Hi Mooly, well to kill the DC offset sent to the headphones? apparently >20mVDC can slowly fry the voice coils, but any DC offset will change the headphones FR anyway...so serious gear always makes sure to suppress it completely.
well, what I don't like about dual opamps on headphones is that they make the center channel blurry and messy...so I'll prolly use OPA211/OPA827 instead of the stock OPA2604?
I also have a dual AD797BR module if that's any good? but I was told that OPA827/AD8620BR were as good as it gets for DC precision.
I also have the usual suspects: LT1363CS8, LT1364, LT1361, LT1213, LT1122 etc etc 😉
OPA2604 was horrid for audio, but if a "DC servo" opamp doesn't color the sound whatsoever then I should be fine?
well, what I don't like about dual opamps on headphones is that they make the center channel blurry and messy...so I'll prolly use OPA211/OPA827 instead of the stock OPA2604?
I also have a dual AD797BR module if that's any good? but I was told that OPA827/AD8620BR were as good as it gets for DC precision.
I also have the usual suspects: LT1363CS8, LT1364, LT1361, LT1213, LT1122 etc etc 😉
OPA2604 was horrid for audio, but if a "DC servo" opamp doesn't color the sound whatsoever then I should be fine?
also, many headphones amps give a loud "pop" when you turn them on/off...this could be completely avoided if you suppress DC apparently: HeadWize - The Kumisa III Headphone Amplifier
20mv won't fry anything, you can work it out 🙂
but I agree it's not good. That's way over an acceptable (to me) offset for a power amp, let alone a headphone amp.
The OPA2604 should hold the output to under 1mv I would say, probably nearer 0.1mv
No gains whatsoever by swapping it... and the opamp must be FET input for a servo. A typical servo has the audio output fed to it via a 1 meg or higher resistor, with a cap connected from opamp output to input (configured as an integrator) to remove ALL audio and just give a DC level that's used to correct and maintain the DC operating point. The audio quality of the device doesn't come into it at all.
Can't agree on that one... but we have been here before 😉
ohhhhhh, so the AC audio doesn't even get through the DC servo op-amp then? why do they even need an opamp to suppress DC then?
so if I'm bored, which other opamp could I try in the list I mentioned above? OPA827? LT1364? THS4031?
the opamp is very close to the tubes, so I'd rather use a high temperature chip such as THS4031ID 🙂
what's a dangerous DC offset for you? 100mV?
Thanks,
PS: I also like how this site sells opamps to swap the stock OPA2604AP of my tube amp: http://www.audiophileproducts.com/opamps
Firestone told me that I could try my dual AD797BR module.
so if I'm bored, which other opamp could I try in the list I mentioned above? OPA827? LT1364? THS4031?
the opamp is very close to the tubes, so I'd rather use a high temperature chip such as THS4031ID 🙂
what's a dangerous DC offset for you? 100mV?
Thanks,
PS: I also like how this site sells opamps to swap the stock OPA2604AP of my tube amp: http://www.audiophileproducts.com/opamps
Firestone told me that I could try my dual AD797BR module.
Last edited:
ok, what about LT1213? Linear Technology - LT1213
they say "Excellent DC Precision", it can work up to 85C...it's not FET, though.
what should I look for in the datasheets to find the best DC Precision exactly?
they say "Excellent DC Precision", it can work up to 85C...it's not FET, though.
what should I look for in the datasheets to find the best DC Precision exactly?
Last edited:
For servos you need an OPamp with high input impedance. OPamps with Fet input provide that. The reason is that you can use a high value sensing resistor ( say 1MOhm) and consequently a much lower value cap ( say 1uF ) to get a low corner frequency.
The 1MOhm also does not load the output. I usually use an OPA134 and offset with that OPamp is much under 1mV. Look at input offset in the spec sheet.
If you need better then 1mV it gets expensive. The OPA627 is the usual choice.
The 1MOhm also does not load the output. I usually use an OPA134 and offset with that OPamp is much under 1mV. Look at input offset in the spec sheet.
If you need better then 1mV it gets expensive. The OPA627 is the usual choice.
Hey guys, what would be a good opamp to replace the LF353N in the Maverick TubeMagic D1? One of the mods is suggesting to replace them with LT1364 and LM4562. Do you think that's good? What do you think might be better?
Thanks! 🙂
Thanks! 🙂
LM4562 has a noise of 2.5nV/sqHz. LM353N has 25nV/sqHz.
LM4562 has SR of 20V/uS. LM353N has 13V/uS.
LM4562 has BW of 55MHz. LM353N has 4MHz.
What do you think, is there a doubt? 🙂
I have swaped almost all the OpAmps in my equipment with LM4562, I like the sound of it a lot.
LT1364 I think is OK for DAC I/V section, but I cannot tell without a schematic. It is a lot noisier that 4562.
LM4562 has SR of 20V/uS. LM353N has 13V/uS.
LM4562 has BW of 55MHz. LM353N has 4MHz.
