Hello all.
Im in the midst of putting the final touches on a 45 amp project. Sadly it seems to have some 60hz noise that I just can't seem to get rid of. I do have access to an oscilloscope and was able to measure ~3mv of 60hz pretty much everywhere in the amp. I see on the inputs of the amp (or 6j5 grid), after the 6j5, and after the 45. Its as if the tubes aren't really amplifying the noise which makes no sense to me.
The ONLY thing that has helped reduce the hum was adding an input transformer to the design. Before the input transformer, the hum was at about 5mv. The voltage level fluctuates a tiny bit depending on where you measure it, but its by in large stable at 3mv. The hum level does not change with volume level. The heater filament wires are all twisted tightly together and shielded. Even if you move them away from the amp, the hum level does not change.
The signal section is all grounded to a ground bus that is grounded to the chassis at 1 point only. The power transformer is safety grounded to the chassis. Moving the transformers and chokes further away from everything else doesn't seem to help.
-----------------
I have built a few other tube amps before and I usually fight with 120hz noise. This is the first time I am seeing 60hz noise like this. At first I thought the noise was due to poor layout, since its just a slap dash prototype, but I moved stuff off the board and really gave everything some space and it made no difference.
I thought about noisy DHT filaments, but it wouldn't explain why I am seeing a fairly constant 3mv across the whole amp including the inputs. Its certainly not PSU noise because I am not seeing ANY 120hz noise on the scope. The only thing I can figure is that it is some sort of ground issue. I will say that this has been the first time I have played with common mode chokes, so maybe I wired the PSU incorrectly?
Any thoughts are appreciated. Sorry in advance for the crappy drawing. I suck at drafting.
Im in the midst of putting the final touches on a 45 amp project. Sadly it seems to have some 60hz noise that I just can't seem to get rid of. I do have access to an oscilloscope and was able to measure ~3mv of 60hz pretty much everywhere in the amp. I see on the inputs of the amp (or 6j5 grid), after the 6j5, and after the 45. Its as if the tubes aren't really amplifying the noise which makes no sense to me.
The ONLY thing that has helped reduce the hum was adding an input transformer to the design. Before the input transformer, the hum was at about 5mv. The voltage level fluctuates a tiny bit depending on where you measure it, but its by in large stable at 3mv. The hum level does not change with volume level. The heater filament wires are all twisted tightly together and shielded. Even if you move them away from the amp, the hum level does not change.
The signal section is all grounded to a ground bus that is grounded to the chassis at 1 point only. The power transformer is safety grounded to the chassis. Moving the transformers and chokes further away from everything else doesn't seem to help.
-----------------
I have built a few other tube amps before and I usually fight with 120hz noise. This is the first time I am seeing 60hz noise like this. At first I thought the noise was due to poor layout, since its just a slap dash prototype, but I moved stuff off the board and really gave everything some space and it made no difference.
I thought about noisy DHT filaments, but it wouldn't explain why I am seeing a fairly constant 3mv across the whole amp including the inputs. Its certainly not PSU noise because I am not seeing ANY 120hz noise on the scope. The only thing I can figure is that it is some sort of ground issue. I will say that this has been the first time I have played with common mode chokes, so maybe I wired the PSU incorrectly?
Any thoughts are appreciated. Sorry in advance for the crappy drawing. I suck at drafting.
Tjj226,
It sounds like you may have a ground loop. It would include the power outlet(s), scope, scope probe, and amplifier. The loop causes current to flow through the scope probe shield and ground clip.
Test for the ground loop this way:
Connect the scope probe ground clip to the amplifier chassis ground. Connect the scope probe tip to exactly the same point on the amplifier chassis ground. That means you are shorting the scope probe ground clip to the probe tip, and both to the chassis ground point.
If you still have 3 mV on the scope trace, that means you have the ground loop I am suspecting.
When current flows through the scope probe shield, it causes a voltage to be induced to the probe wire center conductor.
I notice that your B+ bridge output does not connect to chassis ground until it has passed through the first common mode choke, and then through the two second stage common mode chokes. Why did you use common mode chokes? Is it because there is an extremely high level of interference on your power line?
It sounds like you may have a ground loop. It would include the power outlet(s), scope, scope probe, and amplifier. The loop causes current to flow through the scope probe shield and ground clip.
Test for the ground loop this way:
Connect the scope probe ground clip to the amplifier chassis ground. Connect the scope probe tip to exactly the same point on the amplifier chassis ground. That means you are shorting the scope probe ground clip to the probe tip, and both to the chassis ground point.
