no problem, you could conceivably do it if you used a socket and added the compensation, alternatively if you raised the gain to x4-5 it would work fine. its just i see these chips rolled into unsuitable circuits so often
half your luck having too many ad797, wanna sell some? i need some for regulators
half your luck having too many ad797, wanna sell some? i need some for regulators
Last edited:
They are not that much. I just expect this part to be discontinued soon and decided to keep several pcs for black days. So far I'm most impressed by ADA4637-1BR but they are quite expensive. I'm yet to try LME49990 if it is really that good.
I've read almost the whole thread but I still have this in my head: what are the benefits of SMD vs through-hole, except size/reduced signal traces? Are differences within levels of perception?
I've read almost the whole thread but I still have this in my head: what are the benefits of SMD vs through-hole, except size/reduced signal traces? Are differences within levels of perception?
Hi Guys,
After agdr's question about the pot on the PSU I realized I hadn't posted updated schematics for quite a while.
I've attached a PDF which is the absolute final version of all kits, and this is exactly what will be shipped in each kit.
The schematic has information for each part embedded, so just click on the part you're interested in, and it will tell you everything you need to know about it.
I'll post updated BOM's later today.
Regards,
Owen
After agdr's question about the pot on the PSU I realized I hadn't posted updated schematics for quite a while.
I've attached a PDF which is the absolute final version of all kits, and this is exactly what will be shipped in each kit.
The schematic has information for each part embedded, so just click on the part you're interested in, and it will tell you everything you need to know about it.
I'll post updated BOM's later today.
Regards,
Owen
@nevod: you have actually SEEN the performance of the current amp and read the datasheet of the amp you propose right? can you show any cause whatsoever why using an RF driver for audio presents an improvement? the needed compensation to slow it down. CMRR of this one sits at only 40db, has a typical output common mode offset of 40mV, max single rail supply voltage of 5.5v, so would need a totally separate power supply. max output of only 4.75v p2p, noise is orders of magnitude higher than opa1632, the speed is of no practical benefit whatsoever and that is what it has been tuned for at the expense of good performance where its actually useful for audio
qusp,
No problem, I'm just suggesting. I'm not expert here.
What actually bugs me about the BAL-BAL version is that
1) Input stage is a single diff op-amp, not an instrumentation amp like in BAL-SE. Or is it really pointless and a single op-amp is no worse than triple op-amp?
2) Output stage is 2 separate SE op-amps, not a single diff op-amp. But, I suppose, there are just no such op-amps?
No problem, I'm just suggesting. I'm not expert here.
What actually bugs me about the BAL-BAL version is that
1) Input stage is a single diff op-amp, not an instrumentation amp like in BAL-SE. Or is it really pointless and a single op-amp is no worse than triple op-amp?
2) Output stage is 2 separate SE op-amps, not a single diff op-amp. But, I suppose, there are just no such op-amps?
I can provide some info here. I'm one of those folks who would love to see "quieter" power supply regulators and/or filters if for no other reason than I can.
And a few months ago I made a few suggestions on regulators with lower published noise specs by the manufacturer. Same is true with the O2 headamp in this section. Several other folks have also offered up regulator and filter suggestions.But in the end both opc with the Wire here and RocketScientist, in the case of the O2, have thoroughly convinced me that it simply will not matter by way of their published noise floor measurements on the amplifiers. Opc's measurements are with an Audio Precision analyzer and RocketScientist's are with a similarly capable dScope analyer, both pretty much the best available in the industry by my understanding. And in both cases they have tried other lower noise power supply regulator chips and they only made vanishingly small, if any, noise difference in the amp output. The noise floor opc measured on the SE-SE version is already a rather astonishing -160dBv:
http://www.diyaudio.com/forums/head...-headphone-amplifier-pcbs-18.html#post2790636
So although I would still just love to see my current favorites in the power supply, LT3015 and LT1963A with an order of magnitude lower noise specs, I unfortunately realize it would be purely for fun, because it really doesn't make any measurable difference by way of the real-world tests. I guess another way to look at this is a person can hit the point where any further noise reductions are either (1) too low to measure even with the best equipment and/or (2) too low to hear - and at those levels it becomes a moot issue.
Hope this helps!
Last edited:
Hope this helps!
I have no doubt that the performance is impressive. Though I do think that measuring only noise floor for PS performance is not enough. AFAIK, the advantage of the miltiplier is that regulator doesn't have to handle any HF drain variations, and that any drain variation is indeed quickly supplied, so that is actually more about handling aperiodic signals.
But I may be wrong. Anyways, it's not that it is impossible to toss an multiplier between power supply and the amp itself.
