| Lucas_G |
Dear Dac builders,
I had completed the Doede TDA 1543 DAC a while ago.
http://www.dddac.de/ma_dac21.htm
However, It does not what it is supposed to do.
It is quiet when no digital SPDIFF signal comes in, but as soon as I feed it with some digital data, it starts making a high frequency noise, in which you can faintly hear some parts of the music in the background. The louder the digital signal, the louder the fuzzy noise...:cannotbe:
I have tried everything I could in the past week.
I also asked Doede some questions, but he prefers not to go into detail - which is his good right, of course. He never promised anybody to give support on his design....:apathic:
Now, I am very much hoping that someone else can give advice.
All measurements with my multimeter seem to be ok.
I get 5.04 Volts on VBD and VBA1.
With VBAdjust on 1.25 kOhm I get 7.5 Volt at VBA2.
Here is the scheme:
http://www.dddac.de/pics/dddac1543/pcb1543.gif
I have two questions:
1. In some older schemes, Doede mentions 10 nF caps at the SPDIFF In and Ground. On later schemes he uses 100 nF caps (so that is what I used). Could it be that the use of 100 nf instead of 10 nF is responsible for the fuzz?
2. I did solder all the chips straight in the pcb (except for the Tent clock). I used a 20 Watt soldering iron, and I worked quickly...:smash: Is the CS8412 extremely sensitive to heat? :eek:
Any suggestions extremely appreciated!!!!:att'n:
Regards,
Lucas |
|
|
| Taco |
The CS8412 is not very sensitive to heat, the surface mount version CS8414 survived my punishment with a 25W solderingiron.
The change from 10nF to 100nF isn't your problem I think. But how many TDA1543 you're using?
-I would suggest to try things first without the extra clock. So connecting the CS8412 directly to your DAC's.
-Also what is your spdiff connection, optical or electrical?
-is the output of pin 26 of the CS8412 steady? (data line) |
|
|
| Peter Daniel |
| I observed similar behaviour yesterday in my DAC, when the ground connection from external clock to the DAC board was missing. After connecting ground it was fine. |
|
|
| Duck-Twacy |
| But I haven't even started yet? :confused: |
|
|
| peranders |
| quote: | Originally posted by Lucas_G
2. I did solder all the chips straight in the pcb (except for the Tent clock). I used a 20 Watt soldering iron, and I worked quickly...:smash: Is the CS8412 extremely sensitive to heat? :eek:
|
Heat sensitive? More than usually for an IC? Don't think so. It's plain silicone. 250 deg C for 10 seconds in owen is OK.
I would not very heat sensitive :nod: |
|
|
| Lucas_G |
| quote: | Originally posted by Taco
The CS8412 is not very sensitive to heat, the surface mount version CS8414 survived my punishment with a 25W solderingiron.
The change from 10nF to 100nF isn't your problem I think. But how many TDA1543 you're using?
-I would suggest to try things first without the extra clock. So connecting the CS8412 directly to your DAC's.
-Also what is your spdiff connection, optical or electrical?
-is the output of pin 26 of the CS8412 steady? (data line) |
Hi Taco,
I am using 8 dacs in a tower. I could adjust the voltage to 8.5 V and the output CD to 3.5, just like Doede recommends...
I tried both ways. I have a switch with which I can toggle all three switches at once: either all three closed or open.
I use an electrical SPDIFF. The inner pole connected to the SPDIFF in pin. The outer pole connected to the D-gound, next to the 12 Volt plus in for the Digital domain. I also fitted a 75 Ohm resistor between these two poles.
How can I know whether the ouput of pin 26 of the CD8412 is steady? Can I meusure it?
Thanks!
Lucas. |
|
|
| Taco |
If you measure a stable voltage without great fluctuations then it's oke.
But like Peter said it could also be a ground problem. Have you measured if all the components that should be connected to ground have a low resistance? Are the digital and analog ground connected together?
And have you bypassed the clock, to eliminate that possibility? |
|
|
| Cobra2 |
Your DAC-chips are overheating, use a fan and / or reduce voltage to 7+...
Arne K |
|
|
| finneybear |
I think it's overheating, too. Also, check the way you built the DAC tower. Did you stick copper foil to the top of each dac and ground them all together?
