Yes, I remember that. Whats confusing is the Drip manual shows both XLR pin 1s connected to chassis along with audio ground from PCB
Pultec clone - grounding problems
- Thread starter morls
- Start date
Help Support GroupDIY Audio Forum:
Bo Deadly
Well-known member
Those switch wires are carrying 220 VAC. Even if the wires are twisted, they will still emit electromagnetic radiation. Magnetics (transformers, inductors and chokes) can pickup that noise where and become part of the audio. So no, the PCB mounting legs is not a problem. But currently the wire runs right next to you EQ inductors. If you can bring it out from under the board without disconnecting anything, try watching the spectrum while you move that wire around (being vary careful not to touch anything of course). Try moving it close to the EQ inductors or other parts of the boards and see what happens to the mains noise in reaction to that. Another thing you could try is to remove the switch and let it hang out the back away from the boards when you test.
Also, I don't know if I mentioned this before but you might see a small difference if you put the lid on when you test. You don't have to bolt it down but being steel it provides some electromagnetic shielding. Actually if it does make a difference that would be very interesting because we have to consider that the mains noise could be external such as because you just happen to be unlucky enough to be working right next to a power cable in the wall. If you ground it to the chassis with one screw or an alligator clip, it would also provide RF shielding but that's not the type of noise we're immediately concerned about.
Audio "ground" is what I usually call 0V but sometimes referred to as signal ground. Then you have the chassis ground. The chassis and 0V should be connected at only one point through one fat wire between the 0V preferably right next to the 0V of the power supply filter capacitors and the chassis bolt on the back where earth ground is connected. At least that's how I do it and I test my builds pretty thoroughly and never had a problem (although I haven't built anything with a linear supply in years) and that's what we tell people to do around here and I don't see any reason to change those instructions at this time.
Generally you usually don't have to worry about ground loops with shields connected to the chassis because it's all low impedance and there shouldn't be currents running through those. Exceptions might be made for phantom power and maybe power amps. What is important is that radio frequency EMI is most likely to get in on pin 1 and so you want to shunt that to the chassis over the shortest possible wire. That's why there's a separate chassis bolt for that. Some XLRs actually have a metal spike in the screw hole that stabs the panel and connects pin 1 to the panel for this reason. If you connect pin 1 to your shield and it's connected to the ground plane, then you might be letting RF it a little more than it should. But as long as the wire between pin 1 and the chassis is solid, I'm not sure the extra pin 1 to 0V would be a problem. At least I don't see how it could contribute to the mains hum. But again, you have multiple violations to deal with and it can be difficult to predict what is happening when you have multiple different problems going on.
I would do things iteratively and see what happens at each step so that if you see a change in mains hum you know which change made the most difference. If I were doing this I would gut the whole thing and just rewire it the way I have in the drawing. But for you it might help to understand these things (which is obviously why you and I and everyone else is here). So for now maybe leave them. Or not. But if you do disconnect those points, maybe just desolder and test first before you just cut.
I'm not sure you understood what I suggested. What you need to find out is if the grounds on the PCB are connected. Unless you desolder all of the grounds before measuring with your meter, there is no way you could know that for sure. It could be that the ground plane and 0V are actually not connected. If that is the case, then you will need some additional wires that I did not put in my drawing.
Also, I don't know if I mentioned this before but you might see a small difference if you put the lid on when you test. You don't have to bolt it down but being steel it provides some electromagnetic shielding. Actually if it does make a difference that would be very interesting because we have to consider that the mains noise could be external such as because you just happen to be unlucky enough to be working right next to a power cable in the wall. If you ground it to the chassis with one screw or an alligator clip, it would also provide RF shielding but that's not the type of noise we're immediately concerned about.
Audio "ground" is what I usually call 0V but sometimes referred to as signal ground. Then you have the chassis ground. The chassis and 0V should be connected at only one point through one fat wire between the 0V preferably right next to the 0V of the power supply filter capacitors and the chassis bolt on the back where earth ground is connected. At least that's how I do it and I test my builds pretty thoroughly and never had a problem (although I haven't built anything with a linear supply in years) and that's what we tell people to do around here and I don't see any reason to change those instructions at this time.
