|
Good evening! Big thank you to Hans and everyone else in this forum! I had a lot of fun building these kits and I couldn’t have successfully completed them without the folks who ask and answer questions here. I wanted to make this post to thank everyone but also to detail some of the problems I encountered, the equipment I used, and also how I resolved them.
This is a long post for anyone who’s interested but TLDR: I built 2 QMX+s and 1 QMX and all three needed new PCM1804s.
I purchased a QMX+ in May and a couple weeks later got to work. I had built a couple other kits and antennas previously but this was the most involved so far. I was able to make it through the build and load firmware but I couldn’t run all of the self-tests and the audio was loud static. I didn’t have enough experience or equipment to troubleshoot myself but luckily Jeff Moor W1NC reached out and offered to help. See /g/QRPLabs/message/126274 for my original post. I mailed him my QMX+ and he got it back to me in about a week with several fixes. It had several bad joints, insufficiently scraped enamel, a bad PCM1804, and a bad inductor. Thanks Jeff for your knowledge, experience, and willingness to help the community with your excellent service! I’m extremely pleased with the outcome and my QMX+.
I used:
-
A Weller 70W digital soldering iron
-
63/37 Kester .20 solder
-
30X/60X loupe
-
LED Head Magnifier
-
Crescent shear cutter/pliers set from Home Depot
-
A couple of different helping hands type devices
-
Solder wick & plunger type desolderer
-
Flux paste
I primarily used the conical soldering iron tip with success but in retrospect, I should have primarily used the wedge tip. The head magnifier with 3x magnification, plenty of light, solder wick, flux, and desoldering tool were critical. The loupe was ok but the head magnifier would have been enough. I ruined the side-cutters on solder wick.
Pleased but not content apparently because in July, I purchased another QMX+ so that I could use the lessons learned from the first unit and hopefully successfully build another myself. The kit stayed in the box until my wife & kids went on vacation for about a week. Over a couple of evenings, I had a great time building the kit, using all my new knowledge and experience, but when running the terminal tests, I had bad results on Audio Filter Sweep/RF Filter Sweep/Image Sweep/ADC IQ test. I reworked some of my joints but to no avail. Some example scans, before and after 1804 replacement, are attached.
?
With bad PCM1804
I started diving into groups.io posts and determined that maybe I should invest in more test equipment and that I might have a bad PCM1804 chip. I ordered two more PCM1804 chips (just in case I ruined one). I also ordered a QMX because I was more confident in my kit building skills. In the mean time, I also purchased an LCR meter, a TinySA Ultra (for its signal generator), and an oscilloscope. Probably overkill but I had my eye on them for a while.?
Over the winter holiday break, I built the QMX. I enjoyed and had more confidence in my build but I had very similar problems to my 2nd QMX+ build during self-tests. During this build, I used a grounding strap/bracelet the entire time so I was pretty sure it wasn’t static discharge related. What are the odds I’d have the same problem twice? I found this post /g/QRPLabs/topic/109639794#msg133769, indicating the PCM1804. I did the diagnostic tests from /g/QRPLabs/wiki/37111 and determined that replacing the PCM1804 was my next step. Luckily I had two new PCM1804’s in stock although I previously intended on only needing one.
After doing research here and on youtube for replacing SMDs, I got a ChipQuik kit and a hot air rework station. After doing some tests, I decided I would just use the ChipQuik, which is basically easy-mode for low-skill chip removal. My process was basically, remove the chip using flux and ChipQuick, clean the left-over ChipQuik with fluxed solder wick, clean the board with 99% IPA, flux the new chip legs, place the chip carefully on the board (head magnification essential), put a small amount of solder on the iron tip, tack the first pin, ensure chip alignment, then use the solder drag technique to solder the remaining legs. It doesn’t take much solder and I removed any excess with fluxed wick. I went back over the pins to ensure uniform connections and triple-checked for bridges. I repeated the process for the QMX+.
After replacing the chips, the terminal hardware tests all returned good results on both the QMX and QMX+. For good measure, I compared them with results from the QMX+ that Jeff Moore repaired for me and they were comparable. Once again, I’m super appreciative of all the folks here asking questions and giving helpful information. I’ve learned quite a bit about reading schematics, the function of different parts/sections of these radios, and test procedures. Thanks for everything!
?
