Just completed a build of the attached linestage.
The rca inputs are dc coupled (1M ground resistor) prior to being wired to a 100k stepped attenuator, which is then dc coupled (1M ground resistor) prior to going to the input of the pcb.
All the rca jacks and the stepped attenuator are grounded to the pcb, the rca jacks are also grounded to the chassis (which is in turn grounded to the transformer).
However, there is a humming/buzzing noise even at minimal volume (it is also there regardless of whether I wire in the stepped attenuator or not).
Is anyone able to see a design flaw in this circuit that would lead to such noise?
Note that I used R5/R9 as 2k (schematic shows 28k).
The rca inputs are dc coupled (1M ground resistor) prior to being wired to a 100k stepped attenuator, which is then dc coupled (1M ground resistor) prior to going to the input of the pcb.
All the rca jacks and the stepped attenuator are grounded to the pcb, the rca jacks are also grounded to the chassis (which is in turn grounded to the transformer).
However, there is a humming/buzzing noise even at minimal volume (it is also there regardless of whether I wire in the stepped attenuator or not).
Is anyone able to see a design flaw in this circuit that would lead to such noise?
Note that I used R5/R9 as 2k (schematic shows 28k).
Last edited:
Schematic doesn't show your "1M ground resistors" and attenuators so I can't tell what that's all about.
R2 and R3 should be on the other side of the 100k if they are intended as stoppers.
Not clear how you ran your circuit and chassis grounds. But the input jacks should be grounded only to the circuit, not the chassis. Sounds like a ground loop issue.
R2 and R3 should be on the other side of the 100k if they are intended as stoppers.
Not clear how you ran your circuit and chassis grounds. But the input jacks should be grounded only to the circuit, not the chassis. Sounds like a ground loop issue.
Thanks for the assistance. I've taken out the grounding which links the sleeves of the input and output jacks, but the noise persists. You are right, it certainly sounds exactly like a grounding problem.
Grounding structure is as follows:
(1) RCA input jacks, sleeve is running to the stepped attenuator
(2) Stepped attenuator ground is running to the pcb GND input
(3) GND output on PCB goes to the RCA output jack sleeves
(4) RCA output jack sleeves also connected to the chassis ground, which is the same point as the transformer earth
Is this clear enough, or should I post a picture?
Grounding structure is as follows:
(1) RCA input jacks, sleeve is running to the stepped attenuator
(2) Stepped attenuator ground is running to the pcb GND input
(3) GND output on PCB goes to the RCA output jack sleeves
(4) RCA output jack sleeves also connected to the chassis ground, which is the same point as the transformer earth
Is this clear enough, or should I post a picture?
Hard to diagnose without seeing the physical elements but let's try.
1. Remove any and all connections between circuit ground and chassis ground.
2. By "transformer earth" I hope you mean the metal frame of the transformer is connected to the chassis. No transformer windings should be connected to chassis ground. If your power supply is not on the same pcb, it too should not be connected to chassis ground.
3. I assume it's built on a metal chassis. Make sure chassis is connected to mains ground ("earth").
4. Make sure input and output jacks are connected to the pcb with shielded cable. If the pots are not on the pcb, make sure they are also connected with shielded cable.
5. Short out the input jacks with shorting plugs, At this point you should have no noise. If you still do, it's likely not a grounding issue. Measure the noise frequency and go from there.
6. If the noise stopped, connect circuit ground to chassis at one point only either directly or via parallel RC circuit. You may have to experiment.
Google and read about grounding techniques. For example:
http://www.geofex.com/article_folders/stargnd/stargnd.htm
1. Remove any and all connections between circuit ground and chassis ground.
2. By "transformer earth" I hope you mean the metal frame of the transformer is connected to the chassis. No transformer windings should be connected to chassis ground. If your power supply is not on the same pcb, it too should not be connected to chassis ground.
3. I assume it's built on a metal chassis. Make sure chassis is connected to mains ground ("earth").
4. Make sure input and output jacks are connected to the pcb with shielded cable. If the pots are not on the pcb, make sure they are also connected with shielded cable.
5. Short out the input jacks with shorting plugs, At this point you should have no noise. If you still do, it's likely not a grounding issue. Measure the noise frequency and go from there.
