The input cap is optional for sources that have DC offset. Output cap is a must. Not a problem if the power amp is DC coupled. It is better to have some point where DC loop is broken by a cap. More then once I have seen servo's of DC coupled pre and DC coupled power amp interacting...
Please make sure to use film caps and not ceramic here. 100 pF styroflex/MKP/PPS (in SMD !) are standard values so I would pick those. You made a very good point regarding the filter before the potentiometer !
I think filtering should be done at a lower frequency otherwise it makes not much sense. 1 kOhm and 100 pF ? My calculator is gone and I am fighting with the calculator in Windows 7 so I am taking a guess here.
Please make sure to use film caps and not ceramic here. 100 pF styroflex/MKP/PPS (in SMD !) are standard values so I would pick those. You made a very good point regarding the filter before the potentiometer !
I think filtering should be done at a lower frequency otherwise it makes not much sense. 1 kOhm and 100 pF ? My calculator is gone and I am fighting with the calculator in Windows 7 so I am taking a guess here.
Last edited:
We should retain output cap, but we could probably use two electrolitics in series with negative poles connected to form a bipolar, footprint would be small and filtration very good (my Marantz CD57 mk2 uses two Elna 220uF/16V caps in series at the output). 10uF film cap will be huge, hard to obtain and the pcb will have to "grow".
I would be reluctant to add too high resistor value in series with the input of j-fet that is already at some disadvantage in comparison with bjt when it comes to noise performance. I think that's the reason why Bonsai proposed 100R. In case that we want lower frequency for the low pass filter we could use more capacitance, say 100-220pF polystyrene or polypropilene (footprint will be the same). But, that's for each builder to decide what to use.
For the first version of the pcb I would suggest TH components, j-fet and bjts that is easy to obtain from mainstream distributors. We could always make revision of the pcb that will use SMD. We must consult PMI who will make pcb layout what is his choice.
How about
SMD for all bipolar devices (SOT 23)
TO 92 for the JFET
SMD for the zeners
0.5W metal films for the resistors ( I have a good selection in my parts cupboard!)
Standard TH for caps
For the power dropper resistors feeding the Zener regs, 4W ww types - commonly available and cheap.
Any views?
SMD for all bipolar devices (SOT 23)
TO 92 for the JFET
SMD for the zeners
0.5W metal films for the resistors ( I have a good selection in my parts cupboard!)
Standard TH for caps
For the power dropper resistors feeding the Zener regs, 4W ww types - commonly available and cheap.
Any views?
Footprint for big wire wound resistors should allow ordinary horizontal mount resisitors to be mounted vertically to save pcb space. Vertical mount types are not easy to find.
4 Watt wire wounds ??? This was supposed to be small and we wouldn't use tube stuff isn't it ? 
Separate regs for each channel fulfill same purpose but make better power too. Can be heard at first spot. Simple discrete regs that can handle 50 V input voltage or more are an option too. What is the purpose of a buffer with mediocre power supply feeding an excellent power amp ? Weakest link is weakest link.
BTW Alps RK27 is not very good. On Ebay one can buy DACT 21 stepped attenuators that are way better and cost less. Pads for RK27 allow for DACT 21 to be used.
Better first define design goals more precise...I have more than once made stuff because it needed to be either cheap, too simple or very small that did not fit the bill and were left unused after testing. When DIYing one should go for the best otherwise it has no sense. DIY amp will be more expensive than ready-made anyway. Bonsai, your power amps really deserve good buffer amps.
Maybe design goals should include quality (more on the foreground than size or having certain parts).
Wima MKS2-XL exists in 10 µF. 5 mm pitch and size almost equal to electrolytic caps.... Sounds good, no need for series arrangement or replacement after X time. I have also tried the Vishay 10 µF 63 V as used in VSSA amps. Those are better but they are a tad larger (but not very large). Choosing electrolytic caps in this case is a design error IMO. I will be happy to send you 2 x large value small size film caps for your marantz. I think you will be surprised. This was one of the mods I did a lot in the past with marantz stuff.
