You are getting closer to the solution, vcelkamaja. This is also the hardest part of the de-bugging stage - patience.
Reading through the thread comments since yesterday, I see a lot of good suggestions.
Based on your latest description of the problem, the higher-frequency "buzz" coming from the tweeter, it sounds like you have solved the AC heater-hum problem in both channels and are now hearing another problem that was masked by the AC heater-hum you originally had.
From the photos and your description, it looks like your input coax cables lay across (or very near) the power transformer when you have the amplifier fully assembled. You do not hear the hum or the buzz when you have the amplifier board out of the chassis - and presumably when the amplifier is out of the chassis your input cables are moved further from the power transformer.
A couple of things.
First, you say you are using coaxial cable for your audio input. I assume by this you mean a length of RF coaxial cable (single center conductor with outer shield). This type of cable will not provide magnetic shielding, and can easily pick up noise from the power transformer magnetic field.
Toroids are great, and generate very little magnetic field - but they are not perfect. They will most certainly generate a small residual field, and if your input audio cable runs near your power transformer you will pick up a bit of that noise and feed it into that high-gain 12AX7 stage.
When you grounded the grid of the 12AX7 earlier, you still had hum because the AC heater was the dominant problem. Can you repeat the experiment and ground the 12AX7 input grid (the bottom triode, not the top triode of the SRPP stack)?
If, when you ground your input grid, the noise goes away then replace the coax with twisted pair AND route the input cables well away from the power transformer.
Or, at least try a longer piece of coax and route it well-away from the power transformer. You can make twisted-pair chucking up a couple of lengths of 22 or 24 gauge wire in a drill to get them very tightly twisted - 6 turns per inch minimum. (sorry for not using SI units - dang Americans...).
If you can get shielded twisted pair, so much the better. But the important thing is to get the input cables away from your power transformer.
Second, as I posted earlier - IF your power transformer has only a single 6.3VAC winding, and if you are connecting that winding to both the F1 and F2 connections, then you only need a single bias network. The bias will raise all of the heaters, as the DC bias will pass to all the heaters through the transformer winding.
Of course, if you have two separate 6.3VAC windings then you will need to provide two bias networks.
Note, if you have a single 6.3VAC winding then placing two bias networks will not be the end of the world - but you will have two resistive dividers working against each other for no good purpose. You will still get a bias voltage of some kind, but it's not a good idea.
Also, can you verify whether you are using 220 nanofarad or 220 MICROfarad capacitors? The difference is significant - you should be using at least 220 MICROFarads across the 10k (or 20k or 30k) resistor to ground. You need to filter the bias voltage well, and 220 nanoFarads (0.22 microfarads) will not be sufficient.
As I said earlier, I believe you have eliminated the first hum-problem by using the 30V heater bias circuit. You are now hearing another problem that was likely present all along, but masked by the much louder (?) heater hum.
Finally, a word of caution. DO NOT poke around inside that amplifier with your fingers while it is plugged in!! The voltages in there are lethal - if you accidently get across the B+ this will go from a hobby to a tragedy!!
If you absolutely must wiggle wires while the thing is plugged in or turned on, please at least take some precautions:
1) USE A LONG (15cm or so) PIECE OF DRY, UNPAINTED WOODEN DOWEL,
2) USE ONLY ONE HAND TO MANIPULATE THE DOWEL,
and
3) KEEP YOUR OTHER HAND IN YOUR BACK PANTS POCKET AT ALL TIMES WHEN YOU ARE POKING INSIDE THE ACTIVE AMPLIFIER!!
Seriously.
Us older guys who grew up with tube radios all know this. And the ones who grew up to BE old guys remembered to DO these things!
Do NOT use a screwdriver, your fingers, your friend's fingers or anything else. Accidents can be prevented.
To paraphrase the old adage: "There are old electricians and there are bold electricians - but there are NO OLD BOLD ELECTRICIANS"
Best luck, and please be safe.
Reading through the thread comments since yesterday, I see a lot of good suggestions.
Based on your latest description of the problem, the higher-frequency "buzz" coming from the tweeter, it sounds like you have solved the AC heater-hum problem in both channels and are now hearing another problem that was masked by the AC heater-hum you originally had.
From the photos and your description, it looks like your input coax cables lay across (or very near) the power transformer when you have the amplifier fully assembled. You do not hear the hum or the buzz when you have the amplifier board out of the chassis - and presumably when the amplifier is out of the chassis your input cables are moved further from the power transformer.
A couple of things.
First, you say you are using coaxial cable for your audio input. I assume by this you mean a length of RF coaxial cable (single center conductor with outer shield). This type of cable will not provide magnetic shielding, and can easily pick up noise from the power transformer magnetic field.
Toroids are great, and generate very little magnetic field - but they are not perfect. They will most certainly generate a small residual field, and if your input audio cable runs near your power transformer you will pick up a bit of that noise and feed it into that high-gain 12AX7 stage.
When you grounded the grid of the 12AX7 earlier, you still had hum because the AC heater was the dominant problem. Can you repeat the experiment and ground the 12AX7 input grid (the bottom triode, not the top triode of the SRPP stack)?
If, when you ground your input grid, the noise goes away then replace the coax with twisted pair AND route the input cables well away from the power transformer.
Or, at least try a longer piece of coax and route it well-away from the power transformer. You can make twisted-pair chucking up a couple of lengths of 22 or 24 gauge wire in a drill to get them very tightly twisted - 6 turns per inch minimum. (sorry for not using SI units - dang Americans...).
If you can get shielded twisted pair, so much the better. But the important thing is to get the input cables away from your power transformer.
