《AudioLabs-Tube-DAC3_5b-dac-sm 电路图 维修手册.pdf》由会员分享，可在线阅读，更多相关《AudioLabs-Tube-DAC3_5b-dac-sm 电路图 维修手册.pdf（17页珍藏版）》请在收音机爱好者资料库上搜索。
1、 Tube DAC 3.5B Assembly Manual February 10, 1997 S HELDON D. S TOKES 103 Windy Cove Apt I Hampton VA 23666 http:/www.clarkson.edu/stokessd A u d i o L a b s RadioFans.CN 收音机爱 好者资料库 2 Table Of Contenets: Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2、 . . . . . . . . . . . . . page 3 Theory Of Operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 3 Board Etching Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 6 Parts List . . .
3、 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 7 Schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 10 Component Outlines. . . . . . . . . . . . . . . .
4、. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 12 Ground Plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 13 Signal Plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5、. . . . . . . . . . . . . page 14 Wiring the Circuit Board. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 15 Digital Filter Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 16 Customizing .
6、. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 16 Corrections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 17 RadioFans.CN 收音机爱 好者资料库 3 Introduction: The purpose
7、of this manual is to be a guide to the proper assembly and maintenance of your Tube DAC. The board has been carefully laid out and constructed to make assembly easy and straightforward. The DAC itself has been designed to be very robust and to give years of trou- ble free service. Theory of Operatio
8、n: This DAC consists of four separate parts, the decoder, the digital fi lter, the digital to ana- log conversion, and analog gain and buffering. The fi rst part, is the decoder. The digital audio data enters the DAC via a coaxial input BNC connector. It uses the standard SPDIF standard 75 input imp
9、edance, but the DAC uses a BNC connector rather than an RCA jack. The data stream is capacitively coupled to a pulse transformer. The pulse transformer reduces common mode noise dramatically, and improves the performance of the input circuitry. The output of the pulse trans- former goes to the input
10、 pins of the decoder chip. It also has a 75 resistor in parallel with the chip to set the input impedance. The decoder chip reconstructs the various clocks from the serial data stream. This chip uses a PLL (phase locked loop) to generate the clocks from the data stream. Sampling and status informati
11、on are also extracted from the data stream. The data and clocks then go to the digital fi lter. This fi lter removes any information from the data stream that is greater than half the sampling rate. The data coming out of this chip is oversampled by 8 times. This means that there are now 8 samples i
12、n the time that there was one going into the chip. This will be very useful because it will move the quantization noise generated by the DAC chip 8 times farther away from the audible band. The data and clocks then go to the DAC chips. This is the heart of the DAC. The DAC chip used here are 20 bit
13、models, that actually have two DAC chips inside each physical chip. They are operating in a complimentary “push-pull” fashion. The DAC chips output a very small current that is proportional to the number input. The analog stage must take that current output and convert it into a voltage, amplify it
14、if necessary, and buffer the signal so it can be sent to the pre-amplifi er without serious degradation. The fi rst task is to convert that very small (micro to millivolt) current into a voltage. There have been a myriad of schemes to convert that current to a voltage in other commercial DACs. Very
15、fast op-amps are popular, as are a variety of very clever discrete schemes. The problem with these “active” current to voltage conversion (I/V) schemes is that they typically require many elements. In the case of an op-amp, there may be 40 transistors and a bunch of bulk silicon resistors and capaci
16、tors as well as a large amount of feedback involved in the process. And if that wasnt enough, the current value changes very rapidly, so if the I/V stage isnt designed perfectly, it can produce slewing induced distortion. This DAC uses a much more simple scheme (thanks to Peter Campbell for this one
17、). A small value resistor is placed from the output of the DAC to ground. Thus, via Ohms law (V=IR), a voltage proportional to the current is produced. There is only one user selected, passive element in the signal path. There is no risk of slewing induced distortion, and the signal path is as short
18、 as possible. There is a downside to this approach. The DAC chip was designed to have its output pin always at ground potential. And it has a pair of diodes inside the chip to protect the output from any extraneous voltage on it. So that if the output rises to around 0.7 volts, the diodes start to c
