| Bernhard |
Hi,
is there a schematic that uses 4 TDA1541 ?
Is it possible to put more chips parallel ? Like 8 or 16 or 32 or :devilr:
How is that done ?
I have orderd two S1 today, I will getTDA 1541A N2-S1 they said.
Greetings, Bernhard |
|
|
| Netlist |
I recall seeing a few approaches while googling.
/Hugo :) |
|
|
| Bernhard |
My eyes hurt like hell from "googling" the digital section of this forum.
Please ! |
|
|
| dddac |
| quote: | Originally posted by Bernhard
Hi,
is there a schematic that uses 4 TDA1541 ?
Is it possible to put more chips parallel ? Like 8 or 16 or 32 or :devilr:
How is that done ?
I have orderd two S1 today, I will getTDA 1541A N2-S1 they said.
Greetings, Bernhard |
Every DAC with a current output (like 1541, 1543, pcm63 etc) can be paralleled for improved linearity. Just tie the outputs together and decrease the load resistor by the number of dacs you parallel.
There is no limit in theory, but for twice the effect you need to quadruple the # of dacs again and again !!! 1 dac=22Ohm, 4 dac=88Ohm etc
Although I am not 100% sure, as I have not tried, but I suspect, you can piggy back the chips and increase the decouple C's value by the number of dacs again... 4 dacs--> 390nF etc This is an easier way to build....
good luck!
doede |
|
|
| guido |
| quote: | Originally posted by dddac
Every DAC with a current output (like 1541, 1543, pcm63 etc) can be paralleled for improved linearity. Just tie the outputs together and decrease the load resistor by the number of dacs you parallel.
There is no limit in theory, but for twice the effect you need to quadruple the # of dacs again and again !!! 1 dac=22Ohm, 4 dac=88Ohm etc
Although I am not 100% sure, as I have not tried, but I suspect, you can piggy back the chips and increase the decouple C's value by the number of dacs again... 4 dacs--> 390nF etc This is an easier way to build....
good luck!
doede |
Since you say you need to decrease the load resistor, i guess you mean that with 4 DACs the resistor should be 22/4=5.5 ohm.
The reason that you cannot leave the resistor 22 ohm is that it would mean that the voltage over the DAC output would increase by 4. And the 1541 does not like to see a voltage over it's output. Hence the low resistor value in the first place.
So paralleling them is not going to give a lower-gain gainstage, unless you feed halve of them to give an inverted output signal and then add the outputs there.
GuidoB |
|
|
| Bernhard |
Hi Guido,
is it possible to build your 2-TDA dac without GALs ? Using h-cmos logic ?
Bernhard |
|
|
| guido |
Yes!
You could build all with logic, just make the GAL equasions in hardware. * means "AND", + means "OR", := means clocked output (positive edge iirc). But you have to be carefull with delays due to all the ports (causing glitches etc.).
You can also leave out functionality, like no mute (im not using it either, i'm using the 'mute trick' with the 7210.
And you could remove the choice have differential or not. Then the equasions would be smaller and therefore the amount of logic would be less.
You can also go the other way, use more modern progr. logic to put all into one chip...
As i have a GAL programmer, i used it. It is easy (for me) and a bit flexible. E.g. going to look at DEM reclocking some day and then i probably need to extend the GAL logic further to support it.
For me, i now just use 50 resistors per DAC (two per channel, see the other post) and a 1:15 transformer to get ~2V RMS. After the transformer a buffer is required, AD811 there is ****.
GuidoB
PS, the 1543 can cope with voltage at the output, so there the trick to use multiple of them to get enough output signal without gainstage works. |
|
|
| Bernhard |
Thanks Guido,
another questions as I am a newbee to DAC circuits...
It is possible to connect the outputs of two or more 1541 directly together ? So that I could stack let's say 7 ?
How about 7 separate 1541 and a transformer on each output.
Connecting the secondaries in series, so that we get 10Vrms which following op amps still can handle.
That would be a great thing:
- Better performance of relais volume attenuators because of large signal swing.
- Less voltage gain required for power amps.