6. If the noise stopped, connect circuit ground to chassis at one point only either directly or via parallel RC circuit. You may have to experiment.
Google and read about grounding techniques. For example:
http://www.geofex.com/article_folders/stargnd/stargnd.htm
It just occurred to me that this is a 6111 tube. I don't know what B+ you're running, but if it's less than 63V as suggested by the PS cap, you probably have a non-optimal design. The data sheet suggests much higher B+ and much higher anode loading.
http://frank.pocnet.net/sheets/127/6/6111.pdf
Depending on the B+ voltage and tube current, Chris may be right about more filtering.
http://frank.pocnet.net/sheets/127/6/6111.pdf
Depending on the B+ voltage and tube current, Chris may be right about more filtering.
With such small anode resistors you will have virtually no PSRR - you are listening to your PSU ripple, which is rather crude with no smoothing (just a reservoir cap). Buzz is almost inevitable. You need more smoothing, and check that the PSU ground arrangements are correct so you are not also listening to charging pulses in a ground link.
Re the B+, the pcb calls for 35VAC = 50vdc, I assumed the tube was biased correctly in the circuit? Quite easy to increase the voltage, I ordered a few spare pcbs for testing so can alter the value of the capacitors, too. I've increased the filtering to 3300uF.
I'll try to remove the connection between earth and chassis ground and see if this improves the noise. The chassis is connected to mains earth.
I'll try to remove the connection between earth and chassis ground and see if this improves the noise. The chassis is connected to mains earth.
Under normal bias conditions the 6111 has an anode impedance of 4k. Your bias may differ, but it probably won't be hugely different. 28k anode resistor and 4k anode impedance means PSRR is about 1/8=-18dB. You have 2k instead of 28k, so your PSRR will be 2/3~-3dB. Throwing away 15dB-worth of PSRR will create buzz in any weak design. This is a weak design because it has a crude PSU. Don't increase the reservoir, instead add an RC smoothing section. You need to stop and think about why it is failing, not just slap on sticking plasters.
OK thanks - I ordered a couple of kits and each of them came with 2k for R5/R9, so I assumed it was an improvement on the original circuit, which showed 28k.
I'll take out the 2k for R5/R9, what is the optimal value for these resistors? While we are in the desoldering phase, are there any other improvements to be made?
I assume whoever swapped 2k for 28k was simply trying to drastically reduce the gain and didn't bother to think about other consequences (or maybe didn't even realise that there might be other consequences). A low triode anode load gives poor PSRR, and increased distortion. Maybe he was after 'tube sound'?
The 6111 has a mu of about 20, so that is the natural gain of the valve when used in a well-designed grounded cathode circuit. Far too high for a line stage, but there is no easy way to reduce it without serious knock-on effects:
1. reducing the anode load ruins PSRR and increases distortion,
2. removing cathode bypass (so adding cathode degeneration) increase output impedance so reduces load driving capability, and also ruins PSRR by raising anode impedance,
3. adding feedback by turning it into an anode follower seriously reduces input impedance so might not always be appropriate.
Possibly the best option is to use the original 28k, then follow the stage with an attenuator to throw away most of the gain. This may sound daft. It is daft, but it still may be the best option for someone who has set his heart on a grounded cathode line stage with very little gain. It retains PSRR and distortion (instead of making them worse), reduces output impedance, and allows for easy adjustment of gain. Or just ditch the line stage; in my opinion this is the best improvement.
The 6111 has a mu of about 20, so that is the natural gain of the valve when used in a well-designed grounded cathode circuit. Far too high for a line stage, but there is no easy way to reduce it without serious knock-on effects:
1. reducing the anode load ruins PSRR and increases distortion,
2. removing cathode bypass (so adding cathode degeneration) increase output impedance so reduces load driving capability, and also ruins PSRR by raising anode impedance,
3. adding feedback by turning it into an anode follower seriously reduces input impedance so might not always be appropriate.
