Headphone Guitar Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

PRR

Member
Joined 2003
Paid Member
> if the output is limited to the maximum safe SPL for 32ohm headphones, it will not be capable of driving 400ohm headphones to a normal level.

Headphone efficiency can't fall outside a broad range. And Thévenin predicts a broad optimum. When designing a an amp which could get LOUD in "any" headphone, I came to 7Vrms behind 29 Ohms. If you are not being paid well to give-up some hearing, this should probably be more like 1.4V behind 29 Ohms (1V, 1,5V, 22r, 36r, not critical)
 
I've now had a chance to add an output limiting circuit. I've used a pair of 1N4004 silicon diodes across the output in reverse parallel with a 10ohm series resistor between the amplifier and the output.

In that basic configuration, the clipping was too extreme and occurred too early. I added a 10ohm resistor in series with the diodes to increase the threshold slightly. With that in place, the clipping is now at the right level and starts to occur when it gets too loud.

I have tested the amplifier with guitar and with a signal generator and while it is now capable of being turned up too loud, it is now 'long-term exposure loud' rather than 'instant deafness loud'. Obviously I have not been able to measure the actual SPL but subjectively, I wouldn't be worried if I heard that volume for a short time.

I think this is now an acceptable situation. The output does not clip so much that it destroys the clean headroom of the amplifier. I don't see that there is a problem now with the volume. Yes, it is capable of damaging hearing long-term if turned right up, but that's fine, I just won't turn it up too loud. In the event of a fault, though, it won't produce any instant hearing damage levels.

I can get back to the reverb problem now.

James
 
I just had a minor epiphany.

If you feed 32 ohm and 400 ohm headphones from the same voltage source (meaning zero source impedance), the 32 ohm phones receive much more power.

But if you feed each of them with the same current source (meaning infinite source impedance), the 32 ohm phones receive much less power.

That suggests that there must be a crossover point somewhere in between. In other words, there must be a specific value of source impedance, somewhere between zero and infinity, which will deliver exactly the same amount of power to either 400 ohm or 32 ohm 'phones.

So I wrote down equations for the power delivered to a load, stuck in 32 ohms and 400 ohms for the load, and solved the resulting quadratic equation with a little algebra, to find the magic source impedance that would deliver the same power to both types of headphones. The answer is 113.137085 ohms, or rounding to a sane "close enough" engineering value, about 110 ohms.

Now we have a super-simple solution to the "How do ensure safe SPL levels if the user might use 400 ohm or 32 ohm 'phones?" conundrum. Simply put 110 ohms in series with the 'phones, and you're done. :)

There is a downside: bass response of headphones varies with source impedance, and there is no guarantee that your phones will have their flattest frequency response when fed from a 110 ohm source impedance.

But at least switching headphone impedances won't blow out your ears!

To do a sanity-check on my algebra, I also used Gnuplot to plot the power delivered to the headphones from a 1-volt source with a 113.13708 ohm source impedance. The result is attached; the x-axis is headphone impedance in ohms, the y-axis is power delivered in milliwatts.

Note that x=32 and x=400 both receive the same 1.5 mW of power.

Note also that any headphone between 32 ohms and 400 ohms will be safe to use; power only goes up from 1.5 mW to 2.2 mW at the loudest headphone impedance (which, of course, is also 113.13708 ohms, as per the maximum power transfer equations we've probably all seen.) This is only a 0.7 dB increase, which is negligible as far as ear safety goes.


-Gnobuddy
 

Attachments

  • Pout_vs_Rphones.png
    Pout_vs_Rphones.png
    28.9 KB · Views: 88
Last edited:

PRR

Member
Joined 2003
Paid Member
You found an electrical optimum.

Headphones also vary in sensitivity, and for reasons mostly historical, the low-Z phones tend to be less sensitive and more rugged than the hi-Z phones. And MANY outliers.

I've written this, with much data and graphs, at least twice this week in the Headphones section.
 

Attachments

  • I-V-Z-P-29r.gif
    I-V-Z-P-29r.gif
    14.5 KB · Views: 81
I've now reassembled the amplifier with the new pot values and the output limiter soldered across the output jack. I think I've found a way to orientate the reverb tank that minimises hum. I just need to modify the tank casing slightly so that it will fit in the chassis.
 
