Good idea, it might be easier/better to just go straight to a PCB layout. I have definitely experienced the connection issues and EMF problems associated with breadboard layouts that you mention. I think my last order of PCBs from EasyEDA was only like $20 and they arrived in only a few days
Will probably go with a PBC layout so I can use a big ground plane. Do you not use solder with perf boards or do you just de-solder the components when you want to switch?
Starting to look at ordering parts now... I will probably use Vishay-Dale PTF series metal film resistors for all resistors in the signal path. These are 0.1% tolerance (good for R-L channel balance) and I used them in my power amp with good results. However as they are expensive I don't want to use them where they are not necessary.
I'm wondering how much of an effect on sound quality do the 100K (R1, R26, R35, R36, R29, R33, and R34) and 220K (R27 and R28) resistors have? My feeling is not much, but can anyone confirm this? Would like to use cheaper metal film resistors there if possible.
I'm wondering how much of an effect on sound quality do the 100K (R1, R26, R35, R36, R29, R33, and R34) and 220K (R27 and R28) resistors have? My feeling is not much, but can anyone confirm this? Would like to use cheaper metal film resistors there if possible.
I desolder them. There are perfboards with a ground plane, but they are rare. The main issue is that SMD components don't fit well on a classical 2.54 mm-grid perfboard, although 1206 and 0805 components can usually be soldered between one solder island and the next, and other components can be put on an adapter board first - but I guess you have the same problem with your breadboard.
Those resistors are for instrumentation. You can use them, but the requirements for audio transparency are far more lenient. An Agilent audio analyzer can see the difference, where and if it's applicable, but an audio circuit uses low series resistance values and fairly small currents. The 100k reference resistors have even lower current flow and small thermal noise, making their impact negligible.
For the design of a fresh PCB, might I suggest using SMD parts so the source and feedback resistors can be as close to the op amp input pins as possible. Thin film resistors are preferred over thick film types, though the difference is only significant below 100Hz. The same methodology applies to the power supply decoupling capacitors in that they need to be as close to the op amp package as physically practical.
The layout is higher importance than the brand or tolerance of resistors or interstage coupling capacitors in this circuit.
@Palmtrees. Here is perhaps something to consider with regard to Tone control bypass. The basic idea is by Rod Elliott (project 97) and it would allow you to spare one opamp (U15 in schematic from post #135) and the 4-ganged bypass switch (that is for two channels) could be replaced by a simpler DPDT switch or relay. Below is my drawing with changes using the tone control section of your schematic. I am using a small signal DPDT relay in my preamp, makes no noises or clicks at changeover. The price is slightly increased noise level however negligible IMO, especially considering exclusion of one mechanical contact group from the signal path.
Best of luck to you with your project!👍
Best of luck to you with your project!👍
Any thoughts on good power supply decoupling strategies? I was thinking to have a 100 nF X7R and a 47 uF electrolytic cap between each voltage rail and ground of each op-amp. Would this be enough?
One known best practice is to place one 100nF or 1uF directly across the chips' supply pins and 10uF tantalum or aluminum electrolytics (no low-ESR types!) nearby, one from each rail to GND.
I've maybe mentioned this already, for a PCB I'd definitely go 4-layer or even 6-layer with solid rail planes and GND plane, that makes supply and GND distribution both easiest to design and electrically optimal. Then a dozen or so 1uF 25V X7R from either rail to GND locally at all chips will do.
You don't need more that 2-layer for most analog audio, and its usually considerably cheaper. 6-layer is way unnecessary. For mixed analog/digital 4 layer isn't unreasonable, but you can get away without it unless very space constrained. 2-layer is much easier to modify with no inaccesible buried traces/planes. One trick for prototype boards is to use vias without solder resist (not tented) so they can be used to add bodge wires if needed or as test points - this also means vias should be large enough for a suitable hook up wire.
Mark, in some china based vendors the price differential between 2 and 4 layers is less than 10 USD/Euros per 5 boards. I just ordered a 100 x 70 mm layers ( it is a lm317 337 smt power supply and I wanted more copper area for heat dissipation). I was also surprised
I have looked at diverse Gerbers from top designers, all fully analog preamps, and all were two layers. Since the cost differencial is so small 4 layers are tempting even if the noise improvement minimal. I have not layed any 4 layers audio board so I was considering for my next project, using the inner layers for pwr and hand.
I have looked at diverse Gerbers from top designers, all fully analog preamps, and all were two layers. Since the cost differencial is so small 4 layers are tempting even if the noise improvement minimal. I have not layed any 4 layers audio board so I was considering for my next project, using the inner layers for pwr and hand.
Well, you don't need a PCB at all by that token, because point-to-point on solder terminal strips did also work well for decades ;-)
We've heard that same argument 20 years ago when 2-layer got affordable vs. single layer.
Like @Frabor said, the cost penalty for 4/6-layer vs. two or one is already negligible even for mass production let alone a DIY one-off. 6-layers is pretty much ideal for modern analog design when board space might be limited and when high-speed parts are used. The ability to modify the PCB is still fully there with a clever layout.
Well done, I absolutely agree, a very elegant solution. 👍
And to put another inverting buffer in front of that and nothing else is needed except the selector and the input volume potentiometer. And throw out those coupling electrolytes, if he already uses modern OPAs with a small DC offset. And it will sound much better than the original idea. Much much better. With less OPA, he can afford to buy some quality discrete OPA, instead of spending it on a bunch of unnecessary circuits. 😎
Last edited:
When you have two low-ESR caps of very different values, like the 100nF ceramic and a very low ESR electrolytic (newer polymer types, for example) this often forms a ringing tank circuit at MHz frequencies, the 100nF interacts with any trace/wiring inductance plus the intrinsic inductance of the large 'lytic. The impedance as seen by the chip can rise to several ohms at certain frequencies. Things can definitely go wrong here, degrading performance.
A medium ESR (~ a few ohms), OTOH, provides a dissipative path (snubber) for the ringing energy.
IME, Power- and Ground planes completely solve this, provided there are enough "stitching" capacitors all across the board. Those double up as the load capacitors at the output of the voltage regulators which should be on-board and must handle ultra-low ESL gracefully, of course.
I have found a nasty tendency on many datasheets of capacitors, that unless the cap is low ESR, when they proudly display the ESR value, many tend not to include the ESR in the datasheets, making it a guessing game what manufacturers and families of the non-low ESR have a medium ESR, unless a resistor is added, and that may be another guessing game.
I measure the ESR, I mean I always check each element before installation, but I'm not sure how much it will change when the capacitor is under voltage for a while? Especially the ones that have been in the warehouse for a long time.