- Improved linearity because of 7 outputs. Does it matter if it is parallel or in series ? I guess not.
The 2x7 decoupling caps, are they for analog or digital ?
I have "glass-k" caps 100n, gold plated leads and small size, but leads are magnetic.
To split the data signal into left and right, I guess I need that I2S ?
Problem for me, because I use Ultra Curve...
Bernhard |
|
|
| guido |
| quote: | Originally posted by Bernhard
Thanks Guido,
another questions as I am a newbee to DAC circuits...
It is possible to connect the outputs of two or more 1541 directly together ? So that I could stack let's say 7 ?
How about 7 separate 1541 and a transformer on each output.
Connecting the secondaries in series, so that we get 10Vrms which following op amps still can handle.
That would be a great thing:
- Better performance of relais volume attenuators because of large signal swing.
- Less voltage gain required for power amps.
- Improved linearity because of 7 outputs. Does it matter if it is parallel or in series ? I guess not.
The 2x7 decoupling caps, are they for analog or digital ?
I have "glass-k" caps 100n, gold plated leads and small size, but leads are magnetic.
To split the data signal into left and right, I guess I need that I2S ?
Problem for me, because I use Ultra Curve...
Bernhard |
Think some people combine more then one together, have a look how they do it. If you just put all together, the current is times 7, so the resistor needs to be lower times 7 (approx) so the voltage over the DAC output is not too high (silence is -2mA, so with 7 that would be -14mA).
7 transformers is expensive!
The caps should go to analog GND.
Split into l and r is indeed designed for i2s, but other formats should be possible i think. Dont know you source.
I would recommend to start with one and then go to two in the normal way
Greetings,
Gudo
sorry, in a hurry... |
|
|
| Bernhard |
7 transformers is for one channel... :xeye: We have two... :xeye:
But I could get them eventually.
The caps, do they decouple a digital or analog voltage ?
So are audiophile or "digital" caps needed ? |
|
|
| Bernhard |
Got my first two (out of 7 :wiz: ) S1 today :)
And they are not made in Taiwan :D |
|
|
| Elso Kwak |
| quote: | Originally posted by Bernhard
Got my first two (out of 7 :wiz: ) S1 today :)
And they are not made in Taiwan :D |
Hi Bernard,
Why is it I get more and more the feeling the "S1's" and crowns are stamped on much later????:rolleyes:
:confused: :confused: :confused: |
|
|
| Bernhard |
| quote: | Originally posted by Elso Kwak
Hi Bernard,
Why is it I get more and more the feeling the "S1's" and crowns are stamped on much later????:rolleyes:
:confused: :confused: :confused: |
Elso...
it seems to me that you always suspect something...
Tell me why you did not like the OSCON :cannotbe: and I tell you why these are stamped much later :devilr:
Bernhard :clown: |
|
|
| dddac |
| quote: | Originally posted by Bernhard
Thanks Guido,
It is possible to connect the outputs of two or more 1541 directly together ? So that I could stack let's say 7 ?
How about 7 separate 1541 and a transformer on each output.
Connecting the secondaries in series, so that we get 10Vrms which following op amps still can handle.
Bernhard |
Hi Bernhard,
Why nou use the outputs all parallel with a (22/7) 3E3 Rload and use ONE transformer to go to 10 Volt (or whatever voltage)?? You would need a 1:100 transformer though, causing a Zout from 33 kOhm, ok for a tube stage next...
regards
doede |
|
|
| Bernhard |
Doede,
because transformers with such high ratio are not so good, thats what I heard about them.
Anyway I am not decided yet what to do, the only thing I know is to use 7 chips :D
Bernhard |
|
|
| Elso Kwak |
| quote: | Originally posted by Bernhard
Elso...
it seems to me that you always suspect something...
Tell me why you did not like the OSCON :cannotbe: and I tell you why these are stamped much later :devilr:
Bernhard :clown: |
Hi Bernard:
Oscons: leads too short. Letters somewhat faded (used?)
