A nod to Pano and his original thread, focused on loudspeaker watts (Or volts).
In this instance I'd like to seek opinions for headphone systems instead.
How many Volts do you really need?
Hacking open some cheapo in ear phones and testing the limits of my simple headphone amplifier I have been shocked at how little voltage I actually need.
Covering the full gamete of headphones in a head amp isn't easy (kudos to those who have done so already)
Thus I suggest the bounds of tests should be based on impedance, minimum 8 Ohm maximum....Say 100 Ohm?
I.e..no really high Z phones, no esl phones or the like.
Starting with el cheapo ear buds at 32 ohm I used a maximum of 3.3 Volts at earsplitting dBs. (Louder than comfortable)
In this instance I'd like to seek opinions for headphone systems instead.
How many Volts do you really need?
Hacking open some cheapo in ear phones and testing the limits of my simple headphone amplifier I have been shocked at how little voltage I actually need.
Covering the full gamete of headphones in a head amp isn't easy (kudos to those who have done so already)
Thus I suggest the bounds of tests should be based on impedance, minimum 8 Ohm maximum....Say 100 Ohm?
I.e..no really high Z phones, no esl phones or the like.
Starting with el cheapo ear buds at 32 ohm I used a maximum of 3.3 Volts at earsplitting dBs. (Louder than comfortable)
Last edited:
My headphones are 32 Ohm, sensitivity +95dB at 1 mW.
When I design headphones amp
I set the gain so 1 Vrms input gives +26dB Output. That is 20mW out at 32 Ohm.
Voltage at output then is: 1.13 Vpeak = 0.80 Vrms.
+95db +26dB = 121 dB for potentiometer max at 1 Vrms in.
+121dB is nothing I would recommend listening to.
You probably quickly will damage your hearing.
When I design headphones amp
I set the gain so 1 Vrms input gives +26dB Output. That is 20mW out at 32 Ohm.
Voltage at output then is: 1.13 Vpeak = 0.80 Vrms.
+95db +26dB = 121 dB for potentiometer max at 1 Vrms in.
+121dB is nothing I would recommend listening to.
You probably quickly will damage your hearing.
Last edited:
It really depends on the phones I use at any given time. My go to OB-1’s are rated 96dB/1mW 55ohms but measure 99dB/1mW (how often does a manufacture understate sensitivity?). Too bad they are not made anymore - at least I have two pairs. My HE400i’s are rated 93dB at 50ohms and measure closer to 90dB. My gen1 KZ 4-way bal armature hybrid ZS5 IEMs are rated 114dB/1mW but a staggeringly low 8ohms. My DT880-250s are rate 96dB by measure 93dB. So it all depends, but typically 0.35vrms (1vpp) is a very loud level for me on most phones except the HE400i and DT880’s which are more like 1vrms (2.8vpp) to sound loud.
So it’s all dependent on phones. But 3vpp is max needed on most phones.
So it’s all dependent on phones. But 3vpp is max needed on most phones.
Thanks for the comments so far 🙂
Hi.
Line-up thanks for sharing your experiences. I consider 26dB gain maybe a little high?
I settled on 10V/V then lowered to 7.5V/V which seems to be spot on for an input maximum level of 0.707 Vrms. I believe I can lower gain even further.
All of this is notepad and empirical lab work and the output devices are parts bin budget items (bd139)
I can not listen to the output through phones about about 1.5 Vrms, but i test the head amp with a load up to more than 3 volts before clipping occurs.
XRK,
That really is a very low voltage requirement (this project inspired by many others, O2, cmoy etc and the acquisition of some sennheiser hd650 which seem to eat the amperes)
My output stage is biased quite strongly, depending on bias setting anywhere between 150 and 300mA measured on one rail. So in a hand wavy way I guess I'm into class A region of the BD139 🙄
Just tested again with 7 Ohm load and I can achieve 1Vrms from 125mVrms input so perhaps my £0.50 headphone amp isn't so bad.
Hi.