19、on- RadioFans.CN 收音机爱 好者资料库 4 duct. This has the nasty side effect of clipping the signal. So the thing to do is keep the voltage generated across the I/V resistor as low as possible. But if the voltage is too low the signal to noise ratio will be compromised, because the signal will require too muc
20、h gain. So there is a trade off. It turns out that the PCM-63 DAC chips from Burr Brown work very well using this resistor I/V scheme. There is also a “Bipolar offset pin” on the PCM-63 DAC chip. This subtracts 2 mA of cur- rent from the output pin when it is connected to it. The reason they provide
21、 this feature is that a zero signal audio condition really is half scale in digital number terms (the audio signal must swing negative as well as positive). So the BPO pin sucks off half of the total possible output cur- rent. So the zero signal current with the BPO pin attached is zero. This may se
22、em like a great idea, but it has a downside too. The current source is also made up of a bunch of transistors and bulk silicon resistors and capacitors. These things are best left out of the audio path. So the BPO pin is shorted to ground where it pulls 2 mA of current off the ground plane. This mea
23、ns that the zero signal current is now 2 mA and it can be as high as 4 mA. So the resistor value must be lower than it normally would have been if the BPO pin was used. This zero signal current is not a problem because the analog stage is capacitively coupled, and a DC offset on the input is fi lter
24、ed out. A FFT analyzer was used to determine the resistance value for the I/V con- verter where the signal starts being clipped by the diodes. It turned out that a resistor larger than 116 caused harmonics to start to be generated from a pure tone. This design uses a 100 value. This provides a healt
25、hy 2 mV signal for amplifi cation. The signal to noise ratio is not as high as it could be, it is very good all the same. With a good pair of tubes the DAC is dead quiet. The small voltage must now be amplifi ed and buffered. This is done with a SRPP (series regulated push pull) vacuum tube stage. T
26、he SRPP stage can be thought of as a standard resistance coupled triode gain stage, but instead of a large plate resistor, a constant current source is inserted between the power supply and the plate. This acts to make the gain stage much more linear, lowers the harmonic distortion, and provides a m
27、uch lower output impedance than a single gain stage would have. This SRPP stage uses one half of a dual triode for the current source and the other half for gain. The output is capacitively coupled to the output jacks. The signal from the DAC is coupled to the grid of the gain tube via small resisto
28、r. This resistor reduces the likelihood of high frequency instability in the analog section by forming an RC fi lter with the input capacitance. As was stated earlier, the output current changes very quickly, and that creates high fre- quency energy (quantization noise). The continuous audio signal
29、is essentially made up of steps. Most commercial DACs fi lter this energy out. Many different fi lter topologies were tried, and they have all colored the sound and reduced the performance of the DAC in some way. So the analog stage was left unfi ltered. The downstream audio components then act as a
30、 fi lter. Many people have balked at this approach, bringing up the potential for subharmonics being generated in the audio band and such. Those effects havent been noticed to date. And those who have lis- tened to this topology have really liked the sound, many say its better than any other digital
31、 com- ponent they have heard. If some fi ltering is desired, a few pF of capacitance can be added between that plate and the grid of the signal tube. This will act to reduce the bandwidth of the tube, and roll off that high frequency energy. But try listening to the DAC unfi ltered fi rst. Due to th
32、e 8 times oversampling digital fi lter, this noise is at and above 352.8 KHz, which is very far above anything that is audible. The combination of a passive I/V converter resistor and a simple unfi ltered tube stage make this one of the simplest and shortest analog sections on the market today. Its
33、a belief of the 5 designer that a short clean signal path is the best way to get the closest to the music. These various DAC stages all need to be powered. And this is an area that simplicity doesnt always pay. This DAC uses 4 raw supplies and 8 separate regulation stages. It uses a totally separate
34、 transformer for the digital and analog sides as well. The DAC also has a split ground plane that is connected at a point near the mixed mode decoder chip. The digital side of the DAC requires two voltages. Most of the chips require 5 volts, but the DAC chips with their push-pull architecture, also
