Well, my decisions:
No reclocker on the board. The I2S lines will be cut and I will put probably 2 row 10- or 8-pin header footprint there..." but none is supposed to use them actually, everything can still be made with solder (which is normally better as a final solution). I will probably make the additional mounting holes in the board to make easier mounting of the reclocker and the holes will match to those on Elso’s reclocker board. I will also make soon one daughter board with reclocker driven by the can oscillator.
So, CS and TDA will be positioned a bit closer but generally similar as in the previous board and I/V stage will, most generally speaking, take the physical position of the reclocker. This means that the final dimensions of the board will be similar. The previous one was 12.5x12.5cm; if this is important for anyone, its green color is in the reality fresher but my scanner can not capture it.
For the board of such dimensions at this level of interest the price will be about 25EUR, assuming I will not have much troubles and reasons to go again and again with the prototypes.
I have yet to see what will be the analog stage (it won’t change the price anyway).
Pedja
No reclocker on the board. The I2S lines will be cut and I will put probably 2 row 10- or 8-pin header footprint there..." but none is supposed to use them actually, everything can still be made with solder (which is normally better as a final solution). I will probably make the additional mounting holes in the board to make easier mounting of the reclocker and the holes will match to those on Elso’s reclocker board. I will also make soon one daughter board with reclocker driven by the can oscillator.
So, CS and TDA will be positioned a bit closer but generally similar as in the previous board and I/V stage will, most generally speaking, take the physical position of the reclocker. This means that the final dimensions of the board will be similar. The previous one was 12.5x12.5cm; if this is important for anyone, its green color is in the reality fresher but my scanner can not capture it.
For the board of such dimensions at this level of interest the price will be about 25EUR, assuming I will not have much troubles and reasons to go again and again with the prototypes.
I have yet to see what will be the analog stage (it won’t change the price anyway).
Pedja
Just press the "www" button that appears under Pedja's posts! 
BTW I'm still collecting parts to populate the two boards you sent me (I had to order a bunch of transformers!) - one is still untouched - I have to read the whole thread to see if the newer version will be that better to feel the need of an upgrade
BTW the onboard I/V will be discrete or based on AD844?
Cheers
Andrea
PS I only could find OPA660 (actually 3 of them) so I'll try Thorsten's I/V (and even his tube output stage) to see which sound best.
BTW I'm still collecting parts to populate the two boards you sent me (I had to order a bunch of transformers!) - one is still untouched - I have to read the whole thread to see if the newer version will be that better to feel the need of an upgrade
BTW the onboard I/V will be discrete or based on AD844?
Cheers
Andrea
PS I only could find OPA660 (actually 3 of them) so I'll try Thorsten's I/V (and even his tube output stage) to see which sound best.
This is interesting, I was not aware of that the transformer could be a problem.Andypairo said:
I can order locally the transformers with any windings I want for quite modest price and I found lately one manufacturer where the additional windings do not raise the price significantly. I have used previously one transformer for the digital part and another one for the analog stage but I believe also that the analog stage can be fed from the same transformer (but from the independent windings) without much of compromise.
However, as I do not have an intention to sell the kits or any part for this project, I recommend checking the solution for the transformer(s). Ideally, the digital part, excluding a reclocker which might be additionally hooked to the line which feeds, say, the digital part of the CS8412, needs 4 secondary windings of 7VAC and one of 22VAC. I/V in the case of AD844 needs something like 2x15V or 2x18V, probably a bit higher voltage if it will be discrete.
Still don't know. Will have some info about this in a few days.
Pedja
Pedja, what is the VA rating on the transformers you're currently using? In North America, it shouldn't be too hard to obtain these custom transformers from a vendor such as Victoria Magnetics.
Anyone know of any vendors for the TDA1541A? I have one that I've been saving for such a project, but wouldn't mind building two.
Anyone know of any vendors for the TDA1541A? I have one that I've been saving for such a project, but wouldn't mind building two.
The bigger one, used for the digital part, is 100VA rated. Five secondary windings are (actually) 8V and two of 1A are feeding CS, two of 1.5A are feeding TDA’s digital part and one of 2A is feeding the reclocker. Secondary winding that feeds the analog part of the TDA is 22V, 2A.bzo said:
The smaller one is 50VA, 2x18V and feeds the AD844 analog stage.
I do not know if this info is any helpful for you but my source (via local distributor) was Reichelt, Germany.