Line-up thanks for sharing your experiences. I consider 26dB gain maybe a little high?
I settled on 10V/V then lowered to 7.5V/V which seems to be spot on for an input maximum level of 0.707 Vrms. I believe I can lower gain even further.
All of this is notepad and empirical lab work and the output devices are parts bin budget items (bd139)
I can not listen to the output through phones about about 1.5 Vrms, but i test the head amp with a load up to more than 3 volts before clipping occurs.
XRK,
That really is a very low voltage requirement (this project inspired by many others, O2, cmoy etc and the acquisition of some sennheiser hd650 which seem to eat the amperes)
My output stage is biased quite strongly, depending on bias setting anywhere between 150 and 300mA measured on one rail. So in a hand wavy way I guess I'm into class A region of the BD139 🙄
Just tested again with 7 Ohm load and I can achieve 1Vrms from 125mVrms input so perhaps my £0.50 headphone amp isn't so bad.
Interestingly....
My HD650 are rated at 103dB for 1Vrms input and present a load of 300 Ohm...
Woe is me...
Don't I wish I had googled that info before now....
My HD650 are rated at 103dB for 1Vrms input and present a load of 300 Ohm...
Woe is me...
Don't I wish I had googled that info before now....
Headphone SPL charts for 90dB and 110dB, by model
This might help -
Out on my project's Google Drive link I have a folder full of headphone SPL (Sound Pressure Level, of course) spreadsheets for different headphone models. For each headphone there is a sheet to hit 90dB SPL, roughly normal maximum everyday listening level, and 110dB SPL. 110 is loud as a freight train, a momentary peak number.
Headphone SPL - Google Drive
So for a given SPL target, the voltage and current (both shown in RMS in the sheets) are all over the map, depending on the impedance and sensitivity of a specific headphone. For each there is an editable excel spreadsheet and a PDF. Same thing, just one is editable if you want to mess with the numbers.
Also - word of caution - some manufacturers list sensitivity in dB/V, others in dB/mW. And a very few don't say which, you have to ask. Makes a big difference. On the spreadsheet there is a place for both. Just make sure you delete any entry in the other.
Here is an example, for the HD 650. For 90dB SPL the sheet shows a voltage requirement from the amplifier of 0.22V(rms) and a current requirement per channel of 0.75mA(rms). Same thing to hit 110dB SPL is 2.2V(rms) and 7.46mA(rms). Notice that with this choice of SPL numbers the 110 is just 10x the 90, so you could really do it in your head. 🙂
In general, most of the cans made in the last 5 years or so are all low impedance to match up with cell phones and ipods, and very high sensitivity (easy to drive, not much voltage) to reduce drive requirements. Some of the very old headphones with both low impedance and low sensitivity are very hard to drive. Lol, head speakers, essentially 🙂
This might help -
Out on my project's Google Drive link I have a folder full of headphone SPL (Sound Pressure Level, of course) spreadsheets for different headphone models. For each headphone there is a sheet to hit 90dB SPL, roughly normal maximum everyday listening level, and 110dB SPL. 110 is loud as a freight train, a momentary peak number.
Headphone SPL - Google Drive
So for a given SPL target, the voltage and current (both shown in RMS in the sheets) are all over the map, depending on the impedance and sensitivity of a specific headphone. For each there is an editable excel spreadsheet and a PDF. Same thing, just one is editable if you want to mess with the numbers.
Also - word of caution - some manufacturers list sensitivity in dB/V, others in dB/mW. And a very few don't say which, you have to ask. Makes a big difference. On the spreadsheet there is a place for both. Just make sure you delete any entry in the other.