Generally you usually don't have to worry about ground loops with shields connected to the chassis because it's all low impedance and there shouldn't be currents running through those. Exceptions might be made for phantom power and maybe power amps. What is important is that radio frequency EMI is most likely to get in on pin 1 and so you want to shunt that to the chassis over the shortest possible wire. That's why there's a separate chassis bolt for that. Some XLRs actually have a metal spike in the screw hole that stabs the panel and connects pin 1 to the panel for this reason. If you connect pin 1 to your shield and it's connected to the ground plane, then you might be letting RF it a little more than it should. But as long as the wire between pin 1 and the chassis is solid, I'm not sure the extra pin 1 to 0V would be a problem. At least I don't see how it could contribute to the mains hum. But again, you have multiple violations to deal with and it can be difficult to predict what is happening when you have multiple different problems going on.
I would do things iteratively and see what happens at each step so that if you see a change in mains hum you know which change made the most difference. If I were doing this I would gut the whole thing and just rewire it the way I have in the drawing. But for you it might help to understand these things (which is obviously why you and I and everyone else is here). So for now maybe leave them. Or not. But if you do disconnect those points, maybe just desolder and test first before you just cut.
I'm not sure you understood what I suggested. What you need to find out is if the grounds on the PCB are connected. Unless you desolder all of the grounds before measuring with your meter, there is no way you could know that for sure. It could be that the ground plane and 0V are actually not connected. If that is the case, then you will need some additional wires that I did not put in my drawing.
Matt Syson
Well-known member
As I think has been said a few times already (I have not read the whole thread in detail) but the mains 'pair' to and from the switch (it should be 2 pairs if it is a 2 pole mains switch) where the pair is twisted tightly so that all the 'go' current is tightly twisted with the 'return' current otherwise the 'balancing' (cancellation) will not happen.. Then as I believe was taught at 'valve nursery school' the twisted pair(s) should be routed as tight into the corners of the chassis as possible. Having mutiple audio ground/chassis ground points taken to a random sprinkling of bolts to the chasis is almost doomed to failure, not least as the chasis is probably steel so is liable to pick up stray mains magnetic fields so giving different 'ground' potentials which are then imprinted onto the audio path. The concept of ground follows signal starts to get very complicated when working in 3 dimensions with real distance between circuit elements. Every 'amplifying' device (subsystem) amplifies the DIFFERENCE between it's so called input signal and the reference (ground or antiphase signal for a balanced circuit).
Simply copying the schematic of any piece of gear can only yield good results if the layout including grounding and power is reproduced faithfully. Having said that, not all 'originals' managed to achieve minimal interference/noise.
Good luck and keep safe
Matt S
Simply copying the schematic of any piece of gear can only yield good results if the layout including grounding and power is reproduced faithfully. Having said that, not all 'originals' managed to achieve minimal interference/noise.
Good luck and keep safe
Matt S
Winston OBoogie
Well-known member
Sure. The parts that sound fantastic to you are the parts that are original ie. the basic circuit topology.
The parts that aren't so fantastic sounding are the patrs that aren't original i.e. noisy PCB's.
Still, if you are happy with the boards then I'm OK with that.
I'm just saying that, as someone who has laid out, and then built more individual PCB based tube gear than had hot dinners, these don't inspire confidence in me,
No reflection or slight on you intended at all.
Last edited:
Winston OBoogie
Well-known member
Onwards and upwards:
For sure.
Your power switch wiring, as Abbey and also Matt said above, should be a tightly twisted pair and needs to hug the chassis and corners.
Where're the heater supply runs for the tubes, is it AC or DC? If it's AC and just a pair of traces on the underside of the PCB then you're asking for big trouble.
AC heaters have to be tightly twisted pairs of wire, also run tight to the chassis and hugging the corners. As far away from tube electrodes and signal runs as possible.
Clean DC can be run on the PCB, but you still need to pay attention to where it's going and how it gets to where it needs.
Where are the individual stage bypass caps located for each stage? Are all the filter caps on the right side of the PCB as we look at the pics you've posted?
For sure.
Your power switch wiring, as Abbey and also Matt said above, should be a tightly twisted pair and needs to hug the chassis and corners.
Where're the heater supply runs for the tubes, is it AC or DC? If it's AC and just a pair of traces on the underside of the PCB then you're asking for big trouble.
AC heaters have to be tightly twisted pairs of wire, also run tight to the chassis and hugging the corners. As far away from tube electrodes and signal runs as possible.
Clean DC can be run on the PCB, but you still need to pay attention to where it's going and how it gets to where it needs.
Where are the individual stage bypass caps located for each stage? Are all the filter caps on the right side of the PCB as we look at the pics you've posted?
I've decided to rewire everything, and have pulled out the boards to clean everything up. This is one of the first builds I did, must be over 5 years ago now, and it's good to revisit. I've found another potential noise source - the transformer and inductor leads that are beneath the board are very long, and I'm going to clean them up. Should I make them as short as is practical, and twist pairs where possible?
