| spencer |
Attahed is the schematic with Jfet IV and low noise regulator enhancement. PCB layout completed and project in progress.
Note that Parts is still available from Ruach and I brought two sets SM5842 and PCM63PK from him.
Shall update project progress and any comment is welcome.
Spencer |
|
|
| spencer |
Component layout on pcb.
Note that the pdf file is too big to attach here. |
|
|
| MashBill |
Looks nice. Keep us posted on your progress.
Bill |
|
|
| xavier1000 |
Your project sounds interesting. I am planning something similar to mod my Rotel rcd991 but re-using the existing PMD100 instead of your SM5842 filter. My idea is to replace the dac and iv stage but not the filter. I am getting the data after the PMD100 in the rotel and send it to a new pcb (inside the rotel) with 2 PCM63K per channel (working in differential) and balanced jfet IV stages.
BTW After seeing your component layout, it seems you are using an external clock (tentlabs?). Is it right?
Please keep us updated.
Xavier |
|
|
| spencer |
| SM5842 can take clock input and I have a jumper J9 to select either 8414 MCK or a external clock (tent clock for me). The SM5842 has a free running mode and I hope it works. I am making my pcb now and I hope I can receive the boards by mid next week. |
|
|
| spencer |
| Boards received and will start soldering. See attached. |
|
|
| spencer |
8414 to 8412 converter board.
I have 100 pcs for sales. US$1 each and postage is about US$1 for airmail for 5pcs or below and $2 for 5 - 10pcs. Interested please mail me. |
|
|
| Johnwkkwan |
Hi,
I am looking for a DAC too !
Will you have extra PCB and parts to share in future.
Please keep us posted on your progress.
John |
|
|
| spencer |
John,
Thanks for your interest but the key parts PCM63, SM5842 and K170/J74 are purchased from Ruach in Singapore.
I only sell the Pulse Transformer and 8414 converter board in the trading post.
I am stuffing the board and hope to finish it in a week. |
|
|
| spencer |
| Under listening test and break in. |
|
|
| spencer |
| Digital input and select. |
|
|
| spencer |
| Digital filter, clock and digital part regulator |
|
|
| spencer |
| DAC and Jfet IV - Left channel |
|
|
| spencer |
| DAC and Jfet IV - right channel |
|
|
| spencer |
| Jfet IV regulator and power supply - left |
|
|
| spencer |
| Jfet IV regulator and power supply - right |
|
|
| spencer |
There is a miner bug in this design when using the on board clock.
There is a little "click" sound when SM5842 reset the data input due to sync issue between on board clock and 8414 generated clock. I can think of few ways to eleminate the bug.
1. Use TCXO like NP but the cost to get a TCXO is not cheap.
2. Feedback the on board clock back to the cd player to cd player clock has to be in slave with the on board clock. This requires a modification of cd player and choose one player that can use 11.2896MHz clock.
3. Select another receiver IC that will accpet a external clock of 11.2896MHz.
Any body has any suggestion?
Anyway, without using the on board clock, the DAC itself can player very good sound and music to me. |
|
|
| spencer |
I received a few enquiry about buying my D1 clone pcb and I am considering to sell it if the demand is more than 10 pcs. The price My offering will be US$25 per board plus $8 postage fee by registered airmail. Those interested please note down your name.
The board that I am going to sell will be the revision 2 which has the following changes:
1. Add a clock output socket pad.
2. Allow more space for all TO-220 heatsinks.
3. Add one more relay to select digital input.
4. Short circuit the Iout to BPO and thus no more jumper is required.
5. Enlarge some pad and connector hole sizes.
6. Add a ground connection for the ground plan on top.
Board size is 8 x 10 inches, FR4 double size plate through board. Solder mask and silk screen printing on top side.
p/s: For those who has my ver 1 board, I will offer them at US$15 per board but the postage fee is still $8.
Regards,
Spencer |
|
|
| spencer |
More improvement:
7. Jumper to select the clock supply or 5842 supply for the 8414 IC. Clock regulator is not used in ver 1.
8. Use LM317T and LM337T regulators for all DACs.
9. Split the ground plan in sections: 8414, 5842, Input relay and logic IC, DAC left and DAC right.
10. Relocate the retification section (diodes and caps) of the 30V suply so that the last few filter caps will be close to a centre ground point.
