Noise Issue on Note Decay

[Phil_S]

I didn't see the IEC. I agree the short mains ground is OK. Look at the pic. Do all of the PT wires come through this one hole? I really don't like the primary and secondary bundled like that. But, heck, I'm not saying I know better than Hiwatt, lol.

[FourT6and2]

I didn't see the IEC. I agree the short mains ground is OK. Look at the pic. Do all of the PT wires come through this one hole? I really don't like the primary and secondary bundled like that. But, heck, I'm not saying I know better than Hiwatt, lol.

No, that's just the secondaries to the output jacks/impedance selector. Like most Hiwatt DR504/103 amps, the primaries exit through a hole over by the narrow turret strip, on the other side of the OT housing.

I'll upload more photos in the next post.

[FourT6and2]

IMG_6955.jpeg

IMG_6967.jpeg

IMG_6956.jpeg

IMG_6957.jpeg

IMG_6958.jpeg

IMG_6959.jpeg

IMG_6960.jpeg

IMG_6961.jpeg

IMG_6962.jpeg

IMG_6963.jpeg

[FourT6and2]

IMG_6964.jpeg

IMG_6965.jpeg

IMG_6966.jpeg

IMG_6968.jpeg

IMG_6969.jpeg

IMG_6970.jpeg

IMG_6971.jpeg

IMG_6972.jpeg

IMG_6973.jpeg

IMG_6974.jpeg

[FourT6and2]

IMG_6975.jpeg

IMG_6976.jpeg

IMG_6977.jpeg

IMG_6978.jpeg

IMG_6979.jpeg

[FourT6and2]

Used a tone generator and confirmed the underlying hum is indeed 60hZ. Not a harmonic, like 120hZ, but 60hZ. Volume of hum doesn't change with settings on amp. Preamp tubes are all DC, but power tubes are still AC. Heaters are yellow wires. The CT is green and looks to be grounded at one of the power tube cathode grounding points (along with one of the 1ohm bias resistors). I could probably move this over to one of the filter cap grounds instead, but would need to splice/lengthen it.

[FourT6and2]

You can also see that the PT's HV CT (black) is fused and then grounded right at the chassis under the fuses. Maybe this should go to the negative terminal of that first filter cap as well...

[Stevem]

The HV center tap should go right to the ground of the first node filter, then a wire from there should bring that ground to the chassis.

[R.G.]

You can also see that the PT's HV CT (black) is fused and then grounded right at the chassis under the fuses. Maybe this should go to the negative terminal of that first filter cap as well...

This alone will cause 60hz hum. There is no maybe about it - the CT must go to the negative terminal of the first filter cap.
Stevem is absolutely right.

For low noise and hum, the AC mains safety ground should go to its own chassis grounding bolt/stud; this bolt should not be used for anything else.
For low RF reception, the chassis nas to be connected to the signal ground, and for lowest noise, this will be at exactly one point. It can be at an input jack, or at the first filter cap negative. If it's at the FFC, the jacks really ought to be isolated and the signal ground wire run to the local ground of the first tube they serve. An RF bypass cap from the jack "ground" to chassis bypasses RF from the cable.

I noticed that there is a DC-DC converter used in the power supply. That may be OK, but this adds the possibility that the converter is a rich source of both radiated and conducted EMI inside the chassis itself.

My knee-jerk reaction to the schematic was that the maker didn't pay a lot of attention to good guitar amp wiring practice and design. The bias pot, for instance, doesn't have a safety resistor from the wiper to protect the output tubes from losing bias if the wiper goes open. There are others.

[martin manning]

There is no maybe about it - the CT must go to the negative terminal of the first filter cap.

To be clear, I believe it is fine for the PT CT and the negative terminal of the reservoir/first filter to go to the same point on the chassis. You just don't want the high return current pulses traveling across any length of chassis.

[FourT6and2]

You can also see that the PT's HV CT (black) is fused and then grounded right at the chassis under the fuses. Maybe this should go to the negative terminal of that first filter cap as well...

This alone will cause 60hz hum. There is no maybe about it - the CT must go to the negative terminal of the first filter cap.
Stevem is absolutely right.