I've also taken some photos of the boards. I'm certain that these noise issues are 100% caused by my work, and not the boards. Each of the control PCBs has a layer of copper on each side, with the traces in between. The main PCB has a thicker ground trace on the outermost part of the board.
Have also drilled the from plate to move the power switch to the same side as AC inlet, which will also help.
I'm also planning to use a choke, there is a large resistor between the two PSU filter caps, which aren't installed. The silkscreen specifies Hammond 158M, rated at 10H. Are there other options?
I've also taken some photos of the boards. I'm certain that these noise issues are 100% caused by my work, and not the boards. Each of the control PCBs has a layer of copper on each side, with the traces in between. The main PCB has a thicker ground trace on the outermost part of the board.
Have also drilled the from plate to move the power switch to the same side as AC inlet, which will also help.
I'm also planning to use a choke, there is a large resistor between the two PSU filter caps, which aren't installed. The silkscreen specifies Hammond 158M, rated at 10H. Are there other options?
Attachments
Last edited:
warpie
Well-known member
- Joined
- Feb 7, 2009
- Messages
- 1,598
A bit off topic but regarding the mains switch wiring... I guess that twisting is only important if the PSU is internal right? In an external PSU there's no real benefit of twisting the wires correct?
I'm only asking because I have a piece of gear that I built ages ago and te PSU is in its own enclosure.
I need to measure it again but I remember seeing a bit of hum. Although not at 50hz but rather at 100, 200, 300 etc..
I'm only asking because I have a piece of gear that I built ages ago and te PSU is in its own enclosure.
I need to measure it again but I remember seeing a bit of hum. Although not at 50hz but rather at 100, 200, 300 etc..
Correct.
This is a sign of either insufficient value caps, or rectifier current polluting the audio ground.
Many tube units (particularly vintage or recreated vintage) rely on a heftypiece of copper wire, or worse chassis to establish ground connections, without respect for the "ground follows signal" rule. The resulting noise performance was satisfactory in the context of the days, recording to tape, mastering to disc, AM broadcast, but today, with 16-bit digital audio being the norm, the imperfections stand out.
Winston OBoogie
Well-known member
The underneath view of the amplifier PCB's shows that they are 6.3V AC heaters.
That 6V3 current (which is twice that of 12V6 wired heaters) is just looped around all the signal traces and allowed to spew/radiate its noise into any and all sensitive areas. It would be considered bad practice if these runs were just looping around each valve socket, let alone all of the signal traces and amplifier circuit.
Pretty basic stuff really.
You can't easily switch this over to running clean DC heaters either since the traces are shared with the rectifier diode valve.
I'm sorry, but in my humble opinion, this will be a noisy PCB no matter how you wired it.
You'd need to do surgery to make this optimum.
I doubt this reply will be liked and I'm sorry about that. But maybe, if anyone, it should be Drip who is shot rather than a messenger.
That 6V3 current (which is twice that of 12V6 wired heaters) is just looped around all the signal traces and allowed to spew/radiate its noise into any and all sensitive areas. It would be considered bad practice if these runs were just looping around each valve socket, let alone all of the signal traces and amplifier circuit.
Pretty basic stuff really.
You can't easily switch this over to running clean DC heaters either since the traces are shared with the rectifier diode valve.
I'm sorry, but in my humble opinion, this will be a noisy PCB no matter how you wired it.
You'd need to do surgery to make this optimum.
I doubt this reply will be liked and I'm sorry about that. But maybe, if anyone, it should be Drip who is shot rather than a messenger.
Attachments
Last edited:
Why have you chosen to move the filter caps from their designated position to the middle of the board?
The net result is that the rectifier current circulates in the ground trace.
I’ll have to take your word for it, as I don’t have any understanding of circuit design. I’ll do my best to get the wiring right, and see what the results are.
I haven't moved any caps around, but I have removed the two motor run caps which were incorrectly mounted. I may well be mistaken but I took these to be the filter caps? The directions for using a choke are on the silkscreen between the footprints of these two caps, which I haven't remounted yet.
Attachments
Winston OBoogie
Well-known member
Fair enough. I posted a screen shot from one of Morgan Jones' books on building tube amps so you might take his word for it. Also so that you might see what surgery you could do while you have the board out of the chassis.
I'll second Abbey's point that you want your main filter caps next to the rectifier to keep that current loop small.
However, it's generally considered best practice to locate an individual stage's bypass cap next to its respective stage to keep that current loop small.
Dc heathers should be the simple fix. Cut the pcb and rewire ac heather sound tricky...