Here is an example, for the HD 650. For 90dB SPL the sheet shows a voltage requirement from the amplifier of 0.22V(rms) and a current requirement per channel of 0.75mA(rms). Same thing to hit 110dB SPL is 2.2V(rms) and 7.46mA(rms). Notice that with this choice of SPL numbers the 110 is just 10x the 90, so you could really do it in your head. 🙂
In general, most of the cans made in the last 5 years or so are all low impedance to match up with cell phones and ipods, and very high sensitivity (easy to drive, not much voltage) to reduce drive requirements. Some of the very old headphones with both low impedance and low sensitivity are very hard to drive. Lol, head speakers, essentially 🙂
Last edited:
But OTOH, I am designing an amp that can drive 2.5wrms into a 32ohm load, and this should be sufficient for driving the flagship HE6 (84dB and 50ohm). Not that I have an HE6, but I want the amp to drive any load. So I am using 8x OPA1688 in parallel as buffers (ea capable of 75mA or 600mA total with 28vpp) with +/-18v rails.
Cheap as Chips OPA1688 Low-THD Muscle Amp
Last edited:
XRK
I've seen your recent thread, also wrenchone's thread and in all honesty both threads inspired me to start again with building a headphone amp.
I have LT1210, LME49600, BUF34 for IC options and have almost built them all....
But something stopped me.
My old lunch hour veroboard amp, which while pretty near devoid of technical flourishes, managed to produce a watt or two with perhaps 10 components, before audible and scope visible clipping. (No THD measurement capability here)
One op amp buffered with output transistors. As much NFB as I can get.
This time, instead of LF353 and TIP41/42 I have TL071 and BD139/140. I just make enough current gain to get to 1Vrms on 7 Ohm load. I figure I'm running into current limitations of the op amp and low current gain device with low Z.
I've seen your recent thread, also wrenchone's thread and in all honesty both threads inspired me to start again with building a headphone amp.
I have LT1210, LME49600, BUF34 for IC options and have almost built them all....
But something stopped me.
My old lunch hour veroboard amp, which while pretty near devoid of technical flourishes, managed to produce a watt or two with perhaps 10 components, before audible and scope visible clipping. (No THD measurement capability here)
One op amp buffered with output transistors. As much NFB as I can get.
This time, instead of LF353 and TIP41/42 I have TL071 and BD139/140. I just make enough current gain to get to 1Vrms on 7 Ohm load. I figure I'm running into current limitations of the op amp and low current gain device with low Z.
Agdr,
Thanks for the excel sheet and general advice here.
I was going to write more but for the moment toddler feeding comes first and I also pretty much ran out of useful words.
Thank you again.
Thanks for the excel sheet and general advice here.
I was going to write more but for the moment toddler feeding comes first and I also pretty much ran out of useful words.
Thank you again.
This is something I actually needed to know and measure when designing my headphone amp because it runs on just 6.8v dc.
Using Sony MDR-V7's a maximum output on music of just 0.8v peak to peak (and so less than 0.3 vrms) was more than sufficient.
Using Sony MDR-V7's a maximum output on music of just 0.8v peak to peak (and so less than 0.3 vrms) was more than sufficient.
Yes, my Pocket Class A with a BF862 and single MOSFET like IRF610 output still satisfies most needs for most headphones.
Alternatively, a BF862 front end and a KSC1845/2SA1837 CFP output is also very good.
Alternatively, a BF862 front end and a KSC1845/2SA1837 CFP output is also very good.
Thanks Mooly for your insight.
I like it loud but I like retaining my hearing too!
I think that 2 Vrms and enough current to drive into 8 Ohms should be plenty.
I have considered adding another TL072, as paralleled buffers to drive the output stage.
But then I could also just change the putout to a compound follower pair
I have some bzx53 and compliment coming in the post at some point so I'll try them.
I really need to look into modern audio BJT, as well and FET. These rare ones may be great but I'm still amazed I didn't smoke a single device in tweaking this little amp!
XRK,
I have some to220 case 2SJ...something...laying around somewhere.
Come to think I also have a few LME49810. I know these are meant to drive an output stage but....
I'm sure they make enough current for 2 Volts into 8 Ohms, and a boatload of voltage headroom 😀
I like it loud but I like retaining my hearing too!
I think that 2 Vrms and enough current to drive into 8 Ohms should be plenty.
I have considered adding another TL072, as paralleled buffers to drive the output stage.
