Today I feel like the biggest idiot in town.
I have below the desk the first part of an amplifier for my active 3-way. I made a couple of months ago the first 2 LM3886 pcb's and put them in an old 19" rack I had around. At that time I also wired up the transfo, a little board with fuses, the rectifier/capacitor pcb, mounted the LM3886 pcb's to their heatsinks and mounted the heatsinks on the chassis. Because I missed some connectors it then went under the desk and other projects took over.
Now I pulled it out, fired up the psu and checked if that was ok. Made the wiring for both LM3886 pcb's and connected the first one.
Switched on, no bang, no smoke, nothing got hot and the protection relay activated after a couple seconds. So all good! I connected a loose speaker and a function generator and fired up again.
Nothing, absolutely nothing. Even the relay didn't activate. So I pulled out the scope. Signal was at the input. Nothing at the output but 1V offset. Strange. Connected the second module.... idem ditto.
Ok, then signal tracing with the scope. I touch the first pin of the LM3886.... hey, this seems to move!
Ok, so the idiot writing usual mounted the LM3886 to the heatsink, then the pcb with angle brackets to the heatsink and as a last thing HE SHOULD HAVE SOLDERED THE LM3886 TO THE PCB!
I feel so stupid... Going to solder them tomorrow and see if they are OK.
I have below the desk the first part of an amplifier for my active 3-way. I made a couple of months ago the first 2 LM3886 pcb's and put them in an old 19" rack I had around. At that time I also wired up the transfo, a little board with fuses, the rectifier/capacitor pcb, mounted the LM3886 pcb's to their heatsinks and mounted the heatsinks on the chassis. Because I missed some connectors it then went under the desk and other projects took over.
Now I pulled it out, fired up the psu and checked if that was ok. Made the wiring for both LM3886 pcb's and connected the first one.
Switched on, no bang, no smoke, nothing got hot and the protection relay activated after a couple seconds. So all good! I connected a loose speaker and a function generator and fired up again.
Nothing, absolutely nothing. Even the relay didn't activate. So I pulled out the scope. Signal was at the input. Nothing at the output but 1V offset. Strange. Connected the second module.... idem ditto.
Ok, then signal tracing with the scope. I touch the first pin of the LM3886.... hey, this seems to move!
Ok, so the idiot writing usual mounted the LM3886 to the heatsink, then the pcb with angle brackets to the heatsink and as a last thing HE SHOULD HAVE SOLDERED THE LM3886 TO THE PCB!
I feel so stupid... Going to solder them tomorrow and see if they are OK.
WHS. I love my protective gear. This dust mask works a treat when cutting wood, particularly MDF. These push blocks are way more effective than any push stick, and for narrow cuts the Grripper is my tool of choice. Ear defenders are obligatory of course: what's the point of building systems for music if you leave your hearing at the table saw? Safety goggles came in useful when a Dremel cutting blade disintegrated before touching the workpiece and a shard bounced off the lens.
The usual precautions apply: dry runs with the table saw or router before powering them up to make sure everything has enough clearance, that sort of thing.
The usual precautions apply: dry runs with the table saw or router before powering them up to make sure everything has enough clearance, that sort of thing.
I think we all have these moments. One of my best was getting the footprint wrong on all 20 SOT-23 devices on a board I designed a couple of years ago, and discovering it as I soldered transistor #20 to the board.
The SOT-23 transistors were not a big loss, cannot say the same for the time wasted soldering 38 transistors to a pair of boards, the boards themselves and countless expensive passive parts that went in the trash at the end of it. No hope of salvage, but the replacements have served me well for over a year now. And I learned an important lesson, I now double and triple check the footprints on different days. (I don't have anyone who can review layouts for errors for me.)
I've also soldered mislabeled parts in the wrong location, but those are usually pretty easy to find. (railed outputs)
The SOT-23 transistors were not a big loss, cannot say the same for the time wasted soldering 38 transistors to a pair of boards, the boards themselves and countless expensive passive parts that went in the trash at the end of it. No hope of salvage, but the replacements have served me well for over a year now. And I learned an important lesson, I now double and triple check the footprints on different days. (I don't have anyone who can review layouts for errors for me.)
I've also soldered mislabeled parts in the wrong location, but those are usually pretty easy to find. (railed outputs)
Well, soldered everything up now and I fear I have destroyed both LM3886. Both go to -V offset at powerup. 😭
I’ve had quite a few PCBs with layout errors. Usually benign enough that a blue wire or drilling new holes for a component fixes it. But then there was the one where the negative rail wouldn‘t come up. Copper pours connected on both sides of the capacitor. Ended up using a dremel to hog out a ring around the positive terminal. Long before that was the base-emitter of a driver transistor shorted. One channel didn’t quite bias up the same as the other but it worked so I used it. A little distortion at war volume but ran fine, and used it that way for two years. Then one night I realized how hot the predriver was getting. At least the trace that needed cutting wasn’t on the heat sink side. Replaced the predriver for good measure too. Then guess what - it biases up properly with the same pot setting as the other channel.
Does one consider having to replace 44 op amps because the negative regulator went into oscillation “my fault”? Then the 7815 and 7915 got replaced with zener shunt regulators and a pair of pass transistors. No way I’m doing that TWICE.