I have Doede's board, too. Still have not started the work yet. Am thinking a good way to build the tower. Probably I will make my own PCB. Piggyback two 1543 together as a set then mount them on both sides of the PCB. Two sets on top, two sets on the bottom. |
|
|
| Lucas_G |
| quote: | Originally posted by finneybear
I think it's overheating, too. Also, check the way you built the DAC tower. Did you stick copper foil to the top of each dac and ground them all together?
I have Doede's board, too. Still have not started the work yet. Am thinking a good way to build the tower. Probably I will make my own PCB. Piggyback two 1543 together as a set then mount them on both sides of the PCB. Two sets on top, two sets on the bottom. |
I took small pieces of aluminium (6 mm wide, 10 mm long, 2 mm thick) which I glued on top of each 1543. This gave me a tower of appr. 7 cm lenght. On the sides I glued two strips of 10 mm width and 10 cm lenght.
Actually it does cool quite well. I stick two larger heatsinks on the sidebars for extra cooling. Then the temp stays well below 40 degrees Celsius.
If it were a matter of heat, it should perform well during the first minutes, shouldn't it?
I am now also contemplating to make my own pcb (a kind of veroboard) with p2p wiring. Your idea of 4 times 2 tda's appeals to me...
Also, I would never again solder the chips straight on the PCB; When there are problems (such as I have now) one keeps asking himself whether the chips themseves are still ok...:bawling:
Regards,
Lucas. |
|
|
| Lucas_G |
There is maybe one thing I should mention:
I have ordered BG caps, but since delivery takes some time, I used som cheap quality 20uF standard (or worse) quality caps. Could they be too bad to prevent oscillating? Or am I already too much influenced by the Blackgate-Mob?:D
Lucas |
|
|
| Jax |
I don't like the grounding on this board.
The return path for the 3 digital signals to the 1543 goes all the way to the battery and back to the board again. This is bad design, it may work but will depend on how the ground connections are routed.
I would try to remove GNDA1 and GNDA2 and only use GNDD from the power supply.
Connect a short wire between pins 18 and 21 on the 8412. This is the single point connecting GNDD and GNDA1 together.
Connect a wire from the 1543 pin 4 to the jumper with GNDD on it going to the XO and 4040.
It's easy fixes and can be undone. Give it a try.
Other things to consider is adding 100n ceramic SMD capacitors on the 8412, one between pins 7 and 8 and one between pins 16 and 17. The 20u is not a capacitor at high frequencies, BG or not. |
|
|
| guido |
Are you sure you are using TDA1543 and not TDA1543A?
Just a thought.
If the DAC's are stone cold when you power up the dac and it goes wrong right away, it can't be overheating i guess.
Replace the DACs with just one and see what happens.
GuidoB |
|
|
| Fabian |
Lucas,
I've also built Doede's dddac. In the beginning all I could hear was a lot of noise but almost no music. Quite similar to your description. It took me more than a day to find the problem: LT1085 just didn't work (although it measured well). After replacing with a LM317T it worked fine.
BTW: Doede recommends 3.85V, not 3.5V as you mentioned in post #6
Fabian |
|
|
| Jax |
LT1085 needs at least 150uF on the output if a bypass capacitor is used on the adjust pin. A 10uF is also needed on the input pin to ground since the distance to the source is significant. Else it may oscillate.
LM317 is more stable but noisier and need no capacitor at all on the output. |
|
|
| jewilson |
You should be using a buffer somthing like a 74ACT244 to drive all those DACs. That's to much of a load on the 8412.
Also, check to make sure you have set modes M0,M1,M2, M3 correctly. The can change with different DACs.
Also, change filter cap on the 8412 to a .047uf film cap, don't use an electrolytic there. |
|
|
| finneybear |
Lucas, if the problem existed already when the DACs were cool, then I am running out of clue as well. Probably you just have to try the simple but dumb way... replace the parts one by one to see how it goes.
The Doede board looks good to me already. Yet if I have to make a new board myself, definitely I will add more buffers to the 8412 and reroute the ground.
