Paia Headphone Amp PSU question

[lars-musik]

Hi,

slightly off-topic because you cannot stomp it, but I'll ask my layman's question anyway:
Could somebody explain the power supply of that circuit: Paia Headphone Distribution Amp to me?



It looks to me like there's somehow a positive and a negative rail coming from a 12V DC source. But obviously I am missing something. What's the difference between the V- and the Ground pad?

[stonerbox]

I'm nothing but a noble noob but to me it looks a lot like some sort of filtering going on and really not much else. Check out the BeavisAudio 'Huminator' there's a lot of documentation on it both here and on other forums.

http://www.next.gr/circuits/project-Huminator-l34756.html

Here's also a video on power filtering by the user DIY Guitar Pedals.
https://www.youtube.com/watch?v=julbbj4R3Ko

[Kipper4]

Theres nothing special about this prvwer supply except its 12v instead of 9v
basicaly the designer has made a 1/2voltage V Bias rail that is called ground.

Top rail =12v
Ground (V Bias) = 6v
Bottom rail =0v

Sure theres an extra cap from the +12 to Gnd.
and led power on indicator

To make it even quiter I'd put a series 50-100 ohms resistor between the switch1 and the 100uf cap
Then I'd put a 100pf-4.7nf cap in parallel with the 100uf cap this should help lessen any RF inteferance.

[lars-musik]

Thank you both very much. Now that you explained it I see it, too. I didn't realise that the term "ground" is negotiable.

Could you kindly look over the schematic for one channel (I think I am going to do a single channel modular layout with onboard jacks and pots, so that I can add or subtract channels as needed).

The volume pot not going to ground but to VB looks unfamiliar to me. Is this right?

[Kipper4]

Yer looks ok.
V Bias is sometimes referred to as signal ground, voltage reference, Vb
If your doing a single channel mono headphone amp why not use the other half of the op amp as a V Bias buffer.
So you build your power supply as is and then instead of going directly to the + pin on the op amp as you normally would.
You feed V Bias to the unused op amps + pin and connect the inverting input pin and output together and then go from you buffered VB to the other Vb connections like the volume pot, pin 5 of the audio op amp.
This makes vb very stable and uses the unused half of the op amp. Or you could simple use a single op amp and not bother buffering vb.

[lars-musik]

It is meant as a stereo headphone distribution amp for monitoring the recoding console in our rehearsal room, so stereo it will be. Here are the little modules I plan (4 – 8 pieces - not sure yet - sooner or later I am going to record a brass section, so maybe 6 channels would be in order) to mount in an old 19" rack. One PSU (mounted by the switch) and the modules mounted by pot and jack.



Sorry I have to ask again: The input and output sleeves should be on (my) "Ground". The op amps and the volume pots should be on "VB" (Vref or whatever) right? Because in the original schematic they are all marked as "Ground" (meaning "VB" following  the before mentioned replies).



I am confused....

[Kipper4]

Yep I think you got it
Volume pots to (your) vb
Sleeve to (your) ground
Also the + non inverting op amp inputs to (your) vb

[samhay]

I would suggest you bump up the value of the 470R Vbias resitors by a significant amount (I usually use 2 x 10k) as these will suck quite a lot of current as drawn (~ 10 mA).

[Kipper4]

Good call Sam I didn't spot that one.

[lars-musik]

Great. That I will do. Then I'll quit whining and just build it. Thanks a lot!

[duck_arse]

before the smoke emits, make very sure you have your "12V dc" isolated from everything (or at least anything signal ground is connected to). because your new signal ground is at 6V, relatively speaking, if you ground the (-) of the supply ....... and you'll then have 6V at the opamp outputs, so further .......

[lars-musik]

before the smoke emits, make very sure you have your "12V dc" isolated from everything (or at least anything signal ground is connected to). ...

Oh my. I guess I'll draw up a complete schematic and ask for your advice on it. Right now I'm at my in-laws and not supposed to stick my nose into my computer.  Sunday  or Monday I will get to it.

[ashcat_lt]

Sorry I have to ask again: The input and output sleeves should be on (my) "Ground". The op amps and the volume pots should be on "VB" (Vref or whatever) right? Because in the original schematic they are all marked as "Ground" (meaning "VB" following  the before mentioned replies).

No!  All of the ground symbols on this schem here should be connected together, and to what you're calling VB above.  The only thing (!!!) to which the bottom of the 12V supply connects is the negative supply pins on the opamps.  Yes, the chassis and jack sleeves are at the mid-point of the supply, and you have to make sure that that no part of that power supply is common to any other device connected to this or you'll get a some "undesired current flow" like duck_arse said. 

