The 35V are there because the device is floating in relation to earth. The measurement means nothing since they do not share a common reference point. Perhaps only through the interconnects. But do you really want a reference through some thin and long wires? And is your pre or amp earthed? 35V are a lot for a voltage drop accross an interconnect.
As Andrew said, connect the common to earth.
Having no robust earth reference in a DIY device is very bad and dangerous practice, and illegal in many areas.
If you have ground loops, use one of the many options available to break the earth connection without removing it (thermistor, diodes or whatever)
As Andrew said, connect the common to earth.
Having no robust earth reference in a DIY device is very bad and dangerous practice, and illegal in many areas.
If you have ground loops, use one of the many options available to break the earth connection without removing it (thermistor, diodes or whatever)
Last edited:
I've discovered that people who live in 100 and 120 V areas have much more relaxed views on the earth connection than people who live in 220 and 240 V areas. None of my equipment has earth connection because none of the wiring in my house has an earth connection, it's all two wire. I still manage to sleep at night.
As dimkasta says, the 35V is largely meaningless. Probably. If there is an high impedance between earth and chassis (chassis is disconnected from earth) then the smallest induced currents will produce a large voltage. This voltage would disappear as soon as the earth connection is made, and is likely harmless.
If you are going to go ahead and earth the phono preamp, you will probably need to isolate the circuit COM/GND from the chassis with a small resistor, diodes. etc.
As dimkasta says, the 35V is largely meaningless. Probably. If there is an high impedance between earth and chassis (chassis is disconnected from earth) then the smallest induced currents will produce a large voltage. This voltage would disappear as soon as the earth connection is made, and is likely harmless.
If you are going to go ahead and earth the phono preamp, you will probably need to isolate the circuit COM/GND from the chassis with a small resistor, diodes. etc.
Call me skeptical. 35v is not trivial to my mind (and in my experience). There's a problem with that much float, and that's the first place to work on. At the very least, as others are saying, once you connect the interconnects, I can see NO WAY to get rid of the hum. My experience is it will find a way to drain off, through the preamp or amp circuit itself if not those units' ground systems, and you will have a BIG ground loop with a lot of current running through it. That or you'll blow some parts in the next components depending on their design.
If ALL components are floating, then the 35v float MAY be meaningless in a technical sense, but you'll still get big hum unless all components float at the same voltage, no? At least that's my understanding, though I freely admit I'm no electronics guru.
And rjm, sounds like your home makes an earth ground somewhat pointless, but if that unit goes to someone WITH an earth ground, they may believe themselves safer than they are. Perhaps for 35v it's not a big deal - I don't know. But should that 118v (in my case) mains come loose and touch the case, turning on your amp could be a tad uncomfortable. That I DO have experience with.
BTW, when I had about 35v on the case of an amp (the user was protected by some wood casing) and I connected a floating battery-powered line stage to the amp with a separate ground wire in order to resolve a hum problem, that ground wire insulation began to melt almost instantly. Serious current.
A melting wire shows there is something weird in your wiring.
A ground (read earth/chassis connection) wire should have virtually 0 current moving in it under normal operation. It should not be used for returning currents. Just for protection in case of an accident.
The 0V COM return point should be the where you return all your currents, and the wire should be able to handle any moving currents the way your V+ wire does.
A ground (read earth/chassis connection) wire should have virtually 0 current moving in it under normal operation. It should not be used for returning currents. Just for protection in case of an accident.
The 0V COM return point should be the where you return all your currents, and the wire should be able to handle any moving currents the way your V+ wire does.
@CarlP. All my builds have 3 wire plugs and earthed power supplies, I'm just saying that at home they don't connect to anything so they don't *do* anything in my case presently.
If you put a transformer in a floating metal enclosure and connect the enclosure to the secondary winding center tap or equivalent the case can float up to tens of volts relative to earth pretty easily. Of course no current is drawn and its perfectly safe. Any resistance to earth will immediately bring it to zero volts. The reason it floats up to some voltage is the leakage current from the primary to the secondary windings is non-zero even if it is very small.
If you put a transformer in a floating metal enclosure and connect the enclosure to the secondary winding center tap or equivalent the case can float up to tens of volts relative to earth pretty easily. Of course no current is drawn and its perfectly safe. Any resistance to earth will immediately bring it to zero volts. The reason it floats up to some voltage is the leakage current from the primary to the secondary windings is non-zero even if it is very small.
its ok guys, that is why I am bench testing this and finding whats up. There are several commercial audio products that do not have an earth ground. This is like Richard said, its a floating ground.