With new PCM1804:
?
|
It's remarkable that you have had three PCM1804 failures even while using antistatic procedures. I built a QMX and a QMX+ without using any precautions, and did not have a chip failure. The PCM1804 mystery continues.
--
73, Dan - W2DLC
|
My assumption is that the Chinese manufacturer does not install original chips, but copies that in many cases do not work well.
?
73, Zrinko Ivan, 9A7AM
|
Hi Zrinko Ivan
My assumption is that the Chinese manufacturer does not install original chips, but copies that in many cases do not work well.
It isn't how it works. While the PCBs are manufactured and SMD assembly is in China, we don't use a Chinese manufacturer in the way that people submit a PCB and BOM to JCLPCB or someone (not saying there's anything wrong with JCLPCB per se).?
I have a trusted colleague in Shenzhen, a sub-contractor who I have worked closely with since 2014. Many of the parts are ordered direct from the manufacturers e.g. NXP Semiconductor, or Analog devices or from US distributors such as Digikey, Mouser and Arrow. Other parts like resistors and capacitors are sourced locally in Shenzhen from reliable sources but all the parts are tested by me and then the same suppliers used in all manufacturing batches. Everything is traceable and documented.?
Over the years there have occasionally been problems, things were particularly difficult during the global semiconductor shortage post Cv19. We've been able for the most part to quickly identify and resolve any issues that arise.?
The PCM1804 problem is a rather unique one that so far remains a mystery. Progress is being made. On QDX and on current Rev 2 QMX+ PCBs, we're able to test the board INCLUDiNG PCM1804 before shipment and I think we have a very high success rate at detecting failed chips. There was about a 15% failure rate in the batch, which were rejected here and not shipped to kit constructors. I don't think it's an ESD problem caused by the kit constructors.
So I think I'm making progress on the problem and hope to solve it finally got the QMX which is the remaining issue.?
73 Hans G0UPL
|
Wow - a 15% failure rate sounds exceedingly high.
toggle quoted message
Show quoted text
On Jan 1, 2025, at 7:48?AM, Hans Summers via groups.io <hans.summers@...> wrote:
Hi Zrinko Ivan
My assumption is that the Chinese manufacturer does not install original chips, but copies that in many cases do not work well.
It isn't how it works. While the PCBs are manufactured and SMD assembly is in China, we don't use a Chinese manufacturer in the way that people submit a PCB and BOM to JCLPCB or someone (not saying there's anything wrong with JCLPCB per se).?
I have a trusted colleague in Shenzhen, a sub-contractor who I have worked closely with since 2014. Many of the parts are ordered direct from the manufacturers e.g. NXP Semiconductor, or Analog devices or from US distributors such as Digikey, Mouser and Arrow. Other parts like resistors and capacitors are sourced locally in Shenzhen from reliable sources but all the parts are tested by me and then the same suppliers used in all manufacturing batches. Everything is traceable and documented.?
Over the years there have occasionally been problems, things were particularly difficult during the global semiconductor shortage post Cv19. We've been able for the most part to quickly identify and resolve any issues that arise.?
The PCM1804 problem is a rather unique one that so far remains a mystery. Progress is being made. On QDX and on current Rev 2 QMX+ PCBs, we're able to test the board INCLUDiNG PCM1804 before shipment and I think we have a very high success rate at detecting failed chips. There was about a 15% failure rate in the batch, which were rejected here and not shipped to kit constructors. I don't think it's an ESD problem caused by the kit constructors.
So I think I'm making progress on the problem and hope to solve it finally got the QMX which is the remaining issue.?
73 Hans G0UPL
|
Hi Hans,
?
Yes, the problem is very strange.
In my case the PCM1804 worked almost correctly. The fault was only in the unequal amplitude of the I and Q signals. If I had not used the diagnostic tools in the terminal I would not have noticed the problem immediately.
What is common in all cases is that the problem is present from the first power up, and fault never appears later. That is why I suspect a fault in the PCM1804 manufacturing process.
?
73 de Zrinko Ivan, 9A7AM
|
A 15% failure rate for any semiconductor device is extremely high. Googling PCM1804 failures shows posts about high-level audio noise that randomly occurs and goes away.
Tony
On Wed, Jan 1, 2025 at 10:26?AM Zrinko Ivan Vukoja via <ivan.9a5czi= [email protected]> wrote: Hi Hans,
?
Yes, the problem is very strange.
