How much DC offset would you consider harmful to HP and after which point would you throw the commercial amp out?

[jr41]

Can I ask how 10mV was decided as the safe limit? I'm following some of what is being discussed here, but I'm quite new to the concepts.
 
I understand that an increase in current results in an increase in heat and that there is a point where the conductor will break down. I've seen shots of Grado drivers, and the wiring from the cable to the magnet is extremely thin - not much beyond human hair thickness perhaps.
 
Though one thing which is confusing me about this (and this is really getting down to basics), is that I've read higher impedance headphones are less susceptible to damage from DC offset. I understand higher impedance headphones need higher voltage to drive - to overcome their higher resistance (impedance) - but since (I think) we're saying too much current through a very fine conductor is the ultimate cause of DC offset damage, I'm struggling to understand, therefore, where voltage values come into the equation? Surely it would be better to specify dangerous current levels, thus negating the fact headphones of different impedances have differing levels of susceptibility to DC offset voltages. I hope this makes some sense!
 
To take a real world example, I measured the DC offset from my iphone 4 line out recently and it was 10mV. I've got it hooked up to my AMB Mini³, which has no decoupling caps, and at listening levels the DC offset is amplified to 50mV by the mini³. I'm driving Sennheiser HD 25-1 II, which are 70ohm. (By the way, I couldn't measure the DC offset from the iphone with music playing as it confused my DMM - I've read this is a common limitation, but for a few seconds after pausing the music my DMM read 10mV.) Apologies if this last paragraph is getting a bit off topic.

 

[FallenAngel]

I think 10mV was the rule of thumb after amplification, not before.

 

[wdahm519]

I think staying under 25mV is good.  Under 10mV is better.  Keep in mind, when you measure DC offset with no headphones plugged in, its going to be a a lot higher, and it may start oscillating from 20mV to 80mV.  
 
I was measuring my Bijou's offset and it was doing that.  I plugged the headphones in and it dropped to 1.XX mV.  

 

[JoetheArachnid]

It's 00:42 here. Why did I just spend the last 40 minutes obsessively tweaking my DC offset trimpots? Because of this F&*^ing thread.
I checked on impulse, and though one channel was 0.6mV, the other had drifted to nearly 26mV. It doesn't help that it changes by +/-5mV or so depending on the temperature, whether the lid is off the case, whether the screwdriver is touching the trimpot...
Broadly: Grach. Damn you, thread.

 

[DouglasQuaid]

Anything over 20mV offset will not touch my Grados.

 

[svyr]

I'm not really happy with what I got off the D100 either. The XLR/RCA outs were fine and measure near 0-10mv DC, but HP out was between 30-60mv on one channel and 80-160 on another. (unlike MD11 where it was constant, here it was sort of oscillating between the two values. Unfortunately the HP terminals are sealed, so I can't measure the offset on HP plugged in. (edit: hmmm, actually, I can on the driver terminals on DT48e and it's about 0mv on those...for D100. For MD11 however, even with DT48e connected, it's still 140mv on one channel and 130mv on the other))

I'm confused...are we supposed to measure with a load connected, or just the DMM? (load connected seems to make sense, but then you need dodgy HP and an unsealed terminal to measure or an open unit (not a problem for DIY but I'd need to void the warranty on the MD11 if I didn't have DT48e where the driver terminals are exposed )

 

[wdahm519]

Quote:

I'm not really happy with what I got off the D100 either. The XLR/RCA outs were fine and measure near 0-10mv DC, but HP out was between 30-60mv on one channel and 80-160 on another. (unlike MD11 where it was constant, here it was sort of oscillating between the two values. Unfortunately the HP terminals are sealed, so I can't measure the offset on HP plugged in. (edit: hmmm, actually, I can on the driver terminals on DT48e and it's about 0mv on those...for D100. For MD11 however, even with DT48e connected, it's still 140mv on one channel and 130mv on the other))

I'm confused...are we supposed to measure with a load connected, or just the DMM? (load connected seems to make sense, but then you need dodgy HP and an unsealed terminal to measure or an open unit (not a problem for DIY but I'd need to void the warranty on the MD11 if I didn't have DT48e where the driver terminals are exposed

)

 
I did both.  I measured without a load, and it started oscillating from 20mV to 80mV slowly, each channel.  I feel like this method is useless because it doesn't tell you anything.  I plugged in some cheapo AD-700's and offset dropped to 1.?? mV.  
 
The only way to get a proper reading is to have a load connected.
 
 

 

[svyr]

 
I did both.  I measured without a load, and it started oscillating from 20mV to 80mV slowly, each channel.  I feel like this method is useless because it doesn't tell you anything.  I plugged in some cheapo AD-700's and offset dropped to 1.?? mV.  
 
The only way to get a proper reading is to have a load connected.
 