To do as you've proposed might actually be easier and more foolproof, but you'll need coupling caps in series with the jacks - before the pot on the input, and either before or after the resistors at the output.

[PRR]

> suggest you bump up the value of the 470R Vbias resitors by a significant amount (I usually use 2 x 10k) as these will suck quite a lot of current as drawn (~ 10 mA).

This is a "power" amplifier. Real current is sucked from the output (to shake your eardrums). That current must return somewhere half-way between the two power rails. Since there is no "hard ground", the resistor divider *must* be pretty substantial.

In fact 10mA hardly seems like enough for multiple channels of headphones.

And so-what if it eats 10mA or 100mA? It is surely powered by a wall-wart, surely rated for more current than that. (To get Really LOUD levels in *one* headphone you want 100mA available, though this one only gives you 40mA.)

Do it *just* like PAiA did it. John was no fool. His stuff may be unconventional but it DOES work.

[samhay]

Paul - fair point, but to be fair it's a real amp that uses a 5532 op-amp(s).

[lars-musik]

Maybe a stupid question.  But is there a technical reason why the circuit shouldn't be powered by a  'real' symmetrical +/- 12V PSU?

[samhay]

No reason I can see not to use use a bipolar supply if you have one. The 5532 op-amps are good to +/- 22 V (safer / conventional to stay within +/- 18V).

[PRR]

> is there a technical reason why the circuit shouldn't be powered by a  'real'

The technical reason is that John Simonton never used a $2 solution if he could find a 50-cent solution.

Moreso in this case. John was selling kits, cheap. +/-12V supplies are fairly rare and generally full-price. 12V supplies are use in bulk for answering machines, burglar alarms, etc, and often found cheap on the surplus market. 

______________________________
WARNING- you hit my hot-button.

> symmetrical +/- 12V PSU?

The *total* supply, as shown, is 12V. Not 24V.

You CAN put +/-22V into this circuit. These specific opamps will take it.

You probably do not want anywhere near 24V peak-peak (much less 40V p-p) in *headphones*.

Especially modern (post-iPod) phones, which are all low-Z to suck power from a 3V supplied amplifier.

I *am* arguing with myself. Some years back I needed to monitor recording quality in very loud environments, and with a collection of various older (pre-iPod) headphones. I also needed to ensure the monitoring system did not distort before the recording system. I calculated that for VERY LOUD level in "all" headphones you "could" want 7V or 0.1A rms (10V 0.14A peak). I built several amps to this spec and used them extensively. AFAICT, I *never* got near those power levels, even working around large brass ensembles and percussion orgies.

Since then the iPod has dominated headphone design. Phones are mostly low-Z and good efficiency so that they can get LOUD from a 3V power amplifier. Of course this "loud" is a compromise with battery life and lawsuits for deafness. Serious ear-abuse may need more than 3V supply. 6V is "only" 6dB up. 12V supply is 12dB up.

What is 12dB? Say that the iPod and appropriate headphones is a Honda Civic. Good performance. 100 horsepower? 12dB up from that is 1,600 horsepower!! 4X the power of any street car (exclude street-legal race cars like a Lamborghini WTF). More power than many WWII fighter planes. More power than Garlits had when he went 200MPH in the 1/4 mile.

Engines are expensive, so we only buy as much as we need. Headphone amps are teeny so a little over-kill is quite affordable and fun.

But for 16-100 Ohm headphones, it is hard to see how you could *ever* want more than 12V total supply.

(300 Ohm phones, and a few high-buck very-low-efficiency headphones, are exceptions.)

Supplying more voltage than you can possibly use in the phones is just more waste-heat in the amplifier. If you run +/-12V supply and only put 3V peaks in the phones, the chips have to burn-off 3 times more power than is delivered into the phones. Like the 1600HP engine in the Civic, dragging the brakes to hold speed down, except worse because you "can" reduce the air to an engine, but the chip must take the full supply voltage all the time.

If we "must" second-guess Simonton's headphone amp design........

This is an older design, and headphone fads have drifted since then.

This was not meant to be The Pinnacle of headphone amps. (IIRC the whole kit was under $50; you can't buy one audiophile cable for that money.)

The '5532 is exceptionally clean (when used as intended), pretty hefty for a dual opamp in a DIP8, and inexpensive. With higher-Z phones it works very well. For 32 Ohm phones (iPod world) its 40mA output will not make more level than an iPod will. (40mA peak in 32 Ohms is 1.28V peak, or 2.5V p-p, like a 3V battery.) A fair number of fanatics feel that something more is better for low-Z phones.

32 Ohms part-Watt is actually an awkward zone for available chips. Op-Amp convention is 2K load, and while the '5532 will pull down below 500 Ohms, 32 is outside its comfort-zone. (Hence the 100 Ohm padding resistors in the Simonton amp.) We have a wealth of 8 Ohm many-Watt chips for LoudSpeaker work, but even if the chip is cheap the over-kill is awkward. Many of these won't even wake-up below +/-9V. Once woken, the high idle current (needed for good low-level liniarity in few-Ohm loads) means heat and power problems out of line with headphone demands. Opamps can drive booster transistors, but this gets into all the idle current control and blow-up protection problems of Real Loudspeaker Amps (Cordell's 640-page book covers the highlights but isn't all you want to know, and I am finding it a long chew).

LM380 family, especially the LM386, "looks" appropriate, though marginally clean. However these chips were cheap when new, and commodity '386s tend to be processed for low idle current rather than fine sound. They need the R-C network on the output to not freak-out under some combination of load and signal.

TI has a DSL chip which is fabulous for higher voltages, less-fab for 6V or 9V work, and AFAIK is only available in the teeny-leg (SMD) packages.

There are many opamp plus single transistor plans which run class-A (hot). Promising, but my experiments show that closing the NFB loop around opamp and small power transistor tends to go unstable in the 10MHz band. Not that you can hear 10MHz, but while slamming the rails with RF the opamp can't do a proper job of controlling the audio.

The "simple best" plan may be to parallel '5532s. Do it right, with isolating resistors, or the two amps get in a fight and the headphone loses.

___________________________
The Ground/Vref hack in reply #3 this thread has DC troubles. The inputs *and* outputs are DC-referenced to +6V, but the inputs come from another box which is surely zero VDC referenced, and while leaving +6V on the phone jack shell is possible (with insulated jacks) and finger-safe, it isn't good practice (especially once you break the phone plug and replace with metal shell, with other grounded metal objects around). Either add a lot of blocking caps (large ones at the headphone end) or go back to a both-polarity supply.

A 6VAC transformer and a voltage-doubler (or 12VCT and FWB) gives +/-8V DC which is probably ample for any sane headphone needs. Better opamps and such will eat the raw (un-regulated) DC without audible buzz in the phones, if the main caps are pretty big.

[tubegeek]

I like this article a lot, it's a careful description and test of several small discrete-transistor audio amplifiers that outperform the ubiquitous LM386 amp-on-a-chip.

Homo Ludens: Small audio amplifiers

[lars-musik]

Thank you all for your help.

@Paul: This is a very thorough analysis of the power requirements of this amp in combination with ipodish speakers and ears, thanks for that. It seems it sometimes pays to hit your "hot-button". So, this time not doubting John Simonton's circuit I built one channel for testing purposes and it absolutely does what it should (at least on my workbench, let's see how it holds up in a loud recording situation in our rehearsal room).

32 Ohms part-Watt is actually an awkward zone for available chips. Op-Amp convention is 2K load, and while the '5532 will pull down below 500 Ohms, 32 is outside its comfort-zone. (Hence the 100 Ohm padding resistors in the Simonton amp.)

We are using some low-z Shure SH-440 headphones with an impedance of 44 Ohms. As I haven't cut through the math, I'll just ask: is there something to improve the circuit with such low-z headphones regarding possible distortions? Raising/lowering the before mentioned resistors (just asking – not doubting!!)

The "simple best" plan may be to parallel '5532s. Do it right, with isolating resistors, or the two amps get in a fight and the headphone loses.

How would that work? Is it possible to make that second op-amp switchable in case we ever get our hands on high-z headphones?



And one more: Some time ago I changed my main mixing desk from an old Studiomaster Pro Line Gold that slowly deteriorated to a newer RAMSA model. However, the EQs of the Studiomaster were quite nice, so I took it apart, cleaned it even repaired one dead channel by sheer luck and would like to re-instate it. However, the phones out is labeled "MIN 400 Ohm".

Could you calculate the load of the distribution amp and tell me, if there's a chance these two will communicate as intended?



As my contribution to this whole thread here are my layouts for a modular version of the PAIA Headphone amp. The channels are meant to be connected in series by 5pin headers and plugs and I guess you can plug as many together as your power supply allows (I might be wrong there, though).

The pcbs  are mounted directly by the jack and pot and the power switch, respectively:

http://i1244.photobucket.com/albums/gg578/lars-musik/Paia_Single_Channel_PLUS_PSU.jpg