Possibly the best option is to use the original 28k, then follow the stage with an attenuator to throw away most of the gain. This may sound daft. It is daft, but it still may be the best option for someone who has set his heart on a grounded cathode line stage with very little gain. It retains PSRR and distortion (instead of making them worse), reduces output impedance, and allows for easy adjustment of gain. Or just ditch the line stage; in my opinion this is the best improvement.
OK I'll try the 28k and report back - mu of 20 isn't too bad for a linestage (depending on the vrms of the source of course).
I might also increase the value of R2 & R3 and put them on the other side of R1 & R8 - as 'DGTA' suggested, they appear to be intended as grid stoppers, but the values are neither high enough nor are they in the correct position.
I might also increase the value of R2 & R3 and put them on the other side of R1 & R8 - as 'DGTA' suggested, they appear to be intended as grid stoppers, but the values are neither high enough nor are they in the correct position.
Update - no luck, disconnected the connection between circuit ground and mains earth, also increased the anode load to 28k, but the buzz persists.
What is the next step - without access to a scope, all I can really suggest for a measurement of the noise is that it is about 440Hz - close to the 'middle c' sound in music. Does this help at all?
The other possibility is that it could be a pin misplacement? The filament is fine, the gain and sound is excellent - but given that there are no markings on the pins, could one of them be misplaced and result in this kind of noise?
What is the next step - without access to a scope, all I can really suggest for a measurement of the noise is that it is about 440Hz - close to the 'middle c' sound in music. Does this help at all?
The other possibility is that it could be a pin misplacement? The filament is fine, the gain and sound is excellent - but given that there are no markings on the pins, could one of them be misplaced and result in this kind of noise?
No. With 2 channels and 1 LED, it is improbable that the led is part of the circuit, perhaps it is a power indicator. On the other hand, a kit where the components don't match the schematic is a red flag of a poor quality supplier. Anyway, that circuit should work and not make noise regardless of led biasing or not.
First make sure the noise you hear is in the line stage and not the subsequent power amp.
Then put an ac voltmeter on the power supply rail and see how much ripple you have. The ripple should be 100Hz, not 440. Make sure your AC voltmeter can take the DC present on the rail, if not use a high voltage cap of a few uF in series with the leads. If you have significant ripple, you can test for that. Use a high voltage cap to connect that ripple to your power amp and see if it's the same noise.
First make sure the noise you hear is in the line stage and not the subsequent power amp.
Then put an ac voltmeter on the power supply rail and see how much ripple you have. The ripple should be 100Hz, not 440. Make sure your AC voltmeter can take the DC present on the rail, if not use a high voltage cap of a few uF in series with the leads. If you have significant ripple, you can test for that. Use a high voltage cap to connect that ripple to your power amp and see if it's the same noise.
Hey,
Why do you call the thread with "6111/6N16B" which are two different miniature tubes when the thread is only about 6111?
Whereas 6111 is a really unlinear tube, 6N16B is acceptable in that respect.
If you have the chance to switch from the mediocre 6111 to the better 6N16B(but at descent anode-voltage), do it!
As the whole circuit is suboptimal I suggest you start from the beginning using the Ua/Ia-curves to find the resitorvalues. Together with a realistic B+ of over 100V it might work, allthough with an all to high gain.
Another idea could be to use one pcb per channel and reconstruct to a GG stage with some NFB to get gain down to an acceptable level.
Why do you call the thread with "6111/6N16B" which are two different miniature tubes when the thread is only about 6111?
Whereas 6111 is a really unlinear tube, 6N16B is acceptable in that respect.
If you have the chance to switch from the mediocre 6111 to the better 6N16B(but at descent anode-voltage), do it!
As the whole circuit is suboptimal I suggest you start from the beginning using the Ua/Ia-curves to find the resitorvalues. Together with a realistic B+ of over 100V it might work, allthough with an all to high gain.
Another idea could be to use one pcb per channel and reconstruct to a GG stage with some NFB to get gain down to an acceptable level.
Last edited:
I have a whole bunch of 6n16b awaiting my attention, so that is easy. I also have two spare pcbs, so the idea to use one per channel is a distinct possibility. I also have a transformer which has secondaries of:
60v-0v
60v-0v
6.3v-0v
6.3v-0v
This will get us a B+ of 85v dc, could this be enough to work with?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.