 

in that case D100 is fine, MD11 is most certainly not

 

[jr41]

I don't really want to de-case my mini³ to measure DC offset from my iphone with a load connected (I'm measuring the output from the iphone lineout - but this is still subject to a small load, i.e. the mini³). I don't want to hook up my Grado Labs SR225i to my Mini³  and iphone 4 until I know the DC offset is less than 10mV.
 
I'm building a M³ at the moment, so I'll measure DC offset from the iphone 4 lineout during testing.
 
I've yet to come across any definitive guidelines on dangerous DC offset levels and all the influencing factors.

 

[svyr]

I don't really want to de-case my mini³ to measure DC offset from my iphone with a load connected (I'm measuring the output from the iphone lineout - but this is still subject to a small load, i.e. the mini³). I don't want to hook up my Grado Labs SR225i to my Mini³  and iphone 4 until I know the DC offset is less than 10mV.
 
I'm building a M³ at the moment, so I'll measure DC offset from the iphone 4 lineout during testing.
 
I've yet to come across any definitive guidelines on dangerous DC offset levels and all the influencing factors.

relative max voltages DC+Signal above derived from the impedance and max rated input power sounded fairly definitive re: will this burn out my HP.
I think you're a bit hung up on the 10mv thing...

 

[scottiebabie]

cant speak much for dedicated headamps nor porty headamps but im a big vintage gearhead so checking & adjusting Bias current & DC offset is part & parcel of every amp after i bring it home. most if not all (havent read it all so i cant say) service manuals states static measurements with unplug source at AUX, zero/min vol with all ancilliary controls off & amp should be warmed up for atleast 10mins prior.
 
DC offset should be as close to 0mv as possible - not very probable unless the amp is output cap coupled.
 
Bias current settings according to each mfgrs specs as indicated in service manual - theres quite a bit of play room with Bias if overheating &increased power consumption isnt an issue with possible substantial sonic benefits eg. improved dynamics, bass, clarity, quickness, soundstage, coherance, decreased distortion, etc...etc but this will be specific to each individual amp of which only our own ears can be the final arbiter.
 
once adjustment is made, i usually let it run for a few minutes to have the final stabilized reading. FWIW, i use alligator clip lead cables to attach the DMM to the amps appropriate test lead points if for nothing else, i can read & adjust on the fly. more importantly, i dont accidentaly short any parts nor (MOST IMPORTANTLY) electrocute myself as readings can only be done with a 'live' amp so USE APPROPRIATE CAUTION!
 
some of the vintage stuff i bring home shows ori DC readings up over 100mv-200+mv at amp output (usually read at speaker taps but some amps have internal test leads too) with no detrimental effect on speakers or headphones even at fairly high output - usually btwn 1-3o'clock. just saying this as IMO i think DC offset is highly overrated plus i havent personally nor read of any transducer blowing as a direct result of DC. theres a fairly high built-in tolerance of DC in transducers & more likely than not, most blown drivers are caused by amp clipping thru inappropriate over driving IMHO.
 
ps:though it doesnt apply on single transducer speakers nor headphones but wont the inline caps at transducer crossover circuits also serve as DC blockers too? i think its more appropriate to worry about symetric readings on both channels to preserve channel/stereo coherence rather than worry about the lowest possible setting - again just my 2cents ofcos after playing with a fair number of vintage amps/receivers. 
 
most research i've done says any DC offset upto 50mv is considered normal & 'safe' under practical real world conditions. alot of amps deviate upwards to a few hundred mv with no apparent detriment to speakers (atleast structurely). 0mv is best ofcos!

 

[High_Q]

I've tried measuring audio signal out of my DMM one time, and thought occurred to me:  Wait a minute, I'm measuring a audio signal(it was fluxuating), maybe I should use a O-Scope!!!   One thing I don't understand is, how is offset of 10mV(thats 100th of a volt) dangerous if the signal swings way beyond that if you raise the gain?   
 
All these concepts may help with understanding headphone impedance.
 
From wiki:
 
 

Complex impedance

Impedance is represented as a complex quantity 

 and the term complex impedance may be used interchangeably; the polar form conveniently captures both magnitude and phase characteristics,

where the magnitude 

 represents the ratio of the voltage difference amplitude to the current amplitude, while the argument 

 gives the phase difference between voltage and current and 

 is the imaginary unit. In Cartesian form,

where the real part of impedance is the resistance 

 and the imaginary part is the reactance 

.

Where it is required to add or subtract impedances the cartesian form is more convenient, but when quantities are multiplied or divided the calculation becomes simpler if the polar form is used. A circuit calculation, such as finding the total impedance of two impedances in parallel, may require conversion between forms several times during the calculation. Conversion between the forms follows the normal conversion rules of complex numbers.

 
 

Inductor

 

For the inductor, we have the relation:

This time, considering the current signal to be

it follows that

And thus

This tells us that the ratio of AC voltage amplitude to AC current amplitude across an inductor is 

, and that the AC voltage leads the AC current across an inductor by 90 degrees.