If you still have 3 mV on the scope trace, that means you have the ground loop I am suspecting.
When current flows through the scope probe shield, it causes a voltage to be induced to the probe wire center conductor.
I notice that your B+ bridge output does not connect to chassis ground until it has passed through the first common mode choke, and then through the two second stage common mode chokes. Why did you use common mode chokes? Is it because there is an extremely high level of interference on your power line?
I used common mode chokes simply because I haven't used them before and I wanted to know how they would impact the power supply.
And yes, they are grounded the way they are because kevin from K&K audio was saying that in order to get the benefit of common mode rejection, I have to ground the last capacitor in the chain. I asked him if that changed whether I did and LCLC or an LCLCLCLCLCLC chain and he assured me that you always ground the last capacitor.
The only thing I am not sure about is how the ground works with the split rail design. I have a nagging feeling that is where the issue lies.
I will try shorting the probes on the top plate like you suggested and report back.
And yes, they are grounded the way they are because kevin from K&K audio was saying that in order to get the benefit of common mode rejection, I have to ground the last capacitor in the chain. I asked him if that changed whether I did and LCLC or an LCLCLCLCLCLC chain and he assured me that you always ground the last capacitor.
The only thing I am not sure about is how the ground works with the split rail design. I have a nagging feeling that is where the issue lies.
I will try shorting the probes on the top plate like you suggested and report back.
They are all wired for common mode rejection as the data sheet suggests.
http://www.lundahl.se/wp-content/uploads/datasheets/1673.pdf
If you have wired them as the bottom picture in that datasheet then that is OK, but that is not a common-mode choke. It is series-aiding (i.e. a normal choke), with a little extra common-mode rejection. In most cases unnecessary, as the normal serial connection is good enough.
The wiring of the second choke introduces coupling between the two channels, but this should do little harm as you essentially only have one PSU anyway.
A true common-mode choke will not get rid of ripple, because ripple is mainly differential mode. Hence it is good that you have not used common-mode chokes.
3mV of hum everywhere suggest some problem with ground reference. Basically your amp and your oscilloscope are using different grounds, with 3mV potential difference between them.
The wiring of the second choke introduces coupling between the two channels, but this should do little harm as you essentially only have one PSU anyway.
A true common-mode choke will not get rid of ripple, because ripple is mainly differential mode. Hence it is good that you have not used common-mode chokes.
3mV of hum everywhere suggest some problem with ground reference. Basically your amp and your oscilloscope are using different grounds, with 3mV potential difference between them.
Ok, cool. So then the PSU is probably wired correctly (in theory), and the issue is probably in my ground schema.
I should also probably point out that the top plate is steel and not aluminum. IDK what kind of difference that makes. Some people says its better, some people say it causes issue, but its what I had on hand to make the prototype.
I should also probably point out that the top plate is steel and not aluminum. IDK what kind of difference that makes. Some people says its better, some people say it causes issue, but its what I had on hand to make the prototype.
Steel chassis will conduct the magnetic interference from the power transformer to the output transformer. But that is only observed by measuring the output transformer secondary.
If you put spacers between the power transformer and the top panel, and put spacers between the output transformer and the top panel, you can reduce the coupling due to the steel chassis top. But you need to take care mounting the power transformer and the output transformer. They both need metallic / galvanic contact to the top panel. Use a screw, then a lock washer, then the transformer mounting foot, then a spacer, then the top panel, then a lock washer, then a nut. The lock washers need to go through the paint of the transformer mounting foot, and go through any paint or anodization of the bottom of the top panel.
But the problem you seem to have is a constant hum of several points in the circuitry, versus the chassis as the reference/ground. That is back to the issue of the amp grounding, versus the scope grounding.
If you put spacers between the power transformer and the top panel, and put spacers between the output transformer and the top panel, you can reduce the coupling due to the steel chassis top. But you need to take care mounting the power transformer and the output transformer. They both need metallic / galvanic contact to the top panel. Use a screw, then a lock washer, then the transformer mounting foot, then a spacer, then the top panel, then a lock washer, then a nut. The lock washers need to go through the paint of the transformer mounting foot, and go through any paint or anodization of the bottom of the top panel.
But the problem you seem to have is a constant hum of several points in the circuitry, versus the chassis as the reference/ground. That is back to the issue of the amp grounding, versus the scope grounding.