Separate regs for each channel fulfill same purpose but make better power too. Can be heard at first spot. Simple discrete regs that can handle 50 V input voltage or more are an option too. What is the purpose of a buffer with mediocre power supply feeding an excellent power amp ? Weakest link is weakest link.
BTW Alps RK27 is not very good. On Ebay one can buy DACT 21 stepped attenuators that are way better and cost less. Pads for RK27 allow for DACT 21 to be used.
Better first define design goals more precise...I have more than once made stuff because it needed to be either cheap, too simple or very small that did not fit the bill and were left unused after testing. When DIYing one should go for the best otherwise it has no sense. DIY amp will be more expensive than ready-made anyway. Bonsai, your power amps really deserve good buffer amps.
Maybe design goals should include quality (more on the foreground than size or having certain parts).
Wima MKS2-XL exists in 10 µF. 5 mm pitch and size almost equal to electrolytic caps.... Sounds good, no need for series arrangement or replacement after X time. I have also tried the Vishay 10 µF 63 V as used in VSSA amps. Those are better but they are a tad larger (but not very large). Choosing electrolytic caps in this case is a design error IMO. I will be happy to send you 2 x large value small size film caps for your marantz. I think you will be surprised. This was one of the mods I did a lot in the past with marantz stuff.
Last edited:
If we want this pcb to be appealing to most potential group buyers it should be practical. It is not supposed to be uncompromising in quality. We see that even the most basic circuit (without cascoding or bootstraping) has distortion 0,000...% and decent drive capability. With separate RC filtering for each psu rail for each channel we should have good noise performance too. I think that is good enough for most diyers.
Practicality criteria are:
1. Cheap (yes, Wima MKS2 is small but it cost 4.- Euro each. Apart from Alps RK27 pot that's more than all semis in the circuit. Is it practical to use such part?)
2. All parts TH. Most builders are not excited about using SMD parts.
3. All parts available from one mainstream distributor (DACT 21 is realtively cheap good quality part, but separate e-bay order just for the pot + all expenses, waiting, etc... not practical. I do not have Pay Pal account so for me that is not option)
PS
It seems that DACT 21 has the same footprint as Alps RK16, which is different from Alps RK27. But using DACT 21 means that we loose important RK 27 property, which is to hold the whole pcb. With DACT 21 we would have to run some kind of wires from the pot to the pcb.
I know how warm is LED resistor in my preamp that drops only 18V. In this circuit we have to drop 30V or more. It's nice to have at least 2W dropping resistor.
Practicality criteria are:
1. Cheap (yes, Wima MKS2 is small but it cost 4.- Euro each. Apart from Alps RK27 pot that's more than all semis in the circuit. Is it practical to use such part?)
2. All parts TH. Most builders are not excited about using SMD parts.
3. All parts available from one mainstream distributor (DACT 21 is realtively cheap good quality part, but separate e-bay order just for the pot + all expenses, waiting, etc... not practical. I do not have Pay Pal account so for me that is not option)
PS
It seems that DACT 21 has the same footprint as Alps RK16, which is different from Alps RK27. But using DACT 21 means that we loose important RK 27 property, which is to hold the whole pcb. With DACT 21 we would have to run some kind of wires from the pot to the pcb.
I know how warm is LED resistor in my preamp that drops only 18V. In this circuit we have to drop 30V or more. It's nice to have at least 2W dropping resistor.
Hi,
I suggest to design for a more common supply voltage range of +-12V to +-15V for TH and +-5V to +-9V for SMD.
Especially SMD would run into heatpower dissipation problems as useable JFETs only come in SOT23 casings allowing not much more than 100mW reasonably.
Anybody using higher rails would be free to add a dropper stage before the circuit anyway.
