It was never really designed to... More like at low levels, and high efficiencies 32R will work, but it's really designed around my 470R cans. This WCF give the circuit more raison d'être.
Could you sim the original 6N3 based SE circuit with a 470R load? I want to why it satisfies me
I know HP isn't R though. Thevelin proves it, right?
No problem. This has been fun.
I thought Grado cans were very low impedance, like 32 ohms?
I think a WCF is a nice way to make a tube OTL headphone amp for Sennheiser HD580, 600 or 650, Beyerdynamic DT880, or other 125 to 600 ohm headphones.
The only problem I see with the WCF is that it works best into a narrow range of load impedance. I could see designing one specifically for a 300 ohm load, to work with the nice Sennheiser open-back cans. Yeah, that's what I should do with those boards I got from you.
I might have to cut some traces, though...
I thought Grado cans were very low impedance, like 32 ohms?
I think a WCF is a nice way to make a tube OTL headphone amp for Sennheiser HD580, 600 or 650, Beyerdynamic DT880, or other 125 to 600 ohm headphones.
The only problem I see with the WCF is that it works best into a narrow range of load impedance. I could see designing one specifically for a 300 ohm load, to work with the nice Sennheiser open-back cans. Yeah, that's what I should do with those boards I got from you.
I might have to cut some traces, though...
I am following this interesting thread for a while now and am quite puzzled about the amount of power one is putting into the various headphone impedances:
32mW into 32 Ohm is 115 dB SPL which is indicated as "very loud".
13.3mW into 300 Ohm is also indicated as "very loud".
Is it really necessary to design a headphone amplifier which must be capable of producing this output power?
Last edited:
Oh yes, they can handle 1W but the point is : 1W into your 300(?) Ohm headphone ..... will it not blow out your ear-drums?
I have to disagree. A headphone amplifier is a power amp, but you have to think about what the load is the power amp will be driving.
A headphone amp is a low power amp, especially for low impedance headphones (16 to 50 ohms). Figure a max power output of 100mW is going to be enough for almost all headphones. But what does that mean?
If a headphone presents a 32 ohm impedance and sensitivity of 94dB/1mW, then to get to 115dB SPL the headphone amp will need to deliver 128mW into that 32 ohm load.
If we accept 100mW as the maximum power output we actually need, the driving amplifier has to be able to swing 1.8V RMS into the 32 ohm load and sink 56mA into that load as well.
The 6N6P WCF can easily deliver the necessary 1.8V RMS output, but it can only, possibly, maybe just barely sink the necessary 56mA.
________________
If the headphones present a 600 ohm load but are not very sensitive then the problem becomes primarily one of swinging enough *volts* across the load.
If these 600 ohm headphones have a sensitivity of 96dB for 1mW RMS input (like Beyerdynamic DT880), then we'll need 64mW to reach 114dB SPL or half the power required by the 32 ohm headphones. In terms of voltage and current, that is 6V RMS from the driving amplifier, with 10.7mA current sunk from the amp. That's more than 3X the voltage needed, but only about 15% of the current when compared to an amp driving 32 ohms headphones to the same SPL.
So as you can see, the higher impedance load requires less power, but a higher voltage swing to be driven.
Tubes can swing lots of volts, but driving current into a low impedance load is definitely not what tubes are good at.
That's why a 6N6P WCF is going to work a lot better into a 300, 470 or 600 ohm headphone than into a 32 headphone.
What you need for those 32 ohm headphones is a little, low-power power amp. It's gotta drive a fair amount of current into that low impedance load.
--
Sensitivity is 114dB? Is that for 1mW input, or for 1V input?
Seriously, sensitivity of 114dB for 1mW input means you don't need anything more than 1mW power to drive them. Is that correct?
I'll bet that 'sensitivity' spec is 114dB maximum SPL before exploding.
Let's say that 32 ohm headphone has a quite sensitive spec of 103dB for 1mW.
You'll only need 16mw to get to 115dB SPL, at which point either you're about to blow up the headphones or you're about to go deaf.
16mW into a 32 ohm load equals about 0.72V RMS, which means about 22.5mA current sunk.
The 6N6P WCF can easily supply the 0.72V RMS (no sweat) but that 22.5mA of current is going to make it work pretty hard. But it should be OK. The price for stressing the 6N6P WCF's current capabilities will be higher THD. That's all.
And if all that is correct, then 700mV into those headphones should be LOUD AS HECK !!!
--
Last edited:
Yes, exactly. A 3.3 or 5 volt battery supply can deliver lots of current, but can't swing much in the way of volts. A low target impedance works best in that situation, which is why earbuds often have a super-low 16 ohm impedance, and most cans for portable use have 32 ohm impedance.
An OTL tube headphone amp will have trouble driving such low impedance loads, but it can still work out if the low impedance headphones also happen to have very high sensitivity (lots of dBs of SPL from very little power input). It sounds like your particular HD 407 cans are so sensitive they'll work well enough with just about anything driving them.
An OTL tube headphone amp will have trouble driving such low impedance loads, but it can still work out if the low impedance headphones also happen to have very high sensitivity (lots of dBs of SPL from very little power input). It sounds like your particular HD 407 cans are so sensitive they'll work well enough with just about anything driving them.
What is a Broskie Hybrid Cathode Follower? Like this?
That won't provide any more current than the standing current of the cathode follower. So if you bias that ECC99 for 20mA or plate current, then the circuit will only be able to sink 20mA current into the load. That defines the power output of your CF headphone amp. I'd say you want at least 30mA to successfully drive 32 ohm cans.
______________
The attraction of the WCF is that, once set up as a true push-pull amplifier, it can sink twice its standing current into the load. I.e., a WCF biased at 20mA plate current can sink up to 40mA into the load.
If a WCF has a gain stage in front of it, NFB can be used to lower the Zout to the point where it might be able to drive 32 ohms well enough.
A WCF has a plate resistor 'Rp' (actually a current sense resistor as the top tube's plate load) whose resistance value can be adjusted to make the WCF work best into a limited range of load impedances.
For instance, a higher value of Rp will get the WCF working best into a higher impedance load (like 300 ohms) while a lower value of Rp will get the WCF working better into a low impedance load (like 32 ohms) at the expense of worse performance into the higher-impedance load.
What if a WCF pcb had a space for a parallel pair of Rp, so you could switch between a lower and higher value there?
--
That won't provide any more current than the standing current of the cathode follower. So if you bias that ECC99 for 20mA or plate current, then the circuit will only be able to sink 20mA current into the load. That defines the power output of your CF headphone amp. I'd say you want at least 30mA to successfully drive 32 ohm cans.
______________
The attraction of the WCF is that, once set up as a true push-pull amplifier, it can sink twice its standing current into the load. I.e., a WCF biased at 20mA plate current can sink up to 40mA into the load.
If a WCF has a gain stage in front of it, NFB can be used to lower the Zout to the point where it might be able to drive 32 ohms well enough.
A WCF has a plate resistor 'Rp' (actually a current sense resistor as the top tube's plate load) whose resistance value can be adjusted to make the WCF work best into a limited range of load impedances.
For instance, a higher value of Rp will get the WCF working best into a higher impedance load (like 300 ohms) while a lower value of Rp will get the WCF working better into a low impedance load (like 32 ohms) at the expense of worse performance into the higher-impedance load.
What if a WCF pcb had a space for a parallel pair of Rp, so you could switch between a lower and higher value there?
--