Second, as I posted earlier - IF your power transformer has only a single 6.3VAC winding, and if you are connecting that winding to both the F1 and F2 connections, then you only need a single bias network. The bias will raise all of the heaters, as the DC bias will pass to all the heaters through the transformer winding.
Of course, if you have two separate 6.3VAC windings then you will need to provide two bias networks.
Note, if you have a single 6.3VAC winding then placing two bias networks will not be the end of the world - but you will have two resistive dividers working against each other for no good purpose. You will still get a bias voltage of some kind, but it's not a good idea.
Also, can you verify whether you are using 220 nanofarad or 220 MICROfarad capacitors? The difference is significant - you should be using at least 220 MICROFarads across the 10k (or 20k or 30k) resistor to ground. You need to filter the bias voltage well, and 220 nanoFarads (0.22 microfarads) will not be sufficient.
As I said earlier, I believe you have eliminated the first hum-problem by using the 30V heater bias circuit. You are now hearing another problem that was likely present all along, but masked by the much louder (?) heater hum.
Finally, a word of caution. DO NOT poke around inside that amplifier with your fingers while it is plugged in!! The voltages in there are lethal - if you accidently get across the B+ this will go from a hobby to a tragedy!!
If you absolutely must wiggle wires while the thing is plugged in or turned on, please at least take some precautions:
1) USE A LONG (15cm or so) PIECE OF DRY, UNPAINTED WOODEN DOWEL,
2) USE ONLY ONE HAND TO MANIPULATE THE DOWEL,
and
3) KEEP YOUR OTHER HAND IN YOUR BACK PANTS POCKET AT ALL TIMES WHEN YOU ARE POKING INSIDE THE ACTIVE AMPLIFIER!!
Seriously.
Us older guys who grew up with tube radios all know this. And the ones who grew up to BE old guys remembered to DO these things!
Do NOT use a screwdriver, your fingers, your friend's fingers or anything else. Accidents can be prevented.
To paraphrase the old adage: "There are old electricians and there are bold electricians - but there are NO OLD BOLD ELECTRICIANS"
Best luck, and please be safe.
to rfengineer2013: Thanks a lot for your detail answer and suggestions.
Regarding 6.3 VAC windings there are two separate for each channel and therefore I setup 2 independent elevated nodes. Each consisting 100k + 20 k resisots and 220 n capacitor parallel to the lower one. I have tried to add 1000 u / 63 V as well (to both channels) but there was no difference. Anyway I will attache them there again.
As for the input cable it is coaxial - single wire surrounded by shielding. You are right that it would be better to use twisted pair. I tried to move the wires farer from the power transformer but no audible difference. I guess it is not input wires as they both run next to each other from input connectors to input terminal on the PCB and the buzz is in one channel only. Actully the second channel is silent. I mean really silent, there is not even low level of buzz.
I will try to play with input wires again but maybe it is a good idea to try to shortcut input tube grid to GND first to check if it is before or after the input tube.
I are probably right that the buzz was masked by 100 Hz hum before.
What makes it very strange to me is that even if the PCB is pretty symetrical as well as transformer's arrangement etc. this buzz is in one channel only.
Also thanks a lot for your safety advices. I use wooden stick to move the wires and I'm not touching anything when powered on.
I really like the idea of keepin AC heater voltage rather than moving to DC. So let me try to ground the input tube grid and see what happens.
Regarding 6.3 VAC windings there are two separate for each channel and therefore I setup 2 independent elevated nodes. Each consisting 100k + 20 k resisots and 220 n capacitor parallel to the lower one. I have tried to add 1000 u / 63 V as well (to both channels) but there was no difference. Anyway I will attache them there again.
As for the input cable it is coaxial - single wire surrounded by shielding. You are right that it would be better to use twisted pair. I tried to move the wires farer from the power transformer but no audible difference. I guess it is not input wires as they both run next to each other from input connectors to input terminal on the PCB and the buzz is in one channel only. Actully the second channel is silent. I mean really silent, there is not even low level of buzz.
I will try to play with input wires again but maybe it is a good idea to try to shortcut input tube grid to GND first to check if it is before or after the input tube.
I are probably right that the buzz was masked by 100 Hz hum before.
What makes it very strange to me is that even if the PCB is pretty symetrical as well as transformer's arrangement etc. this buzz is in one channel only.
Also thanks a lot for your safety advices. I use wooden stick to move the wires and I'm not touching anything when powered on.
I really like the idea of keepin AC heater voltage rather than moving to DC. So let me try to ground the input tube grid and see what happens.
Thank you for the kind words, Vcelkamaja. I am very glad you are taking the safety precautions, too
You are right about the symmetry of the board, and it is certainly puzzling why one channel has the buzz and the other is clean. You are doing all the right things, taking things one step at a time until you eliminate the problem.
It's time consuming and can be quite frustrating - welcome to the wonderful world of debugging electronic designs
This is also an excellent learning experience. If you decide to build more electronics projects (warning: the experience can be habit forming
) the things you learn here will help you get through the new set of problems on the next project.
We're all rooting for you here - it will be your success, and in an odd way everybody here will feel almost as good as you do when you solve the problem!
Plus, we will all learn from your experience too!
Collecting the "bag of tricks" is part of this game, and you are collecting a lot of them here! We are all learning something new here, too - from the newbies to the old hands. You never see all the problems, and forums such as this one are excellent places to learn from other peoples experience.
This is a huge, distributed trade school, as it were and I'm sure even the masters pick up new things here from time to time.
Wishing you many happy hours of music once you get your amplifier "on the air"!