In my case the PCM1804 worked almost correctly. The fault was only in the unequal amplitude of the I and Q signals. If I had not used the diagnostic tools in the terminal I would not have noticed the problem immediately.
What is common in all cases is that the problem is present from the first power up, and fault never appears later. That is why I suspect a fault in the PCM1804 manufacturing process.
?
73 de Zrinko Ivan, 9A7AM
|
Yes, remarkable that the PCM1804 saga continues. I built 4 QMX and one QMX+. Only one had an 1804 failure. First Rev 1 and Rev 3 (mini) had/have no failures. Second Rev 1 had SMPS (5V), then Mplx chip failures. Both replaced and all well. Rev 4 had 1804 failure - replaced and all is well. In fact, that is my go-to mid-band rig for portable operations. QMX+ no problems. In addition, 2 QDXs before had no issues with the 1804s. Both QDXs sold since I have the QMXs.?
--
73, Dan? NM3A
|
To be clear again: I'm saying 15% of the batch of 1000 QMX+ Rev 2 PCBs was rejected during our testing here. The percentage of Rev 2 QMX+ with faulty PCM1804 that reach actual kit builders should be near zero.?
QDX is a similar story. We test for it here.?
QMX is the outstanding problem. We will find a solution to that too.?
So in other words, QDX and QMX+ are purely my own internal problem now. Issues for the kit constructor should be very rare. QMX is still an issue I'm working on and needs to join QDX and QMX+.?
73 Hans G0UPL
Yes, remarkable that the PCM1804 saga continues. I built 4 QMX and one QMX+. Only one had an 1804 failure. First Rev 1 and Rev 3 (mini) had/have no failures. Second Rev 1 had SMPS (5V), then Mplx chip failures. Both replaced and all well. Rev 4 had 1804 failure - replaced and all is well. In fact, that is my go-to mid-band rig for portable operations. QMX+ no problems. In addition, 2 QDXs before had no issues with the 1804s. Both QDXs sold since I have the QMXs.?
--
73, Dan? NM3A
|
I understood that; you are handling it before it hits customers for everything except the QMX, however even you having to deal with a 15% failure rate on a component is nuts (for you to have to deal with).
toggle quoted message
Show quoted text
On Jan 1, 2025, at 8:52?AM, Hans Summers via groups.io <hans.summers@...> wrote:
To be clear again: I'm saying 15% of the batch of 1000 QMX+ Rev 2 PCBs was rejected during our testing here. The percentage of Rev 2 QMX+ with faulty PCM1804 that reach actual kit builders should be near zero.?
QDX is a similar story. We test for it here.?
QMX is the outstanding problem. We will find a solution to that too.?
So in other words, QDX and QMX+ are purely my own internal problem now. Issues for the kit constructor should be very rare. QMX is still an issue I'm working on and needs to join QDX and QMX+.?
73 Hans G0UPL
Yes, remarkable that the PCM1804 saga continues. I built 4 QMX and one QMX+. Only one had an 1804 failure. First Rev 1 and Rev 3 (mini) had/have no failures. Second Rev 1 had SMPS (5V), then Mplx chip failures. Both replaced and all well. Rev 4 had 1804 failure - replaced and all is well. In fact, that is my go-to mid-band rig for portable operations. QMX+ no problems. In addition, 2 QDXs before had no issues with the 1804s. Both QDXs sold since I have the QMXs.?
--
73, Dan? NM3A
|
On Wed, Jan 1, 2025 at 08:52 AM, Hans Summers wrote:
?
QMX is the outstanding problem. We will find a solution to that too.?
?
The QMX boards could be tested with some sort of bed of nails test fixture that provided the appropriate test signals for IC403 couldn't they?
Use the CLK2 signal attenuated by R513 with a suitable level and phase shift, then the Rx Audio and Image tests should work.? That's enough to test most of the Rx side on an unpopulated board.
?
I guess you don't see the PCM1804s until they arrive on boards.? Has your trusted supplier been caught by someone flogging clone ICs?
?
Chris, G5CTH
|
> I understood that; you are handling it before >? it hits customers for everything except the? > QMX, however even you having to deal with > a 15% failure rate on a component is nuts > (for you to have to deal with).
Nuts, yes! Welcome to my life.? QMX is the outstanding problem. We will find a solution to that too.?
The QMX boards could be tested with some sort of bed of nails test fixture that provided the appropriate test signals for IC403 couldn't they? Use the CLK2 signal attenuated by R513 with a suitable level and phase shift, then the Rx Audio and Image tests should work.? That's enough to test most of the Rx side on an unpopulated board.
Yeah like my QCX-mini test fixture:
It took me 3 R&D-Weeks to build that. I shudder to think how many actual elapsed weeks it would take me to do that now. Arrrgghhhh. But yes it's an option I'm considering... The complexity and number of test pins could be greatly reduced if looking for one specific fault rather than trying to test every aspect of the entire radio.?
I guess you don't see the PCM1804s until they arrive on boards.? Has your trusted supplier been caught by someone flogging clone ICs?
Nope. That would be too easy... Anyway I have never seen anything other than absolute loyalty in the last 11 years working with my colleague in Shenzhen and I'm sure he would never cheat me. He knows what side his bread is buttered on. Issues further along the supply chain (in other words him being cheated, or suppliers of his suppliers) are harder to be so sure of. A messed up world.?
73 Hans G0UPL 73 Hans G0UPL
|
Hans,
?
Looking at web hits for PCM1804 failures other than on QM* products, I only see low level noise issues.
What's being reported in this forum is complete failure.
A 15% failure rate is nuts.
?
When final test detects a problem with the PCM1804, do they replace it with stock you buy from directly from TI?
Do these parts also fail in the same way at a 15% rate?
It could be your assembly house is sourcing what they believe are parts from TI, but they are counterfeit.
?
Have you looked at the PCM4202?
Claims to be pin and function compatible with the PCM1804, $3.20 US direct from TI at 1K quantity.
From the dates in the datasheet, it came out a couple years after the PCM1804.
Being almost identical, it makes me wonder why they bothered to do this.
Page 23 of the datasheet shows differences between the two parts, none of which look like a problem.
?
The PCM1804 and PCM4202 are both over two decades old, TI has the much newer
PCM1820 and TAA5242 parts that might be considered for future QRP Labs products.
Those are very small with a 0.5mm pin pitch and a center ground pad, much harder to work with.
But perfect for a possible QMXnano, to be stuffed into the traditional Altoids tin.? ;-)
?
Jerry, KE7ER
?
?
On Wed, Jan 1, 2025 at 11:22 AM, Hans Summers wrote:
Nuts, yes! Welcome to my life.?
|
If boards with a failed PCM1804 never fail when the part is replaced, that points to counterfeit parts.? And I'd not have them stuffed at that assembly house, make it?a hand add at QRP Labs instead.
?
Jerry, KE7ER
?
?
On Thu, Jan 2, 2025 at 07:56 AM, Jerry Gaffke wrote:
When final test detects a problem with the PCM1804, do they replace it with stock you buy from directly from TI?
Do these parts also fail in the same way at a 15% rate?
It could be your assembly house is sourcing what they believe are parts from TI, but they are counterfeit.
|
Jeff Moore has never seen a PCM1804 fail after he’s replaced it and he replaces quite a few of them. Jeff is sourcing replacement devices from Digikey and they’re also available directly from TI.com. I tend to agree there’s something amiss in the CM’s procurement process.
Tony
If boards with a failed PCM1804 never fail when the part is replaced, that points to counterfeit parts.? And I'd not have them stuffed at that assembly house, make it?a hand add at QRP Labs instead.
?
Jerry, KE7ER
?
?
On Thu, Jan 2, 2025 at 07:56 AM, Jerry Gaffke wrote:
When final test detects a problem with the PCM1804, do they replace it with stock you buy from directly from TI?
Do these parts also fail in the same way at a 15% rate?
It could be your assembly house is sourcing what they believe are parts from TI, but they are counterfeit.
|
If boards with a failed PCM1804 never fail when the part is replaced, that points to counterfeit parts.?
No, it doesn't; counterfeit parts is only one option; there could also be ESD problems causing it, either during assembly, or subsequent handling somewhere along the line; or there could be problems during our bootloader flashing and testing procedures.? ? And I'd not have them stuffed at that assembly house, make it?a hand add at QRP Labs instead.
Just when I thought I was already having the maximum amount of fun possible per day...
|
On Wed, Jan 1, 2025 at 11:22 AM, Hans Summers wrote:
QMX is the outstanding problem. We will find a solution to that too.?