If the earth ground on my preamp makes no difference in hum, then yes, obviously I will use it. For now, with testing trying to track down other sources, I have left it disconnected.
I am swapping out my twisted pair wires with some actual RCA shielded cables. I noticed that touching the insulation on the input or output wires increased noise a ton! so, gonna try that this evening and finish my transformer shield/separator.
trying the simple things first before I splurge on a separate enclosure to house the toroid and diode bridges.
Ground loops happen, no biggie. I will get it working, and I thank you guys for the input! ;-)
If the earth ground on my preamp makes no difference in hum, then yes, obviously I will use it. For now, with testing trying to track down other sources, I have left it disconnected.
I am swapping out my twisted pair wires with some actual RCA shielded cables. I noticed that touching the insulation on the input or output wires increased noise a ton! so, gonna try that this evening and finish my transformer shield/separator.
trying the simple things first before I splurge on a separate enclosure to house the toroid and diode bridges.
Ground loops happen, no biggie. I will get it working, and I thank you guys for the input! ;-)
@scottjoplin well yes but you never, ever use neutral as a substitute for earth. That's pretty much the ground zero rule of electronics to prevent newbies from killing themselves. I assume though you were thinking more along the lines of why the circuit should be floating up to 35 V when neutral is close to zero volts. Which now that I think about it is a very good question for a transformer with 12 V secondaries and only ~18 V DC on the supply rails. I appreciate that 35 V is |V+|+|V-| but still ...
@Erik the reason it gets noisy when you touch the wires is your body is conductive and acts as a (high impedance) earth connection. In the absence of a real (low impedance) earth connection and in the case of some finite power supply noise leaking into the circuit common, that noise is going to flow up the circuit to drain off through you.
Are you sure you get more noise touching the output wires too though? Usually this happens at the input, its normal, but I'm surprised this kind of ground loop can be heard equally on the output as well. That's rather alarming, to be honest.
After you try some shielded cables, I suggest you look at connecting the chassis at points other than the board GND pad. Try the power supply for example, or the input RCA. And also look at the transformer, try a different type if you have one lying around, or at try powering the circuit from a different, external power supply.
One final thought: are you sure the primary windings are connected properly, and perhaps you can try switching phase (neutral to the black dot of the primary windings, or live to the black dot of the primary windings) sometimes transformer leakage is worse from one side of the coil than the other.
@Erik the reason it gets noisy when you touch the wires is your body is conductive and acts as a (high impedance) earth connection. In the absence of a real (low impedance) earth connection and in the case of some finite power supply noise leaking into the circuit common, that noise is going to flow up the circuit to drain off through you.
Are you sure you get more noise touching the output wires too though? Usually this happens at the input, its normal, but I'm surprised this kind of ground loop can be heard equally on the output as well. That's rather alarming, to be honest.
After you try some shielded cables, I suggest you look at connecting the chassis at points other than the board GND pad. Try the power supply for example, or the input RCA. And also look at the transformer, try a different type if you have one lying around, or at try powering the circuit from a different, external power supply.
One final thought: are you sure the primary windings are connected properly, and perhaps you can try switching phase (neutral to the black dot of the primary windings, or live to the black dot of the primary windings) sometimes transformer leakage is worse from one side of the coil than the other.
Last edited:
I don't know if my approach would help you but I built aa AD797 phono-stage and in trying to get the last little bit of hiss out of it I slid all my twisted pairs, power in, input and output into some copper braid and grounded each end of the individual pieces to the case. I also slid the majority of the wires in Richard's designed power supply and grounded them the same. The umbilicals were made with 18 gauge wire twisted and each was also placed in the copper braid. The umbilicals were grounded at the PS and the phono-stage. This approach turned out way quieter than without the shielding. You have to have the volume up to unlistenable ear-splitting levels to hear anything and then it is a small amount of tube rush sound that you can't hear from the listening position. I need to separate my mains earth from the other grounds in the power supply and I'll get that done on Saturday. I don't have pics of it all but here are a couple of the phono-stage. Not the PhonoClone obviously. I am using shielded cables from the cartridge to the phono-stage (Litz) and unshielded VH-Audio DIY fine silver interconnects from phono stage to pre-amp.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Yes, of course. And that was my point. The case of the amp had 35v on it (this was not due to drift because of leakage or less than perfect transformer center taps and the like - it was due to one power rail connected to the case inadvertently, as I explained). My point was simply that in my experience 35v shows something other than drift, and in my case that 35v traveled through the ground wire I attached and into the line stage. There WAS indeed serious current, again as I explained in my earlier post, and that was what melted the wire insulation.