The 4x2 idea goes this way: A is for the top, B for the bottom. The arrangement looks like: ABAB. This will ensure any two adjacent sets will always share one side to have the shortest signal path. Now I will have to come out a good way to connect the right signals together. Probably through p2p shielded silver wires? Another good thing about this topolgy is that there will be no signals go between the pins. I can put ground lines around them instead.
It will be easy to install heat sinks as well. Probably no need for a fan anymore. |
|
|
| Elso Kwak |
| quote: | Originally posted by Lucas_G
Dear Dac builders,
I had completed the Doede TDA 1543 DAC a while ago.
http://www.dddac.de/ma_dac21.htm
However, It does not what it is supposed to do.
It is quiet when no digital SPDIFF signal comes in, but as soon as I feed it with some digital data, it starts making a high frequency noise, in which you can faintly hear some parts of the music in the background. The louder the digital signal, the louder the fuzzy noise...:cannotbe:
I have tried everything I could in the past week.
I also asked Doede some questions, but he prefers not to go into detail - which is his good right, of course. He never promised anybody to give support on his design....:apathic:
Now, I am very much hoping that someone else can give advice.
All measurements with my multimeter seem to be ok.
I get 5.04 Volts on VBD and VBA1.
With VBAdjust on 1.25 kOhm I get 7.5 Volt at VBA2.
Here is the scheme:
http://www.dddac.de/pics/dddac1543/pcb1543.gif
I have two questions:
1. In some older schemes, Doede mentions 10 nF caps at the SPDIFF In and Ground. On later schemes he uses 100 nF caps (so that is what I used). Could it be that the use of 100 nf instead of 10 nF is responsible for the fuzz?
2. I did solder all the chips straight in the pcb (except for the Tent clock). I used a 20 Watt soldering iron, and I worked quickly...:smash: Is the CS8412 extremely sensitive to heat? :eek:
Any suggestions extremely appreciated!!!!:att'n:
Regards,
Lucas |
Hi Lucas,
Just a short advice if not already done:
Check all supply voltages.
Check with a scope if BCK, WS, and DATA are present at outputs on the CS8412 when playing a CD.
If you have a lock light see or measure if the CS8412 is locked to the datastream. Check very carefully if you have the right mode for the CS8412. TDA1543 needs mode#2 I2S outputformat. (table 3 in the datasheet)
Also check as already suggested if the DAC works without the "reclocking" The latter requires also special setting of the CS8412. Again see the datasheet.
Hope this helps.
;) :cool: |
|
|
| Lucas_G |
| quote: | Originally posted by Jax
I don't like the grounding on this board.
The return path for the 3 digital signals to the 1543 goes all the way to the battery and back to the board again. This is bad design, it may work but will depend on how the ground connections are routed.
I would try to remove GNDA1 and GNDA2 and only use GNDD from the power supply.
Connect a short wire between pins 18 and 21 on the 8412. This is the single point connecting GNDD and GNDA1 together.
Connect a wire from the 1543 pin 4 to the jumper with GNDD on it going to the XO and 4040.
It's easy fixes and can be undone. Give it a try.
Other things to consider is adding 100n ceramic SMD capacitors on the 8412, one between pins 7 and 8 and one between pins 16 and 17. The 20u is not a capacitor at high frequencies, BG or not. |
Hi Jax,
I see your point regarding grounding.
Doede wanted to remain all options open for using several powersupplies...
But I will try your suggested fix.
I have only 100nF polyester at hand. Is a ceramic obligatory?
Regards,
Lucas |
|
|
| Lucas_G |
| quote: | Originally posted by guido
Are you sure you are using TDA1543 and not TDA1543A?
Just a thought.
If the DAC's are stone cold when you power up the dac and it goes wrong right away, it can't be overheating i guess.
Replace the DACs with just one and see what happens.
GuidoB |
Hi Guido,
It is the TDA1543N (ordered from Reichelt).
It is bit difficult to replace the DAC-tower with just one; I made a stable tower with the help of heat-paste-glue...!:dead:
But I may have to maltreat the board anyway...:bawling:
Regards,
Lucas |
|
|
| Lucas_G |
| quote: | Originally posted by Fabian
Lucas,
I've also built Doede's dddac. In the beginning all I could hear was a lot of noise but almost no music. Quite similar to your description. It took me more than a day to find the problem: LT1085 just didn't work (although it measured well). After replacing with a LM317T it worked fine.