TDA1541: S1 and crown too sloppy printed if compared with the ones in Peters thread. The second crown is on the corner of the ship. Am I too suspicious? I have seen some TDA1541AS1's original from Jocko taken out of Philips players and from my supplier here when they were available. Also had a look into a Marantz CD-80 with a TDA1541AS1 recently.
:xeye: |
|
|
| Bernhard |
| quote: | Originally posted by Elso Kwak
Am I too suspicious? |
Yes. |
|
|
| Bernhard |
The S1 is the same like the A only.
Just the S1 are s(1)elected and thus stamped.
I believe after production they were all tested and if they met specifications as linearity 1/2 LSB, they got the crown.
Same thing with Ortofon pickups.
Know why there are R1 too ? :devilr: |
|
|
| Bernhard |
R1 means r(1)ejected  |
|
|
| tbla |
| quote: | | I get more and more the feeling |
....me too...:crying: |
|
|
| Bernhard |
| quote: | Originally posted by tbla
....me too...:crying: |
I see no problem.
Imagine you shake an apple tree.
The apples that fall down you sort into three categories.
The ones with little animals inside :dead: you stamp R1.
The ones which look acceptable :o you do not stamp at all.
The ones which look very nice :D you stamp S1.
Did you really think you get something much different when you pay a few bucks more for S1 ? :confused:
This ugly millipede in it's el cheapo plastic body... :cannotbe:
They should have made a beautyful ceramic with gold plated leads and stamped it S3 :xeye: |
|
|
| Terry Demol |
| quote: | Originally posted by Bernhard
7 transformers is for one channel... :xeye: We have two... :xeye:
But I could get them eventually.
The caps, do they decouple a digital or analog voltage ?
So are audiophile or "digital" caps needed ? |
Hi Bernard,
I have done a fair bit of development on transformer
I-V in DACs and transformers in general. So I will give
some experiences and hopefully you will not waste
any money :-)
As stated elsewhere here, the higher the ratio the worse
the performance, however this also depends on the tranny
impedance and source impedance. If these are low a high(ish)
ratio tranny can perform well. I have used a Lundahl high
quality 1:5 mic IP transformer for I-V and it was way too
coloured. Nice and warm though, ultimately not satisfying,
a bandaid for poor dac designs.
There is too much information to try and squeez into 1 post,
so your best plan of attack is to go to Jensen transformers
website and download all their mic IP and MC IP tranny
data sheets. Compare the ratios against LF distortion when
working from a specified sorce Z. All jensen trannies are
designed to work from a specified source Z and into a specified
load with a specified snubber. They are also the lowest measuring
distortion trannies currently available.
Since they give full FR and THD data for all transformers
you can comparatively see which designs will be best
suited for your dac.
Work out how many dacs you need to parrallel into what load
to give a certain voltage swing. This load will be the source
Z driving tranny. Also remeber that for TDA1541, it is designed
to work into 0 ohm load, so the more voltage swing the more
the dac will produce additional distortion. It's a balancing act.
I have done all this many times and have basically come to
the conclusion that it is very difficult to design an optimum
transformer I-V that REALLY works well, however it can be done.
Do the research, get aquainted with the parameters and
devices and then get back to us. Right now you are making
guesses at best and will not get the best result.
As an end note, if you want super transparency, the only
transformer I would think of using is a Jensen or possibly a
Stevens and Billington. I haven't had any experience with S & B,
however I know Thorsten has used them so maybe speak to him
which one is their best and whether it is suited to your
application. If you look at the price of Jensens best transformers
you will get a shock, but this is not the place to be penny
pinching.
Good luck,
Terry |
|
|
| Bernhard |
Some Jensens from catalog is too expensive, I will look for surplus stuff, have seen pieces that were made for Rhode&Schwarz...
Another question: Is a double sided pcb with upper layer as groundplane necessary or can I make p2p wiring ?
Bernhard |
|
|
| Bernhard |
Is it possible to implement a digitalfilter / oversampling ?
Any schematic around ?
I read that the non-os versions of TDA1541 could make problems with solid state amps and it is recommended to use them only with tube amps, because RF...
