|
Post by oohms on Oct 19, 2010 5:23:50 GMT
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Oct 19, 2010 8:35:34 GMT
That DC servo will do the trick. Nicely done and a good example of when to use circuits like this.
Giving some improvement tips (not meant as critique):
the DC servo feedback point you now have is coming from a voltage devider (I asume the 60 Ohm is replacing a HP in a simulation and is not actually there IRL) I would have fed back the DC from eQ1. Reason being the DC feedback will not be load dependent but always be correct. Now the voltage on eQ1 can vary with the load attached. I know it is minute but can be remedy-ed easily.
changing the output resistance from 2.2 to 68 Ohms like you did is recommended indeed. The 100mA idle current is just enough to bring full power to a 32Ohm headphone (that is with the 68 Ohm series resitance) With the 2.2 Ohm as in the original you would run a serious risk of destroying headphones due to asymmetrical clipping (and thus DC on the output) your DC servo won't counteract this entirely with 2.2 Ohm output resistance but with 68Ohm this cannot happen. output = 2V at 32 Ohms (0.14W) or 5.4V at 300 Ohm (0.09W) which is sufficient in most cases.
It is effectively a 2 stage amplifier design with 10x gain with an open loop gain of about 5000x to 7000x which equates to 40dB reduction in distortion. These configurations generally have a THD of around 0.1% (which is inaudible for most b.t.w.) The 68 Ohm reduces the distortion even further as the emitter follower output stage is loaded far less.
|
|
|
Post by oohms on Oct 19, 2010 10:51:50 GMT
I'll amend that, thanks for the tip! I really appreciate your feedback btw, as i only just know enough to do the basics ie: an output resistor is good for limiting current into headphones, and that a DC servo can solve the DC offset problem, so i borrow the circuit from another project A good learning experience
|
|
FauDrei
Been here a while!
Posts: 489
|
Post by FauDrei on Oct 19, 2010 11:53:16 GMT
Hmm... I've asked this in another thread but haven't got an explicit answer... Which characteristics make opamp better suited for DC servo operation? - offset voltage?
- noise under 1kHz?
- PSRR?
- CMRR?
- slew rate?
- ...
According to those characteristics, is there any "go to" opamp for DC servo operations?
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Oct 19, 2010 12:01:31 GMT
Hmm... I've asked this in another thread but haven't got an explicit answer... Which characteristics make opamp better suited for DC servo operation? - offset voltage?
- noise under 1kHz?
- PSRR?
- CMRR?
- slew rate?
- ...
According to those characteristics, is there any "go to" opamp for DC servo operations? Valter The LF411 is very often used in this application. It's quoted specifications may provide some answers ? Alex
|
|
|
Post by oohms on Oct 19, 2010 13:50:41 GMT
I was always under the assumption that DC servos were supposed to operate at subsonic frequencies? So the only real difference between opamps would be the input bias current which results in different offset voltages
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Oct 19, 2010 15:08:29 GMT
low noise* and low offest voltage might be preferable. The DC offset correction on the output of the amp is as good as the input offset of the opamp as the amp has maximum gain for DC and zero gain for AC signals (AC is filtered out only the average DC, up to a point and within a certain timeframe, and DC offset remain). high PSRR is recommended if you don't have a stabilised power supply feeding it. slew rate and FR is not of importance as Oohms correctly points out. It only works for DC and very low frequencies (below 1Hz in general).
* All HF noise, above 100Hz in this case, coming from the IC is filtered out by R13 and C10. Increasing C10 will lower this. When these parts are not mounted the noise is also of importance.
The input bias current (if a transistor input IC is used) may be of importance. This can be solved by not connecting the grounded input directly to ground but with the same resistance as is used in the + input. Adds noise though if the resistors have high values. FET and MOSFET inputs have such low input currents this is of no relevance.
stabilisation within 10mV is already more then enough so you don't have to be anal about the used opamp and parts surrounding it. NO audio is passing through it.
|
|
FauDrei
Been here a while!
Posts: 489
|
Post by FauDrei on Oct 19, 2010 21:57:07 GMT
I was around/under 1mV with OPA2134, just wandered if OPA2227/OPA2228 might be better there.
2227/2228 have lower offset voltage, noise, PSRR and CMRR.
2134 has lower input bias current (FET) and higher slew rate
So, I assume all three would work OK, 2227/2228 being slightly better.
Thanks guys. I got my answers.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Oct 21, 2010 17:21:19 GMT
Only OPA2227 is suited and the GND reference pin must NOT be connected with a series resistor to ground but directly like in Oohms schematic. OPA2227 will have 50 times smaller DC offset (theoretically) as OPA2134. In practice it will be closer to 10 times smaller.
|
|
|
Post by tripmaster on Nov 1, 2010 10:56:36 GMT
Hi Oohms
I fancy having a go at one of these headphone amps. Are you still pleased with it?
Apart from adding a dc servo have you modded it in any other way?
Richard
|
|
|
Post by radelius on Mar 12, 2015 19:21:16 GMT
I have just built a pair of this, nice sounding but to much dc drift on the output. I'm going to build a servo and this one seems nice, a question a have: When a do some calculations on the filter I get app. 16Hz for the first pole R12/C11 and 0.5Hz R11/C12. Why not put in 330k on R12? I'm a bit unsure when it comes to inverting OP:s. Staffan
|
|