And, sorry if I've really overlooked that.. Are there any more kit runs planned? In 3-4 months a need for a headphone amp may arise..
Excellent answers all around! Both qusp and agdr have covered the questions exactly as I would have
I'm seeing a trend in your questions however, and I'm going to guess that you're placing an importance on the "speed" of the amplifier.
The design presented here is extremely wide bandwidth, and if you're focusing in on slew rate as a measure of the amplifier's ability to deliver dynamics, then you're not really looking at things the right way.
For the part you referenced, the slew rate is indeed a massively high 13,000V/us. In order to require a slew rate that high, you either need to be swinging an enormous amount of voltage in the audio band, or you need to be dealing with extremely high frequencies at normal voltages. Slew rates that high are not required for audio because the signals quite simply do not change at that rate. No matter how fast or how great the transient, you will not hit the slew rate of The Wire within the meaningful (or even within the better part of the non-meanignful) audio band. Slew rate is not a measure of the amplifiers ability to respond to dynamics in music.
Think of it as adding horsepower to a vehicle. You could say that 80hp is enough to get you through 99% of your driving needs, but sometimes it's fun to have more like 200hp. And if that's a little more fun, then why not say 400hp. If you really like going fast, you could push to a fairly unreasonable 1000hp. At this point, you're making some serious sacrifices in order to have all that power. The car will be less drivable in normal conditions, and it will be less efficient in terms of fuel usage, but you could still potentially make the argument that it's of some benefit since it does indeed allow you to accelerate a little faster, and obtain a higher top speed than say 400hp. Now try making the same argument for 100,000 horsepower. The car will be nearly impossible to drive correctly, and will act in an unstable manner all of the time. The additional horsepower will only serve to tear your wheels to pieces, and provides nothing in terms of additional speed because you've exceeded the grip limitations of your tires. Sure it sounds cool, but all you've done is spent a fortune on an engine and you have something that performs worse in every conceivable way than a car with a more reasonable amount of horsepower.
Slew rate is the same thing. It's good to have enough, and nice to have more than enough, but it quickly becomes detrimental to have way too much. 13,000V/us is WAY too much.
Same applies to the power supply. Those HF transients you're talking about are not delivered by the supply, they're delivered buy the 10uF ceramic bypass caps placed less than 1mm away from the pin, and the local capacitance created by the large supply planes sitting on top of the GND plane. The bulk supply and regulators have very little to do with the HF transient response of this particular amp.
Regards,
Owen
I'm seeing a trend in your questions however, and I'm going to guess that you're placing an importance on the "speed" of the amplifier.
The design presented here is extremely wide bandwidth, and if you're focusing in on slew rate as a measure of the amplifier's ability to deliver dynamics, then you're not really looking at things the right way.
For the part you referenced, the slew rate is indeed a massively high 13,000V/us. In order to require a slew rate that high, you either need to be swinging an enormous amount of voltage in the audio band, or you need to be dealing with extremely high frequencies at normal voltages. Slew rates that high are not required for audio because the signals quite simply do not change at that rate. No matter how fast or how great the transient, you will not hit the slew rate of The Wire within the meaningful (or even within the better part of the non-meanignful) audio band. Slew rate is not a measure of the amplifiers ability to respond to dynamics in music.
Think of it as adding horsepower to a vehicle. You could say that 80hp is enough to get you through 99% of your driving needs, but sometimes it's fun to have more like 200hp. And if that's a little more fun, then why not say 400hp. If you really like going fast, you could push to a fairly unreasonable 1000hp. At this point, you're making some serious sacrifices in order to have all that power. The car will be less drivable in normal conditions, and it will be less efficient in terms of fuel usage, but you could still potentially make the argument that it's of some benefit since it does indeed allow you to accelerate a little faster, and obtain a higher top speed than say 400hp. Now try making the same argument for 100,000 horsepower. The car will be nearly impossible to drive correctly, and will act in an unstable manner all of the time. The additional horsepower will only serve to tear your wheels to pieces, and provides nothing in terms of additional speed because you've exceeded the grip limitations of your tires. Sure it sounds cool, but all you've done is spent a fortune on an engine and you have something that performs worse in every conceivable way than a car with a more reasonable amount of horsepower.
Slew rate is the same thing. It's good to have enough, and nice to have more than enough, but it quickly becomes detrimental to have way too much. 13,000V/us is WAY too much.
Same applies to the power supply. Those HF transients you're talking about are not delivered by the supply, they're delivered buy the 10uF ceramic bypass caps placed less than 1mm away from the pin, and the local capacitance created by the large supply planes sitting on top of the GND plane. The bulk supply and regulators have very little to do with the HF transient response of this particular amp.