How would the wiring changes I've added to this pic work?
Attachments
Winston OBoogie
Well-known member
Better yes 
Twist the AC heater wires tightly by clamping a pair in a drill's chuck. Run the twisted pairs directly soldered to the tube pins (cut those traces there too of course) and run them perpendicular to the pcb and down to the chassis.
I see too that an awful lot of hum will potentially have been picked up by your audio transformer leads as they crossed over those heater traces.
Twist the top and bottom of each audio transformer winding together as tightly as possible, keep the leads as short as practicable, and keep those leads well away from your twisted AC heater runs.
You should maybe elevate the heaters of the audio tubes by about 40V DC too. The original Pultec schematic shows this I'm sure. In your case, you can create a centre tap with a pair of 100 ohm resistors from each side of your heater voltage. Use the junction of the two 100R's to inject your elevating voltage .
Otherwise, use a "humdinger" pot across your heaters and return the wiper to ground.
Twist the AC heater wires tightly by clamping a pair in a drill's chuck. Run the twisted pairs directly soldered to the tube pins (cut those traces there too of course) and run them perpendicular to the pcb and down to the chassis.
I see too that an awful lot of hum will potentially have been picked up by your audio transformer leads as they crossed over those heater traces.
Twist the top and bottom of each audio transformer winding together as tightly as possible, keep the leads as short as practicable, and keep those leads well away from your twisted AC heater runs.
You should maybe elevate the heaters of the audio tubes by about 40V DC too. The original Pultec schematic shows this I'm sure. In your case, you can create a centre tap with a pair of 100 ohm resistors from each side of your heater voltage. Use the junction of the two 100R's to inject your elevating voltage .
Otherwise, use a "humdinger" pot across your heaters and return the wiper to ground.
Last edited:
Matt Syson
Well-known member
Hi
That heater trace is almost guaranteed to make it hum. I would suggest for a bit of simplicity to mount a bridge rectifier and large 'smoothing' cap right at the transformer, or at least well away from anything else and feed with DC. You will need a dropper resistor to 'lose' a bit of voltage and then another large cap. I think schottky diodes may be beneficial as they are a bit less 'noisy' than ordinary silicon and they won't lose so much voltage which then makes the cap/res/cap arrangement better to reduce ripple. Some Marshall guitar amps use this idea where the first couple of valves use DC with the latter and output valves running on AC.
All the transformer wires should be tightly twisted as pairs so each 'end' of any given winding should be twisted with it's 'opposite'. This will take a bit of thinking and concentration to get right.
The inductor wires should also be twisted tightly, probably as a complete bunch because at any one time there will be the common and whichever tapping is being used.
Of course when most valve gear was built originally mobile phones did not exist and anything like a 'switchmode' supply was a rare beast although mercury rectifiers had their own 'excitement'. Later selenium rectifiers were introduced (around 1940 I guess) but they were quite 'soft' so not much high frequency noise produced. It is a game of 3 dimension tetris.
Matt S
That heater trace is almost guaranteed to make it hum. I would suggest for a bit of simplicity to mount a bridge rectifier and large 'smoothing' cap right at the transformer, or at least well away from anything else and feed with DC. You will need a dropper resistor to 'lose' a bit of voltage and then another large cap. I think schottky diodes may be beneficial as they are a bit less 'noisy' than ordinary silicon and they won't lose so much voltage which then makes the cap/res/cap arrangement better to reduce ripple. Some Marshall guitar amps use this idea where the first couple of valves use DC with the latter and output valves running on AC.
All the transformer wires should be tightly twisted as pairs so each 'end' of any given winding should be twisted with it's 'opposite'. This will take a bit of thinking and concentration to get right.
The inductor wires should also be twisted tightly, probably as a complete bunch because at any one time there will be the common and whichever tapping is being used.
Of course when most valve gear was built originally mobile phones did not exist and anything like a 'switchmode' supply was a rare beast although mercury rectifiers had their own 'excitement'. Later selenium rectifiers were introduced (around 1940 I guess) but they were quite 'soft' so not much high frequency noise produced. It is a game of 3 dimension tetris.
Matt S
Matt Syson
Well-known member
As pointed out the heater should be 'biassed'to around 40 Volts DC or so which reduces the strain cathode to heater on some valves and can reduce hum a bit (I think) depending the exact details of each valve circuit.
Winston OBoogie
Well-known member
I agree that DC would be best. However, we have a 6X4 H.T. rectifier here which has its heater fed from the same 6V3 AC, so it's not as simple as just rectifying the AC right at the transformer.