But then I could also just change the putout to a compound follower pair
I have some bzx53 and compliment coming in the post at some point so I'll try them.
I really need to look into modern audio BJT, as well and FET. These rare ones may be great but I'm still amazed I didn't smoke a single device in tweaking this little amp!
XRK,
I have some to220 case 2SJ...something...laying around somewhere.
Come to think I also have a few LME49810. I know these are meant to drive an output stage but....
I'm sure they make enough current for 2 Volts into 8 Ohms, and a boatload of voltage headroom 😀
Last edited:
I just added the MDR-V7's to the headphone SPL spreadsheets at the link in my post #6 above. 🙂 Hard to track down the specs on those now, but found them here
Some MDR-V7 info | Headphone Reviews and Discussion - Head-Fi.org
63 ohms and 106dB/mW. Interestingly, those are exactly the same specs as the Sony MDR7506 already in the SPL charts.
I played with the SPL number on the spreadsheet and 0.3V(rms) works out to be around 107dB SPL.
Forgot to mention... scroll all the way down on the SPL spreadsheet and the author, Rob Robinette, includes a health/hearing chart related to SPL loudness. Exposure times vs. SPL levels.
Last edited:
Many years ago, for an insane situation (quality monitoring recordings in the same room with large bands), I asked myself to design for 120dB SPL "in any headphone".
120dB SPL is Very Frikkin LOUD!! Deafening. It would only be used for seconds at a time, a couple minutes in my lifetime. However when I needed to check a loud passage for clipping, I needed to know the clip was NOT in the headphone amp but the signal to tape/digits.
As xrk971 says, "depends on the phones". Sensitivity varies a LOT. Impedance has ranged all over the place, making V/I a wide target.
Using RANE's older paper, the attached is what I came up with. I decided that I would live without the up-right outliers, they were not useful for location recording (too costly, too fragile, or too darn dull).
For 120dB SPL, I get 7V or 250mA RMS to cover most-all headphones.
For continuous listening (not spot-monitoring), 110dB SPL is still a LOT. This needs 2.2V 80mA RMS, or 3.1V 110mA Peak.
6.2V peak-peak plus a few Volts for amplifier losses means you can design an AMPLE headphone driver with 8 or 9 (or 7) Volt supply.
The 110mA peak need can be covered with TO-92 BJTs or a few opamps.
Several-WATT amps for (all but a few) headphones is, IMHO, like using a 2KW PA amp for apartment loudspeakers. Or the 5-ton truck to bring the beer home. If it makes you happy, OK; but it is not necessary.
120dB SPL is Very Frikkin LOUD!! Deafening. It would only be used for seconds at a time, a couple minutes in my lifetime. However when I needed to check a loud passage for clipping, I needed to know the clip was NOT in the headphone amp but the signal to tape/digits.
As xrk971 says, "depends on the phones". Sensitivity varies a LOT. Impedance has ranged all over the place, making V/I a wide target.
Using RANE's older paper, the attached is what I came up with. I decided that I would live without the up-right outliers, they were not useful for location recording (too costly, too fragile, or too darn dull).
For 120dB SPL, I get 7V or 250mA RMS to cover most-all headphones.
For continuous listening (not spot-monitoring), 110dB SPL is still a LOT. This needs 2.2V 80mA RMS, or 3.1V 110mA Peak.
6.2V peak-peak plus a few Volts for amplifier losses means you can design an AMPLE headphone driver with 8 or 9 (or 7) Volt supply.
The 110mA peak need can be covered with TO-92 BJTs or a few opamps.
Several-WATT amps for (all but a few) headphones is, IMHO, like using a 2KW PA amp for apartment loudspeakers. Or the 5-ton truck to bring the beer home. If it makes you happy, OK; but it is not necessary.
Attachments
Thanks. The numbers sound right because the 0.3vrms I mentioned was the approximate figure derived from the peak levels observed. So actual average listening levels would be considerably lower. It all fits 🙂
- Status
- Not open for further replies.