11. Looking at input digital i/p section to see if relay can be eliminated or find a better way to reduce jitter introduction. This can be a daughter board put on top of the digital input section as it will take a longer time to develope.
Thanks for some of the feedback and the pcb count is now up to 5 persons.
Regards,
Spencer |
|
|
| spencer |
I received feedback that other members are not able to post reply in this thread. Not sure if it is still the case now?
I received more than 10 enquiry about the ver2 board and I will build about 20 pcs to offer to those interested. It will take 2-3 weeks time. Update will be posted here when ready.
Thanks for the interest.
Spencer |
|
|
| spencer |
Hi,
Changes on ver2 board has been freezed and design has released to pcb manufacturer today. Here are the changes:
Marking on board:
BAL PCM63 DAC
2007-4-v30
DIY Fun - CSK
1. Add a clock output socket pad (J20). This is reserve for the clock to feedback back to CD transport.
2. Allow more space and add holes for all TO-220 heat sink (total 15 T0-220 parts).
3. Add one more relay to select digital input (K2). Use jumper to select either XLR or RCA input (J8).
4. Short circuit the PCM63 Iout to BPO and thus no more jumper is required (J1-4 do not need anymore).
5. Enlarge all hole/pad sizes.
6. Add a ground connection for the ground plan on top near main filter capacitors (J19).
7. Jumper to select the “clock regulator” or “5842 regulator” for the 8414 IC (J17). Thue the regulator for 8414 will be isolated from 5842 when the on board clock is not used.
8. Use LM317T and LM337T regulators (better noise performance) for all DAC chips.
9. Add de-emphasis LED indicator (D32).
10. Split the ground plan in sections: 8414, 5842, Input relay and logic IC, DAC left and DAC right.
11. Relocate the rectification section (diodes and caps) of the 30V supply so that the last few filter caps will be closed to a centre ground point J19.
12. Add “+” mark for all E-capacitors for polarity identification.
13. Add resistor and VR for all PCM63 to trim the MSB (this adjustment is optional and reserve for diyer to play around the trimming of MSB).
14. Add Trimmer pot (R8, 9, 17 & 18) to adjust Jfet IV output level to match “+” and “-“ cycle. This is also optional and need accurate voltmeter and test CD. Do not solder the parts if diyer decided not to trim the output.
Spencer |
|
|
| spencer |
Version 2 Circuit 2
I need to split into two files as it is too big to post in single file. |
|
|
| spencer |
| Version 2 Bill of Material |
|
|
| spencer |
| PCB received, testing in progress. |
|
|
| spencer |
| I shall also test a SM5847 to SM5842 converter board which is a drop in for the overpriced SM5842. Note SM5847 is still under production by NPC and the price is much cheaper than SM5842. |
|
|
| spencer |
| Short of 6 heatsinks and will be completed soon. |
|
|
| spencer |
| Jfet IV and regulator part |
|
|
| spencer |
| Ver 2 started singing and further testing in progress. |
|
|
| spencer |
Stuffing and Adjustment Procedures:
1. Solder resistors, LED and diodes
2. Solder film capacitors except the BIG 10uF cap
3. Solder all jumper pins
4. Solder IC sockets
5. Solder bead, inductors
6. Solder Pulse transformer
7. Solder all pots (VR)
8. Solder small E-caps (22uF to 330uF)
9. Solder Regulator Jfet / Transistors / TL431 etc.. (do not solder jfet IV fets)
10. Solder connectors if used
11. Solder big E-caps (1000uF and 2200uF)
12. Solder all relays
13. Solder IRF610/9610 with heat sink
14. Verify the +/- 20V regulator (across R59 and R60) is working and adjust the output voltage.
15. If cannot adjust the voltage to 20V, solder a 220k resistor and parallel to R47, R48, R65 or R66. Do this when needed.
