Hello,
I have an ordinary KT88 PP amplifier that is silent with shortened inputs and 96 dB headphones on the output, but I hear little mains + harmonics noise with 102 dB headphones. It is the same noise that I hear on the chassis and on PE.
The amplifier uses hum bucking resistors HBRL/HBRR, the chassis is connected to the 0 V reference point and the 0 V reference is connected with a bridge rectifier to PE.
I need to drop mains noise by 6 dB or more 🙂.
Should I try a bigger choke (now 470 µF - 2.6 H - 470 µF) or
a Feed Forward Shunt Regulator (IXCP 10M90S or EL34 ?) or
a capacitance multiplier (gyrator) - B+ 500V - or
...something else ?
Thank you for helping.
I have an ordinary KT88 PP amplifier that is silent with shortened inputs and 96 dB headphones on the output, but I hear little mains + harmonics noise with 102 dB headphones. It is the same noise that I hear on the chassis and on PE.
The amplifier uses hum bucking resistors HBRL/HBRR, the chassis is connected to the 0 V reference point and the 0 V reference is connected with a bridge rectifier to PE.
I need to drop mains noise by 6 dB or more 🙂.
Should I try a bigger choke (now 470 µF - 2.6 H - 470 µF) or
a Feed Forward Shunt Regulator (IXCP 10M90S or EL34 ?) or
a capacitance multiplier (gyrator) - B+ 500V - or
...something else ?
Thank you for helping.
Mains frequency hum is not connected to main smoothing/choke issues as they are twice the fundamental frequency. Eg; 50 hZ mains = 100 hZ from bridge rectifier etc.
Hum bucking resistors sound interesting as a hum bucking guitar pickup uses out of phase difference for signal. How does your resistor do that?
Hum bucking resistors sound interesting as a hum bucking guitar pickup uses out of phase difference for signal. How does your resistor do that?
I have 50 Hz + 100 Hz + 150 Hz + .... the rest buried in the noise. without the 2H6 choke I could here it using the 96 dB headphones. After installing the choke it was dead silent with the 96 dB, but I can still hear it with the 102 dB headphones.
Sorry not what you expect. The HBRR/HBRL seem to originate from this link:
Ground Loop again
and are presented in more detail here :
The dozens schemes to wire an amp...
I see, a local abbreviation, not an actual science. That's why it is not in any text books.
If you use a star configuration and ground from one point chassis, input commons and outputs then there will be no hum.
There are a few manufacturers that ground the chassis with a bonded earth, (as we are required to do for safety) and then use a low value, say 10R resistor and connect the star ground to the chassis, at one point only. In troublesome installations, a pair of diodes are also used to clamp the ground to chassis earth.
If you use a star configuration and ground from one point chassis, input commons and outputs then there will be no hum.
There are a few manufacturers that ground the chassis with a bonded earth, (as we are required to do for safety) and then use a low value, say 10R resistor and connect the star ground to the chassis, at one point only. In troublesome installations, a pair of diodes are also used to clamp the ground to chassis earth.
Well I did everything and I agree for < 96 dB, but for 102 dB or more my amplifier has a different view 😕
So it's silent for 96dB but has too much and objectionable to the point of trying to fix the amp's power supply, with 102dB. Just my opinion but it could be poor quality headphone shield/ground/plug/jack connections or you may be overly sensitive looking for perfection. Is the impedance the same for both sets of HP's? Power ratings equal? General quality high? ..... Toss the 102's. No, don't tell me, those are the expensive one's.
You need to use a resistive pad between your headphones and the amp. Try a 20 ohm resistor in series with a 1 ohm resistor to pad down the output (and the noise). You have about 5,000 times the power you need, so having this issue isn't at all surprising.
@20to20
The headphones are just an easy way to illustrate the problem. FFT shows the same.
@audiowize
What I really would like to know is, if the current situation is already the practical limit in terms of noise performance or if I have to invest more, like increasing the choke or an active regulator or something else to reach the in practice achievable noise floor.
O.k. I don't tell you 🙂, but I have cheaper ones with 108 dB and that is no fun at all to listen too.
The headphones are just an easy way to illustrate the problem. FFT shows the same.
@audiowize
Thank you very much for the good advice, but your solution is too simple 😀 (I hope I found the right words - my English is not so good).
What I really would like to know is, if the current situation is already the practical limit in terms of noise performance or if I have to invest more, like increasing the choke or an active regulator or something else to reach the in practice achievable noise floor.
ground loops, ground loops, ground loops:
Inputs and first stage; B+; middle stages.
Steel chassis is bad, because of power transformers, chokes, and output transformers.
Even with aluminum chassis, large enough spacing, proper rotational orientation, are both important.
Very low ripple DC on all DHT filaments.
Those are my favorite things to avoid, fix, or at least pay attention to.
My power amplifiers have less than 100uV hum and noise, and I only use loudspeakers, not headphones.