Regards,
Owen
Last edited:
several reasons not to do that.
• for starters, an instrumentation amplifier usually has (perhaps by definition has to have) a differential input and a single ended output, how would this be useful for a balanced output amp? you would need to have 2 of them and supply it with a dual differential dac output to get a single balanced output.
• 3 x opamps for starters doesnt make sense for balanced output unless performing a balanced-SE conversion, or if one of them was a differential I/O chip like the opa1632, noise performance with 2 series active elements will compound and be worse than the single diff chip. thermal and therefor common mode coupling/matching will be worse and thus noise performance will suffer again. this is handling 2 phases of the same channel, not 2 different channels. there is no benefit at all to using separate chips as could be argued (in a rather academic fashion) when talking about a left and right channel.
• signal routing of the phases would be less than ideal and thus noise and CMMR would again suffer.
the whole thing of balanced audio is that its supposed to be balanced, for best CMMR the phases should ideally be exposed to the 'elements' equally and coupled as tightly as practical electrically, thermally and physically, there is no better way than to have them on the same die. Besides the opa1632 is an extremely high performance chip
doesnt exist for audio, perhaps there are some RF or ADSL line drivers that provide a high current differential output, but then we would be in the same territory as both myself and opc have outlined. the lme496X0 is arguably the single best performing unity gain audio buffer IC that is purpose built for the job at hand. its certainly been my preference
but i have to say, didnt you just argue with yourself with those 2 questions?
Last edited:
all of the amps are set for unity gain as stock, this will be how they were measured.
both should be 1k0 as previously, as in the new schematic that amp would spin out of control, DC offset off the charts lol. better fix that Owen
actually looking again, it has a weird kind of symmetry, its definitely a mistake, but the way its drawn is hurting my head figuring out how it would behave. it looks as if the non-inverting FB would be 10x, but the inverting FB would be 1/10x, but yeah for some reason the way its drawn is playing games with my morning head
Last edited:
L-Train:
It is not a schematic error... those are definitely the correct values for a unity gain amplifier. R15 and R19 were not originally the correct value and need to be 10k for unity gain.
This breaks down as follows:
R13 and R14 form an L-pad that gives a gain of 0.909
The gain of the amplifier section is 1+(R16/R15) which is 1+(1k/10k) or 1.1
The two together mean you have 0.909*1.1 = ~1
The amp was measured with a gain of 1 using exactly what is show in the schematic, but would be perfectly fine with a higher gain if needed. The noise floor might go up slightly, but what's show in the measurements is just the AP noise floor, so it might not even be a measurable difference if the gain is kept around 2x or 3x.
After using this amp and measuring it, I would actually agree that it's the only one of the amps you could really make a case for having more gain with. Many SE sources don't have much drive, and it might be reasonable to up the gain to 2x in some cases. For example, the AP measurements stop at 7.2VRMS since that's the highest voltage the generators can do in SE mode. This amp is good for roughly 9.2VRMS which might be more attainable with a gain of 4x if you really need all that voltage swing.
For a gain of 2x change R15 and R19 to 833 ohms
For a gain of 3x change R15 and R19 to 435 ohms
For a gain of 4x change R15 and R19 to 295 ohms
Cheers,
Owen
It is not a schematic error... those are definitely the correct values for a unity gain amplifier. R15 and R19 were not originally the correct value and need to be 10k for unity gain.
This breaks down as follows:
R13 and R14 form an L-pad that gives a gain of 0.909
The gain of the amplifier section is 1+(R16/R15) which is 1+(1k/10k) or 1.1
The two together mean you have 0.909*1.1 = ~1
The amp was measured with a gain of 1 using exactly what is show in the schematic, but would be perfectly fine with a higher gain if needed. The noise floor might go up slightly, but what's show in the measurements is just the AP noise floor, so it might not even be a measurable difference if the gain is kept around 2x or 3x.
After using this amp and measuring it, I would actually agree that it's the only one of the amps you could really make a case for having more gain with. Many SE sources don't have much drive, and it might be reasonable to up the gain to 2x in some cases. For example, the AP measurements stop at 7.2VRMS since that's the highest voltage the generators can do in SE mode. This amp is good for roughly 9.2VRMS which might be more attainable with a gain of 4x if you really need all that voltage swing.
For a gain of 2x change R15 and R19 to 833 ohms
For a gain of 3x change R15 and R19 to 435 ohms
For a gain of 4x change R15 and R19 to 295 ohms
Cheers,
Owen
Last edited: