Do you get hum if you short the P101 input terminals?
Anyways, if you need to put your ear right against the speaker to hear anything I guess it's good enough.
Sparks at the caps is a bad thing. You need to solve that whatever it is.
Also, install a speaker protection circuit to protect them from any possible thump or DC presence there.
Anyways, if you need to put your ear right against the speaker to hear anything I guess it's good enough.
Sparks at the caps is a bad thing. You need to solve that whatever it is.
Also, install a speaker protection circuit to protect them from any possible thump or DC presence there.
No, all there is with a shorted P101 input is some white noise. It's barely audible on the speakers, I can only really hear it with the IEMs. Even with my ear pressed against the tweeter it's extremely faint.
Yeah, I figured... They haven't appeared again since though. Really weird stuff, because they weren't between the terminals or anything like that. They seemed to be on the plastic isolation. If they come back, I guess I'll just have to replace the PSU caps. Right now they appear to be working fine though, they're no hotter than usual and even when the sparks were there, they seemed to have no effect on the operation of the PSU.
I'll report back when I've tried the other cable, thanks again for your help! I might also wire up the right channel today, my guess is there will be a lot less or no hum with the good cable (I still have some terminated cables that I can use for the right channel, but I ran out of the little terminals to make one of the right length for the left one; more should arrive tomorrow) since the pot->amp distance is so much smaller.
Last edited:
You should NOT hear any hum nor any noise at the speaker when the input is shorted.
Even measuring should detect no hum+noise, i.e. <0.1mVac This is completely inaudible at the speaker with 90dB/W @ 1m sensitivity.
0.1mVac into 8ohms is 1.25nW. That would be 59dB below audibility @ 1m or ~30dB below audibility @ 0.1m
Listening to hum and noise at the speaker is not a valid comparative test.
Using sensitive headphones (after you have confirmed <10mVac and <5mVdc) is a better audibility test.
The headphone audibility test is useful for differentiating, hum from buzz, from white noise, from pink noise.
You can play safe and use a pair of DC blocking capacitors to protect the headphone voice coils. 22uF for 300ohms passes 50Hz with a loss of ~1dB.
Last edited:
Well, after listening to some youtube videos of pink and white noise I'm pretty sure it's white noise the amp is producing. I'm not 100% certain though, the noise is really quiet even with earphones.
I don't have anything to reliably measure the AC noise, unfortunately.
Also, at 0.1m there is no chance whatsoever of hearing anything from the speaker. I really have to press my ear against the tweeter to be able to barely hear it, and even then it's very difficult to make out. And on the earphones it's also very quiet.
Shame your DMM cannot be set to 200.0mVac. That allows us to compare actual measurements.
We can't compare volume, or the lack of it, from unknown speakers/headphones.
The earphones are the standard crap you get with Samsung phones, 32 Ohms impedance. The speaker is an Energy Point 1e, specified as 86dB at 1m and 2.83V.
I might go to the university tomorrow and take the amp with me, there's a student lab with pretty good scopes and it might be open. That should give us a more detailed picture.
Just re did this test on mine with the volume pot all the way down , and the reading is below the meter sensitivity.
An externally hosted image should be here but it was not working when we last tested it.
I just noticed something else: Moving the wire that supplies the 12V to the soft start board causes microphonic noises when it scrapes against the bottom plate of the chassis. However, feeding the P39 with 12V from an external supply makes no difference to the white noise, neither does removing all ground connections of anything that doesn't belong to the P101 circuit. It really seems like the noise is either produced or picked up by some part of the P101.
The student lab is closed today, but when I go there I'll also get the chance to hook up the P101 to a lab supply, so I can rule out the PSU as the source.
The student lab is closed today, but when I go there I'll also get the chance to hook up the P101 to a lab supply, so I can rule out the PSU as the source.
Well, I just replaced the cable with a twisted pair shielded one and the hum remains. I didn't change anything else, so the white noise is obviously also still here.
I'm really starting to thing it's impossible to get this amp noise-free. My grounding scheme seems to be completely loop-free and all the signal paths are close-coupled with their return paths. I don't know what else I could be doing. It just won't stop making noise, and I'd feel pretty dumb closing up an amp that still produces measurable/audible hum and white noise.
I also noticed that it seems to be possible to send either stage into oscillation (?) by arranging wires a certain way - for example, if the 12V return from the P39 board has a shoddy connection to star ground, the preamp will start humming very loudly (the noise varies with the volume control. And if certain ground wires are run right next to the small-signal cables to the P101 so that they touch, it'll start humming very loudly too. I'm guessing this is oscillation because it behaves like an on/off switch - either it's humming or it's not, which is unlike the buzzing picked up by the small-signal connections that will vary gradually with the positioning of the wire.
The lab was closed again today, so still no measurements.
I'm really starting to thing it's impossible to get this amp noise-free. My grounding scheme seems to be completely loop-free and all the signal paths are close-coupled with their return paths. I don't know what else I could be doing. It just won't stop making noise, and I'd feel pretty dumb closing up an amp that still produces measurable/audible hum and white noise.
I also noticed that it seems to be possible to send either stage into oscillation (?) by arranging wires a certain way - for example, if the 12V return from the P39 board has a shoddy connection to star ground, the preamp will start humming very loudly (the noise varies with the volume control. And if certain ground wires are run right next to the small-signal cables to the P101 so that they touch, it'll start humming very loudly too. I'm guessing this is oscillation because it behaves like an on/off switch - either it's humming or it's not, which is unlike the buzzing picked up by the small-signal connections that will vary gradually with the positioning of the wire.
The lab was closed again today, so still no measurements.
That's very strange. The star is also firmly attached and making good connection with the chassis, isn't it?
Yes, I just re-tightened it. The problems with the small-signal and ground wires only occurred before I rearranged everything - now the small-signal and ground wires are routed so far apart that there's no chance of them touching. When everything has a good connection, there are no problems right now, I just thought it a bit odd that the P39 ground connection should have such an effect.
I'm really wondering where the buzzing is coming from now... Perhaps it's picked up by the pot itself? I can't seem to get it to change in volume by moving the pot->P101 cable now, except if I move it really far to the left into the noisy area around the transformers/PSU. And the white noise is pretty strange too, I haven't found any mention of this from other P101 builders, but then again it's really quiet so perhaps nobody noticed it...
I also managed to break my soldering iron today while trying to repair a Wire PSU. I'd removed the regulator and was trying to wick the solder out of the holes when it just snapped. Those PSUs keep breaking or just not working right after assembly on me for some reason - I think the TPS7A3301 regulators must be extremely sensitive to static discharge, this one crapped out after I'd used it to power the P39's relay for a while. The rail I wasn't using to power the P39 just stopped working and all I did was move it across my desk twice. It's a shame because those regs are also super expensive at $7 a pop and I have to get them from Mouser. And replacing the regulators once the PSU is assembled is quite the pain in the backside as well.
I'm really wondering where the buzzing is coming from now... Perhaps it's picked up by the pot itself? I can't seem to get it to change in volume by moving the pot->P101 cable now, except if I move it really far to the left into the noisy area around the transformers/PSU. And the white noise is pretty strange too, I haven't found any mention of this from other P101 builders, but then again it's really quiet so perhaps nobody noticed it...
I also managed to break my soldering iron today while trying to repair a Wire PSU. I'd removed the regulator and was trying to wick the solder out of the holes when it just snapped. Those PSUs keep breaking or just not working right after assembly on me for some reason - I think the TPS7A3301 regulators must be extremely sensitive to static discharge, this one crapped out after I'd used it to power the P39's relay for a while. The rail I wasn't using to power the P39 just stopped working and all I did was move it across my desk twice. It's a shame because those regs are also super expensive at $7 a pop and I have to get them from Mouser. And replacing the regulators once the PSU is assembled is quite the pain in the backside as well.
Ground connections always have to be good and firm otherwise there'd be issues.
Regarding the star, use 1 long bolt firmly attached to chassis (remove the chassis anodizing right around the bolt or it will NOT make a good contact), then use 1 nut and washer per not more than 2 ring style wire connectors.
Also, use nylon or fiber washers to isolate the pot from the chassis. Tape will probably fail under pressure upon time and make a short to chassis.
The level of white noise depends on the amplifier design and components chosen to build it. It's unrelated to the ground scheme.
Regarding the voltage regulator problem I don't think it's about static discharge. Check out the power dissipation and keep below the maximum ratings stated in its datasheet.
Regarding the star, use 1 long bolt firmly attached to chassis (remove the chassis anodizing right around the bolt or it will NOT make a good contact), then use 1 nut and washer per not more than 2 ring style wire connectors.
Also, use nylon or fiber washers to isolate the pot from the chassis. Tape will probably fail under pressure upon time and make a short to chassis.
The level of white noise depends on the amplifier design and components chosen to build it. It's unrelated to the ground scheme.
Regarding the voltage regulator problem I don't think it's about static discharge. Check out the power dissipation and keep below the maximum ratings stated in its datasheet.
Attachments
The voltage regulator wasn't even being used when it broke - I wasn't even using the rail that failed, and all the other rail had to do was supply the 50mA or so of relay coil current for the P39, which these regulators can take easily. You can see the heatsinks in the pictures, they should hold up fine up to at least 0.5A (that would be 4W of dissipation, these sinks are probably 10-15° C/W so that should not make any problems as the regs are rated at 100°+). I must be doing something wrong with these PSUs but I've no idea what it is, I've had trouble with every single one I've built now and always had to swap out regulators.
Thanks for the tip about the plastic washers for the pot - I'm gonna get a few when I get the chance.
The grounding is already pretty much the way you described it, although I'm not using so many washers and nuts yet because it gets annoying when you have to redo the connections - when (if) I ever solve the buzz problem or just decide that it's at a tolerable level (I'm guessing that's the only choice I've got at this point) I'll do it properly though.