I´d also suggest to not include the Poti, but to design jumpers instead.
This would leave the user the choice of using any Poti he likes, or to omit with a Poti alltogether.
If one designs a SMD layout it should preferrably be SMD throughout and not mixed up SMD/TH.
The Poti value may be chosen higher than 10k, say 50k.
This would rise the input impedance from <10k to ~33k, thereby loading the source less.
Noisewise there´s no significant loss in noise figure here with higher Poti ohmage.
The input filter as I suggested in #12 filters only >2MHz.
I wouldn´t raise the bandwidth further to keep HF safely out of the circuit.
jauu
Calvin
I suggest to design for a more common supply voltage range of +-12V to +-15V for TH and +-5V to +-9V for SMD.
Especially SMD would run into heatpower dissipation problems as useable JFETs only come in SOT23 casings allowing not much more than 100mW reasonably.
Anybody using higher rails would be free to add a dropper stage before the circuit anyway.
I´d also suggest to not include the Poti, but to design jumpers instead.
This would leave the user the choice of using any Poti he likes, or to omit with a Poti alltogether.
If one designs a SMD layout it should preferrably be SMD throughout and not mixed up SMD/TH.
The Poti value may be chosen higher than 10k, say 50k.
This would rise the input impedance from <10k to ~33k, thereby loading the source less.
Noisewise there´s no significant loss in noise figure here with higher Poti ohmage.
The input filter as I suggested in #12 filters only >2MHz.
I wouldn´t raise the bandwidth further to keep HF safely out of the circuit.
jauu
Calvin
Good ideas. I would also place Poti and buffers + integrated PS at the back of the case in the vicinity of input circuitry and use an axle extension rod. An extra mini toroid (Nuvotem: 10 to 15 Euro) won't break the bank. Why waste energy and convert it in useless heat by feeding the circuit high + and - voltages ? I already see the heat coloring the PCB ....Dropper resistors mean that DC wiring has to made to the front panel which is not practical. Also reliability is in question.
"good enough" is not good enough in DIY audio A film cap has so much advantages that I would not even think of using electrolytic caps in the signal path. I buy the Wimas in the hundreds and then prices go down. If it must be cheap then it has no sense to build it. Prices of parts are decisive in ready made gear and we all know what that leads too.
Sorry but I am out of this thread as I feel I create too much unwanted noise. IMO DCB1 Mezmerize is a more appealing choice as it is also class A, has shunt PS, input relay switching, muting relay and volume control on one single beautiful and relatively small board. Besides that it is not expensive to build. Having built at least 4 of them I can tell it sounds very very good. A practical buffer with many win-win points.
"good enough" is not good enough in DIY audio
A film cap has so much advantages that I would not even think of using electrolytic caps in the signal path. I buy the Wimas in the hundreds and then prices go down. If it must be cheap then it has no sense to build it. Prices of parts are decisive in ready made gear and we all know what that leads too.Sorry but I am out of this thread as I feel I create too much unwanted noise. IMO DCB1 Mezmerize is a more appealing choice as it is also class A, has shunt PS, input relay switching, muting relay and volume control on one single beautiful and relatively small board. Besides that it is not expensive to build. Having built at least 4 of them I can tell it sounds very very good. A practical buffer with many win-win points.
Last edited:
It's very easy to transform this simple project into diyer's nightmare by two simple steps:
1. use hard to obtain expensive parts from many different sources
2. use miniature SMD parts that many diyers find difficult to solder
We do not want to compete with Nagra Kudelski in miniaturization game, don't we?
1. use hard to obtain expensive parts from many different sources
2. use miniature SMD parts that many diyers find difficult to solder
We do not want to compete with Nagra Kudelski in miniaturization game, don't we?