BTW: Doede recommends 3.85V, not 3.5V as you mentioned in post #6
Fabian |
Hi Fabian,
Now that is some news!:xeye:
I will order some LM317s too.
Regards,
Lucas
| quote: | Originally posted by Jax
LT1085 needs at least 150uF on the output if a bypass capacitor is used on the adjust pin. A 10uF is also needed on the input pin to ground since the distance to the source is significant. Else it may oscillate.
LM317 is more stable but noisier and need no capacitor at all on the output. |
Hi Jax,
So this expensive LT1085 is pretty difficult ot handle...:att'n:
I will try your solution too. If two persons point at the 1085 being the problem, it is worthwile to try this.
Thanks.
Lucas |
|
|
| Lucas_G |
| quote: | Originally posted by jewilson
You should be using a buffer somthing like a 74ACT244 to drive all those DACs. That's to much of a load on the 8412.
Also, check to make sure you have set modes M0,M1,M2, M3 correctly. The can change with different DACs.
Also, change filter cap on the 8412 to a .047uf film cap, don't use an electrolytic there. |
Hi Jewilson,
It DOES work with 8 TDA's in Doede's own setup...:bigeyes:
Lucas |
|
|
| Lucas_G |
| quote: | Originally posted by finneybear
Lucas, if the problem existed already when the DACs were cool, then I am running out of clue as well. Probably you just have to try the simple but dumb way... replace the parts one by one to see how it goes.
The Doede board looks good to me already. Yet if I have to make a new board myself, definitely I will add more buffers to the 8412 and reroute the ground.
The 4x2 idea goes this way: A is for the top, B for the bottom. The arrangement looks like: ABAB. This will ensure any two adjacent sets will always share one side to have the shortest signal path. Now I will have to come out a good way to connect the right signals together. Probably through p2p shielded silver wires? Another good thing about this topolgy is that there will be no signals go between the pins. I can put ground lines around them instead.
It will be easy to install heat sinks as well. Probably no need for a fan anymore. |
Hi Finnybear,
Yes, I guess I will alaso have to exchange the 8412...:cannotbe:
Desoldering the current one will be a nice job...
BTW, with the aluminium tower that I constructed, you can also get away without a cooling fan, as long as you attach some decent heatsinks (3 x 5 x 1 cm) on the sides of it.
Regards,
Lucas |
|
|
| rfbrw |
Can you test the CS8412 on its own? If you have a LED connected to pin 25 it will go out when the CS8412 locks to the data.
ray |
|
|
| jewilson |
| quote: | | It DOES work with 8 TDA's in Doede's own setup |
Lucas_G
I do not know what the input impedance of the DAC looks like now, but you have severely reduced the input impedance by a bunch. Crystal has not rated the output current for the clock and data outputs on this part, however you should not be driving this many gates from the 8412, it is just a poor design practice to do that.
Next, issue with this implementation you are slowing the down both devices due to the load. This will reduce the rise time at the DAC's input and can make the converter more susceptible to jitter.
Try adding a buffer there; you might be surprised to here the difference. Also, ground the input to the unused gates. |
|
|
| finneybear |
| Lucas, I am afraid that the heat is still a concern. The chip may be a bit cooler outside but the internal junction temerpature may be still too high. DAC chips performance is very sensitive to heat actually. Since I will redo the PCB, the new 4x2 arrangement will give me some edge on both heat dissipation as well as signal feed quality. I want to create a topology such that the signal paths from 8412 to the 8 1543 all have equal length. A buffer or two between them. The output of the reclock circuit is a concern, too. |
|
|
| Lucas_G |
| quote: | Originally posted by finneybear
Lucas, I am afraid that the heat is still a concern. The chip may be a bit cooler outside but the internal junction temerpature may be still too high. DAC chips performance is very sensitive to heat actually. Since I will redo the PCB, the new 4x2 arrangement will give me some edge on both heat dissipation as well as signal feed quality. I want to create a topology such that the signal paths from 8412 to the 8 1543 all have equal length. A buffer or two between them. The output of the reclock circuit is a concern, too. |
Hi finneybear,
I will be very interested in such a redo of the pcb!
