Thanks Andy and Jonathan for your replies.
Jonathan, what you say about the power supply reservoir capacitors is interesting. Having looked at their ground paths more closely I understand what you mean. I hope that my PCB in its present incarnation will offer stable performance. I think double sided boards with more carefully defined ground planes are in order in the future!
Andy, considering I only want one PCB, I think PCB Pool works out as (just) being cheaper. They also offer a shorter lead time. However, from what you say, Minnitron would seem to be much cheaper for more than one PCB. Thanks for the info! FWIW, PCB Pool once sent me double the number of boards I asked for free of charge, all of them fine.
Thanks again,
Tim.
Jonathan, what you say about the power supply reservoir capacitors is interesting. Having looked at their ground paths more closely I understand what you mean. I hope that my PCB in its present incarnation will offer stable performance. I think double sided boards with more carefully defined ground planes are in order in the future!
Andy, considering I only want one PCB, I think PCB Pool works out as (just) being cheaper. They also offer a shorter lead time. However, from what you say, Minnitron would seem to be much cheaper for more than one PCB. Thanks for the info! FWIW, PCB Pool once sent me double the number of boards I asked for free of charge, all of them fine.
Thanks again,
Tim.
Tim
Sounds like PCB pool did you proud.
It's quite normal to get extra boards, as the company would usually overmake to account for any rejects. They often send those extra boards, particularly to obvious one-off's or DIY types.
If nothing else, at least it's another avenue to try!
This place is great isn't it - where else can you get critique from well-known audio designers
Andy.
Sounds like PCB pool did you proud.
It's quite normal to get extra boards, as the company would usually overmake to account for any rejects. They often send those extra boards, particularly to obvious one-off's or DIY types.
If nothing else, at least it's another avenue to try!
This place is great isn't it - where else can you get critique from well-known audio designers
Andy.
Tim, don't want to critizise too much I just want to give you some food for thought:
The groundplane is very sliced up and the return currents go a little everywhere. This may not be a problem (you could get reduced performance) but you could make jumpers in stategic places in order to get better and more predictable current paths. One trick to make a better groundplane is to let traces go closer and by that get wider ares with ground.
Using a 2000 V/µs device is can be tricky but since this only is a buffer I'll guess it's not so tricky....
The groundplane is very sliced up and the return currents go a little everywhere. This may not be a problem (you could get reduced performance) but you could make jumpers in stategic places in order to get better and more predictable current paths. One trick to make a better groundplane is to let traces go closer and by that get wider ares with ground.
Using a 2000 V/µs device is can be tricky but since this only is a buffer I'll guess it's not so tricky....
Hi Peranders,
"This may not be a problem (you could get reduced performance) but you could make jumpers in stategic places in order to get better and more predictable current paths."
Thanks for the idea - I shall do that. For instance, I could place a jumper across the power trace running under C10 and C17.
"One trick to make a better groundplane is to let traces go closer and by that get wider ares with ground."
I have my design rules set for a minimum track-track distance of 13th (~0.33mm). It just looks a lot more in the screen shot I posted.
Thanks a lot for your input,
Tim.
PS - I wish I had posted here _before_ I ordered PCBs
"This may not be a problem (you could get reduced performance) but you could make jumpers in stategic places in order to get better and more predictable current paths."
Thanks for the idea - I shall do that. For instance, I could place a jumper across the power trace running under C10 and C17.
"One trick to make a better groundplane is to let traces go closer and by that get wider ares with ground."
I have my design rules set for a minimum track-track distance of 13th (~0.33mm). It just looks a lot more in the screen shot I posted.
Thanks a lot for your input,
Tim.
PS - I wish I had posted here _before_ I ordered PCBs
Peranders (or anyone else),
On your schematic of your headphone amp (very nice looking headamp, BTW), I noticed the CCS after the input opamp and before the BUF634. In reading over various threads, such as the Best Opamp, the concensus is for a 2mA CCS at this point, and this is what you designed.
While this would seem like a good option for you to have, I'd a little confused as to why anything is different at these CCS settings. My confusion concerns two items: the opamp idle current and the opamp output current.
The idle current of most opamps is over 3mA (the AD8610 is about 3mA; the AD845 is 10mA; and most are somewhere in between). I've read that you want the CCS to be about 20% of the maximum output. In the case of the AD8610, it's maximum is 50mA, so 10mA would be a good goal. For the AD845, its output is also 50mA, so it's output is probably already close to 20%.
Does this mean that the AD845 won't be improved since it is possibly already in class A mode? Does it also mean that the AD8610 should be run closer to 10mA rather than 2mA?
Finally, if the idle for the AD8610 is 3 mA, with probably most of this in the output stage, than it would seem to already being running at about 2mA on the output stage. So what would providing a 2mA CCS add? (If the answer, is the AD8610 should have a CCS of 20% of its output, or 10mA, then a CCS on the output starts to make sense.)
Thanks, Robert
On your schematic of your headphone amp (very nice looking headamp, BTW), I noticed the CCS after the input opamp and before the BUF634. In reading over various threads, such as the Best Opamp, the concensus is for a 2mA CCS at this point, and this is what you designed.
While this would seem like a good option for you to have, I'd a little confused as to why anything is different at these CCS settings. My confusion concerns two items: the opamp idle current and the opamp output current.