Less than 100uV works for me.
Inputs and first stage; B+; middle stages.
Steel chassis is bad, because of power transformers, chokes, and output transformers.
Even with aluminum chassis, large enough spacing, proper rotational orientation, are both important.
Very low ripple DC on all DHT filaments.
Those are my favorite things to avoid, fix, or at least pay attention to.
My power amplifiers have less than 100uV hum and noise, and I only use loudspeakers, not headphones.
Less than 100uV works for me.
Last edited:
Perhaps it is best to show the schematic and photos of the amp, as really low levels of hum can come from many different and subtle paths. Just simply describing the amp does not provide sufficient detail for thread viewers to get a balanced view of the amps operations, and then to provide some advice.
I think you need a dedicated headphone amp.
Using a big tubeamp for the task can risk your hearing !! Anything that
goes wrong could create sound that damages your ears permanently !!
One could also mention that running a kt88 amp extra hours will cost
in tubes & electricity, some of this will be recovered by using a small
dedicated headphone amp.
At first you should maybe try to figure out where the hum/noise is coming from. I could give a few more suggestions about what to do but not knowing what you are dealing with this would just add noise to the thread.
The amplifier is a modified Antique Sound Labs AQ-1001 (early version). I got it used, I don't know how good the tubes are, one KT88 needs a lot longer to stabilize than the others.
According to the datasheet of the KT88 the spacing between them is too small and the orientation is wrong. The OPT's are 90° to the PT but there is minimal space between the pots. The axis of the PT windings goes through the center KT88s.
The chassis is stainless steel (best possible metal), but very crowded. I tried to make things right, but I wanted to keep the chassis and the PCB therefore there are compromises.
@6A3sUMMER
Mains on, but without KT88s (to see coupling from PT to OPT) on the left 330 µVrms and on the right channel 230 µVrms. Possible explanation is the drilled mains cable from back to front mains switch on the left side without shielding.
KT88s inserted I measure about 500 µVrms on left channel (with the problematic KT88) and on the right channel, but on the right channel there are every few seconds short bursts up to 10-30 mVrms. I have no clue why ?
@dotneck335
According to my simulations you are right and you are the first one to answer my initial question 🙂.
@trobbins
I attached the schematic, I feel ashamed to show pictures 😀, but if it helps I will make them.
@petertub
Thank you for the warning. I use the headphones only for testing and I have protection diodes.
@schiirrn
Therefore I am back at AC-heating elevated to 60 V + 100nF to GND - there is no difference between 100nF and 100µF. There is no difference using 2 * 220R or a humdinger potentiometer.
The other main source for noise should be the B+. The theoretical cancellation is compromised by not using 100.0000% identical parts and I use the LTP driver with identical resistors to have some second harmonic. I need the second harmonic a little bit higher than the third to be happy 🙂. Adding the 2H6 choke made an improvement of about 6dB. Therefore my initial question about bigger choke or active regulation etc.
Then there is the PT which is close to OPT and KT88...see above.
And ofcourse it could be the tubes...changing them around I got results from 300 µVrms to 1500µVrms...
@all
Please tell me how to identify from where the hum/noise is coming from or whatelse I should test.
Do I need a new set of KT88 (or KT120) because of the 'slow' KT88 on the left and the bursts on the 'good' right channel ?
Is it necessary to make a bigger chassis with more spacing ?
According to the datasheet of the KT88 the spacing between them is too small and the orientation is wrong. The OPT's are 90° to the PT but there is minimal space between the pots. The axis of the PT windings goes through the center KT88s.
The chassis is stainless steel (best possible metal), but very crowded. I tried to make things right, but I wanted to keep the chassis and the PCB therefore there are compromises.
@6A3sUMMER
Thank you for the information, unfortunately my DMM reports already 80 µVrms without mains power.
Mains on, but without KT88s (to see coupling from PT to OPT) on the left 330 µVrms and on the right channel 230 µVrms. Possible explanation is the drilled mains cable from back to front mains switch on the left side without shielding.
KT88s inserted I measure about 500 µVrms on left channel (with the problematic KT88) and on the right channel, but on the right channel there are every few seconds short bursts up to 10-30 mVrms. I have no clue why ?
@dotneck335
According to my simulations you are right and you are the first one to answer my initial question 🙂.
@trobbins
I attached the schematic, I feel ashamed to show pictures 😀, but if it helps I will make them.
@petertub
Thank you for the warning. I use the headphones only for testing and I have protection diodes.
@schiirrn
The amp continues to run for some time without mains power. One of my first tests was to switch off mains and the noise was immediately gone. I couldn't help thinking lets change from AC heating to DC, tried DC on input + driver tubes and AC on KT88s and finally tried DC for all tubes (using external PSU to avoid overloading the PT) - no change.
Therefore I am back at AC-heating elevated to 60 V + 100nF to GND - there is no difference between 100nF and 100µF. There is no difference using 2 * 220R or a humdinger potentiometer.