Maybe we should also consider to place 4 times 2 chips on the underside of the pcb. With alu strips between them, we could press the aluminium on to the bottomplate of the enclosure...
Regards,
Lucas |
|
|
| finneybear |
Hi Lucas,
The 1543s can not be all on the same side otherwise some signals will have to go between the pins. ABAB is still the best way to go, I think? Every two sets can still share one heat sink. A memory chip heat sink can do the job. I still have to sit down and think about a way to get equal signal path length. A tiny clock skew to different 1543, for instance, will have impact on the sound image, etc. Even the DAC tower solution has this problem. |
|
|
| dddac |
Hi Lucas,
I see you are still trying to get your problem fixed....
Looking at all the reactions, you may be understand now, why I cannot give support from a distance..... I had one DDDAC1543 builder with the problem that nothing worked !! The solution was to remove a small splitter of metal on the datatrack !! There is no way you can find things like this through email.
I also noticed that your request for help post has also encouraged people to start proposing redesigns and tweaks, not to mention statements on how the dac is supposed to work and not work.... This is clearly not helping you and you seem not to have to the means to do so, which I regret. If you like you can send the DAC to me and I will fix it for you.
In the meantime: many builders have finished the dac and are very happy with the results, even with poor ground designs or cs8412 loads which are too high ;) Don't get yourself distracted. The set up works fine and gives a much better result than the cost would justify....
Oh, I use the 1085 myself......... with 22uF Black Gate. The datasheet recommends this value (10-22). If you use a cheap electrolite, this might proof to be too low. so have you tried a 220uF (just to make sure) at the output from the 1085?
any way my offer stands.....
take care
doede |
|
|
| Edwin Dorre |
Hello,
I have problems as well:
When I put my dac tower in, the LT1085 does not hold his voltage, without dac tower I have the 8.5V, with dac tower it drops to 1.87V or so.
When I remove the dac tower and put in one dac, and change Rsource to about 2k and change Rref to get to 3.85V DC output I get a nasty osclillation of about 50Mhz 2v top-to-top on the output. However when I turn on my cd-player the dac works. When I turn down the input voltage (I run of a stabilized variable power supply instead of a battery) to about 8V (the LT1085 does not have to do anything anymore) the oscillation stops! And the line out looks nice! When I put the tower back in, everything stops, 1.87V on VA2, nasty oscillations on the power supply rail etc.
So there is something wrong with this LT1085 thingy! By the way I use dirt as sh*t 22uf 35V elco's instead of BG because the BG's are in backorder! I will raise the elco's to 100uf or so tonight...
Edwin |
|
|
| Ola |
Hi Doede!
I can't enter Your dac page, explorer crashes.
?
Best regards
Ola |
|
|
| dddac |
OK, I think I found the problem. It is 100% the LT1085....
When it starts cold, there is oscilation which can be heard as hard noise. In my DAC at home after a minute or so it is gone, I thought it was the DAC biasing and did not cared too much, as every thing worked fine after warm up. My tube amps have to warm up also, eh?
BUT, it might be, that in some other (critical) situations, the 1085 does NOT STOP making noise !!!
I have just measured it and just check the scope pictures....
Please do note the difference in scale !! When the warm up is done, the rest-noise is at an average of -90dB. This is ok enough, although I am sure some will diasagree, :rolleyes:
The cold pictures show clear hard stuff !!!-15dB is not particulary a good noise floor level ;)
ANYWAY: try a larger C at the output of the 1085 or take a LM317.....
Will make notes at the WEB Site as well....
Take care
doede |
|
|
| dddac |
above was the cold start NF
below is the cold start HF |
|
|
| dddac |
and now the warm situation,
note the different SCALE !!!!
just to make sure: These are FFT-spectrum images from the OUTPUT of the 1085..... |
|
|
| dddac |
| quote: | Originally posted by Ola
Hi Doede!
I can't enter Your dac page, explorer crashes.
?