16. Then solder the Jfet IV FETs section module by module.
17. Power up again and adjust Pot R1 until Point A to 10Vdc, Adjust Pot R2 until Point B is 0Vdc, Repeat adjustment on R1 and R2 when necessary.
18. Then adjust Pot R8 until voltage across R14 is 5Vdc.
19. Repeat to solder other Jfet IV section and adjust voltages as above.
20. Solder the LM317 and 337 with heat sink in pairs and verify the regulator output voltage is 5V +/-0.1V at the PCM63 IC socket positions.
21. Repeat and verify all regulators to PCM63 is either +5 or -5V.
22. Verify the +20/-20V regulators are still at correct voltage. This voltage should not change more than 0.2V or else something may be wrong.
23. Now solder the 7805 and 7808 in the digital sections and verify all output voltage correct. You can check voltage at the IC sockets pins to double confirm.
24. Solder the 8412/8414 chips in position. Plug in 5842 chip.
25. Short the pins for 8414 power to clock regulator. Also select the input jumper to RCA socket (right side two pins shorted).
26. Power up the digital section, the error LED should be ON after 1 second. Then inject a SPDIF signal to the RCA digital in socket, the Error LED should be OFF now.
27. Use a scope to check if there is pulsed at CLK, LE, DOL and DOR nets. If don’t have scope, use a DC voltmeter and check voltage is about 2.5Vdc.
28. Power down and go back to DAC section.
29. Plug in PCM63 first chip.
30. Verify the Iout is at 0V dc (point B), if not, fine tune the R2. Note that this voltage should not change much after the first adjustment. Check point A still at about 10V.
31. Repeat for other PCM63 one by one.
32. Then check the current drain at R81 and R82 (5.1 ohm resistors). The voltage drop across R81 is about 0.97V while the voltage at R82 is about 1.15V.
33. Solder the 4 BIG film caps 10uF.
34. If everything is fine, plug in transport and you should hear music from the output.
35. Perform normal testing with a test CD.
36. Warm up the set for 1 hrs and adjust the voltage at point A and B again when needed.
Notes:
1. The “+” mark on D10 LED is wrong polarity.
END |
|
|
| spencer |
| Wave form at 100Hz, output voltage about 1.9Vrms single ended. |
|
|
| spencer |
| Output at 20kHz. Output level at 1.73 to 1.68Vrms for left and right Single ended output. |
|
|
| spencer |
For those interested, can email me. 10 pcs board are still available and the rest already booked by other members.
Price of ver2 board is US$25. Shipping with registered mail is $8. Thus total will be $33.
8414 to 8412 converter board is $1.
I shall release a SM5847 to SM5842 converter board soon and price of this gold plated board is $5. Pins and 4 ceramic smd capacitors will be included. I shall post photo when I receive the chip and tested working. I am able to provide 6 SM5847 IC for sales and price is $20 each.
Selected parts are avaliable but with limited quantity. See list attached.
Thanks for all your interest. |
|
|
| spencer |
| 5847 to 5842 converter bare board. |
|
|
| spencer |
| SM5847 converter board half done. Need to source a good pins and hope to finished it by Sunday or coming Monday. |
|
|
| spencer |
| Still waiting for the pins! |
|
|
| spencer |
| OK got pins and put on and everything is working fine. |
|
|
| spencer |
| Some more photo of 5847 converter |
|
|
| spencer |
All the SM5847 chips have been sold out.
The sound for SM5847 is more harmonics while the SM5842 is thicker a bit in sound. Both sound good to me.
Thanks. |
|
|
| spencer |
Shipped 10 boards and still pending 6 pcs for confirmation. Please contact me if I did not send a email to you.
Thus left 4 pcs ver2 boards for other diyer member.
Also, only limited parts available for those interested.
Also I would like to know if any one interested for a PCB with regulator and jfet IV (Single Ended design) only. I estimate the board will be priced at $5 each channel. This board will be great for CD player and DAC upgrade.
Thanks. |
|
|
| spencer |
Update part list that still available.
All SM5847 chips have been sold out.