35、require -5 volts. These voltages come from a full wave rectifi ed sup- ply that feeds a pair of adjustable regulators, with their adjust pins heavily decoupled. Adjustable regulators were chosen over fi xed regulators because they are much better at keeping digital noise from being fed back out of t
36、hem and back into the power line, only to come into the analog stage. The analog stage requires several voltages. The DAC chips analog section also requires plus and minus 5 volts. There is another full wave bridge supply feeding four adjustable regulators, thus each DAC chip has a dedicated regulat
37、or pair feeding it plus and minus power. The tube stage requires two power supplies as well. First the tubes require heater power. The heaters also have a regulated DC supply. An adjustable regulator is also used here as well, so the fi lament voltage can be changed by simply changing a resistor. Th
38、e heater supply actually fl oats about 85 volts above ground, so the heater to cathode voltage is not exceeded in the tubes. The fi nal power supply is the high voltage for the tubes. There are many opinions out there on that is the best way to supply power to tubes. One school of thought is to use
39、passive fi ltering via RC and inductive fi lters. This can be very musi- cally satisfying, but creates a power supply that has a high impedance. This high impedance sup- ply actually degrades the performance of the constant current source portion of the SRPP stage. Regulated supplies typically have
40、amazingly low output impedances, but they can be electrically noisy (especially on transient demands), and some folks complain that they lack musicality. A semi-regulated power supply design was chosen, this gives the advantages of a regulated supply, with the quietness and musicality of a passive R
41、C supply. First the high voltage from the trans- former is full wave rectifi ed and fi ltered. A string of zener diodes is biased via a resistor from this voltage. A capacitor is placed across the zener string to maintain the voltage precisely during varying loads, and the capacitor also fi lters ou
42、t any diode noise. A pass transistor is referenced from the zener string. The drain of the transistor is attached to the raw supply. The gate is attached to the top of the zener string. The source is then sitting at the zener potential. This arrangement can provide much more regulated power than the
43、 analog stage could ever use. The output of the transistor then goes through a pair of small resistors and is fi ltered with a big pair of fi ltering caps. This fi nal RC stage gives additional noise fi ltering and reserve power for the analog stage. All the rectifi ers are solid state. Many people
44、seem to prefer tube rectifi cation. Tube rec- tifi cation produces much less electrical and RFI hash than solid state rectifi ers. But tube rectifi ers are very limited in how much capacitance can be used after them. And they generally make higher impedance supplies than solid state rectifi ers. So
45、what would be ideal is to have a solid state rectifi er that wouldnt produce all that electrical hash. Well the secret is to absorb that hash with ceramic capacitors hung across each leg of the diode bridge. Each rectifi er bridge has 4 ceramic caps to absorb that diode hash. There are also capacito
46、r power supply bypass caps right next to each chip where the power enters. The digital side uses tantalum capacitors, and the analog side uses fi lm capacitors. The fi nal capacitors in the high voltage supply also are bypassed with fi lm caps. 6 Board Etching Tips: The artwork is printed onto trans
47、parency fi lm from a laser printer, print it three times. Cut out two of the prints with about a quarter inch of clear space around the circuit board image. Then carefully tape these two copies to the uncut one after carefully aligning the traces of the overlay to the uncut sheets traces. When fi ni
48、shed, there should be three perfectly stacked copies. This increases the contrast of the fi nal image. When a transparency is printed with a laser printer, there are usually holes in the black printed parts. And the blacks arent all that black when it is held up to the light. Overlaying makes the bl
49、acks much more black, and gets rid of the holes. Now the artwork is ready to use. This method uses GC positive sensitized boards and developer. The FR-4 fi berglass 1 Oz. grade board works very well (they can be gotten local electronics stores). The board emulsion is sensitive to UV light, A good source of UV to expose the board is a GE sunlamp. The sunlamp is hung so the bottom of the bulb is about 12 above the board. The exposure time is 9 minutes. With a yellow incandescent bug light-bulb on, pull the
免费区 