Now the price of 25EUR is not in question. The price for shipping by the registered mail for the European countries is about 8EUR and it is probably the same whether the one or two boards are sent. It should not be much higher for the rest of the world.Rotellian said:
Yes, I2S protocol will be used and lines will be cut and will have the pads.
Stay tuned for a few days.
As I dealt with some issues the previous time I will now just move the solutions. It should not take much time to do the rest. Will see...machinow said:
Pedja
Jean-Paul, I saw you have mentioned in the other thread latch-up problem if separate supplies are used for the analog and digital side of CS8412. I always used separate supplies and never had a problem with it. Did you, or anyone else, have? I could put in two just_in_case diodes relatively easily considering the space but I am simply not convinced this can really be the issue and I just do not like to put anything not really needed (problems are almost always around the corner).
Pedja
Pedja
I/v
To be clear from the start, I like what AD844 used as a common base stage delivers soundwise and I like the simplicity of its implementation. What bugged me was the AD844’s harmonic distortion figure. Its internal circuit is just fine but for such an application somewhat higher bias would result in the lower distortion. So I tried similar discrete circuit, i.e. symmetrical diamond transistor with its “emitter” running at some 3.5mA. This circuit’s measured harmonic distortion at 1kHz/0dBFs is 0.012% and it is almost entirely the 2nd harmonic (about -79db below the signal); else than barely visible 4th harmonic the rest was below the noise floor of the measurement setup at about -110dBFs. With the -6dBFs (50%) signal the distortion drops to 0.006%. I will post the circuit and graphs to my site soon. Btw, this stage surely might be made for somewhat better specs.
<review> Sonically, I would rather talk about the different results than about one being preferred to the other. This lower distortion seems to really sound cleaner and one could find better resolution as well, but I doubt that anyone would say that the AD844 per se is not clean. Because of the less grain, the discrete stage was better for female voices (Diana Krall, Laura Fygi), though I doubt that anyone would find the AD844 per se being grainy. At the other hand I would certainly rather listen to Robbie Robertson (Contact From The Underworld Of Red Boy ) using AD844. The AD844 was generally more dynamic and fuller to me. </review> There could be more reasons for this, for instance the presence of the output buffer or slightly different supply, but I can not (once again) escape the feeling that there is the truth in the claims the monolithic is generally preferable “layout” to the discrete one. I could now proceed building/tweaking this discrete stage but I am afraid it might finish with the project which will be indigestible for some. If anyone can urge to guys in Analog to make the same or similar chip running at bit higher current, the result might be all we ever needed about the I/V… (I had a post to National asking them if they have a plan to make one chip with externally available transconductance node using the circuit of or like the LM6181 and with somewhat higher current but, despite of the fact I am registered at their site for a few years, I got the reply with the questions about my company, expected production volume etc.
– if I’ll have one day enough nerves I will proceed with this correspondence.)
Finally, I should mention that the distortion of the AD844 could be better realized if it is considered as a form of clipping and it decreases as the signal level decreases; the graph you can see in my site is done using 0dBFs signal. With -6dBFs (50%) the harmonic distortion is 0.09% and with -12dBFs (0.25%) it is below 0.04% comprising mainly 2nd and 3rd harmonic.
So, this is the situation:
Contra AD844:
- Measured distortion is higher than it is usual for the solid state circuits.
Pro AD844:
- Regardless of the measured distortion it sounds excellent as it is.
- Easy to put together.
- The layout of this I/V will be compatible with classic I/V stage based on any applicable single opamps, so anyone who considers the specs more important than I do, can achieve them quite easily.
- Can drive almost any load. You can use it without its internal buffer as well. Though I liked it better with it.
Contra discrete:
- Adds to complexity.
- Additional buffer adds to complexity even more.
Pro discrete:
- Measures well still being common base based I/V.
Hence I am planning to go with AD844 for this board. Once I might make and offer another board with a discrete I/V but that will not happen in the next half year and probably not this year at all and this only might happen, no promise of it.
____________________________________
As about the protection diodes for CS8412, I’ll put the pads so anyone who considers them necessary can use them. The active divider’s decoupling caps of TDA will be as in the previous board, WIMA MKP10 0.1u/250V, and the same can be used in CS’s filter (there will be enough space here for 0.22u/250V if one wants to go with wildmonkeysects solution). MSB decoupling caps footprints are somewhat fatter and are intended for WIMA MKS4 2.2u/63V. Output caps will be Black Gate.
