Discussion in 'Digital: DACs, USB converters, decrapifiers' started by TomB, May 20, 2021.
Looking good man!
Meanwhile, Doug is working on the wood sides. He shared some pics with me and gave permission for me to post them (and hopefully he doesn't mind me copying his description, too).
Referring to the wood he had available, Doug started his e-mail by describing the thickness of the available wood pieces he had in stock:
"The only difficulty was that those were 9mm thick, and such small pieces can't go in a planer. And it isn't really safe to use power tools on small pieces anyway. As a consequence, these were done by hand. Which was fun for me
first pic is a shooting plane used with a shooting board with a fence on it. That allowed me to square the edges and make them 90 degrees:"
"This is a thicknessing jig. basically tracks for the plane to run in. I put different thicknesses of shims under the piece to be planed to adjust final thickness:"
"it works OK:"
TomB: Wow - what precision!!
"Here i trimmed them to length before putting them back on the shooting board to square the ends:"
A little extra explanation - the wood pieces are all off cuts from the last round of 3F's. They were in a box of firewood, but since these sides are so tiny, they are plenty large. 3F's use 9mm thick wood while this dac uses 8mm. They were also all around 8" long and my planer will eat anything shorter than ~12" in length, so these needed to be thicknessed by hand. Moreover, as one gets older, the appeal of large noisy machines that can instantly remove a finger starts to wain, so any excuse to do woodwork by hand is appreciated.
I have that exact same Hakko 936 for just about as long as you. Serves me well everytime.
OK - So, now I have definitely struggled. As mentioned, the next step I was taking was to solder the diodes. Of course, LEDs are diodes. Vaguely remembering about previous difficulty with SMD diodes, I started with them, first. In the Walnut X.3 DAC, as with many of Doug's designs - including amplifiers - LEDs are used to set a constant voltage drop. As it turns out, they are quite excellent in doing this, as long as you pick the right LED. It seems red ones do this best. Doug has 5 LED positions on the Walnut DAC. However, one of them is the classic power-on LED and is a typical through-hole, 3mm LED. That one is perfectly simple and will come near the very last of assembly.
However, the other 4 LEDs are SMD type and are used as constant voltage drops in the buffer circuit and the buffer power circuit. Their function is important. Like any LED, the polarity is absolutely critical. Install in the correct direction, they light up. Install in the other direction, they don't.
Therefore, knowing how to determine the polarity is important. With through hole LEDs, this is easy. In every case, the positive (Anode) lead is longer. Here's a typical diagram of a through-hole LED that many of you are probably familiar with:
Unfortunately, there are no through-hole leads with which to make one longer than the other with SMD LEDs. And that's where everything falls apart quickly.
Here's a link you may find interesting about the state of SMD LED polarity marking in the industry: How to Build a PCB: Diode Polarity (No, It’s Not Obvious) (eeweb.com)
The IPC-610 standard (most widely used inspection standard in the PCB industry) defines marking SMD LED polarity like so:
If we look at the Walnut X.3 PCB, we can see that Doug has absolutely followed this standard. I've marked the negative and positive traces for clarity:
The three dots, with the middle dot displaced in the direction of the Anode. The Anode end represents the positive end of the LED.
In the Walnut DAC X.3, Doug specified the HSMH-C150 LED. Remember, the selection is for the purpose of the circuit, not for lighting. I suppose another LED could've been selected without the following problem, but it might not have the desired characteristic in the circuit.
Here's where the problem comes up and gets out of control. The image below is a sample cut from the HSMH-C150 datasheet:
The top LED image does not apply to our LEDs. On the bottom, though, does that line in the middle with the middle nub correspond to the three square dots on Doug's PCB silkscreen? What does it apply to? Note in the parentheses that it states, "ANODE LINE FOR HSMH-C150," which is completely opposite - apparently - every other LED they make in the series. What about the standard LED schematic diagram floating slightly over to the right in-between the top and middle LED image? What are they saying corresponds to the anode or the cathode?
Here's a pic of the bottom of the actual LED, along with a soldered one on the Walnut DAC PCB:
So, following the image from the data cut sheet, which plainly stated that for the HSMH-C150 the Anode indication was reversed and "assuming" the lines correspond to the square dots on the Walnut DAC's silkscreen, I carefully lined up all the LEDs and proceeded to solder them on the board.
After soldering, I thought to myself, "I really need to test these, because pulling them off will be a real hassle once everything else is soldered." So, I tested them with the diode setting on my Fluke.
As it turns out, almost half of them were wrong and would not light! That's out of a total of 20 LEDs - 4 on each of 5 PCBs. How could this happen after I was so careful in lining them all up? I could see making a mistake with 1 or 2, or even 3 or 4, but almost half of them?
I got out the hot-air rework station that Doug talked me into purchasing a few years ago (a great piece of advice). I removed all of the LEDs that tested badly:
Here's what I removed - 8 of them!
The rework station heats the joints up with hot air, melts the solder on both sides of the chip, and you simply remove the chip with tweezers in the other hand. The de-soldering braid is used to clean the solder on one pad only, since we want solder to stay on one pad anyway to reinstall the part.