What do you think, is there a doubt? 🙂
I have swaped almost all the OpAmps in my equipment with LM4562, I like the sound of it a lot.
LT1364 I think is OK for DAC I/V section, but I cannot tell without a schematic. It is a lot noisier that 4562.
Last edited:
Certainly there is, as there's more to opamps than those simple numbers. The LF353 (note not LM353) is a JFET amp - hence the high voltage noise specs. What you haven't pointed out is its current noise. Post up the specs of the current noise of these two devices would you?
My bad. Typo error, it is indeed LF353 (specs are from that OpAmp).
Are you having high-impedance sources in that schematic? Otherwise, the current noise is irrelevant.
Are you having high-impedance sources in that schematic? Otherwise, the current noise is irrelevant.
Right, I didn't know whether the source impedance was high, which is why I questioned your suggestion. Seems you don't know either 😀 Often people will spec a FET opamp for precisely that reason. Its valves so I'm guessing it is quite a high impedance.
Well, valves have Hi impedance AFTER the OpAmp usually. I guess it is just a plain diff/filter stage (the one on the left in the picture) followed by a buffer (the one on the left) and then a tube (that is bypassable)? DAC looks like it is just voltage out type, not current (so no I/V), but I cannot see clearly.
Last edited:
I can't see at all as I guess the pics are hosted on Flickr which is blocked here. But yes, if the tubes are after the opamp, I agree. And if the LF353 isn't doing I/V conversion and its got low impedances around it, the LM4562 might well be an improvement.
My next stage, an aikido preamp has enough input impedance that I could use even 0.22uf (can't remember the exact input impedance).
I have ordered some OPA134 and a friend is loaning me a pair of OPA627 to try on it.
Thanks.
The opamp is not in the audio path as such at all.
Why use it... because a "servo" made with an opamp holds the output to zero volts within very very close limits. You are not "suppressing DC", you are using the opamp to set the DC conditions within the amp. It takes the output of the amp, removes all audio by means of the integration network (resistor and cap) and this is fed to one of the inputs of the device. The other is referenced to zero volts. The output of the opamp feed a bias or correction voltage into the main amp to constantaly keep the average output DC level at zero.
The smaller the offest the better. For power amps historically 100mv or less was considered OK, that figure reducing as circuit designers became aware how to improve on it. A servo'd amp should hold to within a couple of tenths of a millivolt.
Unless you have an accurate means of measuring the offset... and we are all assuming the OPA2064 is used as a servo implying that the headphone output is DC coupled... then I would'nt alter anything.
It could just as easily be used to hold the voltage on (for example) the anode or cathode of the valve at some preset value... that's still a "servo" function but different to what we are assuming 🙂
ok thank Mooly for the detailed explaination 🙂
I'll roll LT1213 for the hell of it, as its datasheet boasts about "excellent DC precision"...and I've got a nocebo towards OPA/LME chips anyway, using LT/AD opamps makes me feel gooood 😀
I'll roll LT1213 for the hell of it, as its datasheet boasts about "excellent DC precision"...and I've got a nocebo towards OPA/LME chips anyway, using LT/AD opamps makes me feel gooood 😀
Last edited:
Mooly,
You obviously did a great deal of measurements. Have you ever measured the noise components from the earth line and the noise from the chassis ground picked up from the mains / transformers / EMI, etc?
I guess from the mains earth we would get 50/60Hz and rectified 100/120Hz noise, but there are also a whole lot of other junks. I am wondering if they are broadband noise like white noise, or depend on the EMI pickup.
This is totally beyond me because I don't think I am able to find a ground reference point in the first place.
One of my most primitive measurements was to connect the ground of the scope to the chassis and moved the probe around in the air. When it was closed to the transformer it showed me the mains frequency superimposed by possibly wideband noise. Of course, this only gives me a very small picture out of the bigger picture.
Understanding these noise components will help us design the right circuits to reduce the noise in our opamp circuits.
I found connecting the circuit ground to the chassis only at high frequencies, and also connect the chassis to earth only at high frequencies (via safety bridges) give me the lowest noise (subjective) and I want to verify this.
Regards,
Bill
You obviously did a great deal of measurements. Have you ever measured the noise components from the earth line and the noise from the chassis ground picked up from the mains / transformers / EMI, etc?
I guess from the mains earth we would get 50/60Hz and rectified 100/120Hz noise, but there are also a whole lot of other junks. I am wondering if they are broadband noise like white noise, or depend on the EMI pickup.
This is totally beyond me because I don't think I am able to find a ground reference point in the first place.
One of my most primitive measurements was to connect the ground of the scope to the chassis and moved the probe around in the air. When it was closed to the transformer it showed me the mains frequency superimposed by possibly wideband noise. Of course, this only gives me a very small picture out of the bigger picture.
Understanding these noise components will help us design the right circuits to reduce the noise in our opamp circuits.
