| Devil_H@ck |
I made a circuit for my preamp and there are 16 opamps on the board. I know that it's common practice to bypass the supply rails with for example 2.2µF & 100nF. But since there are so many chips on the board (the 16 opamps + 6 buffer ICs + 4 balanced line drivers), routing it on a one-sided board gets a bit though, so I was wondering if all this bypassing is also necessary when there is such an amount of opamps on one board. Is it also possible to put lets say a 470µF cap where the supply rails enter the board and bypass all the opamps with only 100nF for each rail? Or will this have a bad effect on sound? All the opamps are driven into class A, I'm not sure if I have to take this into account for the bypassing.
Any thoughts are highly appreciated. |
|
|
| carlmart |
| quote: | Originally posted by Devil_H@ck
Is it also possible to put lets say a 470µF cap where the supply rails enter the board and bypass all the opamps with only 100nF for each rail? Or will this have a bad effect on sound? All the opamps are driven into class A, I'm not sure if I have to take this into account for the bypassing.
|
It will depend on the opamp you are using. Some are more sensitive to bypassing, particularly the fast chips, and you should use at least 47uF and 100nF close to the power pins on each chip. A good film cap or some SMD caps close to the pins should be the minimum you should use, and perhaps some 47uF caps for each pair of chips.
Bypassing certainly influences the sound, but how much is you who would judge. Try and listen.
Carlos |
|
|
| Sjef |
You could consider something like the 47uF/25V Black Gate PK series. These are very affordable and very very small so they can be placed close enough to the opamps and they don't need any bypassing (better not with these) Try to keep the path between opamp and cap shorter than 5mm, specially with high-speed opams like the OPA627 and BUF634. It's a better solution then using inferior electrolytics with bypass caps wich can cause unwanted resonances.
If routing the PCB is that big a problem just use some jumpers (with some ferrit bread) for the power supply. These opamps you are using really need decoupling caps, otherwise use slower opamps.
If you search a bit they can be found for less than a dollar each. Performance is also very good, comparable with the far more expensive standard and F series or Elna Silmic and that kind. |
|
|
| Devil_H@ck |
Ok, thanks. I indeed plan on using OPA627 or AD8610, maybe OPA134 in the beginning and upgrade from there.
Hm, thinking about it, I might not buy those Black Gate PKs at first, but I'll allow space for it. I already need 46 electrolytics caps for the main stereo board in my preamp, so that already almost €50 for caps alone. I don't know, I'll see. If you know a very cheap source, please let me know. Thanks! |
|
|
| RetroAudio |
| perhaps taking into account the bandwidth of each device as well as the current swinging expectation would help decide how to balance all this bypassing, given that real estate will be an issue. Is bandwidth limiting going to be used (cap used in the feedback network) ? If not, then bypassing with .01-.1 uf might have it's place. Lots of factors at work here. |
|
|
| Sjef |
see www.kyoto-electro.com/pk.html they sell them for $0.76 each.
If you don't want to decouple the opamps then there is no benefit of using opamps like the OPA627. I have experimented a lot with them in an active crossover. Without decoupling things get only worse than with cheaper/slower opamps, a 2134 for example will sound much better when used in such a configuration. so don't spend all your money on them. Building a good amp is not only about buying good opamps, power supply and PCB layout and circuit design are also very important.
If you don't have the money then upgrade the opamps later on. |
|
|
| digi01 |
Sure,I agree with Sjef.
In the analog circuit design,we should reduce the quantity of the passive device.
| quote: | | Building a good amp is not only about buying good opamps, power supply and PCB layout and circuit design are also very important. |
In my view,just like op amp and transistors, the PCB boards is initiative device too;)
-digi |
|
|
| garbage |
| quote: | Originally posted by carlmart
It will depend on the opamp you are using. Some are more sensitive to bypassing, particularly the fast chips, and you should use at least 47uF and 100nF close to the power pins on each chip. A good film cap or some SMD caps close to the pins should be the minimum you should use, and perhaps some 47uF caps for each pair of chips.
|
i am using opa627 and have 4.7uF BG N caps with 100nF ceramic caps on pins 4 and 7.
have not tried other values of caps yet, but would i benefit from a larger value over my BG caps?
i have the opa627/buf634 currently in my passive preamp (it's not passive now ;)). the result so far is bass that goes lower, and greater soundstage. |
|
|
| Doovieman |
| quote: | Originally posted by Devil_H@ck
I made a circuit for my preamp and there are 16 opamps on the board. I know that it's common practice to bypass the supply rails with for example 2.2µF & 100nF. But since there are so many chips on the board (the 16 opamps + 6 buffer ICs + 4 balanced line drivers), routing it on a one-sided board gets a bit though, so I was wondering if all this bypassing is also necessary when there is such an amount of opamps on one board. Is it also possible to put lets say a 470µF cap where the supply rails enter the board and bypass all the opamps with only 100nF for each rail? Or will this have a bad effect on sound? All the opamps are driven into class A, I'm not sure if I have to take this into account for the bypassing.
Any thoughts are highly appreciated. |
Could you post a pic of the schematic? Thanks,
Doovieman |
|
|
| Devil_H@ck |
| quote: | Originally posted by RetroAudio
perhaps taking into account the bandwidth of each device as well as the current swinging expectation would help decide how to balance all this bypassing, given that real estate will be an issue. Is bandwidth limiting going to be used (cap used in the feedback network) ? If not, then bypassing with .01-.1 uf might have it's place. Lots of factors at work here. |
I tried some more and I got one "buffer block" down to 4cm x 2.5cm (1.6in x 1in). I don't think it can be made much smaller (a few mm maybe).
| quote: | Originally posted by Sjef
see www.kyoto-electro.com/pk.html they sell them for $0.76 each.
If you don't want to decouple the opamps then there is no benefit of using opamps like the OPA627. I have experimented a lot with them in an active crossover. Without decoupling things get only worse than with cheaper/slower opamps, a 2134 for example will sound much better when used in such a configuration. so don't spend all your money on them. Building a good amp is not only about buying good opamps, power supply and PCB layout and circuit design are also very important.
If you don't have the money then upgrade the opamps later on. |
Thanks for the link! I'll decouple them ;). I got some sample OPA627s that will use in the main signal chain. The rest I'll probably upgrade later on.
| quote: | Originally posted by Doovieman
Could you post a pic of the schematic? Thanks, Doovieman |
The whole schematic is barely readable on my 19 inch monitor. I split it up in 2 parts, you can see it here: http://members.lycos.nl/anthonyvh/i...eamp-audio-main. This is probably the part you're interested in: http://members.lycos.nl/anthonyvh/i...io-main-big.gif.
See the attachment for the layout I came up with for one buffer block. This "block" is used 6 times in the entire main audio circuit. Does it look a bit decent? For the schematic it is based upon, check my website. The 2 buffer blocks at the top left in the main circuit are the most readable. |
|
|
| Devil_H@ck |
I forgot the attachment :rolleyes:.
The input it at the top left between the 2 big caps. The output is at the top right. The light blue polygon around the parts is a ground plane, so some traces are actually connected to each other but you can't see it. +V is the most lower right trace, -V is the trace with the vertical jumper wire in it. Do you think it will be a problem that the -V for the opamp needs to go through the small SMD traces? I guess not, but I find it looks weird. |
|
|
|