- Sam
You are right about the symmetry of the board, and it is certainly puzzling why one channel has the buzz and the other is clean. You are doing all the right things, taking things one step at a time until you eliminate the problem.
It's time consuming and can be quite frustrating - welcome to the wonderful world of debugging electronic designs
This is also an excellent learning experience. If you decide to build more electronics projects (warning: the experience can be habit forming
) the things you learn here will help you get through the new set of problems on the next project.We're all rooting for you here - it will be your success, and in an odd way everybody here will feel almost as good as you do when you solve the problem!
Plus, we will all learn from your experience too!
Collecting the "bag of tricks" is part of this game, and you are collecting a lot of them here! We are all learning something new here, too - from the newbies to the old hands. You never see all the problems, and forums such as this one are excellent places to learn from other peoples experience.
This is a huge, distributed trade school, as it were and I'm sure even the masters pick up new things here from time to time.
Wishing you many happy hours of music once you get your amplifier "on the air"!
- Sam
I will also add that the buzz appearing after about 10 seconds tells me that the coupling is probably not going directly to the output transformer. Your B+ has silicon diodes, which will "turn on" instantly, but the preamp and PA tubes take much, much longer.
I believe if you were having a direct-coupling problem to the OPT then you would hear the hum or buzz almost immediately - but the 10-second delay tells me it's something in the tube circuits.
The tubes have to warm up before they start operating, and it takes several seconds for the heaters to come up and start the tube operating. Not to belabor the "old guy" theme, but before solid-state became common we all grew up with tube radios and TeeVees and remember well that there was no such thing as "instant turn-on" for those devices. They had to warm up for a distinctly long time before you could hear any audio or even a background hum - exactly the thing you describe in your posts.
We can't yet totally eliminate the possibility of direct coupling to the OPT, but it's pretty low on the list of things. You are doing the right thing at this point - focusing on the 12AX7 input.
Best Regards.
I believe if you were having a direct-coupling problem to the OPT then you would hear the hum or buzz almost immediately - but the 10-second delay tells me it's something in the tube circuits.
The tubes have to warm up before they start operating, and it takes several seconds for the heaters to come up and start the tube operating. Not to belabor the "old guy" theme, but before solid-state became common we all grew up with tube radios and TeeVees and remember well that there was no such thing as "instant turn-on" for those devices. They had to warm up for a distinctly long time before you could hear any audio or even a background hum - exactly the thing you describe in your posts.
We can't yet totally eliminate the possibility of direct coupling to the OPT, but it's pretty low on the list of things. You are doing the right thing at this point - focusing on the 12AX7 input.
Best Regards.
I am also in Thomas' camp in that these tubes are designed to operate with AC on the heaters. If DC gets you out of a problem, then maybe it's the easier thing to do. But the tubes themselves were designed into radios and amplifiers that generally did not use DC for the filaments.
Directly Heated Tubes (DHT) use the filament as the cathode - a whole different animal. But the tubes you are using are indirectly heated cathode types - the filament is not the electron-emitting element, it just serves to heat up the cathode which IS the emitting element (and thus part of the audio signal path).
The unique design element of your amplifier is the SRPP stack - with the 2nd triode in series between the B+ and the plate of the first triode. This is not an uncommon design approach, but needs a bit of "extra care" IMHO.
The upper triode has a cathode that is ~155V above ground (according to the schematic provided for this amplifier). The huge voltage difference between the heater and the cathode in the upper tube makes it (the cathode) sensitive to "voltage ripple" on the filament (which is, in the original design, referenced to ground).
Raising the filament to 30VDC bias reduces the voltage difference between the filament and the upper cathode, and (in theory, at least) should reduce the upper tube sensitivity to the AC on the filament.
That, in a nutshell, is the reasoning behind the recommendation to bias the filament (at least, as I understand it).
The fact that you have eliminated the first-order part of the hum tells me that the bias is doing what it (in theory) should do. As I said earlier, you are on the right track and we will just try to avoid locking in on a solution and keep our minds open to what you are observing and telling us.
Directly Heated Tubes (DHT) use the filament as the cathode - a whole different animal. But the tubes you are using are indirectly heated cathode types - the filament is not the electron-emitting element, it just serves to heat up the cathode which IS the emitting element (and thus part of the audio signal path).
The unique design element of your amplifier is the SRPP stack - with the 2nd triode in series between the B+ and the plate of the first triode. This is not an uncommon design approach, but needs a bit of "extra care" IMHO.
The upper triode has a cathode that is ~155V above ground (according to the schematic provided for this amplifier). The huge voltage difference between the heater and the cathode in the upper tube makes it (the cathode) sensitive to "voltage ripple" on the filament (which is, in the original design, referenced to ground).
Raising the filament to 30VDC bias reduces the voltage difference between the filament and the upper cathode, and (in theory, at least) should reduce the upper tube sensitivity to the AC on the filament.
That, in a nutshell, is the reasoning behind the recommendation to bias the filament (at least, as I understand it).
The fact that you have eliminated the first-order part of the hum tells me that the bias is doing what it (in theory) should do. As I said earlier, you are on the right track and we will just try to avoid locking in on a solution and keep our minds open to what you are observing and telling us.
Reading through Vcelkamaja's postings, I believe he has done that several times.
In the present problem, the buzz seems to be in one channel only and does not follow the 12AX7 when he swaps them.
That is a very good question - and would determine whether the hum is actually gone or whether the "quiet" channel is non-operational and thus quiet...
Again, reading through the thread when both 12AX7s are removed the buzz goes away. This would imply that the output stage is not contributing to the buzz.