The QMX boards could be tested with some sort of bed of nails test fixture that provided the appropriate test signals for IC403 couldn't they?
Use the CLK2 signal attenuated by R513 with a suitable level and phase shift, then the Rx Audio and Image tests should work.? That's enough to test most of the Rx side on an unpopulated board.
?
Yeah like my QCX-mini test fixture:
?
Yes, that's the sort of thing I was thinking of, not surprised you have already tried it..
?
I understand that it's a lot of work to make but if 1 in 7 of the QMX kits fail because of the PCM1804 problem 100% testing seems essential.? I don't think it's reasonable to keep shipping QMXs untested if that level of failure can be expected, it isn't fair on your customers.
?
Would a test rig that picked up the test signal from R518 and passed it through T401 to the inputs of IC403 work?? Your test harness would provide VCC and VDD and ground, that's 6 or 7 pins.? Run the Rx diagnostics through the USB connection.? This bypasses the LPFs and the BPFs but if they are needed they could be added to the test rig and a single band test done.? The alternative seems to be to connect each board manually using test clips, time consuming and error prone.
?
Chris, G5CTH
?
?
|
I suspect ESD or power supply damage to an incomplete board would not be repeated once fully built.? ESD damage, to ICs in particular, is often evident only after a few weeks or months...
|
I have no idea what procedures QRP Labs has before shipping a QMX+,? so take this with as much skepticism as you can muster:
?
My assumption is that they plug in a USB cable to program the STM32F and power the board up, not much more.
That USB cable is just a connection to ground and a couple pins on the STM32F, nothing anywhere near the PCM1804 gets touched.
A fake received signal would have to be driven into pins 7 and 9 of IC403, but that is well isolated from the PCM1804.
The power supplies on the QMX+ are complete, exactly as they will be when shipped to the customer.
I can't imagine any reason for the procedure to physically disturb the PCM1804 in a way that would not happen after a bad chip is replaced.
Any handling at QRP Labs would likely be considerably more benign than that of the typical kit builder who is adding dozens of parts.
My understanding is that the PCM1804 seldom (never?) fails once the PCM1804 has been found to work, either during initial testing at QRP Labs
for recent versions of the QDX and QMX+,? or after Jeff has replaced a PCM1804 (which I'd guess is now mostly on the QMX, since
the QDX and QMX+ have already been tested).? Older kits that did not have firmware loaded at QRP Labs don't have the PCM1804 tested.
?
If the above is correct, it sounds to me llke the assembly house is stuffing bad parts.
?
I could easily imagine that shorting the VCC rail (perhaps due to clearance issues around the front panel) would cause a spike in VCC when the short is removed, blowing the PCM1804.? However, if such failures seldom (never?) occur after the receiver has been found to be working correctly,?
that is probably not an issue.
?
A complete bed of nails would not be required to test the PCM1804 on the currently untested QMX.? I would think it would just require applying VCC and VDD, a USB cable to a host computer that has appropriate test software, and appropriate signals at IC403 pins 7 and 9 that are 180 degrees out of phase.??
Getting this working properly could take several days that Hans does not have, but should not take weeks and a major hardware effort.
?
Jerry, KE7ER
?
?
On Fri, Jan 3, 2025 at 05:42 AM, Bruce Akhurst wrote:
I suspect ESD or power supply damage to an incomplete board would not be repeated once fully built.? ESD damage, to ICs in particular, is often evident only after a few weeks or months...
|
I am, in all likelihood, one of those hit by the? PCM1804 problem.? I have read threads about replacing it
and bought replacement chips and the necessary tools, but I doubt my eyesight and? dexterity are equal to the task.
Wrecking my QMX is the most likely outcome.
?
I would rather restart the build with a PCB known to contain a good PCM1804. Buying a new QMX (which would be my third,
since my first was accidentally destroyed) is expensive. If replacement PCBs (just the part containing the PCM1804) were
available, I would buy one, reusing the SMPS, front panel, and all salvageable components.
?
I think Hans could consider making such replacement PCBs available at or near cost to previous buyers of a QMX,
on a one-for-one basis. I also? feel warning should be given to prospective builders that there is (aparently) a 15%
chance of having to replace a SMD component, something which is at the boundary or beyond of what many of us
can tackle.
?
|
Daniel Conklin
Kaufman, Matt - N3AR
Bruce Akhurst