Now, the DAC board is a little bit harder. The most intersting things I found out after taking a look on the bare board.
1. The circuit paths are a totally chaos. For example:
You surely remember the 4 elCaps in a row, in front of them the 4 voltage regulators and in front of them the other 4 elCaps? At first it seemed to be so: 1 elCap to 1 volt. reg. and following is the next one elCap. If you believe it you are wrong. This is absolutely different. I will try to write the paths in some pics with an image editing prog.
2. The heater run very hot. The mounted voltage regulators run inexactly.
3. Read: http://focus.ti.com/lit/ds/symlink/pcm1742.pdf, the DAC needs Dual-Supply Operation: 5-V Analog, 3.3-V Digital. You have a cheap 2 Zener solution here, but TI says, you need a separate supply for the Digital input if you want to have a good sound.
You can see the connection diagram on page 25. The important facts here as copy:
"The PCM1742 requires a 5-V analog supply (VCC) and a 3.3-V digital supply (VDD). The 5-V supply is used to power the DAC analog and output-filter circuitry, while the 3.3-V supply is used to power the digital filter and serial interface circuitry. For best performance, the 3.3-V supply should be derived from the 5-V supply using a linear regulator, as shown in Figure 28."
There is no "clean" 3,3 V supply. The 2 Zener near the DAC produce 3,3 V but in a very noisy way. So I redesigned the complete PSU on this board.
4. All electrolytics in signal path have to be removed (4 pieces , see above) with anything you want - or can pay.
5. A new clock is a good thing.
6. Except the elCaps I changed all resistors in signal path and the tube area for better ones. Some of them are non-magnetic Vishay-Dale Mil Spec CMF-55, but not all, because they are hard to find here and the most powerful are 1/10 W. Furthermore not all needed values have been available. The other resistors are some of 1, 2 or 3 W metalloxide film from Yageo. The empirical formula is not to take lower power rating but the same value. Minimum power is the power of the old resistor.
7. I had one greater problem in redesigning that you maybe will have too. Please read this now very carefully and avoid this problem:
For the DAC I used a Linear Technology LT1584CT-3.3 / 7A linear regulator. This one has a stability of 0.05 %, is very fast and has a constant voltage of 5 %. Costs are 20 USD.
7 A is absolutely to much for the PCM1742, but it was the only one I could get without lead time. Then I made a great mistake. I connected him where the other linear regulator is connected , the LM 317 - on the big resistor 2,2 Ohms 2W near the two cooler. I have done it because I wanted a single path for the 3,3 V. And I thought, that a 5V regulator can share the same input as a 3,3V. After this procedure it got curious. Player has no more signal! The solder joints seemed to be all right, so I have searched the error for weeks!
.............and then I found it. It was the expensive regulator. It had not 3,3 V on output, it had 3,85 V! But why?
At the end of the big resistor (where the LM317 is connected) there is an input voltage of more than 20V! So the 3.3V regulator and the LM317 share a 20V input? I couldnt believe. At first I thought the 2W 2,2 Ohms resistor is broken and I had a short-circuit. But it was not so. What was the reason?
The reason is that a Linear Technology LT1584CT-3.3 should not be connected on such high input. I found the answer here:
http://www.farnell.com/datasheets/84158.pdf........"input below 7V".
But why has a single 5V regulator (LM317) an input of 20V?
The answer is on the other side of the PCB. The bigger elCap is not only connected to this 2 voltage regulators, no, this cap is also connected to the 7812 regulator on the other side. This regulator produces this voltage only for the delay at the output.
Again studying the datasheet of the PCM1742 I found the answer on page 2:
" ABSOLUTE MAXIMUM RATINGS: Power supply voltage, VDD –0.3 V to 4 V,
and RECOMMENDED OPERATING CONDITIONS:
Digital supply voltage, VDD = 3 MIN, 3.3 NOM, 3.6 MAX. Here is the problem! This DAC dies at 3.85 V input!
The next problem is - by pulling up (slowly and carefully) the cover of the broken DAC the circuits paths were demolished. Between the cover and the chip is epoxy and the chip, the epoxy and the cover are stucked together.
After a longer time I got a new DAC, a new SMD adapter and a new regulator (for 20 USD again :shock: ).
This time I made no more experiments: the 3.3 V regulator I connected directly on the output of the 5 V regulator (on the same cooler of the LM317). The changing of the DAC is really a hard thing and I dont recommend that if it is absolutely not necessary.
The new high precision 5V regulator is a 1,5 A one instead of the old LM317 with 1 A. The heater runned very hot with 1 A, so I searched a solution for it, because I expected now overheat on the 5V regulator. This is a simply 20 W resistor. Pulling out the 2,2R I build in an 6,8R. Now the heater is a little bit cooler, the resistor a little bit warmer. I wanted that the heater is only handwarm, but he isnt yet. At least I added a second 6,8R in serial and an approx. 5R on the other regulator, the 7808 at the top of the pic.
It is a little bit better, but I will take greater values and conduct some experiments. For the 7808 10V input is enough, for the 5V regulator 8V. Its a simple thing to increase the resistors. Better a hot resistor than a hot regulator.
Its ready now! Now its a really good thing! I will try to explain that all with the pics tomorrow.
Have a nice day
Greetings to our new friends, coming to us from Holland and France. I am your neighbour. Coming from Germany. Pics will come tomorrow...... |