Best regards
Ola |
I have a cheap WEB-HOST
:cannotbe: some times there is no site....
Should work ok in the meantime....
br
doede |
|
|
| dddac |
OK, last image on this issue..........
This is a time recorder of the output of the 1085.
After switch-on the oscillations start at aprox 0,6 Volt (!)
As time goes by, the oscillations start to decreas and suddenly disapear after aprox 235 Seconds.....
OK, 4 minutes warm up, never realized it is that much :att'n:
for easy to copy DIY-stuff I may be had better not mentioned the 1085 as an "alternative" for the LM317, which is clearly more stable!
doede |
|
|
| dddac |
I could not stop ;)
So close........., so I did some more checking. I tried a few capacitor values at the output and every thing above 40uF stopped the noises at cold start !!!
My suggestion: Increase the Value of the output C to 47uF - 100uF when a LT1085 is used..........
This can be done emperical of course as you can hear it, no need for scope :D
I am glad as well, as now I have no "warm-up" time any more of 4 minutes.... hahaha !!
Regards
doede |
|
|
| finneybear |
| Good finding, Doede. Now one less thing for me to worry about. Forgot to tell you that I got the board from you last week. Thank you. What a lovely board! The design is very good but now I am trying to see whether I can squeeze a bit more performance out it. :) |
|
|
| Edwin Dorre |
| quote: | Originally posted by dddac
I tried a few capacitor values at the output and every thing above 40uF stopped the noises at cold start !!!
My suggestion: Increase the Value of the output C to 47uF - 100uF when a LT1085 is used..........
|
I still have oscillations with my dac tower... I changed the C to 100uf (jamicon) and the output of the LT1085 ramped up to 5.5V instead of 1.87V and less oscilation (still a lot). Then I put a big 1000uf BG on the 12V voltage rails (for the record, I use a stabilized variable lab power supply for testing instead of an battery) and I get the 8.5V on the LT1085 but with an > 20Mhz oscilation with a top-top of 0.4V . Everything is working except for this oscililation.
This morning (had to test it before work! :cool: ) I still had some thoughts about my power supply and I changed it for a RC car 9.6V battery. Without the 1000uf BG on the voltage rails I still get a awfull lot of oscilcations. With the 1000uf BG I have the same picture as with my lab powersupply. :(
I just bought some 100uf, 220uf, 470uf C's and a lm317T to test more tonight...
So it seems that just ramping up the output C to 100uf is not the answer to everybodies problems :(
Nasty beast this LT1085...
Edwin |
|
|
| rwagter |
| Damn, the lt1085 was the second most expensive part (€6,60) and a lm317t is very cheap.... |
|
|
| Leeuwarden |
| quote: | Originally posted by dddac
My suggestion: Increase the Value of the output C to 47uF - 100uF when a LT1085 is used..........
Regards
doede [/B] |
As far as I understand the value of the output C isn't critical. So why not put in a 100uF just to be sure... Or is this too simple?
Jan |
|
|
| Lucas_G |
| quote: | Originally posted by dddac
OK, I think I found the problem. It is 100% the LT1085....
When it starts cold, there is oscilation which can be heard as hard noise. In my DAC at home after a minute or so it is gone, I thought it was the DAC biasing and did not cared too much, as every thing worked fine after warm up. My tube amps have to warm up also, eh?
BUT, it might be, that in some other (critical) situations, the 1085 does NOT STOP making noise !!!
I have just measured it and just check the scope pictures....
Please do note the difference in scale !! When the warm up is done, the rest-noise is at an average of -90dB. This is ok enough, although I am sure some will diasagree, :rolleyes:
The cold pictures show clear hard stuff !!!-15dB is not particulary a good noise floor level ;)
ANYWAY: try a larger C at the output of the 1085 or take a LM317.....
Will make notes at the WEB Site as well....
Take care
doede |
Hi Doede,
Thanks for your postings.
It is very kind of you to offer me to check the dac for me!
But I am sure that now that we have found the problem, I will manage to get the thing working.:)
Unfortunately (stressed by the lack of luck on that day), I had strifed plus and minus of the two 5 Volt supplies, last week, so I will almost certainly have to swap some parts now. I will start working on it after Christmas since work is keeping me extremely busy now (as is my 10 weeks old baby;) )
Anyway, thanks again for you kind offer, and I look forward to hearing the DAC somewhere in January!