Regards,
~ Sam
Hi Sam - I'm just being cautious given the OP's somewhat inexperience with fault-finding and the growing number of tweaks being made, with more tweaks on the way (ie. a more appropriate level of bypassing the elevated heater bias).
To the matter of basic fault finding, the first steps imho would be to use an isolated battery heater supply; followed by an independant filament transformer supply to the valve base itself; followed by an independant filament transformer supply to the normal pcb heater terminals. The idea being to rule in/out various forms of causes firstly.
To the matter of basic fault finding, the first steps imho would be to use an isolated battery heater supply; followed by an independant filament transformer supply to the valve base itself; followed by an independant filament transformer supply to the normal pcb heater terminals. The idea being to rule in/out various forms of causes firstly.
* remember to disconnect feedback network before troubleshooting.
well seeing what you got into it doesn't surprise me the results you got.
This issue occurs when one of the heater windings is a little off from each other.
this is true: if you remove the heater supply from the noisy channel the noise goes away. when you swap heater supplies it shows up on the other channel.
if so, you can modify the bad heater winding by AC coupling the center tap of the hum balance circuit. On the schematic you'll see the two identical resistors across the heater winding forming a simple voltage divider network that has the center pin (voltage tap 50%) tied to ground.
now if a .1 uf cap is placed between the resistor voltage divider network and ground the noise will get shunt to ground. And the DC bias due to the noise rectification between heater and cathode will be removed.
Some people thinks its nuts/crazy/whatever to not dc reference a heater. But many of those people give up and DC their heaters.
So I won't be surprise when they bash my posting
well seeing what you got into it doesn't surprise me the results you got.
This issue occurs when one of the heater windings is a little off from each other.
this is true: if you remove the heater supply from the noisy channel the noise goes away. when you swap heater supplies it shows up on the other channel.
if so, you can modify the bad heater winding by AC coupling the center tap of the hum balance circuit. On the schematic you'll see the two identical resistors across the heater winding forming a simple voltage divider network that has the center pin (voltage tap 50%) tied to ground.
now if a .1 uf cap is placed between the resistor voltage divider network and ground the noise will get shunt to ground. And the DC bias due to the noise rectification between heater and cathode will be removed.
Some people thinks its nuts/crazy/whatever to not dc reference a heater. But many of those people give up and DC their heaters.
So I won't be surprise when they bash my posting
*in your schematic is interstage cap connected to grid of upper tube, should it be rather on cathode ? (srpp)
*gridleak on 6L6 is 1M, recommended is way smaller -- also on the ECC83 is a bit higher than in datasheets recommended values (i use 100k,330k)
*resistor to screen is 47k, a bit high ?? (datasheet: 6L6 SE 72mA Ia, 5mA Ig2) - you are running it like almost a disconnected screen
*220nF cap in elevated heater - enough ?
/ ECC83 max Uk_f is 180V, it seems you would not violate it if used grounded filament winding (like i do)
or you can increase R18 a bit to lower voltage/
Again thanks all to your ideas and comments.
It seems I'm getting some progress now
While it was heaters elevation what helped in removing 100 Hz hum from both channels it haven't helped with newly appreared buzz in one channel.
It was clear that it is somewhere around the input tube as if I remove it the amp is silent.
Well today I managed to get a new pair of 12AX7 - no change, still the same buzz in one channel.
So I started with inputs again - I shortened CINCH cable in buzzing channel and - the buzz moved to the second channel! Next I shortened both channels (at the CINCH connectors of input cable) and the buzz almost disapeared - it is there but very low level, almost inaudible.
Ok, it seems it is the inputs as if I shortened input cable the buzz is almost gone.
Now I shortened input terminal on the PCB and WOW - the buzz is completely gone from both channels. I tried this thing while analyzing 100 Hz hum in both channels but there was no effect. Only the elevation of AC heaters helped. Now I tried it again to remove the buzz in one channel and it works!
So now the question - if the amp is completely quite with input shortened on the PCB terminal what does it mean? What can be wrong? Is it only input cables between RCA sockets on the back pannel and their grounding? Or anything else can produce the same effect?
If we assume that it is input cables (shielded single wire - coaxial, RCA sockets isolated from the chassis, minus-GND of both inputs cables connected togeter at the input terminl on the PCB) what can I improve? If I use twisted shielded cable from RCA sockets on the panel to the PCB how should I connect their wires properely? (2 x wre + shielding)
It seems I'm getting some progress now
While it was heaters elevation what helped in removing 100 Hz hum from both channels it haven't helped with newly appreared buzz in one channel.
It was clear that it is somewhere around the input tube as if I remove it the amp is silent.
Well today I managed to get a new pair of 12AX7 - no change, still the same buzz in one channel.
So I started with inputs again - I shortened CINCH cable in buzzing channel and - the buzz moved to the second channel! Next I shortened both channels (at the CINCH connectors of input cable) and the buzz almost disapeared - it is there but very low level, almost inaudible.
Ok, it seems it is the inputs as if I shortened input cable the buzz is almost gone.
Now I shortened input terminal on the PCB and WOW - the buzz is completely gone from both channels. I tried this thing while analyzing 100 Hz hum in both channels but there was no effect. Only the elevation of AC heaters helped. Now I tried it again to remove the buzz in one channel and it works!
So now the question - if the amp is completely quite with input shortened on the PCB terminal what does it mean? What can be wrong? Is it only input cables between RCA sockets on the back pannel and their grounding? Or anything else can produce the same effect?
If we assume that it is input cables (shielded single wire - coaxial, RCA sockets isolated from the chassis, minus-GND of both inputs cables connected togeter at the input terminl on the PCB) what can I improve? If I use twisted shielded cable from RCA sockets on the panel to the PCB how should I connect their wires properely? (2 x wre + shielding)
Last edited:
Excellent news!
By shorting the inputs at the CINCH connector (I assume the input connector on your amplifier chassis) and noticing that the buzz is quieter, and then shorting the inputs on the board and noticing that the buzz is gone - this is a very important clue.
I will go back to my original suspicion that the input cables are picking up noise from the power transformer and conducting that noise to your 12AX7 input.
If I were "under the hood" (i.e. - working inside your amplifier), I think I would just make a couple of lengths of twisted pair without shield. Use 24 or 26ga wire if you have it - these wires do not carry current and do not have to be thick. If all you have is thicker stiff, give it a try it won't make any electrical difference except that the resulting twisted pair will be a little stiffer.
Use a drill (on low speed) to twist the wires. 4-5 twists per cm or so, as a guess - not critical, just get the twists good and tight without causing the resulting cable to kink-up. Using a drill makes it quick and easy, and you can get pretty neat twisted pairs quickly. Use two colors, of course, so you can sort out + and - easily on both ends of the pairs.
Once you have two lengths of twisted pair, attach them to your input connectors and route them around the perimeter of the chassis away from the power transformer! I would not take the short-cut of running them directly, as you have the coax now.
The twisted-pair, even unshielded, will give you the self-cancelling properties that the coax does not. Noise induced into the twisted-pair will tend to cancel - which is good - and routing the wires as far from your power trannie as you can (the "long way" around the chassis) will keep the noise pickup from the power transformer to a minimum.
If that gives you a further improvement, but doesn't get you all the way clean, you can try to get some shielded twisted pair or try other routing schemes.
But I think just a simple twisted pair routed away from the power transformer will take care of your noise problem.
BTW - there's lots of information on the web about twisted-pair vs. coax, common-mode noise and differential-mode noise. That's what the twisting is all about - if you're interested, you might take a look at the theory behind all this "black magic" stuff
An exercise for the student, as it were.Good luck, you are almost finished I think! Good on ya for sticking with this thorny little problem
~ Sam
Sam, thanks for your help. I replaced coax cables with tightly twisted pair of wires and tried different routing of the cables between back panel connectors and input PCB connector.
It seems it helps a bit but still not perfectly silent. Depending on the routing way the buzz-hum changes but I haven't the way how to make it completly silent. I tried running the cable for both channels next to each other and interesting enough sometimes one channel is silent while the second one buzz. I also tried to run cable separatetly - left on the left side of the chassis and right on the right side of the chassis but still there is some buzz.
I also tried to decrease input resistance of the amp from 1 M to 27 k to avoid "antenna" effect but no improvement.
It definitely influences the buzz how I route the input cable and it definitely helped to use twisted pairs instead of coax but still not perfect.
I tried some shielding of these twisted pairs using copper foil - grounded and not grounded but no effect.
Anything else that I should try?
It seems it helps a bit but still not perfectly silent. Depending on the routing way the buzz-hum changes but I haven't the way how to make it completly silent. I tried running the cable for both channels next to each other and interesting enough sometimes one channel is silent while the second one buzz. I also tried to run cable separatetly - left on the left side of the chassis and right on the right side of the chassis but still there is some buzz.
I also tried to decrease input resistance of the amp from 1 M to 27 k to avoid "antenna" effect but no improvement.
It definitely influences the buzz how I route the input cable and it definitely helped to use twisted pairs instead of coax but still not perfect.
I tried some shielding of these twisted pairs using copper foil - grounded and not grounded but no effect.
Anything else that I should try?
One more comment - with this new wiring using twisted pair I have also tried to shortcut inputs and it is perfectly silent. By shortcutting input I mean input of the RCA cable (input RCA connector shortcutted --> signal cable ---> output RCA connector connected to RCA jack on the back panel of the amp --> twisted pair of wires --> input terminal on the PCB). Basically it simulates the shortcut on the source of signal including cable between source and amp.
I tried to shortcut one channel while the second channel normally connected to surce. The shortcutted one is perfectly silent while the one that is normally connected to source is buzzing/humming. When I shortcutted the second channel the situation is the same.
I tried to shortcut one channel while the second channel normally connected to surce. The shortcutted one is perfectly silent while the one that is normally connected to source is buzzing/humming. When I shortcutted the second channel the situation is the same.
Thats good news again. Thanks for your patience - I'm not able to hear the differences, but it sounds ( ) like you are making incremental progress. That's the way these noise things go - pushing down the noise a bit here, a bit there until you are satisfied with the result.
As far as the input resistors R6 & R7, the 1 Meg is on the high side but not at all uncommon. Those resistors set the input impedance of the amplifier, which (for line input) should be >47k. Lots of designs use 100k resistors here, but 1Megs are not uncommon either.
Since you tried 27k in those locations without significant difference, I would suggest putting the 1Megs back in - or you could use 100k if you want. The triode grid impedance will be in parallel with each of those resistors, but it is well above the 100k value and will not lower the input impedance of your amplifier (and thus load the music-source you are using)
You might try bypassing the shield terminals on the input connectors to chassis with a couple of 0.01 microfarad, 100V capacitors. Do this right at the connector - you are trying to ground the open "far-end" of the twisted pair, making it "cold" at high-frequency. Right now, the twisted-pair ends are electrically "up in the air" and are antennas. RF-grounding the connector will give a good, low-impedance path for unwanted energy to drain to ground, and will block the low-frequency AC so you don't create another ground-loop
If you are picking up high-frequency noise on the input connector (which floats relative to chassis) it may be coupling into your input triode and generating the noise you are hearing.