We are having first batch of 250 PCBs produced for our DIY Subbu V3 DAC that is mostly SMD and even some very small parts. Not an easy PCB to solder for inexperienced builders. In fact I discourage new DIYers to join such a project. Yet there are quite a lot of DIYers that do this DAC as a first SMD project... (not my choice). I have seen examples of great workmanship by inexperienced "SMD DIYers" with the previous version. "Hard to get" parts come from various sources as well. PCB is smaller than a packet of cigarettes while having 4 power supplies with decoupling etc.
Please see PCB in the first drawing:
http://www.diyaudio.com/forums/grou...-wm8804-spdif-power-supply-pcb-group-buy.html
I think the "DIYers nightmare" is not everybody's nightmare...if one wants to achieve something then just do it. There is more than enough mediocre stuff around so building something very good for a reasonable amount of money is one of my goals in audio. Building standard or below standard stuff automatically makes the design redundant from the start.
Bye !
Please see PCB in the first drawing:
http://www.diyaudio.com/forums/grou...-wm8804-spdif-power-supply-pcb-group-buy.html
I think the "DIYers nightmare" is not everybody's nightmare...if one wants to achieve something then just do it. There is more than enough mediocre stuff around so building something very good for a reasonable amount of money is one of my goals in audio. Building standard or below standard stuff automatically makes the design redundant from the start.
Bye !
Last edited:
So no SMD - ok clear.
JP, the PSRR of the middle circuit I poster up worst case ( so at LF) is - 90 dB and it gets better as frequency goes up. Now, this is using perfect components, so esr is not modeled for example, but this is way better than an IC reg.
I am in the middle of testing a big new preamp I am building, and spent a lot of effort on the PSU. The humble RC filter, CCS and cascoding can work wonders.
JP, the PSRR of the middle circuit I poster up worst case ( so at LF) is - 90 dB and it gets better as frequency goes up. Now, this is using perfect components, so esr is not modeled for example, but this is way better than an IC reg.
I am in the middle of testing a big new preamp I am building, and spent a lot of effort on the PSU. The humble RC filter, CCS and cascoding can work wonders.
I've seen preamp section psu in Yamaha integrated AX492 consisting of simplest shunt reg (resisitor+zener) and RC filters close to opamps. Measured and subjective level of noise is very low! (Almost all preamps using series three terminal regulators have more psu generated noise.) For the simple circuit like this buffer nothing else is necessary. Only for the circuits that consume much more current we could think of three terminal IC regulators.
In industrial applications for portable electronics SMD is great but all the best sounding gear that I heard use only TH parts. I am surprised that builders are willing to accept such drag in diy audio electronics.
In industrial applications for portable electronics SMD is great but all the best sounding gear that I heard use only TH parts. I am surprised that builders are willing to accept such drag in diy audio electronics.
Last edited:
the lowest noise, highest performance parts of just about any type currently available are SMD parts, the layout benefits are many, shortest signal path, lowest parasitics, most compact and built from reasonably priced commonly available parts available at the major vendors.
soon enough you wont be able to build anything if you keep resisting SMD. most of the highest dollar gear in high end audio, instrumentation, video, computing, avionics, space flight, deep space exploration, medical imaging, the highest performance, lowest noise, fastest 'camera' ever built, with unlimited budget, the LHC sensor arrays that recorded the 'God Particle'.. SMD
nothing to do with portable, though it has advantages there too. its not to save money, not to cut corners, not just to save space, not just to automate assembly, but for higher performance
simple low noise circuits need better power supply in general, i'm not following your logic there.