The idle current of most opamps is over 3mA (the AD8610 is about 3mA; the AD845 is 10mA; and most are somewhere in between). I've read that you want the CCS to be about 20% of the maximum output. In the case of the AD8610, it's maximum is 50mA, so 10mA would be a good goal. For the AD845, its output is also 50mA, so it's output is probably already close to 20%.
Does this mean that the AD845 won't be improved since it is possibly already in class A mode? Does it also mean that the AD8610 should be run closer to 10mA rather than 2mA?
Finally, if the idle for the AD8610 is 3 mA, with probably most of this in the output stage, than it would seem to already being running at about 2mA on the output stage. So what would providing a 2mA CCS add? (If the answer, is the AD8610 should have a CCS of 20% of its output, or 10mA, then a CCS on the output starts to make sense.)
Thanks, Robert
CCS
It would be interesting to know whether the CCS brings anything to the AD8610 / BUF634 combination.
In the early days, before the complementary bipolar processes used these days the PNP's used were terrible and largely responsible for the poor performance of the o/p stages.
With the CB, and particularly XFCB process used in the AD8610 and the like, they are much higher quality. Although not class A in operation, the output stages produce very little distortion, particularly when running buffered, i.e. virtually unloaded as is the case here.
I wonder if the difference is audible, and hence whether the extra dissipation is justified.
Have you tried it without the CCS Per?
Andy.
It would be interesting to know whether the CCS brings anything to the AD8610 / BUF634 combination.
In the early days, before the complementary bipolar processes used these days the PNP's used were terrible and largely responsible for the poor performance of the o/p stages.
With the CB, and particularly XFCB process used in the AD8610 and the like, they are much higher quality. Although not class A in operation, the output stages produce very little distortion, particularly when running buffered, i.e. virtually unloaded as is the case here.
I wonder if the difference is audible, and hence whether the extra dissipation is justified.
Have you tried it without the CCS Per?
Andy.
I have found that the AD-8610 and the OPA-627 are improved alot in Buffered applications with an active current source from the Output V- rail. I have heard that Analog devices sound Better if the Current source is referenced to the V+ rail however i have not yet tried the V+ rail.
As far PCB board houses Express PCP is quite reasonable and i Think thay have a $10.00 USD Proto type offer going on now. The URL may be www.expresspcb.com
As far PCB board houses Express PCP is quite reasonable and i Think thay have a $10.00 USD Proto type offer going on now. The URL may be www.expresspcb.com
Have you compared the CCS load approach against operating the o/p stage unloaded?
Many op-amps work much better sonically when they are unloaded at the o/p. The o/p stage is a very real source of distortion. Some series R to decouple any subsequent capacitive load makes a lot of diffference.
Andy.
Yes I have I use a JFET cascode Current source with a small value resistor in series between the Op amp output and the Current source output. I install IC socket Pins so as to be able to remove the resistor while at the same time listening to the effect. Since removing the resistor is quicker and easier I first start out with it Installed then remove it while playing music and note the difference if any. I repeat this several times on different days so as to see if my notse for that day's listening reflect what my Previous notes include. I noticed that most Precision op-amps like the OPA-627/637, AD-825, AD-8065/66, AD-8610/20, OPA-132/134, LT-1122, are improved a lot in the high frequencies with the current source installed. I also found that currents above about 5mA will degrade the Dynamics of these devices. I found 0.5 to 2 mA to be the Ideal range for most of these Op-Amps with the actual value being about 0.7 to 1.5 mA for the AD-8610/20 AD-8065/66. and about 2-3 mA for the OPA-627/637, OPA 134/132. Op-amps like the AD-845 and AD-843 do not improve with a current source however thay do not get degraded ether so I use the current source by default and set it for about 1 mA. These are all referenced to the Negitive supply rail so the results may be different if referenced to the Positive rail.
Hi Guys,
Interesting thread and just at the same time I am working on a portable headphone amp. I have been trying to improve my opamp knowledge recently in pursuit of other projects, and it was recommended to me the a "head amp" was a good way to "buff up"
Well, I have a few questions. I have some OPA627s, OPA637, OPA2134, and OPA2132s. Since I am waiting for my BUF634s to come in I thought about breadboarding a similar amp using the 637s as the amp, the 627s in a unity gain config (as the buffer) and a crossfeed network(Toni Kerrigan's) in between. Alot of what I have learned is out of Headwize articles but there are obvious gaps. Would the 627 work well enough as a buffer? How well would the OPA amps work with a dual 9V battery setup with a TLE2426 as a virtual ground or should I consider the other chips? A CCS might help but I worry about he extra drain on the battery. Thanks in advance.
Per, nice SMD board!
Interesting thread and just at the same time I am working on a portable headphone amp. I have been trying to improve my opamp knowledge recently in pursuit of other projects, and it was recommended to me the a "head amp" was a good way to "buff up"
Well, I have a few questions. I have some OPA627s, OPA637, OPA2134, and OPA2132s. Since I am waiting for my BUF634s to come in I thought about breadboarding a similar amp using the 637s as the amp, the 627s in a unity gain config (as the buffer) and a crossfeed network(Toni Kerrigan's) in between. Alot of what I have learned is out of Headwize articles but there are obvious gaps. Would the 627 work well enough as a buffer? How well would the OPA amps work with a dual 9V battery setup with a TLE2426 as a virtual ground or should I consider the other chips? A CCS might help but I worry about he extra drain on the battery. Thanks in advance.
Per, nice SMD board!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.