I have just done a quick scan of your schematic. You can probably reduce the hiss quite a bit by reducing the 1MEG input resistors on the TL072 amplifier stages and using larger coupling capacitors. I expect that you are hearing a fair amount of Johnson noise.
Re the volume control: the wiper is actually seeing a fairly low input resistance on the gyrator stage. This will affect the overall transfer function of the potentiometer in the way that you describe. You should either raise the input resistance or decrease the pot value.

Sorry I'm rather late to the party here, but the input 1M resistors contribute no voltage noise, they are shunts to ground, not series resistors, so they contribute current noise, 0.13pA/√Hz to be precise. Reflected into a guitar pickup that's basically nothing.

The high value of the resistors in the volume/eq section is the problem. 1Meg pots are for valve circuits, reduce impedances by a factor of 100 in that section so the pots are 10k, then the noise will probably drop by 10--15dB or so.

A 1Meg pot has upto 250k effective source resistance, which is 65nV/√Hz of Johnson noise, this will dwarf anything else in the circuit by miles I reckon.
 
A 1Meg pot has upto 250k effective source resistance, which is 65nV/√Hz of Johnson noise
Very true. At the same time, most guitars have an internal volume pot that is typically between 500k and 250k, so that the source impedance of the guitar itself can go as high as 125k when the guitarist turns down the volume knob! Almost invariably, this turns out to be the single largest source of thermal noise in the entire electric guitar audio chain.

James UK's headphone amp puts about 14 dB of voltage gain between the guitar and the high-Z passive tone stack, which will help to reduce hiss from the tone stack a fair bit - it should be below the hiss from the guitar's own internal volume pot at many guitar volume pot settings.

An audio engineer's solution to the conundrum would be to always keep the electric guitar's volume knob set to maximum, minimizing the guitar's output impedance. But this drastically undermines the capabilities of the instrument, as is vividly demonstrated by Joe Bonamassa in the first two minutes of this video: YouTube

The electric guitar has been around in more or less its current form since 1932, when George Beauchamp's famous Ro-Pat-In Hawaiian "Frying Pan" and less famous Spanish Electric guitars first became available. Astonishingly, that was 88 years ago, in a era when thermal noise from audio electronics wasn't even a consideration. The miracle was that it made sounds at all, never mind any traces of hiss!

So I guess it's not too surprising that when we look at this old system today, we find deep compromises built into the fundamental nature of the electric guitar pickup and amplification system. This is old, primitive audio engineering, but it happens to still be very good at making all the sorts of sounds we've come to expect and love from electric guitars. In fact, most of the development effort in guitar amplification for the past few decades has been put into the struggle to make digital circuits behave as badly as early tube amplifiers did!


-Gnobuddy
 
Noise in input stage

That amount of current noise will give ~ 650uV across the pickup, with typical pots and the full audio range. The pickup itself is typically several H of inductance, a rather high impedance at audio frequencies. Agreed though that the tone controls are the big gorilla.
 
Noise in input stage

I have played guitar for many years and used both solid-state and tube amps. Tube amps, even the best, have a fair amount of hum and hiss. Most solid-state amps have a high level of hiss too, but when played with other instruments on stage, it is inaudible.
I have used my guitars with inputs on traditional pa amps or even direct to recorders with no problems, and their input resistance is well below 1 MEG. If the high input resistance is desirable, the input stage could be bootstrapped, but I am not sure just how much overall difference that would make.
 
That amount of current noise will give ~ 650uV across the pickup
There are a couple of extra decimal places there - that number is more than two hundred times too big...

Say there's a 500k pot in the guitar, set to worst-case 125k source impedance. That appears in parallel with the 1M input resistor of the amplifier, for a combined impedance of 111.11k.

At 20 degrees C, with typical 5kHz guitar bandwidth, that results in a total input thermal noise voltage of about 3 microvolts.

That compares with about 1.3uV of input noise due to the TL072 itself, in the same 5 kHz bandwidth. Depending on where it's set, the guitar's internal volume pot can be noisier than a TL072. Which is one reason why the venerable TL072 remains a perfectly reasonable choice for this application, decades after it first became available.


-Gnobuddy
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.