Without good RF grounding, the high-frequency noise will see the star-ground through a relatively high impedance. The bypass capacitors will give the noise a low-impedance path to chassis. The capacitor from the connector shield-side to chassis will not affect the audio signal.
If you don't have one in 100V rating, use what you have for the experiment and if it eliminates the noise then you can order the parts for your final buildup. The voltage rating is more a long-term safety issue, but the capacitor itself will see only a couple of volts in the circuit under normal conditions.
One important thing is to keep the capacitor leads as short as possible. The capacitor leads look like tiny inductors at RF and can negate the bypassing effects if the leads are long. Not to go nuts with zero-length leads, of course, just try to keep the capacitor connection as short as reasonably possible. Clipping test leads to the capacitor as a test, for instance, won't work.
Routing the two input twisted-pairs together should not cause a problem. I would route both pairs away from the power supply transformer, around the chassis, and not let them drape over the PCB at all.
) like you are making incremental progress. That's the way these noise things go - pushing down the noise a bit here, a bit there until you are satisfied with the result.As far as the input resistors R6 & R7, the 1 Meg is on the high side but not at all uncommon. Those resistors set the input impedance of the amplifier, which (for line input) should be >47k. Lots of designs use 100k resistors here, but 1Megs are not uncommon either.
Since you tried 27k in those locations without significant difference, I would suggest putting the 1Megs back in - or you could use 100k if you want. The triode grid impedance will be in parallel with each of those resistors, but it is well above the 100k value and will not lower the input impedance of your amplifier (and thus load the music-source you are using)
You might try bypassing the shield terminals on the input connectors to chassis with a couple of 0.01 microfarad, 100V capacitors. Do this right at the connector - you are trying to ground the open "far-end" of the twisted pair, making it "cold" at high-frequency. Right now, the twisted-pair ends are electrically "up in the air" and are antennas. RF-grounding the connector will give a good, low-impedance path for unwanted energy to drain to ground, and will block the low-frequency AC so you don't create another ground-loop
If you are picking up high-frequency noise on the input connector (which floats relative to chassis) it may be coupling into your input triode and generating the noise you are hearing.
Without good RF grounding, the high-frequency noise will see the star-ground through a relatively high impedance. The bypass capacitors will give the noise a low-impedance path to chassis. The capacitor from the connector shield-side to chassis will not affect the audio signal.
If you don't have one in 100V rating, use what you have for the experiment and if it eliminates the noise then you can order the parts for your final buildup. The voltage rating is more a long-term safety issue, but the capacitor itself will see only a couple of volts in the circuit under normal conditions.
One important thing is to keep the capacitor leads as short as possible. The capacitor leads look like tiny inductors at RF and can negate the bypassing effects if the leads are long. Not to go nuts with zero-length leads, of course, just try to keep the capacitor connection as short as reasonably possible. Clipping test leads to the capacitor as a test, for instance, won't work.
Routing the two input twisted-pairs together should not cause a problem. I would route both pairs away from the power supply transformer, around the chassis, and not let them drape over the PCB at all.
I just saw this after posting my tome
This certainly sounds like the open-ends of the twisted pair are picking up noise. Shorting the input makes the noise go away, so you are on the right track.
If it is pickup, then the 0.01uF cap from connector-shield to chassis should reduce the noise. I would be tempted to put a 100 picofarad cap right across the twisted-pair, too, but that puts a component across the audio path. For me, if it solved the problem, I wouldn't have any issue - the value is sufficiently low that it will not affect the audio frequency performance. But I'm not what you might call a "purist" either - just a wire-puller
Try the 0.01uF bypass first - it's not part of the audio path and should take care of the problem.
Again, keep the leads as short as possible. I would use ceramic disk caps for these - the disk caps have excellent RF properties. But for this exercise, use what you have - if it works, then you can order whatever parts you need for the final assembly.
Good luck!
~ Sam
Oh, BTW - I was thinking about the voltage rating, and if this works and you order parts then I suggest go with 600V parts. It's probably overkill, but this is more a safety issue in case of some dramatic failure inside the amp. Wouldn't want the capacitor to break down for some reason - and the price difference is negligible. These don't have to be the expensive audio-quality capacitors - just some plain-old garden-variety ceramic caps.
Sam, thanks a lot for your patience and advices. I have a few questions on your recommendations:
1. If the twisted pair helped a bit do you think that shielding it would be useful as well? If so where should I connect on end of the shield? (input RCA jack side or PCB connectors side? connect the shileding to GND - connectors, or star GND or rather to chassis - close to cable end or to the "common chassis point"?)
2. Regarding capacitor between RCA socket minus and chassis - you have mentioned the leads should be as short as possible. No problem on the RCA socket side of the capacitor but how about the side to be connected to chassis? Should I try to connect it to the chassis as close to RCA socket as possible or should I rather connect it to the "common chassis point" which is ~ 3 inches far from the RCA sockets?
3. The last question is for the small capacitor parallel to twisted pair - should I potentially attached it on the RCA socket side of the twisted pari cable (back panel) or rather on the PCB side? I guess it should be at the input so on the back pannel RCA socket but I just wanted to double check
Thanks
1. If the twisted pair helped a bit do you think that shielding it would be useful as well? If so where should I connect on end of the shield? (input RCA jack side or PCB connectors side? connect the shileding to GND - connectors, or star GND or rather to chassis - close to cable end or to the "common chassis point"?)