SMD is really not that hard, but it requires a couple new tools perhaps and some practice for new technique, do not think you have to solder each little pin individually, flux is your friend. I find it much less of a pita than PTH. fine if you dont want to do build with it, but soon enough you will only be able to use NOS, thats the reality of the situation. many of the higher performance discrete parts, even the same dies as were in to92 or similar, are now packaged in SMD only, with the PTH going, or already gone EOL.
it will be better if you just pick a simple, fairly cheap circuit thats easy to troubleshoot, like this one for example … and build it in SMD, stop making excuses caused by reasons other than audio quality. you do want to keep building stuff yes? honestly I cant think of a better starter SMD project to pop your cherry with
like JP, i'm not going to push, I wont hang around, but I post these reminders here on DIYA now and again, as the situation is becoming more and more real
soon enough you wont be able to build anything if you keep resisting SMD. most of the highest dollar gear in high end audio, instrumentation, video, computing, avionics, space flight, deep space exploration, medical imaging, the highest performance, lowest noise, fastest 'camera' ever built, with unlimited budget, the LHC sensor arrays that recorded the 'God Particle'.. SMD
nothing to do with portable, though it has advantages there too. its not to save money, not to cut corners, not just to save space, not just to automate assembly, but for higher performance
simple low noise circuits need better power supply in general, i'm not following your logic there.
SMD is really not that hard, but it requires a couple new tools perhaps and some practice for new technique, do not think you have to solder each little pin individually, flux is your friend. I find it much less of a pita than PTH. fine if you dont want to do build with it, but soon enough you will only be able to use NOS, thats the reality of the situation. many of the higher performance discrete parts, even the same dies as were in to92 or similar, are now packaged in SMD only, with the PTH going, or already gone EOL.
it will be better if you just pick a simple, fairly cheap circuit thats easy to troubleshoot, like this one for example … and build it in SMD, stop making excuses caused by reasons other than audio quality. you do want to keep building stuff yes? honestly I cant think of a better starter SMD project to pop your cherry with
like JP, i'm not going to push, I wont hang around, but I post these reminders here on DIYA now and again, as the situation is becoming more and more real
Last edited:
qusp,
Having never done any smd work how is it that you solder multiple leads without having to heat each junction with an iron? Obviously in mass production they use other methods but how does it work in diy? Does that mean that the boards are already coated with solder and you are only re-flowing the solder? Even then how do you heat multiple connections simultaneously?
Having never done any smd work how is it that you solder multiple leads without having to heat each junction with an iron? Obviously in mass production they use other methods but how does it work in diy? Does that mean that the boards are already coated with solder and you are only re-flowing the solder? Even then how do you heat multiple connections simultaneously?
flux the pads, use a 3-4mm chisel tip, or hoof or well tip; lightly tin the tip; swipe each side, clean up excess or bridges with braid and flux if necessary. with these little sot232 etc parts its dead easy. same with the sot236 or soic8. if your PCB has a decent soldermask and already has a solder layer applied even better, surface tension does most of the work for you.
then you have drag soldering, which takes it another step further with the same idea
HowTo: Drag Soldering Demo - YouTube
I think its this idea that you need to be soldering every single little pin with a fine tip, which is not only the wrolng way to do it, but is also a big part of the fear people have. its the surface tension and flux, with hand soldering and reflow soldering or hot air soldering, that does all the work. dont use too much downward pressure with this technique, just glide along the pins, dont be shy with the solder and flux, you can always clean it up; dont linger, keep moving.
a larger tip is generally preferred to overcome the thermal mass of ground planes etc, using a too small/fine tip makes it much more difficult IMO, the PCB and parts can easily wick all the heat out of the tip and it wont solder/flow properly
then you have drag soldering, which takes it another step further with the same idea
HowTo: Drag Soldering Demo - YouTube
I think its this idea that you need to be soldering every single little pin with a fine tip, which is not only the wrolng way to do it, but is also a big part of the fear people have. its the surface tension and flux, with hand soldering and reflow soldering or hot air soldering, that does all the work. dont use too much downward pressure with this technique, just glide along the pins, dont be shy with the solder and flux, you can always clean it up; dont linger, keep moving.
a larger tip is generally preferred to overcome the thermal mass of ground planes etc, using a too small/fine tip makes it much more difficult IMO, the PCB and parts can easily wick all the heat out of the tip and it wont solder/flow properly
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.