Me again... I just replaced both regulators, being extremely careful to ground myself before touching anything related to them every time. And now it works (almost - I'm getting 11.3V on the neg. rail and 12 on the positive one - I wonder how that's possible since I'm using 1% resistors...)!
Also, the negative side takes quite a while to settle at its final voltage, starting out at like 12.4 and then going down to 11.3 which strikes me as odd - I'm happy I'm at least within 1V of my target voltage though.
Owen, thank you so much for your help! Do you know anything about how these regs are laid out on the inside that could explain how they got fried to give such weird results (+20, -1V)? Also, would a blown C28/C54 explain 20V on one side? Because as I mentioned, with my first build I managed to get one side working correctly by replacing those and cutting around on the PCB trying to cut any shorting traces - which I now know couldn't have done anything to help since I now have a working PSU with an unaltered PCB.
And to anyone else experiencing problems with the new PSU design: It seems these regs are EXTREMELY sensitive to static electricity - it's the only thing that could explain how they both got fried since I was really careful to solder them very quickly. I always grounded myself before touching the regs, but apparently it wasn't enough so when I replaced them, I was even more OCD about being grounded and now it works. Another problem is flux residue - at first, I was getting a fluctuating voltage around 9V on one side, that was fixed by cleaning the PCB with isopropanol and drying it with a hairdryer.
I think these regulators just weren't designed to be hand soldered. On that note, take care to ground the tip of your soldering iron if you have some off-brand piece of crap like I do (not sure if mine's grounded, but before I replaced the regs I touched it to a grounded radiator and now the PSU works, so make of that what you will).
The bottom line is: Buy twice the amount of regulators you'd need to make your PSUs - it's worth not having to put in another order at Mouser, especially if you don't live in the US.
@patchoncas: If your resistors don't show the correct resistance outside of the circuit (i.e. you pulled them off the board to measure), they're definitely blown. Personally, I've never managed to fry an SMD resistor, but I'm sure it's possible. I'm quite certain I did fry one of the 1210 MLCCs on my first PSU though. I always put solder on one of the pads and then heat that up and just slide the part in position with tweezers and when it's in the right place, I remove the soldering iron. That way, you can usually avoid heating things up too much. The trouble with the 1210's is just that they can soak up quite a bit of heat so that technique doesn't work as well as with 0805 parts.
Last edited:
Hi Butizzle,
Very glad to hear you got it up and running!
At this point, I'm a little concerned one of the retailers in the EU has a bad batch of these regs or is selling fakes. I have now soldered at least 40 or 50 of these pretty carelessly (no grounding strap or ESD mat) and have never had a problem. All have come either directly from TI, or from Digikey or Mouser. I have heard of the issue you've described three times now though, and all three times it was in the EU.
Where did you get your parts?
Regards,
Owen
Very glad to hear you got it up and running!
At this point, I'm a little concerned one of the retailers in the EU has a bad batch of these regs or is selling fakes. I have now soldered at least 40 or 50 of these pretty carelessly (no grounding strap or ESD mat) and have never had a problem. All have come either directly from TI, or from Digikey or Mouser. I have heard of the issue you've described three times now though, and all three times it was in the EU.
Where did you get your parts?
Regards,
Owen
I don't think they're blown, because, as I've said, both of them give the same values when diode testing. If one of them was defective, there would be shorts, 0 values or different charging values in the defective one, which doesn't happen.
In the probability that when testing I killed the good reg, there would be a disparity in values still. But the testing checks out.
Could it be the way I wired the transformer? I used one Talema 62053, with the green and red wires powering the -15V rail and the brown and blue wires powering the +15V rail.
All of my parts come from Mouser, they're great because they ship orders for $80+ for free internationally and even pay part of customs/duties. So I doubt I got fakes, but of course a bad batch is always a possibility. I took the ones I'm using now (in the almost working PSU) from the same package as the ones I'd used before, though. Also, I never used a grounding strap or ESD mat, I work on my wood desk and just touch something grounded before handling semiconductor devices.
Do you have any input on why the voltage might be 11.3 on the - rail and 12 on the + ? Is that sort of discrepancy to be expected, or do I still have a bug hiding in my circuit?
Sorry, I misread your last post and then edited it to reflect that - it seems you caught it before I edited it though. Finding out whether it's the transformer should be easy, just check the unregulated voltages after the bridge rectifiers. The electrolytic cap leads are a great point to do that.
As I said in my edited post, if your resistors give the wrong values when off the PCB, they're definitely fried. On the PCB, you can't make any measurements as my results demonstrate - you'll get nonsense readings unless there's a short or a completely open circuit between the leads.
With no load at the output there is a funny behaviour of the circuit (PSU):
Turn it off; put a load at the output to discharge the circuit.
Take the load away and power it on.
Do not touch the regulators, just measure the output voltages.
They should be equal.
Now measure the voltage between "ground" or "+" or "-" and pin 2 of the regulators. By measuring the voltage C17/C55 gets a positive or negative charge.