rjm, I'm sure you have much more experience than I and I see your point that drift, whatever it might be, would easily snap back to 0v, so I defer to you ultimately. But it's hard for me to imagine a transformer either so imbalanced or with so much leakage current that a case wired to the center tap could drift that high. I mean, that voltage is as high or HIGHER than the supply voltage. But again, I have limited experience so I'll shut up and listen for a while.
@Carlp
Indeed. Something is seriously unhealthy.
I have an earthed power supply, unshielded umbilical to the PC4 with the COM connected to the chassis. Circuit common is not earthed until I connect it to the 'scope which, since this is at work not at home, is earthed.
The floating chassis to earth potential is 0.00 V. That's a Plitron sheild/screen toroid for you (very low leakage). There is however 1.6 V AC of noise developed. This signal is 60 Hz, with sharp charging spikes on both phases, unlike the output noise which is 120 Hz. This AC signal of course drops to zero as soon as the case is earthed, through OUT- or elsewhere.
If the case is not earthed but tied to earth with a 2.2 k resistor, the voltage is 125 mV, so about 50 microA of current draining out of circuit COM to earth. I've never deep-dived into these kinds of measurements before but that strikes me as quite a lot. Certainly you would not want that kind of noise current flowing out through the input section of a MC phono stage.
Indeed. Something is seriously unhealthy.
I have an earthed power supply, unshielded umbilical to the PC4 with the COM connected to the chassis. Circuit common is not earthed until I connect it to the 'scope which, since this is at work not at home, is earthed.
The floating chassis to earth potential is 0.00 V. That's a Plitron sheild/screen toroid for you (very low leakage). There is however 1.6 V AC of noise developed. This signal is 60 Hz, with sharp charging spikes on both phases, unlike the output noise which is 120 Hz. This AC signal of course drops to zero as soon as the case is earthed, through OUT- or elsewhere.
If the case is not earthed but tied to earth with a 2.2 k resistor, the voltage is 125 mV, so about 50 microA of current draining out of circuit COM to earth. I've never deep-dived into these kinds of measurements before but that strikes me as quite a lot. Certainly you would not want that kind of noise current flowing out through the input section of a MC phono stage.
This is going to be a longish post.
Ok, so first point... and this will mostly concern the Phonoclone 4 with certain links back to Erik's VSPSX as best as I can draw them... insofar as the noise in my case seems to be pickup at the input wiring section there can be a "two to tango" clause invoked and we can tackle the problem either at the source of the interference or at the receiving end.
So I'm on board with Wntrmute2 and others who have expressed their opinion that shielding all the relevant internal wiring is the most straightforward solution and the one most likely to pay out.
It's also pretty clear to me that whatever is going on in Erik's VSPSX can only be tangentially related to my own problems even if the symptoms and perhaps even the source is the same. I can prove that because with the first stage disabled a phonoclone is a VSPS, basically, and in that configuration I have no hum. My issues involve noise voltages at least an order of magnitude smaller than he is facing.
That out of the way, the fact remains I might have screwed up in the design of the VSPSX and PC4 boards, or the S-Reg circuit, because no one was having these kinds of problems with the earlier X-Reg based boards.
I've looked very carefully at the layout and conclude that there is nothing in the new boards that could be conceived as a problem. The connectivity is nearly identical, and have checked a dozen different variants in modding the PC4 including going back to what is 99% of the PC3 in terms of layout and nothing helped. So I don't think it's that.
I'm left then with the S-Reg itself. I've no doubt that it is working as designed, as simulated, providing 50 microV p-p of ripple on the V+ and V-which, I'm confident, will be mopped up completely by the op amp PSRR.
50 microV is still an order of magnitude larger than the X-Reg however, and it's enough that even a small fraction (~1%) of it coupling into the input section could generate the observed 5 mV of output noise.
So weakness 1 of the S-Reg is the design-point output ripple is higher than the X-Reg and this output ripply might couple from V+, V- into IN+ and IN- to produce hum.