Regards,
Lucas.
P.S. I am always amazed to see how much experience and expretise can come together through such a forum. Without internet people like me could never have undertaken such projects as these...:cool: |
|
|
| Edwin Dorre |
I am spooked,
Perhaps I seeing things I should not looked for...
I still have an oscilation on the outputs! About 5Mhz, always. I have done:
- one dac (changed resistors to 2k2 and 1k5)
- 8 dacs
- changed voltage
- changed powersupply to 9.6V RC car battery pack
- changed C's up to 470uf
- changed all C's to a minimum of 100uf
- removed the clock circuit and changed to mode 2
- disconnected battery checker
- changed to a LM317
- removed and bypassed the complete 8.5V voltage source (LT1085) and put my variable power supply on 8V
HELP?
Edwin
Everyhing is working, both channels have normal output with this oscillation on top of it. I see the following image on the right and left output, which looks like an oscilation of about 5Mhz... Is this normal???
Snif :bawling: |
|
|
| Bricolo |
I'm sorry I can't help but,
Edwin: what's your scope? The screen looks similar to mine. It's a Tek 2232. I got it second hand, and I'm looking for infos, manuals...
Can you help me?
Alex |
|
|
| Edwin Dorre |
Hello,
It is indeed an Tek 2232. I bought it secondhand also, it is in perfect condition and a great help with these kind of problems! I have a manual with the schematics, however I can not help you with a copy because it is almost impossible to copy with all these fold out pages. However I can help you if you have a problems with the scope, please ask me question if you need any help with it!
Edwin |
|
|
| Bricolo |
Hi,
Does your manual explain how to connect it to a printer or a computer, to acquire data?
I'd like to know how to use the connectors on the left side.
Alex |
|
|
| dddac |
I cannot see why you have the oscillations, but I assume from your info, you cannot hear them ?? I also doubt if it will do any thing bad to the music, but you might want to try to put a small Capacitor at the output of the dac (so parallel with Rload) Choose a corner freq of lets say 200kHz ?? 4,7nF with 270Ohm....
you can see if the oscillation goes away,
plan b is a RC at the output. lets say, 27 Ohm in series and then 47nF parallel.....
You can listen to music if you prefer one or the other or no filter at all.
good luck,
doede |
|
|
| rbroer |
Pssssttt,
A clue, 128*Fs
:devilr: |
|
|
| Edwin Dorre |
| quote: | Originally posted by rbroer
Pssssttt,
A clue, 128*Fs
:devilr: |
That's is what I though 128 * 44Khz = about 5.6Mhz ! Which is my oscillation. And that's why I was starting to think it is not 'my' problem but a more generic design problem (or done on purpose because of the lack of filtering) of the DAC .
However why am I the only one seeing and mentioning this behavoir ???
Edwin |
|
|
| finneybear |
Edwin, I am not sure whether I am about to say is correct since I have not designed a DAC chip before. My understanding is that many multi-bit DAC chips have an osc inside locked to the external reference clock through PLL. This osc can run 128X or even higher and is used to control the current source switching.
What you have seen is the switching noise. In a tradtional DAC design, the LPF will block it out yet this is not the case here. As far as your amp, and especially the speaker tweeter can handle it, nothing to worry about. Not sure whether it is a good idea to add a RC filter to output end. This seems to contradict the idea of a filter-less DAC design. Plus most amps will serve as the filter for you anyway. |
|
|
| Lucas_G |
After swapping parts, fitting Blackgates BG-N 33u, and exchanging the lt1085 for a 317, I still had the same problem...:dead:
Then I had to do what I didn't want to do: desoldering the DacTower with the 8 pieces of TDA 1543...:mad:
Then I put in one TDA without heatsink. For the first time I could hear undistorted sound coming from the dac...:rolleyes:
The sound I heard was indeed lively in the midrange. The bass was somewhat lacking in dynamics. The treble however was harsh and undetailed. Not smooth at all, and without any sophistication...:smash:
Compared to my already heavily modified Philips DVD 963 SACD this sound wasn't very promising...:xeye:
I must add however that only ONE TDA was running at its nominal 5 Volt (because of lack of heatsinking), that there hasn't been any break in period, and that the output caps were not a Blackgate N in super-e configuration like it is in the DVD 963 SA.