Also, I've noticed that separating preamp and P101 grounds helps a tiny bit with the buzzing, but nothing major. It's already pretty quiet now so I might just leave it at this.
What would be your guess as to which components are responsible for the white noise? I'm thinking about replacing all the input stage components of the P101 with matched transistors and high quality resistors and caps to get rid of the DC offset (12mV right now with unmatched transistors, so matching the LTP should get me below 5mV or so) and improve sound quality. I'm also a bit skeptical about the use of a ~15cm wire to connect two ground points on the amp (it connects the ground loop breaker from the input GND to PGND on the other side of the board), but I'm not sure whether this might have anything to do with the white noise.
Thanks for the tip about the plastic washers for the pot - I'm gonna get a few when I get the chance.
The grounding is already pretty much the way you described it, although I'm not using so many washers and nuts yet because it gets annoying when you have to redo the connections - when (if) I ever solve the buzz problem or just decide that it's at a tolerable level (I'm guessing that's the only choice I've got at this point) I'll do it properly though.
Also, I've noticed that separating preamp and P101 grounds helps a tiny bit with the buzzing, but nothing major. It's already pretty quiet now so I might just leave it at this.
What would be your guess as to which components are responsible for the white noise? I'm thinking about replacing all the input stage components of the P101 with matched transistors and high quality resistors and caps to get rid of the DC offset (12mV right now with unmatched transistors, so matching the LTP should get me below 5mV or so) and improve sound quality. I'm also a bit skeptical about the use of a ~15cm wire to connect two ground points on the amp (it connects the ground loop breaker from the input GND to PGND on the other side of the board), but I'm not sure whether this might have anything to do with the white noise.
Can you be bothered enough to take the whole assembly out of the Chassis and put the Chassis on the shelf?
Then get each module working properly alone and stage by stage assemble modules together getting each new assembly working properly before adding another module.
If the noise reappears at some extra stage you know where the noise is coming from. It can ONLY be from the added stage, or the effect the added stage has on what is directly connected to it.
Then get each module working properly alone and stage by stage assemble modules together getting each new assembly working properly before adding another module.
If the noise reappears at some extra stage you know where the noise is coming from. It can ONLY be from the added stage, or the effect the added stage has on what is directly connected to it.
I'm already pretty sure about where everything is introduced. The White noise is coming from the P101 (can't say whether it's PSU or the amp but my guess is the amp - taking it out of the chassis won't help me answer this question though) and the buzzing is introduced between in the cable between the pot and the P101 - unplugging it from the pot and shorting it but leaving it in place has the same effect as having it plugged into the pot on 0 volume. Placing the cable (or the whole pot assembly) behind the amp chassis as far as the cables allow silences everything except the white noise completely, but of course I'd like to have it working like that in the chassis. I guess that's not possible with my crappy arrangement though, so I'll probably just leave it the way it is now and buy some inefficient speakers to use it with.
The transistors in the input differential amplifier have to be matched if you expect the stage to perform as expected. PSRR will improve (if actual transistors are not matched ones)
Also use metal film resistors for lowest noise.
Do not use a wire to interconnect 2 ground points in the board. As I stated way above run 2 separate wires to the star.
Also, even though is extra work re arrange the star connections in the way described. This is like what was first, the chicken or the egg; proper star ground connections will keep noises to the lowest level. Just 1 bad ground connection and all the effort would be futile.
Also use metal film resistors for lowest noise.
Do not use a wire to interconnect 2 ground points in the board. As I stated way above run 2 separate wires to the star.
Also, even though is extra work re arrange the star connections in the way described. This is like what was first, the chicken or the egg; proper star ground connections will keep noises to the lowest level. Just 1 bad ground connection and all the effort would be futile.
For fault finding you need to reduce the amount of variables. Is it the transformer, chassis, ground layout, potentiometer, faulty amp or something else? It may seem time consuming but if you follow Andrew's advice it will save you time and produce a better amp.
It must be obvious by now that implementing a magic layout is not the solution.
It must be obvious by now that implementing a magic layout is not the solution.
To put all this in perspective, I do not claim to have developed this ground scheme, and also did not proposed it as the whole and only solution.
It is a star ground scheme in its most strict sense, just that.
If you re read the entire thread you would notice that, in a hum scale of 10 he was stuck at may be 5 with the previous ground scheme, and now he appears to be probably at 9.
It is obvious that implementation and components arrangement is also important and looks like it still is not its best, affecting performance.
Some other issue unrelated to ground layout appears to be going on as well.
It is a star ground scheme in its most strict sense, just that.
If you re read the entire thread you would notice that, in a hum scale of 10 he was stuck at may be 5 with the previous ground scheme, and now he appears to be probably at 9.
It is obvious that implementation and components arrangement is also important and looks like it still is not its best, affecting performance.
Some other issue unrelated to ground layout appears to be going on as well.
- Status
- Not open for further replies.