So, that is the conception. I am starting with drawing and hopefully will finish it till the end of the week or so. Then it will take about 10 days till I receive the prototype boards. Volunteer or two ready to test the prototype can contact me on react_71@yahoo.com and have it for somewhat smaller price, but the offer is valid only for those able to accomplish it for a reasonably short time.
Pedja
To be clear from the start, I like what AD844 used as a common base stage delivers soundwise and I like the simplicity of its implementation. What bugged me was the AD844’s harmonic distortion figure. Its internal circuit is just fine but for such an application somewhat higher bias would result in the lower distortion. So I tried similar discrete circuit, i.e. symmetrical diamond transistor with its “emitter” running at some 3.5mA. This circuit’s measured harmonic distortion at 1kHz/0dBFs is 0.012% and it is almost entirely the 2nd harmonic (about -79db below the signal); else than barely visible 4th harmonic the rest was below the noise floor of the measurement setup at about -110dBFs. With the -6dBFs (50%) signal the distortion drops to 0.006%. I will post the circuit and graphs to my site soon. Btw, this stage surely might be made for somewhat better specs.
<review> Sonically, I would rather talk about the different results than about one being preferred to the other. This lower distortion seems to really sound cleaner and one could find better resolution as well, but I doubt that anyone would say that the AD844 per se is not clean. Because of the less grain, the discrete stage was better for female voices (Diana Krall, Laura Fygi), though I doubt that anyone would find the AD844 per se being grainy. At the other hand I would certainly rather listen to Robbie Robertson (Contact From The Underworld Of Red Boy ) using AD844. The AD844 was generally more dynamic and fuller to me. </review> There could be more reasons for this, for instance the presence of the output buffer or slightly different supply, but I can not (once again) escape the feeling that there is the truth in the claims the monolithic is generally preferable “layout” to the discrete one. I could now proceed building/tweaking this discrete stage but I am afraid it might finish with the project which will be indigestible for some. If anyone can urge to guys in Analog to make the same or similar chip running at bit higher current, the result might be all we ever needed about the I/V… (I had a post to National asking them if they have a plan to make one chip with externally available transconductance node using the circuit of or like the LM6181 and with somewhat higher current but, despite of the fact I am registered at their site for a few years, I got the reply with the questions about my company, expected production volume etc.
– if I’ll have one day enough nerves I will proceed with this correspondence.)Finally, I should mention that the distortion of the AD844 could be better realized if it is considered as a form of clipping and it decreases as the signal level decreases; the graph you can see in my site is done using 0dBFs signal. With -6dBFs (50%) the harmonic distortion is 0.09% and with -12dBFs (0.25%) it is below 0.04% comprising mainly 2nd and 3rd harmonic.
So, this is the situation:
Contra AD844:
- Measured distortion is higher than it is usual for the solid state circuits.
Pro AD844:
- Regardless of the measured distortion it sounds excellent as it is.
- Easy to put together.
- The layout of this I/V will be compatible with classic I/V stage based on any applicable single opamps, so anyone who considers the specs more important than I do, can achieve them quite easily.
- Can drive almost any load. You can use it without its internal buffer as well. Though I liked it better with it.
Contra discrete:
- Adds to complexity.
- Additional buffer adds to complexity even more.
Pro discrete:
- Measures well still being common base based I/V.
Hence I am planning to go with AD844 for this board. Once I might make and offer another board with a discrete I/V but that will not happen in the next half year and probably not this year at all and this only might happen, no promise of it.
____________________________________
As about the protection diodes for CS8412, I’ll put the pads so anyone who considers them necessary can use them. The active divider’s decoupling caps of TDA will be as in the previous board, WIMA MKP10 0.1u/250V, and the same can be used in CS’s filter (there will be enough space here for 0.22u/250V if one wants to go with wildmonkeysects solution). MSB decoupling caps footprints are somewhat fatter and are intended for WIMA MKS4 2.2u/63V. Output caps will be Black Gate.
So, that is the conception. I am starting with drawing and hopefully will finish it till the end of the week or so. Then it will take about 10 days till I receive the prototype boards. Volunteer or two ready to test the prototype can contact me on react_71@yahoo.com and have it for somewhat smaller price, but the offer is valid only for those able to accomplish it for a reasonably short time.
Pedja
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.