Thanks. |
|
|
| spencer |
Update: 13-May-2007
1. The output level at balance mode is about 4Vrms which is too high for my pre-amp (DIY jfet SE pre-amp). Thus I put a 4.7k resistor in parallel to R4, R13, R24, & R34 (solder at the bottom side of pcb). This has no effect to the sound but match the sound level of my pre-amp volume control. Due to this change, all the pot R1, R10, R21, % R31 has to be re-adjust till Point A back to 10Vdc (half of the +ve voltage 20V).
2. I change the 4 output BIG film cap from Vishay 4u7 back to Rifa PHE 426 10uF 250V type. I found the sound improves a lot and sonic is more sweet and relax (less harsh). Thus you should consider to invest on better capacitor at the output coupling.
3. I change the DF from 5847 and 5842 back and fore and find that the 5842 did sound better to me (more relax). No wonder VC America sell the 5842 so high. There is also sonic difference between 5842AP and 5842APT chip, APT is a bit better.
4. My original testing is using a China make 20VA transformer of 24V x 2. I put in the Farnell 9531750 30VA 25V x 2 potted transformer, the pre-regulated voltage is about 30.5V and is very stable. Sonic wide is more stable and sound is more relax also. Anyway I also recommend to use Farnell 9531718 30VA 9Vx 2 transformer.
5. I believe this DAC has lots of potential for diyer to play around to get and improve to better sound. One idea is to use DIR9001 receive IC instead of 8414. 9001 has only 50ps jitter and thus I believe the sound should improve more. I don’t have this converter board yet and I may either build my own or buy one from China diy forum.
6. Thanks for all the support and I hope to receive more comment about this pcb. Thanks. Here is the update photo for reference. |
|
|
| spencer |
Due to the complication of this DAC, 3 members have decided not to take the pcb. Thus there are 3 pcbs left for other members. Note that this pcb is not for non experienced diyers as it required some skills and understanding of electronics to complete it.
Let me know if you are keen. |
|
|
| spencer |
| One more gone, two pcb left. |
|
|
| spencer |
| I have matched some quads (4 pcs) 2SK170BL with Idss > 10mA for sales to those member that have brought my pcb. The Jfet is matched to within +/-0.1mA and are used in location Q3, Q8, Q13 and Q18. |
|
|
| spencer |
| All 20 pcs Ver2 board has been sold out. Thank you for your support. |
|
|
| spencer |
There are some people asking available of this kit details and here it is.
Latest upgade with DIR9001 receiver give better sound on sub sonic and clear back ground music with higher resolution. I shall provide details upgrade later soon. |
|
|
| spencer |
| Here is one photo with DIR9001 converter board. |
|
|
| spencer |
In order to use DIR9001 converter board, two logic need to be change on two ICs.
1. HC86 pin 5 short to pin 7 (pin 5 from 5V short to ground). This is to ensure the de-emphasis function works.
2. SM5842 pin 5 short to pin 10 (pin 5 from ground to +5V). This is to set the input mode of 5842 to 16 bit.
Since the trace on the main board is under a IC socket and cannot easily be cut or jump a wire. I use another layer of IC pins shown and cut one of the pin 5 and then solder a jumper wire to the required pins on the same row of IC pins. This IC pins are first inserted into the original IC socket and then the IC inserted on top of it (show on another photo).
Thus this change can be back ward convertible back to the 8412 receiver without soldering on the main board. |
|
|
| spencer |
Dont forget to set the DIR9001 to 16 bit output mode and also 256Fs. Here is photo of the final main board with all the changes.
Note that the HC86 and 5842 is sitting on two layer of socket pins.
For those interested to do the upgrade, I can provide a converter board later but please give me an email first so that I know how many people are interested. I can provide either bare board or complete assembly of converter board.
Cheers. |
|
|
| spencer |
9001 to 8412 converter board circuit attached. Note that the converter board will also work in other DAC besides the D1 clone.
Also the XTI and XTO are connected to pin 5 & 6 of 8412 socket. Thus external clock per 9001 data sheet can be added off board. |
|
|
| spencer |
| 9001 to 8412 circuit attached. |
|
|
| spencer |
The next upgrade on the Jfet IV part is to reduce the input AC impedance so that the DAC can drive the current to 0V and Zero resistance to ground. This mean more linear and less distortion.