This result is commonly expressed in polar form, as

Or, more simply, using Euler's formula, as

 

[High_Q]

Impdance Z is dependant on frequency w, and what is listed below.
 

Inductance formulae

The table below lists some common simplified formulas for calculating the approximate inductance of several inductor constructions.

Construction​	Formula​	Dimensions​	Notes
Cylindrical air-core coil[6]​		L = inductance in henries(H) μ0 = permeability of free space = 4π × 10−7 H/m K = Nagaoka coefficient[6] N = number of turns A = area of cross-section of the coil in square metres(m2) l = length of coil in metres (m) ​
Straight wire conductor [7]​		L = inductance l = cylinder length c = cylinder radius μ0 = vacuum permeability =4π nH/cm μ = conductor permeability p = resistivity ω = phase rate ​	exact if ω = 0 or ω = ∞
-0+3%	L = inductance (µH) l = length of conductor (mm) d = diameter of conductor (mm) f = frequency ​	Cu or Al l > 100d d2 f > 1 mm2MHz ​
+0-3%	L = inductance (µH) l = length of conductor (mm) d = diameter of conductor (mm) f = frequency ​	Cu or Al l > 100d d2 f < 1 mm2MHz ​
Short air-core cylindrical coil[8]​		L = inductance (µH) r = outer radius of coil (in) l = length of coil (in) N = number of turns ​
Multilayer air-core coil[citation needed]​		L = inductance (µH) r = mean radius of coil (in) l = physical length of coil winding (in) N = number of turns d = depth of coil (outer radius minus inner radius) (in) ​
Flat spiral air-core coil[citation needed]​		L = inductance (µH) r = mean radius of coil (cm) N = number of turns d = depth of coil (outer radius minus inner radius) (cm) ​
	L = inductance (µH) r = mean radius of coil (in) N = number of turns d = depth of coil (outer radius minus inner radius) (in) ​
Toroidal core (circular cross-section)[citation needed]​		L = inductance (H) μ0 = permeability of free space = 4π × 10−7 H/m μr = relative permeability of core material r = radius of coil winding (m) N = number of turns D = overall diameter of toroid (m)   ​

 

 

[High_Q]

 
Quote:

I think staying under 25mV is good.  Under 10mV is better.  Keep in mind, when you measure DC offset with no headphones plugged in, its going to be a a lot higher, and it may start oscillating from 20mV to 80mV.  
 
I was measuring my Bijou's offset and it was doing that.  I plugged the headphones in and it dropped to 1.XX mV.  

Output voltage(without load) is total output from the amp, and the lower value is when loaded with headphone(voltage drop accross it) depends on how much output impedance there is.  But offset or DC voltage drop is real as it deals with Real impedance, which also depends on real part of output impdeance, and heaphonephone impedance.  If output voltage is much greater than headphone impdance, there will not be significant offset on the headphones.
 
Quote:

It's 00:42 here. Why did I just spend the last 40 minutes obsessively tweaking my DC offset trimpots? Because of this F&*^ing thread.
I checked on impulse, and though one channel was 0.6mV, the other had drifted to nearly 26mV. It doesn't help that it changes by +/-5mV or so depending on the temperature, whether the lid is off the case, whether the screwdriver is touching the trimpot...
Broadly: Grach. Damn you, thread.

 

Large-signal models

[edit]Ebers–Moll model

Ebers–Moll Model for an NPN transistor.^[21] * I_B, I_C,I_E: base, collector and emitter currents * I_CD, I_ED: collector and emitter diode currents * α_F, α_R: forward and reverse common-base current gains​

​

Ebers–Moll Model for a PNP transistor.​

​

The DC emitter and collector currents in active mode are well modeled by an approximation to the Ebers–Moll model:

The base internal current is mainly by diffusion (see Fick's law) and

where

1. V_T is the thermal voltage kT / q (approximately 26 mV at 300 K ≈ room temperature).
2. I_E is the emitter current
3. I_C is the collector current
4. α_T is the common base forward short circuit current gain (0.98 to 0.998)
5. I_ES is the reverse saturation current of the base–emitter diode (on the order of 10⁻¹⁵ to 10⁻¹²amperes)
6. V_BE is the base–emitter voltage
7. D_n is the diffusion constant for electrons in the p-type base
8. W is the base width

The α and forward β parameters are as described previously. A reverse β is sometimes included in the model.

The unapproximated Ebers–Moll equations used to describe the three currents in any operating region are given below. These equations are based on the transport model for a bipolar junction transistor.^[22]

 

 

[High_Q]

 
Quote:

I'm confused...are we supposed to measure with a load connected, or just the DMM? (load connected seems to make sense, but then you need dodgy HP and an unsealed terminal to measure or an open unit (not a problem for DIY but I'd need to void the warranty on the MD11 if I didn't have DT48e where the driver terminals are exposed

)

You can deduce from amp's output voltage reading and output impedance of the amp and headphone impedance.  Investigate the output impedance of the amp and headphone impedance and you can figure out what will be outputted to the headphones.