Ok, I got some new and bizarro results.
This time I used a power strip from the wall socket and made sure that my dac, preamp, amp and oscilloscope were all on the same circuit. I am no longer seeing 3mv of 60hz noise everywhere, but I am seeing a ton of high frequency noise.
If I short the two leads on the top plate as 6A3summer suggests, I see 4.6-5mv of Mhz level noise. I thought the tubes might be oscillating, but if I put the lead from ground to the grid of the 6j5, the high frequency noise drops down to less than 2mv. And if I look at the plate of the 6j5, I only see 60hz noise without any high frequency component.
I looked at the noise on the B+ and I am seeing 40-50 mv of weird jagged 5hz noise???? I have never seen anything quite like that on a power supply before. If there is 60hz noise there, it is very small.
Where I did find a ton of 60hz noise was in and around the 45 tube. The grid of the 45 had ~6mv of 60hz and the plate had 300mv.
I threw in a hum pot, and the hum pot either gave no change in hum, or made the hum worse. I am going to throw a couple D cell batteries at the filament and see if the hum doesn't go away.
This time I used a power strip from the wall socket and made sure that my dac, preamp, amp and oscilloscope were all on the same circuit. I am no longer seeing 3mv of 60hz noise everywhere, but I am seeing a ton of high frequency noise.
If I short the two leads on the top plate as 6A3summer suggests, I see 4.6-5mv of Mhz level noise. I thought the tubes might be oscillating, but if I put the lead from ground to the grid of the 6j5, the high frequency noise drops down to less than 2mv. And if I look at the plate of the 6j5, I only see 60hz noise without any high frequency component.
I looked at the noise on the B+ and I am seeing 40-50 mv of weird jagged 5hz noise???? I have never seen anything quite like that on a power supply before. If there is 60hz noise there, it is very small.
Where I did find a ton of 60hz noise was in and around the 45 tube. The grid of the 45 had ~6mv of 60hz and the plate had 300mv.
I threw in a hum pot, and the hum pot either gave no change in hum, or made the hum worse. I am going to throw a couple D cell batteries at the filament and see if the hum doesn't go away.
You have lots of magnetics in a small space. The Lundahls are less sensitive to fields than most, but you also have some other chokes.
This could cause hum because of power transformer, and because of filter choke getting into the other magnetics.
It could also cause magnetic coupling of the tube stages, and maybe even oscillation, like the 5 Hz.
If you look at B+ you may also be seeing line bounce, with the power line voltage jumping as loads in your house and neighbors turn on and off.
You have lots to look for.
This could cause hum because of power transformer, and because of filter choke getting into the other magnetics.
It could also cause magnetic coupling of the tube stages, and maybe even oscillation, like the 5 Hz.
If you look at B+ you may also be seeing line bounce, with the power line voltage jumping as loads in your house and neighbors turn on and off.
You have lots to look for.
1/ Keep heater wires away from audio.
2/ Don't mount transformers next to valves.
3/ Keep input circuit as short as possible.
4/ Decouple each valve stage separately.
5/ Don't run HVAC near audio signals.
6/ Use star grounding where possible.
7/ Elevate heater centre tap to 45 VDC.
8/ Use grid stoppers close to grids..
Some of problems I found causing noise in my early designs.
Definitely not exhaustive.
2/ Don't mount transformers next to valves.
3/ Keep input circuit as short as possible.
4/ Decouple each valve stage separately.
5/ Don't run HVAC near audio signals.
6/ Use star grounding where possible.
7/ Elevate heater centre tap to 45 VDC.
8/ Use grid stoppers close to grids..
Some of problems I found causing noise in my early designs.
Definitely not exhaustive.
Last edited:
Well what gets me is that I have a nice perfect 300mv 60hz sine wave on the plates of the 45s on both channels.
No modulation that I can see, and it is exactly 300mv on the dot. All of the AC wires are shielded to the hilt and are grounded at only one end of the shield. The AC wires are going right past the chokes for the 45, but if I move the wires away, the hum doesn't change.
If it were something like DHT hum or imbalance, I would expect to see slight variances between each channel. But nope. Same nice clean 60hz sine wave on both channels.
Im semi suspicious of the way I have the filament transformers mounted. I have one transformer on top of the other with the steel plate in between them. I tried moving them a while back and it didn't seem to make a difference, but I will move them again and see if I don't discover something.
I have a feeling that when I figure out that issue, a lot of the other issues (5hz noise) will solve themselves.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.