The second concern about the S-Reg is a bit esoteric and concerns how shunt regulators work. In a nutshell as best I can: regulators perform two duties, holding the output voltage constant for a change in input voltage, and a change in output current. Series regulators are nice for controlling change in input voltage, but give a "double punch" (magnifying effect) adjusting against output current, while shunt regulators are the exact opposite: the double punch happens on changes to the input voltage (V++ ripple). This is in addition to the fact that the current draw and therefore the charging current are larger than for series regulation as current is routed to both the circuit and the shunt elements. So the end product here is there is a much higher magnitude of current pulses mirroring the diode conduction in the S-Reg than the X-Reg (or Z-Reg) and so more rectifier-like noise being generated which in turn demands a more careful approach to shielding the inputs from the voltage regulation section.
I can think of two ways to tackle this, one is obviously to shield all the input wires. The other is to add some sort or pre-regulator before the shunt to reduce the ripple on V++ feeding into the board. Simplest is to add extra filter capacitance in the power supply. A CRC filter somewhere perhaps in the main chassis just before the phonoclone boards or perhaps in the power supply. Or, finally, use a +15/-15V or 16 or 18 V regulated power supply of your own to connect to the S-Reg.
Ok, so first point... and this will mostly concern the Phonoclone 4 with certain links back to Erik's VSPSX as best as I can draw them... insofar as the noise in my case seems to be pickup at the input wiring section there can be a "two to tango" clause invoked and we can tackle the problem either at the source of the interference or at the receiving end.
So I'm on board with Wntrmute2 and others who have expressed their opinion that shielding all the relevant internal wiring is the most straightforward solution and the one most likely to pay out.
It's also pretty clear to me that whatever is going on in Erik's VSPSX can only be tangentially related to my own problems even if the symptoms and perhaps even the source is the same. I can prove that because with the first stage disabled a phonoclone is a VSPS, basically, and in that configuration I have no hum. My issues involve noise voltages at least an order of magnitude smaller than he is facing.
That out of the way, the fact remains I might have screwed up in the design of the VSPSX and PC4 boards, or the S-Reg circuit, because no one was having these kinds of problems with the earlier X-Reg based boards.
I've looked very carefully at the layout and conclude that there is nothing in the new boards that could be conceived as a problem. The connectivity is nearly identical, and have checked a dozen different variants in modding the PC4 including going back to what is 99% of the PC3 in terms of layout and nothing helped. So I don't think it's that.
I'm left then with the S-Reg itself. I've no doubt that it is working as designed, as simulated, providing 50 microV p-p of ripple on the V+ and V-which, I'm confident, will be mopped up completely by the op amp PSRR.
50 microV is still an order of magnitude larger than the X-Reg however, and it's enough that even a small fraction (~1%) of it coupling into the input section could generate the observed 5 mV of output noise.
So weakness 1 of the S-Reg is the design-point output ripple is higher than the X-Reg and this output ripply might couple from V+, V- into IN+ and IN- to produce hum.
The second concern about the S-Reg is a bit esoteric and concerns how shunt regulators work. In a nutshell as best I can: regulators perform two duties, holding the output voltage constant for a change in input voltage, and a change in output current. Series regulators are nice for controlling change in input voltage, but give a "double punch" (magnifying effect) adjusting against output current, while shunt regulators are the exact opposite: the double punch happens on changes to the input voltage (V++ ripple). This is in addition to the fact that the current draw and therefore the charging current are larger than for series regulation as current is routed to both the circuit and the shunt elements. So the end product here is there is a much higher magnitude of current pulses mirroring the diode conduction in the S-Reg than the X-Reg (or Z-Reg) and so more rectifier-like noise being generated which in turn demands a more careful approach to shielding the inputs from the voltage regulation section.
I can think of two ways to tackle this, one is obviously to shield all the input wires. The other is to add some sort or pre-regulator before the shunt to reduce the ripple on V++ feeding into the board. Simplest is to add extra filter capacitance in the power supply. A CRC filter somewhere perhaps in the main chassis just before the phonoclone boards or perhaps in the power supply. Or, finally, use a +15/-15V or 16 or 18 V regulated power supply of your own to connect to the S-Reg.
Last edited:
For me at least it is apparently that simple: I replaced the unshielded twisted pair wires connecting the input RCA to the boards with shielded twisted pair and the hum is reduced to the level of having the dummy load connected at the board input.