Also sine waves look like staircase...
Bernhard |
|
|
| jean-paul |
| quote: | | I read that the non-os versions of TDA1541 could make problems with solid state amps and it is recommended to use them only with tube amps, because RF... |
Just try non os and you'll discover that it isn't that bad although when you measure the DAC you'll fall of your chair.
I was worried about that too but never experienced desastrous effects.
BTW R1 means R(elaxed ) specs IIRC. That's the common explanation anyhow. Rejected parts wouldn't have made it onto the market. Not with Philips in that time. Rejected parts are usually thrown away. |
|
|
| Terry Demol |
| quote: | Originally posted by Bernhard
Is it possible to implement a digitalfilter / oversampling ?
Any schematic around ?
I read that the non-os versions of TDA1541 could make problems with solid state amps and it is recommended to use them only with tube amps, because RF...
Also sine waves look like staircase...
Bernhard |
If you are using transformer I-V that will be sufficient
LPF. Even into SS amps.
Cheers,
Terry |
|
|
| Bernhard |
Ok, thanks, still if there is a known schematic for TDA oversampling...
For sure R1 specs in the Philips data sheets are worst of all 1541A.
Without R1 is better. |
|
|
| jean-paul |
Actually one item of the specs is better for R1. Forgot which one.
| quote: | | Ok, thanks, still if there is a known schematic for TDA oversampling... |
Sure there are, they were used with SAA7220P/B ( 4 x fs ) digital filter as a standard in millions of cdplayers of all kinds and brands.
Just find a SAA7220P/B ( Google for the datasheet ), check a schematic of any cdplayer with TDA1541 and the SAA and connect the CS8412 to 3 right pins of the SAA7220P/B...Just like you do it now to the TDA. Feed the SAA with a sturdy supply as it throws out garbage on the supply lines. Do a Search for that here for more info.
I don't know if a SAA7220P/B can drive 4 TDA's. Silly plan IMO, why would you use 4 or more of these excellent chips *and* use oversampling to degrade the results !?!?!
There are not much alternatives for SAA7220P/B unfortunately. I quit searching for one. |
|
|
| rfbrw |
| quote: | Originally posted by jean-paul
There are not much alternatives for SAA7220P/B unfortunately. I quit searching for one. |
Various NPC SM584x
TI DF1704/6 |
|
|
| jean-paul |
| I should have specified I2S in, I2S out types. |
|
|
| rfbrw |
| quote: | Originally posted by jean-paul
I should have specified I2S in, I2S out types. |
A little logic and you are in business. |
|
|
| Bernhard |
Can I use 250Vdc rated cap for the tube DAC output stage ?
I have so beautiful 1µ/250V Siemens MKY hermetically seald :bawling: |
|
|
| Bernhard |
| Is bigger better ? Can I use 1µF if there is no space problem ? |
|
|
| Bernhard |
Seems that I will not use 7 but only 4 chips, so I have still the option to go balanced.
Maybe upgrade to 8 later.
This is my first try to get a DAC board layout, there is a ground plane on the upper side.
Just no space left for the TL431s, is it ok to put them on another board ?
PSU decoupling caps will still be on the DAC boad.
Or I put the TL431s on the backside.
The white caps are stacked 3 on 4, MKP 1839 1µF 1%, .
 |
|
|
| tbla |
mr. walt jung ended up with 1uF - as best compromise....... |
|
|
| Bricolo |
| quote: | Originally posted by tbla
mr. walt jung ended up with 1uF - as best compromise....... |
Did he also test different cap types? And what was the result? |
|
|
| tbla |
| 1uF polycarbonat - its in one of the pooge articles...... |
|
|
| Bricolo |
| quote: | Originally posted by tbla
1uF polycarbonat - its in one of the pooge articles...... |
I'm downloading all the articles on his website. But I'm not sure if the pooge articles are on it. |
|
|
| Bernhard |
4 pre-regulators for 5V -5V -15V for TDA1541 and 12,6V for tube heaters.
12 TL431 regulators for TDA1541.
 |
|
|
|