Does one consider having to replace 44 op amps because the negative regulator went into oscillation “my fault”? Then the 7815 and 7915 got replaced with zener shunt regulators and a pair of pass transistors. No way I’m doing that TWICE.
Worse thing I saw was when a colleague designed a pcb and put the connections to a SSOP20 in mirror on it. He had to mount the IC's upside down with a drop of glue and then connect all pins with wire-wrap wire to the pcb. On all 5 IC's 🙂
Just replaced the LM3886 on the first pcb. Fingers crossed.
Just replaced the LM3886 on the first pcb. Fingers crossed.
Any other solder connections need redoing?
What voltages do you get at the pins?
Using the datasheet design? Something proven or something new?
What voltages do you get at the pins?
Using the datasheet design? Something proven or something new?
UPDATE2: monday morning when the post office opens I'm going to send back my degree in electronic engineering. I'm clearly becoming too old to think. 😭
There was nothing wrong with other solder connection. I touched some that weren't nice (used lead free solder) but other than that nothing going on. I even measured between pins of components to rule out solder connections. Next I wired up a smaller (and lower voltage) PSU as it isn't comfortable working inside the chassis. I meant to measure the voltages at each pin/point.
I got -10.3V just like with the other psu. But then my brain kicked in for a second... what can cause an amp to go to V-? I know it isn't oscillating. It means there is "some" voltage across the +input and -input. Then I looked at my pcb 😱 and felt ashamed and angry.
I have designed the pcb with the intention of having the signal GND and power GND separate. So in fact, the input circuit, the GND connection of the LM3886 and feedback GND were floating. Meaning any offset got the full open loop gain. No wonder it hit the rail....
Made a connection from the signal GND to the power GND on the board and offset is now 3mV which is perfect.
I feel such a fool... That's why when you do a project you should go to the end, not stop halfway and then let it gather dust for 6 months. You will have forgotten details like that and spend too much time chasing ghosts.
There was nothing wrong with other solder connection. I touched some that weren't nice (used lead free solder) but other than that nothing going on. I even measured between pins of components to rule out solder connections. Next I wired up a smaller (and lower voltage) PSU as it isn't comfortable working inside the chassis. I meant to measure the voltages at each pin/point.
I got -10.3V just like with the other psu. But then my brain kicked in for a second... what can cause an amp to go to V-? I know it isn't oscillating. It means there is "some" voltage across the +input and -input. Then I looked at my pcb 😱 and felt ashamed and angry.
I have designed the pcb with the intention of having the signal GND and power GND separate. So in fact, the input circuit, the GND connection of the LM3886 and feedback GND were floating. Meaning any offset got the full open loop gain. No wonder it hit the rail....
Made a connection from the signal GND to the power GND on the board and offset is now 3mV which is perfect.
I feel such a fool... That's why when you do a project you should go to the end, not stop halfway and then let it gather dust for 6 months. You will have forgotten details like that and spend too much time chasing ghosts.
Delighted to hear that you have sorted it out. I think you can keep that degree. The human in you made the mistake, the EE in you found and fixed it.. 😀
We've all had those moments. 😀
We've all had those moments. 😀
Well, well... I finished the four amplifiers in the housing and connected it to the crossover and then the speakers. I have sound so no complaints here. But I also had a 50Hz on all speakers. Ok, very low level so the tweeter didn't fry and you only hear it between tracks. But where does it come from. Most obvious is a ground loop.
So I disconnected the safety ground to the starpoint. And got a far worse hum! Strange, I would have expected it to disappear. Ok, maybe I should connect it to the input pcb. Again worse hum. Removed all but one input cinch. Hum gone. Ok, so change the grounding as per Douglas Self book: safety ground to input pcb and then the low level ground of each amp to the starpoint from the amp pcb.
Before powering up, a quick sanity check:
Ok, so I made a pcb where I could make various ground/earth connections. From each cinch shell to the ground, a single point off the input pcb to connect a spade, from the pcb ground to the chassis mounting... All fine and dandy, but when you do that, don't forget that you need at least a single ground connection to your audio source!
So the ONLY connection to the digital crossover was safety earth. That's why it made more hum when that was broken.
Now I still wonder how I got sound at all if none of the cinches had any ground connection.
Next: build another 4 of those amplifier boards and this time the first time right.
So I disconnected the safety ground to the starpoint. And got a far worse hum! Strange, I would have expected it to disappear. Ok, maybe I should connect it to the input pcb. Again worse hum. Removed all but one input cinch. Hum gone. Ok, so change the grounding as per Douglas Self book: safety ground to input pcb and then the low level ground of each amp to the starpoint from the amp pcb.
Before powering up, a quick sanity check:
- each input of the amplifier boards goes to starpoint: check
- safety to starpoint: check
- no safety ground to starpoint if I break the connection at the input pcb: check
- from the shell of each cinch input to starpoint: no connection?????????
Ok, so I made a pcb where I could make various ground/earth connections. From each cinch shell to the ground, a single point off the input pcb to connect a spade, from the pcb ground to the chassis mounting... All fine and dandy, but when you do that, don't forget that you need at least a single ground connection to your audio source!
So the ONLY connection to the digital crossover was safety earth. That's why it made more hum when that was broken.
Now I still wonder how I got sound at all if none of the cinches had any ground connection.
Next: build another 4 of those amplifier boards and this time the first time right.