So maybe it will all improve dramatically. :clown:
Now I must consider building a new DacTower, or start a new pcb with 8 pieces of TDA sitting in a rectangular shape with a heatsink on top of them. We'll see...:D |
|
|
| Edwin Dorre |
| quote: | Originally posted by Lucas_G
After swapping parts, fitting Blackgates BG-N 33u, and exchanging the lt1085 for a 317, I still had the same problem...:dead:
Then I put in one TDA without heatsink. For the first time I could hear undistorted sound coming from the dac...:rolleyes:
|
I had 100% the same!
One dac is working, 8 not, even with a LM317 it did not work... You only get to 2.7V output on the Lm317 of LT and lots and lots of oscillation. The only way for me to get it working with 8, is to have an extra 220uf on the + and - power rail very close to the LT1085 it self. I soldered it on the + and - power rails pins parallel to the + and - 12V wires from the battery. I also raised the output C to about 100uf !
This way it works at my place!
Ok, two with the same problem, hardly a coincedence. :xeye:
E. |
|
|
| Lucas_G |
| quote: | Originally posted by Edwin Dorre
I had 100% the same!
One dac is working, 8 not, even with a LM317 it did not work... You only get to 2.7V output on the Lm317 of LT and lots and lots of oscillation. The only way for me to get it working with 8, is to have an extra 220uf on the + and - power rail very close to the LT1085 it self. I soldered it on the + and - power rails pins parallel to the + and - 12V wires from the battery. I also raised the output C to about 100uf !
This way it works at my place!
Ok, two with the same problem, hardly a coincedence. :xeye:
E. |
Hi Edwin,
In my case I could easily reach 8.5 Volt though.
I was thinking whether these DacTowers were so badly prone to oscillations... Maybe it is because the distance from the upper TDA on to the capacitor is some 6 cm. In my case the tower is even a bit taller because I put 2 mm thick aluminium between each tda for better cooling. This makes the tower somewhat taller than that of Doede...:confused:
I think I will try it again with an extra cap soldered on the upper (ultimate) tda. See whether that helps.
Regards,
Lucas |
|
|
| Edwin Dorre |
| quote: | Originally posted by Lucas_G
I think I will try it again with an extra cap soldered on the upper (ultimate) tda. See whether that helps.
|
I did not solder an extra cap on the tower. I put it were you connect the 12v from the battery for the dac power. So the cap is in front of the LT1085...
E. |
|
|
| Lucas_G |
I had ordered some new TDA 1543 chips, so I finally decided to build a new Dac-tower...
And indeed, it had been the old DAC-Tower that prevented this Doede Dac to work from the start. Maybe one of the TDAs has been faulty:cannotbe:
For the first time I could hear this design with 8 paralell TDAs.
The sound is indeed much better than with one TDA.
I found the sound with 5 Volt supply quite relaxed, but also a bit less controlled in the treble.
Then I upped the voltage to 8 Volt. It does alter the sound quite heavily. There is more definition, but it becomes also a bit leaner and with less ambience in the midrange.
I do like its dynamics in the midrange, but the treble is still way more primitive than with my modded Philips 963:smash:
Still not a bad DAC. I will fix it into its new enclosure and use better cabling. I will also try various output capacitors and finetrim the output reference by listening to it. Let's see how much improvement can be made to it. I don't think it will ever beat the 963 DAC, but I will give it a chance.
Regards, and thanks to Doede!
Lucas |
|
|
| Bas Horneman |
Hi Lucas,
Have you tried the Wildmonkeysects loop filter mod yet?
Tonight I am going to test my Doede dac...hope it works...
But I am allready starting to look at possible mods...the wildmonkeysects one looks promising..
as a preventive measure I have also added little 20R resistors as stoppers for pin 1, 2 and 3 of the 1543. And 100R after the clock signal of guido tent's oscillator.
Regards,
Bas |
|
|
|