I have done this measurement as below to lower the impedance from 46.8 to 23.5 ohm.
1. Q3/Q8/Q13/Q18 use K170 x 1 give 46.8ohm
2. Q3/Q8/Q13/Q18 use K170 x 2 give 35.8ohm
3. Q3/Q8/Q13/Q18 use K170 x 3 give 29.3ohm
4. Q3/Q8/Q13/Q18 use K170 x 3 AND R3/R12/R23/R33 change from 33r to 13.2ohm (parallel a 22R resistor at the bottom) give only 23.5 ohm ac impedance at the jfet IV.
Conclusion:
The more Jfet in parallel will give lower impedance.
Increase the lower current source will also give lower impedance. Note that NP use 10mA for the mosfet.
But there is a limit on the max current can be set:
1. Idss of Q15.
2. Max power dissipated by Q15. Thus reducing the -20V supply is another way to increase the current source at the bottom.
3. The upper current source must also able to provide sufficient high current to drive the input and also the 1.5k resistor to ground.
4. Another idea is to use BJT current source to provide higher current source but that would not be labelled Jfet IV any more.
Thus diyer can try to see what the sound will change by playing with the value of R3/R12/R23/R33 or parallel more jfet to Q3/Q8/Q23/Q33.
Cheers |
|
|
| spencer |
Next trial is to change all the 8 pcs PCM63 5V regulators from LM317/337 to LT1085/LT1033. Attached is the photo after change. A 22uF 10V polymer capacitor are also added to the output of the LT regulators at the bottom of the pcb. LT give more accurate voltage compare to normal LM regulators (range from 4.98 to 5.02V with 1% resistors 100R/300R).
A simple change is also able without changing the regulators is to add a 10uF capacitors from pin 1 of all regulator to the ground. It will also improve the noise rejection of all regulators according to the LM317/337 data sheet. Beware of the polarity of E-cap when doing the change.
Cheers. |
|
|
| spencer |
After few hour listening, I feel the sound is too tense. Thus I do the following changes.
1) Remove the tan cap 22uF on the +/-5V analogy supply of PCM63, change to a Panasonic FC cap 100uF 50V.
2) Put the 4 pcs 22uf Tan cap in parallel of the 300R resistor of the regulator ground resistors (analogy part).
3) Put 10uF 16V tan cap parallel to the 300R resistor of the regulator ground resistors (digital part).
The sound become more mellow and relax again.
Conclusion, the capacitor at the analogy 5V supply of PCM63 affact the sound a lot. Try to use the cap that suit your taste. |
|
|
| spencer |
| Photo show the changes at the bottom. |
|
|
| spencer |
I go back to LM317/LM337 as the sound of LT regulators are too aggressive for me!
For you information, I am working on a Ver3 pcb and would like to know how many people are interested so that I know how many board to order later. The board will only be offer if the the result is better than ver2! It will have all the upgrades from Ver2.
Attached is a layout which is still changing everyday.
Spencer |
|
|
| spencer |
| 9001 bare board has arrived and evaluation underway. Still waiting parts from Farnell! |
|
|
| spencer |
9001 to 8412 converter board completed.
One change is to add a 750ohm resistor (temporary leaded) at the pin 9 and 10 of the converter board, ie the input pins of the receiver IC 26C32 to move the DC voltage to about 2.5Vdc.
Listening test is underway. |
|
|
| spencer |
| For the trial, I go back to single K170 for the input Jfet and also the lower current source back to 3.5mA (R3/R12 back to 33ohm). THat is the original design values and it gives me more sweetness on the music. |
|
|
| spencer |
The ver3 board has been released to pcb factory and here are the changes relative to ver2.
1. Able to use DIR9001 daughter board upgrade beside 8414 daughter board.
2. Add one more K170 footprint on the input of Jfet IV so that two BL grade fets can be used instead of V grade K170 (can’t find).
