'm confused.

Regarding my circuit, I can't make it simpler without degrading the quality... We were aiming for below 100uV, so I did it.

In order to determine what PSU we need exactly, tell me how many db until hum is inaudible. -40? -5? -300db!? If you do this I'll sim it up and we'll have solid data.

- keantoken
 
Last edited:
Okay, what exactly did Hugh mean by "private"?

I thought we were going to do this by Email.

I'm confused.

Regarding my circuit, I can't make it simpler without degrading the quality... We were aiming for below 100uV, so I did it.

In order to determine what PSU we need exactly, tell me how many db until hum is inaudible. -40? -5? -300db!? If you do this I'll sim it up and we'll have solid data.

- keantoken


Hello

Your goal are very good, but there is maby beginers who read this thread and could not go into very much more than the basic schematic gived by Hugh.

So I presume that is another reason that we need to keep it as simple as possible.

A more advance version can be done by those who want.

Bye

Gaetan
 
Last edited:
Disabled Account
Joined 2007
Hi Gaetan,
I'm with you here, nothing wrong with this and will give excellent performance.
I'd do this: Rect > bypass cap > CRCR > LM317 > bypass cap > decoupling cap.
Simple and very effective.

Hello

And maby a choke or resistor in a pi filter after the rectifier.

Bye

Gaetan

I meant: Rect > bypass cap > RCRC > LM317 > bypass cap > decoupling cap.

RCRC is resistor, cap, resistor, cap

:)
 
Keep up the discussion on the board, KT, it's fine, and everyone can see what's happening.

My preference for a basic, effective power supply would be:

20Vac transformer, with minimum 250mA output.
UF4004 full wave bridge rectifier with single 3,300uF filter cap.
Single LM317 set up to 24V.
TWO 5VA 2 x 6Vac pcb mount trafos, secondaries set up as common mode chokes, one for each channel.
Final 1000uF quality electro at output of each choke, bypassed with 2.2uF film cap.

Bifurcating the output is useful since it decouples the supply to each channel. Not exactly cheap, but low on parts count, reliable and difficult to make assembly errors.

Hugh