Now the outputs show up values from 12 to 20V depending on C17/C55's charge.
It takes a few minutes for the output voltage to reset.
With a load at the output this will not happen (output voltage keeps stable).
Maybe kind of that is your issue?
Turn it off; put a load at the output to discharge the circuit.
Take the load away and power it on.
Do not touch the regulators, just measure the output voltages.
They should be equal.
Now measure the voltage between "ground" or "+" or "-" and pin 2 of the regulators. By measuring the voltage C17/C55 gets a positive or negative charge.
Now the outputs show up values from 12 to 20V depending on C17/C55's charge.
It takes a few minutes for the output voltage to reset.
With a load at the output this will not happen (output voltage keeps stable).
Maybe kind of that is your issue?
Last edited:
I just tried it with my (newly assembled!) Wire headphone amp attached to the output, and it's still asymmetrical - the negative rail keeps rising very slowly towards 12V though. I let it run for about 3 minutes, it started at 11.3, rising rather quickly but then it decelerated and after that time it ended up at 11.7 when I disconnected it again, so maybe you're right...
Edit: I just left it running for like 10 minutes and at the end, the neg. rail was at 11.87V with the positive rail staying put at 12.04 all the time. I didn't check it when I first plugged it in though, so I don't know where it started out. What's bugging me about this stuff is that the positive rail won't budge at all, it's frozen at 12.04, whereas the negative rail has this "warming up" period...
I also think that someone should update the build wiki to reflect that:
-there's a new PSU design with a different BOM (I know the BOM's already on the wiki, but the new PSU isn't mentioned anywhere in the article body) and that there may be issues with the regulators - either a bad batch or just that they're very sensitive
-cleaning the flux residue is NOT an optional step. I forgot to clean my amp board when I first tried it out and almost toasted my speaker (I'll be using it as a preamp and I didn't have any headphone jacks laying around so I just jury rigged one channel into the integrated amp it'll be used in and tried it on my speakers). Turns out there was very ugly noise of about 2V in amplitude (might be more, the multimeter obviously couldn't follow it) at the output, with no input attached! After a quick cleaning with some iso, that stopped and now it's working as it should. The PSU board also didn't work correctly until I cleaned it, as I mentioned above.
On another note, I'll see about taking the whole thing to university and hooking some stuff up to an oscilloscope so I can figure out if there's something still bugging out. For example, there could be some noise from the PSU - that difference in output voltages makes me very suspicious. My DMM says 0-1mV AC on both rails, but I don't quite trust it on these matters.
Last edited:
Hi !
Just to share experience
:
Just received all the component from mouser. Started to solder...My solder tip is round, around 1,5mm. As i didn't have the 1,24k resistor on mouser for psu, i managed to solder a bridge between the two pad, with two smd resistor to get the same value.
Soldering worked not too bad for the first 15 component, until, one smd resistor didn't wanted to stay in place, and small composite plastic from composite tweezer burnt on it. Small part of composite tweezer stayed glued to the resistor, trying to solder again, the composite glued the resistor to the solder tip, and so one of the resistor pad get off trying to separate from the tip^^... resistor, is, ******* dead.
So, i'll try to work on my solder tip to make it with a more suitable form before restart, ordered some cardas solder, flux, and another smd resistor on ebay, and i'll use my wife's precision metallic tweezer
Just to share experience
:Just received all the component from mouser. Started to solder...My solder tip is round, around 1,5mm. As i didn't have the 1,24k resistor on mouser for psu, i managed to solder a bridge between the two pad, with two smd resistor to get the same value.
Soldering worked not too bad for the first 15 component, until, one smd resistor didn't wanted to stay in place, and small composite plastic from composite tweezer burnt on it. Small part of composite tweezer stayed glued to the resistor, trying to solder again, the composite glued the resistor to the solder tip, and so one of the resistor pad get off trying to separate from the tip^^... resistor, is, ******* dead.
So, i'll try to work on my solder tip to make it with a more suitable form before restart, ordered some cardas solder, flux, and another smd resistor on ebay, and i'll use my wife's precision metallic tweezer
Don't know if it helps;
here are some diode testing values (working TPS7A33)
Measured with Agilent U1273A
from pin (polarity) # to pin (polarity) # value
1- # 2+ # OL
1- # 3+ # 1.54V
1- # 4+ # OL
1- # 5+ # OL
1- # 6+ # OL
1- # 7+ # OL
2- # 3+ # 0.82V
2- # 4+ # 2.37V
2- # 5+ # OL
2- # 6+ # 2.44V
2- # 7+ # 2.44V
3- # 4+ # OL
3- # 5+ # OL
3- # 6+ # OL
3- # 7+ # OL
4- # 5+ # OL
4- # 6+ # 0,77V
4- # 7+ # 0.77V
5- # 6+ # OL
5- # 7+ # OL
6- # 7+ # OL
1+ # 2- # OL
1+ # 3- # OL
1+ # 4- # OL
1+ # 5- # OL
1+ # 6- # OL
1+ # 7- # OL
2+ # 3- # OL
2+ # 4- # 0.76V
2+ # 5- # OL
2+ # 6- # OL
2+ # 7- # OL
3+ # 4- # 0.79V
3+ # 5- # OL
3+ # 6- # 0.79V
3+ # 7- # OL
4+ # 5- # OL
4+ # 6- # OL
4+ # 7- # OL
5+ # 6- # OL
5+ # 7- # OL
6+ # 7- # OL
My PSUs work pefectly; no voltage difference/ shifting / instability.