I found connecting the circuit ground to the chassis only at high frequencies, and also connect the chassis to earth only at high frequencies (via safety bridges) give me the lowest noise (subjective) and I want to verify this.
Regards,
Bill
Hi Bill,
There is no easy one size fits all answer to this.
When you have the probe in free space it's just picking up radiated em fields... it doesn't really tell us much at all.
Ground references... what we really mean is a point we define as being our reference in say an amp, and then making sure all vital points are reference to that.
The problems are worse with amps that consume more power (such as power amps) and amps that output significant power into a low impedance load... again power amps.
Imagine your PSU. Forget the regulators. You have bridge feeding two reservoir caps. There is a common conection between the two caps that becomes ground.
Problem... assume we draw a constant current from the PSU. There are large currents flowing in that connection as the bridge rectifier continually charges those caps. If the caps are large, the charging currents are correspondingly higher, but flow for a shorter time. That connection has resistance, so those current pulses create a volt drop (corresponding to the charge current waveform) along that wire.
If you connect an input ground and (say) a feedback return to that wire then even if they are separated by only a mm, that is enough to creat a volt difference between the two and so noise and hum is created.
So we have to define a ground that's clean, and that's easy, all we do is take a short "tap" off that common point and call that (the end of it) our star point or ground reference. It's clean because no charging current flows in it. So our input ground has to be reffered to that point, so too the feedback returns etc.
You have to think of every conductor, and piece of print as a potential problem. With a good grounding scheme you can make "wiring" dissappear.
The problem is much less severe with opamps because the power draw and current supplied to any load is so small. The power supply rules still apply though.
So that gives you a "clean" circuit.
You then have the question of whether to ground the star to the chassis, the chassis to mains earth etc. Again no easy answer.
It has to be safety first, so that means chassis to mains earth. Do you connect the star here too... a lot depends on partnering equipment... no hard and fast rules. If you do, then that in itself shouldn't be a problem. Where a problem arises is if the CD player etc also does this. Now you have a "loop" via interconnects and grounds that can pick up hum/noise. So sometimes the answer is to "break" that by isolating the audio ground of one component, perhaps by not connecting the star earth to chassis but via say a 10 or 100 ohm resistor. Each case is different unfortunately... you only have to look at the hum problems with commercial stuff when connected together sometimes.
There is no easy one size fits all answer to this.
When you have the probe in free space it's just picking up radiated em fields... it doesn't really tell us much at all.
Ground references... what we really mean is a point we define as being our reference in say an amp, and then making sure all vital points are reference to that.
The problems are worse with amps that consume more power (such as power amps) and amps that output significant power into a low impedance load... again power amps.
Imagine your PSU. Forget the regulators. You have bridge feeding two reservoir caps. There is a common conection between the two caps that becomes ground.
Problem... assume we draw a constant current from the PSU. There are large currents flowing in that connection as the bridge rectifier continually charges those caps. If the caps are large, the charging currents are correspondingly higher, but flow for a shorter time. That connection has resistance, so those current pulses create a volt drop (corresponding to the charge current waveform) along that wire.
If you connect an input ground and (say) a feedback return to that wire then even if they are separated by only a mm, that is enough to creat a volt difference between the two and so noise and hum is created.
So we have to define a ground that's clean, and that's easy, all we do is take a short "tap" off that common point and call that (the end of it) our star point or ground reference. It's clean because no charging current flows in it. So our input ground has to be reffered to that point, so too the feedback returns etc.
You have to think of every conductor, and piece of print as a potential problem. With a good grounding scheme you can make "wiring" dissappear.
The problem is much less severe with opamps because the power draw and current supplied to any load is so small. The power supply rules still apply though.
So that gives you a "clean" circuit.
You then have the question of whether to ground the star to the chassis, the chassis to mains earth etc. Again no easy answer.
It has to be safety first, so that means chassis to mains earth. Do you connect the star here too... a lot depends on partnering equipment... no hard and fast rules. If you do, then that in itself shouldn't be a problem. Where a problem arises is if the CD player etc also does this. Now you have a "loop" via interconnects and grounds that can pick up hum/noise. So sometimes the answer is to "break" that by isolating the audio ground of one component, perhaps by not connecting the star earth to chassis but via say a 10 or 100 ohm resistor. Each case is different unfortunately... you only have to look at the hum problems with commercial stuff when connected together sometimes.
Bill... you might like to read/look at this,
http://www.diyaudio.com/forums/solid-state/101321-3-stage-lin-topology-nfb-tappings.html
it's not too long and I did some drawings etc at the time part way through the thread. This was regarding a specific issue with some PCB's and grounding on D Selfs "blameless amp" but it's all relevant.
http://www.diyaudio.com/forums/solid-state/101321-3-stage-lin-topology-nfb-tappings.html
it's not too long and I did some drawings etc at the time part way through the thread. This was regarding a specific issue with some PCB's and grounding on D Selfs "blameless amp" but it's all relevant.