3. Add one more pad for J74 current source in case J74V grade Idss 15mA fet is not found.
4. Enlarge pad sizes of passive parts like resistors, capacitors and inductors etc… on bottom size.
5. Increase pcb thickness to 2mm with gold plated through holes. White mask with black text.
6. Add more pitch size selection of output couple capacitor – 10uF (max 2 inches).
7. Add ground plan on the bottom size of digital section to improve grounding.
8. Add LM317 for pre-regulation for digital ICs and then LT1117-5V regulator for each digital IC.
9. Only use LM336-5V voltage reference at +/-18V low noise regulators in analogy supply.
10. Eliminate 90 deg turn on trace for digital signal paths.
11. Correct D10 polarity mark “+”.
12. Separate the digital supply of PCM63 from the analogy supply.
13. Two ground plans (D & A) and single point of connection at the centre location L21.
14. Add more via hole to improve the grounding at top and bottom ground plan of digital section.
15. Add ground points J19/J23/J24/J17/J26 so that diyer can add additional star ground if required.
16. Add TO-220 type fred diode pad for user upgrade.
17. All more pre-filtering capacitor pad.
18. Move the Jfet IV next to the current output of PCM63 to eliminate any noise pickup or leakage.
19. Add output relay for power off mute and power on delay.
20. Change analogy transformer from 24V x 2 to 18V x 2. Less heat on the low noise pre-regulator.
21. Add selection of 8412/14 output at 16 or 18 bit mode and also SM5842 input at 16 or 18bit mode.
22. Improve all pot tuning to clockwise direction for higher level or voltage etc…
23. Add jumper (or hard wire) for all LM317/337 regulators and thus easy to connect additional dc source for comparison.
24. One heat sink type – easy to source parts. |
|
|
| spencer |
| D1 Ver3 PCB is received today and under assembly and testing. Stay tune. |
|
|
| spencer |
| Under testing and listening! |
|
|
| spencer |
I have built two set with a bit different parts in the last 1.5 week and they are working fine. This kit is sure more difficult to build compare to ver2 due to very dense parts placement and more wirings for the PCM63 digital supply (either from analogy or digital side).
OK, the pcb/kit is ready to go! Here is thet BOM and pirce. The other files like schemetic/assembly instruction/pcb silkscreen top are too big to post here and I shall send to you when you order the kit.
Once again this is a project for experienced person and think twice before you order.
Picture will follows later as I am busy for work this week.
Enjoy. |
|
|
| spencer |
There will be some more delay on the D1V3 board release as I find some quality issue on the boards and I need to return to the pcb factory to perform a 100% open/short test. It will take about 2 more weeks time!
Here are two more photo on the two sets that I built so far. The one using normal E-cap parallel with film cap does not sound as good as the one that I use 18uF Rifa film cap. I added a shield on the 18uf film output so that there will be less interference among them to prevent any noise or even oscillation. |
|
|
| spencer |
| 1st set with copper shield on the output film cap. |
|
|
| spencer |
I have started shipping some of the D1V3 boards but it is slow as the time to do kitting is not easy.
I have tried PCM63P-K2 on my DAC and the K2 chips sound more dynamics than the standard K version. Here is one picture of the chip. It is a used chip selling for US$35 per pc and is hard to find!
I shall contact those that have pre-order the pcb in the next 1~2 weeks. |
|
|
| spencer |
| Here is the update ckt 1 |
|
|
| spencer |
Finally I have time to split the assmebly manual into two parts:
Part 1 here. |
|
|
| spencer |
New toys in D1V3.
I have hard wired a converter socket for PMD100 in my DAC and the sound is quite different from SM5842.
PMD100 sound softer, warm and sweet. Bass is lesser than SM5842.
Spencer |
|
|
| spencer |
PMD100 converter board has been tested successfully in D1V3. Note that this board is only for D1V3 user.
Bare PCB is US$5. $6 will include pcb x 1, pins x 2, 0.1uf x 2, resistor x 1, LED x 1, IC socket x 1.
I have some PMD100 chips (brand new never used) selling with the full kit for $30 to D1V3 diyer.
Spencer |
|
|
|