2. Regarding capacitor between RCA socket minus and chassis - you have mentioned the leads should be as short as possible. No problem on the RCA socket side of the capacitor but how about the side to be connected to chassis? Should I try to connect it to the chassis as close to RCA socket as possible or should I rather connect it to the "common chassis point" which is ~ 3 inches far from the RCA sockets?
3. The last question is for the small capacitor parallel to twisted pair - should I potentially attached it on the RCA socket side of the twisted pari cable (back panel) or rather on the PCB side? I guess it should be at the input so on the back pannel RCA socket but I just wanted to double check
Thanks
Last edited:
No problem, Vcelkamaja - always glad to help!
As for shielding the twisted pair, if you do that you should only connect one end to ground. The easiest to try would be grounding at the PCB end - but I think you already tried that. Grounding the shield to the negative side of the input connector would be next - and if that doesn't work then grounding the shield to the chassis right at the connector would be third.
What you don't want to do is ground the shield at both ends - that will set up a ground loop and probably negate any positive aspects of the shield.
The capacitor from the minus terminal of the input should go to chassis as close to the input connector as possible. You might need to use a solder-lug, but for the experiment just make as good a contact as you can with the capacitor lead to chassis. Clamp it with a clothespin or something - be sure it makes good contact, if it works then you can put in a solder lug and make a permanent connection.
The capacitor provides a low-impedance path to the (high-frequency) EMI to give it a path off the twisted-pair. Running a wire back to your star ground will create what looks like an inductor to the EMI - and will reduce the effectiveness of the bypass capacitor.
The final option of putting a cap across the input terminals is probably unnecessary - so far, I have not seen any of the experienced folks here do that so it's likely not necessary. If you want to try the experiment, though, I think it won't hurt the audio performance as the value is quite low and the capacitor will be a very high impedance at the audio-frequencies.
Still, to the purists here I probably speak heresy by suggesting putting a capacitor across the inputs!! But hey, I'm an RF guy at heart and it probably shows...
The twisted-pair is terminated on the PCB by a resistor, what you are trying to do is to keep the connector-end of the twisted pair from floating "in the wind". Tying one side of the twisted pair to chassis at the connector using a capacitor will pull the twisted-pair back down - the EMI, which is high-frequency, will see a low-impedance path off of the twisted-pair and follow that instead of crawling into your amplifier input.
So the capacitors should be put right at the connectors and taken to chassis by as short a path as possible.
Of course, I am assuming that you have a metal chassis or at least a metal top-plate to connect the capacitor to.
And keep an open mind - there is still a possibility that there is something else going on, but my understanding of the symptoms is that the input cable is picking up some noise and the noise is either being amplified or is mixing down into the audio range in your input triode. Either way, getting the input cable as noise-free as possible is what we're trying to do.
Which brings to mind a question - if you disconnect the twisted-pair at the PCB input, does the noise go away? I don't remember if you did that, but if not it might be a good check before you go to more trouble fiddling with the input cable.
If you disconnect the input cables at the PCB and the noise goes away in both channels, then it's safe to assume the noise is being picked up by the input cable and proceed with the capacitor bypass experiments.
If the noise is still there, then we're on the wrong path...
Hope this helps! Good luck!
~ Sam
As for shielding the twisted pair, if you do that you should only connect one end to ground. The easiest to try would be grounding at the PCB end - but I think you already tried that. Grounding the shield to the negative side of the input connector would be next - and if that doesn't work then grounding the shield to the chassis right at the connector would be third.
What you don't want to do is ground the shield at both ends - that will set up a ground loop and probably negate any positive aspects of the shield.
The capacitor from the minus terminal of the input should go to chassis as close to the input connector as possible. You might need to use a solder-lug, but for the experiment just make as good a contact as you can with the capacitor lead to chassis. Clamp it with a clothespin or something - be sure it makes good contact, if it works then you can put in a solder lug and make a permanent connection.
The capacitor provides a low-impedance path to the (high-frequency) EMI to give it a path off the twisted-pair. Running a wire back to your star ground will create what looks like an inductor to the EMI - and will reduce the effectiveness of the bypass capacitor.
The final option of putting a cap across the input terminals is probably unnecessary - so far, I have not seen any of the experienced folks here do that so it's likely not necessary. If you want to try the experiment, though, I think it won't hurt the audio performance as the value is quite low and the capacitor will be a very high impedance at the audio-frequencies.
Still, to the purists here I probably speak heresy by suggesting putting a capacitor across the inputs!!
But hey, I'm an RF guy at heart and it probably shows...The twisted-pair is terminated on the PCB by a resistor, what you are trying to do is to keep the connector-end of the twisted pair from floating "in the wind". Tying one side of the twisted pair to chassis at the connector using a capacitor will pull the twisted-pair back down - the EMI, which is high-frequency, will see a low-impedance path off of the twisted-pair and follow that instead of crawling into your amplifier input.
So the capacitors should be put right at the connectors and taken to chassis by as short a path as possible.
Of course, I am assuming that you have a metal chassis or at least a metal top-plate to connect the capacitor to.
And keep an open mind - there is still a possibility that there is something else going on, but my understanding of the symptoms is that the input cable is picking up some noise and the noise is either being amplified or is mixing down into the audio range in your input triode. Either way, getting the input cable as noise-free as possible is what we're trying to do.
Which brings to mind a question - if you disconnect the twisted-pair at the PCB input, does the noise go away? I don't remember if you did that, but if not it might be a good check before you go to more trouble fiddling with the input cable.
If you disconnect the input cables at the PCB and the noise goes away in both channels, then it's safe to assume the noise is being picked up by the input cable and proceed with the capacitor bypass experiments.