At this point, I checked the loose LEDs with my Fluke. The symbol on the bottom of the LEDs was reversed according to the data sheet (which itself indicated it was reversed)! So, not only is there no consistent industry standard, but there is an inconsistent standard with the same mfr with different LED series. Worst of all, these LEDs were purchased at different times over the last couple of years, meaning: there is inconsistency with the same LED from the same mfr at different times!!!!!
What to do? I decided to go back to the meter and simply test every LED, either loose or soldered on the PCB, until all of them were correct. At least I knew the red probe on my meter was Positive and I knew which side of the LED pads on the PCB were Positive.
A final check of all of the PCBs proved that the LEDs would all light with the correct polarity from the meter probes:
This was all maddening, to say the least - but it's over. So, expect quite a bit more done with the next update - after I recover a bit this weekend.
I built an amp once where I had the SMD LEDs backwards because I assumed the markings were consistent. This led to a 7V DC offset and replacing the voltage regulator. After a lot of troubleshooting I eventually figured out it was because the LEDs were reversed.
learned my lesson. ALWAYS test LEDs before mounting. Don't trust datasheets
I would agree, but I wonder how the Pick-and-Place is done? All of these were taped. I'd be surprised if they did something other than refer to the datasheets.
Just an FYI - I haven't disappeared the last couple of days, just been steadily working. Anyway - after you've seen several dozen SMD parts, you've seen them all. So more pics are not necessarily adding anything useful.
I'm happy to report, however, after soldering over 350 individual SMD parts, the SMD for the 5 PCBs is finished. I'll have some pics and start in on the through-hole parts tomorrow.
Sorry that I'm running a bit behind, as usual. I expect to catch up next week. We're having some home repairs done and I've been babysitting contractors all week, while trying to keep working on the Walnuts.
Here's a pic of the preliminary rinsing, once the SMD work is finished. I do this because all the flux is on the top side of the PCBs and once the through-hole parts get installed, it's really difficult to clean the top of the PCB, because so much is in the way with the through-hole parts..
The Amanero USB boards also arrived from Italy a couple of days ago:
Doug's wood has also shipped, but I haven't received it, yet - probably Monday will be when the package is delivered. I'll have to finish the wood myself, but I've done that before and already have his recommended finishing oil.
Well, I apologize for another delay again. In preparation for beginning the through-hole soldering, I was gathering the USB header sockets together and discovered something troubling:
In through-hole soldering, you generally want to solder the lowest parts first, ending with the tallest. The socket headers for the Amanero USB boards are generally the lowest height through-hole parts. I was checking those and discovered the discrepancy above.
I had a small length left from the original Walnut X.1 DACs that I had ordered from DigiKey a couple of years ago. That header is the furthest one on the left. The one on the right, and the remaining one in the tube, were both ordered from Mouser about 3 weeks ago.
The dimensional discrepancy is important, because the sockets largely dictate the height of the USB board and the USB connector hole is already cut into the back plate. Upon checking the datasheets from both Mouser and DigiKey, and measuring with my calipers: the DigiKey part is correct, but the Mouser one is not.
This incident is vanishingly rare for Mouser. I don't think I've ever had them send me the wrong part in 15 years of ordering many $thousands of dollars of parts from them. After sharing the evidence with them, they were very helpful and the correct headers are on the way.
There you have evidence of reasons why Mouser and Digikey are at the top of my list for small quantity component acquisition for my lab.
Yes - I should add that I've had plenty of wrong parts received, but until the above, they were all my fault in picking the parts.
So, ... I apologize for the absence over the last week. Contractors have continued to interrupt me constantly throughout the day over the last week. Rather than continue progress on the Walnut X.3 DACs, I gave up in the worry that I'd get distracted in the middle of continuing to assemble the DACs, then mess something up. The interruptions continued with work going on until 10:30 PM last Friday night - heavy machinery and all.
Here's a pic that gives you all a small idea of what we've been going through at our home over the last week or two:
That's my yard on one side of the driveway. What you're looking at is a 24-inch diameter, steel corrugated storm drain pipe that's buried under my lot. (If we knew it had been there, we'd have re-considered purchasing the house 6 years ago.)
Since the night of Hurricane Sally coming through the Atlanta area last year, I've had a sinkhole in the middle of the yard above that pipe. The hole was big enough to swallow me and my riding lawnmower with room to spare. I won't bore you with the crappy political details, but suffice to say I've done everything I could with my HOA and the County Government. I even got the Engineer Of Record (EOR) for the neighborhood's original storm water drainage design to survey the situation and recommend solutions.
This type of pipe installation is not pressurized pipe, but meant for gravity drainage. Trouble is, I have hills behind me and three other homes whose back yards drain into this pipe, including my own. The EOR surmised that rainfall was enough that a backup from the outlet at the drainage easement across the street, combined with the total flows of the four houses through this pipe, arrived somewhere in the middle of that pipe, leaked out, and created the hole in my yard.