The other main source for noise should be the B+. The theoretical cancellation is compromised by not using 100.0000% identical parts and I use the LTP driver with identical resistors to have some second harmonic. I need the second harmonic a little bit higher than the third to be happy 🙂. Adding the 2H6 choke made an improvement of about 6dB. Therefore my initial question about bigger choke or active regulation etc.
Then there is the PT which is close to OPT and KT88...see above.
And ofcourse it could be the tubes...changing them around I got results from 300 µVrms to 1500µVrms...
@all
Please tell me how to identify from where the hum/noise is coming from or whatelse I should test.
Do I need a new set of KT88 (or KT120) because of the 'slow' KT88 on the left and the bursts on the 'good' right channel ?
Is it necessary to make a bigger chassis with more spacing ?
Attachments
Last edited:
Try grounding the grids of the KT88 with clipleads or whatever you have and see if the noise goes away.
That the noise is gone when the amp is switched off to be expected as it now basically runs on batteries and all AC except for the input signal is gone. A picture of the internals would be interesting.
That the noise is gone when the amp is switched off to be expected as it now basically runs on batteries and all AC except for the input signal is gone. A picture of the internals would be interesting.
The schematic is sort of trying to show a 0V layout. Does that mean the output stage cathodes go to the 0V star first and not to the negative terminal of C14, and the same with the output transformer B+ not going directly to the C14 positive terminal ?
Why is the input stage not directly decoupled by C1? You show a 22 ohm ground rail resistor.
Do you only have a DMM, or do you have at least a scope, and preferably a spectrum analyser?
Have you balanced the output stage idle current?
Why is there feedback from anode to grid on the output stage sides?
Why is the input stage not directly decoupled by C1? You show a 22 ohm ground rail resistor.
Do you only have a DMM, or do you have at least a scope, and preferably a spectrum analyser?
Have you balanced the output stage idle current?
Why is there feedback from anode to grid on the output stage sides?
@schiirrn
The left channel drops to 290 µVrms, right channel drops to 330 µVrms.
Balancing the bias doesn't have any effect, the only thing that matters is how much current is drawn. I can have one tube at the standard 50 mA and if I increase the other tube the noise is rising, on the other hand, if I reduce bias, I get the lowest noise reading not when both channels daw the same amount of current, but when one tube is cut off. I can further reduce the noise by reducing the other tube as well. In short, output noise is proportional to bias current.
@trobbins
Actually one wire from C14 plus terminal (mounted on the left side) runs to the relay (mounted on the right side), from there back to the center and from there split into 4 wires, 2 (one L+R) to the OPT and 2 (one L+R) to the PCB.
Modify the UL40-S2 into a Super-Triode amplifier
https://www.mennovanderveen.nl/cms/images/onderzoek-ontwikkeling/publicaties/CFB-CFBH-SSCR_universal_test-amp.pdf
https://www.mennovanderveen.nl/cms/images/onderzoek-ontwikkeling/publicaties/download_3.pdf
I don't understand why I should put the grid at 0 V, this is fixed bias, I guess I should turn the KT88 off by maximum negative bias - at least that is what I did.
The left channel drops to 290 µVrms, right channel drops to 330 µVrms.
Balancing the bias doesn't have any effect, the only thing that matters is how much current is drawn. I can have one tube at the standard 50 mA and if I increase the other tube the noise is rising, on the other hand, if I reduce bias, I get the lowest noise reading not when both channels daw the same amount of current, but when one tube is cut off. I can further reduce the noise by reducing the other tube as well. In short, output noise is proportional to bias current.
@trobbins
Yes, it is the usual 2-channel one PSU amplifier. What I have not drawn is the protection for the KT88s. The voltage over L+R R11 and R12 is monitored by a comparator which opens a single relay to cut off all 4 KT88s from the single PSU.
Actually one wire from C14 plus terminal (mounted on the left side) runs to the relay (mounted on the right side), from there back to the center and from there split into 4 wires, 2 (one L+R) to the OPT and 2 (one L+R) to the PCB.
The HBR to break the ground loop, look at the links in post #3, it works.
I can borrow a scope with a mediocre FFT and I have downloaded ARTA demoversion.
Yes, didn't make any difference, see above.
Because Menno van der Veen says this is Supertriode 😀. I don't like the typical harmonic distortion pattern in UL mode (dominant third harmonic), I always used those amplifiers in Triode mode. This modification dropped the third harmonic and I like that - triode sound with UL power.
Modify the UL40-S2 into a Super-Triode amplifier
https://www.mennovanderveen.nl/cms/images/onderzoek-ontwikkeling/publicaties/CFB-CFBH-SSCR_universal_test-amp.pdf
https://www.mennovanderveen.nl/cms/images/onderzoek-ontwikkeling/publicaties/download_3.pdf