If the noise is still there, then we're on the wrong path...
Hope this helps! Good luck!
~ Sam
No problem, Vcelkamaja - always glad to help!
As for shielding the twisted pair, if you do that you should only connect one end to ground. The easiest to try would be grounding at the PCB end - but I think you already tried that. Grounding the shield to the negative side of the input connector would be next - and if that doesn't work then grounding the shield to the chassis right at the connector would be third.
What you don't want to do is ground the shield at both ends - that will set up a ground loop and probably negate any positive aspects of the shield.
The capacitor from the minus terminal of the input should go to chassis as close to the input connector as possible. You might need to use a solder-lug, but for the experiment just make as good a contact as you can with the capacitor lead to chassis. Clamp it with a clothespin or something - be sure it makes good contact, if it works then you can put in a solder lug and make a permanent connection.
The capacitor provides a low-impedance path to the (high-frequency) EMI to give it a path off the twisted-pair. Running a wire back to your star ground will create what looks like an inductor to the EMI - and will reduce the effectiveness of the bypass capacitor.
The final option of putting a cap across the input terminals is probably unnecessary - so far, I have not seen any of the experienced folks here do that so it's likely not necessary. If you want to try the experiment, though, I think it won't hurt the audio performance as the value is quite low and the capacitor will be a very high impedance at the audio-frequencies.
Still, to the purists here I probably speak heresy by suggesting putting a capacitor across the inputs!! But hey, I'm an RF guy at heart and it probably shows...
The twisted-pair is terminated on the PCB by a resistor, what you are trying to do is to keep the connector-end of the twisted pair from floating "in the wind". Tying one side of the twisted pair to chassis at the connector using a capacitor will pull the twisted-pair back down - the EMI, which is high-frequency, will see a low-impedance path off of the twisted-pair and follow that instead of crawling into your amplifier input.
So the capacitors should be put right at the connectors and taken to chassis by as short a path as possible.
Of course, I am assuming that you have a metal chassis or at least a metal top-plate to connect the capacitor to.
And keep an open mind - there is still a possibility that there is something else going on, but my understanding of the symptoms is that the input cable is picking up some noise and the noise is either being amplified or is mixing down into the audio range in your input triode. Either way, getting the input cable as noise-free as possible is what we're trying to do.
Which brings to mind a question - if you disconnect the twisted-pair at the PCB input, does the noise go away? I don't remember if you did that, but if not it might be a good check before you go to more trouble fiddling with the input cable.
If you disconnect the input cables at the PCB and the noise goes away in both channels, then it's safe to assume the noise is being picked up by the input cable and proceed with the capacitor bypass experiments.
If the noise is still there, then we're on the wrong path...
Hope this helps! Good luck!
~ Sam
Mods - I don't know why this double-posted, but one of my tomes per page is probably sufficient Please remove the 2nd if possible - Thanks!
As for shielding the twisted pair, if you do that you should only connect one end to ground. The easiest to try would be grounding at the PCB end - but I think you already tried that. Grounding the shield to the negative side of the input connector would be next - and if that doesn't work then grounding the shield to the chassis right at the connector would be third.
What you don't want to do is ground the shield at both ends - that will set up a ground loop and probably negate any positive aspects of the shield.
The capacitor from the minus terminal of the input should go to chassis as close to the input connector as possible. You might need to use a solder-lug, but for the experiment just make as good a contact as you can with the capacitor lead to chassis. Clamp it with a clothespin or something - be sure it makes good contact, if it works then you can put in a solder lug and make a permanent connection.
The capacitor provides a low-impedance path to the (high-frequency) EMI to give it a path off the twisted-pair. Running a wire back to your star ground will create what looks like an inductor to the EMI - and will reduce the effectiveness of the bypass capacitor.
The final option of putting a cap across the input terminals is probably unnecessary - so far, I have not seen any of the experienced folks here do that so it's likely not necessary. If you want to try the experiment, though, I think it won't hurt the audio performance as the value is quite low and the capacitor will be a very high impedance at the audio-frequencies.
Still, to the purists here I probably speak heresy by suggesting putting a capacitor across the inputs!!
But hey, I'm an RF guy at heart and it probably shows...The twisted-pair is terminated on the PCB by a resistor, what you are trying to do is to keep the connector-end of the twisted pair from floating "in the wind". Tying one side of the twisted pair to chassis at the connector using a capacitor will pull the twisted-pair back down - the EMI, which is high-frequency, will see a low-impedance path off of the twisted-pair and follow that instead of crawling into your amplifier input.
So the capacitors should be put right at the connectors and taken to chassis by as short a path as possible.
Of course, I am assuming that you have a metal chassis or at least a metal top-plate to connect the capacitor to.
And keep an open mind - there is still a possibility that there is something else going on, but my understanding of the symptoms is that the input cable is picking up some noise and the noise is either being amplified or is mixing down into the audio range in your input triode. Either way, getting the input cable as noise-free as possible is what we're trying to do.
Which brings to mind a question - if you disconnect the twisted-pair at the PCB input, does the noise go away? I don't remember if you did that, but if not it might be a good check before you go to more trouble fiddling with the input cable.
If you disconnect the input cables at the PCB and the noise goes away in both channels, then it's safe to assume the noise is being picked up by the input cable and proceed with the capacitor bypass experiments.
If the noise is still there, then we're on the wrong path...
Hope this helps! Good luck!
~ Sam
Mods - I don't know why this double-posted, but one of my tomes per page is probably sufficient
Please remove the 2nd if possible - Thanks!
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.