I conclude then that the S-Reg is fundamentally noisier than the X-Reg in terms of radiated noise, and one has to be proportionately that much more careful with shielding.
There still is a little hum left, audible with the volume turned above 12 o'clock. On the bench at least the amount of output noise seemed to be very sensitive to the placement of the input cables relative to each other and the amount of stress on them, I did not find any obvious correlation with anything specific.
Since it's become an issue, I'm finding myself hampered by the small case I used. Doing it again I'd use Hammond 1590F instead of 1590D so has to keep cables well separated. I'd also shield the output cables, and just be as fastidious as I could with braiding the power wires and so forth.
I conclude then that the S-Reg is fundamentally noisier than the X-Reg in terms of radiated noise, and one has to be proportionately that much more careful with shielding.
There still is a little hum left, audible with the volume turned above 12 o'clock. On the bench at least the amount of output noise seemed to be very sensitive to the placement of the input cables relative to each other and the amount of stress on them, I did not find any obvious correlation with anything specific.
Since it's become an issue, I'm finding myself hampered by the small case I used. Doing it again I'd use Hammond 1590F instead of 1590D so has to keep cables well separated. I'd also shield the output cables, and just be as fastidious as I could with braiding the power wires and so forth.
Braiding increases the gaps between the wires.
One gets smaller gaps by twisting the wires together.
Low loop area equals lower interference, both as a transmitter and as a receiver.
Twist your wires for lower interference.
Surround your twisted wires in a screen connected to chassis at both ends for even lower interference. And use screened star quad for lowest interference.
The downside to star quad is higher capacitance, so this can demand a better source and/or shorter cabling.
One gets smaller gaps by twisting the wires together.
Low loop area equals lower interference, both as a transmitter and as a receiver.
Twist your wires for lower interference.
Surround your twisted wires in a screen connected to chassis at both ends for even lower interference. And use screened star quad for lowest interference.
The downside to star quad is higher capacitance, so this can demand a better source and/or shorter cabling.
braid is when the three or more wires are folded over each other. That is the form that increases the air gap between the wires.
Twisting can be applied to 2 or more wires. One wants the closest distance between all the wires. Three can be touching all the way along their length.
If you had two separate secondary windings giving separate +ve and -ve pairs then I would twist each secondary winding + & -pair for minimum air gap and then series connect at the receiver/PSU output.
Twisting can be applied to 2 or more wires. One wants the closest distance between all the wires. Three can be touching all the way along their length.
If you had two separate secondary windings giving separate +ve and -ve pairs then I would twist each secondary winding + & -pair for minimum air gap and then series connect at the receiver/PSU output.
Last edited:
There are two COM pads on the board to accept four wire (2 pair) connection from the power supply, but my old chassis are all three wire with COM from each bridge rectifier tied at the power supply. So it is natively three wire until I retrofit.
However, as an approach there is a sense of diminishing returns creeping in and I think that it would be a good idea to start tackling the source of the noise not with shielding but by preventing it getting to the S-Reg in the first place. Which is to say, implementing some sort of series pre-regulator. The noise being picked up is a copy of the charging current through C4,5 as mirrored by the rest of the S-Reg shunt section, so if there was no ripple going into the S-Reg there would be no charging current noise. (The more cynical among you might well ask what the point of having the S-Reg there is at all then, the answer is for output current regulation.)
A simple 12V Zener and dropping resistor would suffice, but a Z-Reg would also be perfect for the job. So the strategy is to lower the S-Reg output voltage to get enough headroom that it will work from the 11.5 V output of the Z-Reg and see how that works out.
However, as an approach there is a sense of diminishing returns creeping in and I think that it would be a good idea to start tackling the source of the noise not with shielding but by preventing it getting to the S-Reg in the first place. Which is to say, implementing some sort of series pre-regulator. The noise being picked up is a copy of the charging current through C4,5 as mirrored by the rest of the S-Reg shunt section, so if there was no ripple going into the S-Reg there would be no charging current noise. (The more cynical among you might well ask what the point of having the S-Reg there is at all then, the answer is for output current regulation.)
A simple 12V Zener and dropping resistor would suffice, but a Z-Reg would also be perfect for the job. So the strategy is to lower the S-Reg output voltage to get enough headroom that it will work from the 11.5 V output of the Z-Reg and see how that works out.