+15.05V
- 14.99V
Regards Martin
here are some diode testing values (working TPS7A33)
Measured with Agilent U1273A
from pin (polarity) # to pin (polarity) # value
1- # 2+ # OL
1- # 3+ # 1.54V
1- # 4+ # OL
1- # 5+ # OL
1- # 6+ # OL
1- # 7+ # OL
2- # 3+ # 0.82V
2- # 4+ # 2.37V
2- # 5+ # OL
2- # 6+ # 2.44V
2- # 7+ # 2.44V
3- # 4+ # OL
3- # 5+ # OL
3- # 6+ # OL
3- # 7+ # OL
4- # 5+ # OL
4- # 6+ # 0,77V
4- # 7+ # 0.77V
5- # 6+ # OL
5- # 7+ # OL
6- # 7+ # OL
1+ # 2- # OL
1+ # 3- # OL
1+ # 4- # OL
1+ # 5- # OL
1+ # 6- # OL
1+ # 7- # OL
2+ # 3- # OL
2+ # 4- # 0.76V
2+ # 5- # OL
2+ # 6- # OL
2+ # 7- # OL
3+ # 4- # 0.79V
3+ # 5- # OL
3+ # 6- # 0.79V
3+ # 7- # OL
4+ # 5- # OL
4+ # 6- # OL
4+ # 7- # OL
5+ # 6- # OL
5+ # 7- # OL
6+ # 7- # OL
My PSUs work pefectly; no voltage difference/ shifting / instability.
+15.05V
- 14.99V
Regards Martin
Hi Owen,
Have you ever tried to measure the distortion with a good notch filter after your amplifier? I wouldn't be surprised if the distortion is even lower than what you have measured in post no. 2 and 3. Perhaps these measurements just show the limitations of the AP system?
The reason I ask is that I am thinking of using a similar concept for the generator output of an audio analyzer that I am working on. One objective is of course to get a very low distortion, even with low load impedances. In this case it should be well below the distortion of the DUT.
Have you ever tried to measure the distortion with a good notch filter after your amplifier? I wouldn't be surprised if the distortion is even lower than what you have measured in post no. 2 and 3. Perhaps these measurements just show the limitations of the AP system?
The reason I ask is that I am thinking of using a similar concept for the generator output of an audio analyzer that I am working on. One objective is of course to get a very low distortion, even with low load impedances. In this case it should be well below the distortion of the DUT.
You can use the filtering method to notch out the fundamental and compare the harmonics to the original signal.
I believe some distortion measuring equipment uses this method.
You have to know how much of the 2nd and 3rd harmonics are attenuated with the fundamental filter.
If you know this you can correct the readings to what the unattenuated harmonics should read.
I believe some distortion measuring equipment uses this method.
You have to know how much of the 2nd and 3rd harmonics are attenuated with the fundamental filter.
If you know this you can correct the readings to what the unattenuated harmonics should read.
JensH,
Owens most recent measurement posts are here http://www.diyaudio.com/forums/showthread.php?p=3699976
He's changed a few things in his setup but AFAIK hasn't used a notch in any of his measurements.
Chris
Hey, has anyone ever had a weird scratching noise on one channel? I'm getting this weird noise that sounds a bit like someone's crumpling newspaper or something, the same noise you get from a bad/dirty contact on a tube amp. I'm getting this only on the right channel. It changes in volume with the volume pot setting (an Alps RK27 wired after the input RCA sockets) and disappears when the input on that channel is shorted. Could this be a faulty potentiometer? I'm thinking that's not it, because moving/shaking the potentiometer and/or wiggling the wiring has no effect on the noise, which sometimes changes in intensity for no obvious reason. I guess I'll have to remove the pot from the signal path to find out...
This is really bothering me, as it's loud enough to be heard over the music (and even if it wasn't it would drive me insane...).
This is really bothering me, as it's loud enough to be heard over the music (and even if it wasn't it would drive me insane...).
Yep, that is a textbook description of a bad pot.
They won't make noise when you rustle the wiring, only when you turn them, and often is just in one channel. Shorting the input makes it go away because with the input grounded there is no voltage across the bad pot element - nothing to hear.Yep, that is a textbook description of a bad pot.
Aw crap, now I'll have to shell out another 15 bucks for a new pot. Can this be caused by overheating one of the contacts? If so, I think I'll go for a few of AMB's RK27 mounting boards as well - don't wanna risk breaking another pot.