If you look closely toward the bottom of the pic, you'll see one of the welded and bolted steel bands, actually broken. The bolted clamp actually sheared off from one side of the band. We believe this was the cause of the sink hole.
Anyway, the hole has now been filled with 10 tons of rock, permeable membrane and perforated pipe to collect any ground flows - in addition to replacing that one band and another one. They are going to begin restoring the top soil and re-turfing the yard this weekend, but it's been h*ll for several days around here.
Just wanted you to know that I hadn't disappeared or met with troubling obstacles on the Walnut DACs - regular life and its problems intervened.
Since they finished filling the hole at 10:30 PM last Friday night, I've been free to resume work on the Walnuts.
We last left with all of the SMD work completed and the tops of the PCBs cleaned and rinsed. The next parts to install were the through-hole parts. The lowest height parts on the boards are the USB board connector strips. The correct replacement strips from Mouser came in late last week, so I got those installed.
The Amanero USB boards actually used two rows of connectors, but the strips come in single rows (at least the ones I purchase). It's a bit tricky to get them soldered correctly, but there's an easy trick: use an Amanero board, plug it into the connectors, then solder the connectors in place on the PCB while the Amanero board is still plugged in. This ensures that the connectors are absolutely correctly oriented.
Next up are the TO-92 parts. I'd call them all transistors, but they're not all transistors: some are actually voltage regulators. Doug's PCB design uses the one-leg offset pad for TO-92 transistors. Unfortunately, some of the transistors/regulators are only available on tape. This means they come with kinks in the legs to facilitate installation in inline, three-hole pads. So, the first step is to bend all the leads back to "normal" transistors to fit the pad types on Doug's PCB:
Next, I insert all of the TO-92 parts at once into a PCB, being careful to ensure each different TO-92 part is in the correct location. While doing this, I push the parts down until the leads start offering a pretty good resistance. This usually ends up with most of them at the same height. Then I flip my building board (it's literally a wood board), upside down and place it over the top of the TO-92 parts inserted into the PCB. Like a sandwich, I hold the top and bottom and rotate the two back so that the board is on my work table with the PCB upside down on top:
I proceed to solder one lead each on each TO-92 part, until all of them have one lead soldered. Then I turn the PCB over to check each part and it's orientation. I make corrections by slightly bending them as necessary, then turn the PCB back over and finish soldering the remaining leads:
These are all 5 PCBs shown with all the TO-92 parts and the USB connector sockets all soldered in place.
Jeez, that no excuse man, get your shit together!
But seriously, everything is looking really great, can't wait to see the final results!
Quickly moving on (after losing all the time last week), I went to the small electrolytic capacitors, next. In a similar fashion to the TO-92 parts, I inserted all of the capacitors of one type and rating all at once. Flipped the wood board over and placed it on the top of the capacitors, and while holding both together, flipped both of them back over again. If you simply try to turn over the PCB, the capacitors will fall out.
Here's the "forest" of leads for the smaller Panasonic FM capacitors in the output stage and its power supply section:
Just like with the TO-92 parts, I solder one lead only on each capacitor, then flip the PCB over to check each capacitor and its orientation. After making corrections, if necessary, then I flip the PCB back over and solder all the remaining leads. Here's one PCB with those capacitors finished:
Note also, that I've already soldered the ferrite inductors in the primary, linear-regulated power supply on the far side of the PCB.
Again, quickly moving on, I finished the remaining four PCBs for the smaller Panasonic FM capacitors and went straight to the polymer caps. Once finished with those, I added the single Wima FKP film cap, the RCA jacks, the power switch, and the power Molex header on all 5 PCBs:
If you look closely at the nearest PCB, you'll see that I've also installed the TO-220 transistors, regulators, and rectifiers:
Once I have those complete, only the large electrolytics and transformers are left.
I expect to have the PCBs finished and tested this weekend and casework should begin next week.
I'm not a municipal engineer though have been tangentially in that field... I don't think you're even supposed to use that style of pipe when you're encountering the flow levels that you're describing. Anyhow, glad to hear the issue has been resolved and you'll be able to have a hard again.
Well, this thread took a strange turn.
(yes, I am a 12 year old boy in a 42 year old shell)
Well y'know when someone landscapes the yard just right... *chef's kiss*
Before anyone else questions this again .... let me assure you that in the last 9 months, I have overturned every single rock in this scenario. The design was approved and stamped by the county storm water division and follows every code in the GA manual on stormwater drainage.
I also happen to be a Registered Professional Engineer in the State of Georgia - since 1992. So, I'm not going to respond to anymore comments on the Engineering Principles and Practices involved.
I only posted the above as an apology for being absent for a week. A few people have already invested in these DACs and I wanted them to know there were good reasons for my delay in building and that I hadn't run off with the money.
Ah fair enough. I'm a P.Eng. as well but I'm sure our local codes and bylaws are very different. Also, I'm pretty sure Georgia doesn't have to contend with harsh winters that freeze the ground 6ft deep and cracks water pipes
Separate names with a comma.