That should be easy enough to try. But that first cap isn't grounded to chassis directly. It feeds down the string of filter cap negative connections, eventually going to ground on the other side of the chassis, where the NFB is grounded.

For low noise and hum, the AC mains safety ground should go to its own chassis grounding bolt/stud; this bolt should not be used for anything else.

Ok, that's already the case.

For low RF reception, the chassis has to be connected to the signal ground, and for lowest noise, this will be at exactly one point. It can be at an input jack, or at the first filter cap negative. If it's at the FFC, the jacks really ought to be isolated and the signal ground wire run to the local ground of the first tube they serve. An RF bypass cap from the jack "ground" to chassis bypasses RF from the cable.

Ok, that's already the case. At least I believe it is. I will lift that star ground and check for continuity to chassis with it in the air, to double check that there is no other ground hiding somewhere.

I noticed that there is a DC-DC converter used in the power supply. That may be OK, but this adds the possibility that the converter is a rich source of both radiated and conducted EMI inside the chassis itself.

Correct. But that's not something I can change.

My knee-jerk reaction to the schematic was that the maker didn't pay a lot of attention to good guitar amp wiring practice and design. The bias pot, for instance, doesn't have a safety resistor from the wiper to protect the output tubes from losing bias if the wiper goes open. There are others.

Yeah I agree. I think they were trying to keep the original Hiwatt DR504 layout. What other things caught your eye?

[R.G.]

There is no maybe about it - the CT must go to the negative terminal of the first filter cap.

To be clear, I believe it is fine for the PT CT and the negative terminal of the reservoir/first filter to go to the same point on the chassis. You just don't want the high return current pulses traveling across any length of chassis.

I'd say it another way: as long as the PTCT is tied directly to the first filter cap negative, where the chassis connects is negotiable. In the case of twist-lock caps, this works out great.

The PTCT wire carries the big pulses of rectifier current, with peaks of up to several times the DC current going out of the FFC. To the extent that there is any resistance between the PTCT and the FFC negative, there will be a voltage across that resistance. Imagine removing the PTCT from the exact FFC negative pin and inserting a few inches of copper wire. If you connect the chassis to the junction of the FFC negative and the wire to the PTCT, the chassis will be at the voltage of the FFC negative. If you connect the chassis to the PTCT end of the inserted wire, the chassis and FFC now are at different voltages because of V=I*R. This makes no difference at all, unless somewhere else in the amp you use the chassis for a signal ground. An input jack, for instance.

The tube that the input jack feeds has its cathode and probably its grid referenced to the FFC by the tube's power wiring. The jack's shield contact rides the chassis voltage. The difference between the chassis and the FFC negative is now applied right to the grid-cathode voltage and gets amplified.

Depending on the exact size and routing of the conductors, the induced voltage may be big enough to be a problem, or may be below the noise floor. Many amps do get away with doing this. So - yeah it may be fine. Depends on the particulars of the amp, chassis, transformer, wire gauge, etc. I always mess up complicated things, so I look for ways to "just do it this way" and not have to hunt for OK-enough setups.

[FourT6and2]

There is no maybe about it - the CT must go to the negative terminal of the first filter cap.

To be clear, I believe it is fine for the PT CT and the negative terminal of the reservoir/first filter to go to the same point on the chassis. You just don't want the high return current pulses traveling across any length of chassis.

I'd say it another way: as long as the PTCT is tied directly to the first filter cap negative, where the chassis connects is negotiable. In the case of twist-lock caps, this works out great.

The PTCT wire carries the big pulses of rectifier current, with peaks of up to several times the DC current going out of the FFC. To the extent that there is any resistance between the PTCT and the FFC negative, there will be a voltage across that resistance. Imagine removing the PTCT from the exact FFC negative pin and inserting a few inches of copper wire. If you connect the chassis to the junction of the FFC negative and the wire to the PTCT, the chassis will be at the voltage of the FFC negative. If you connect the chassis to the PTCT end of the inserted wire, the chassis and FFC now are at different voltages because of V=I*R. This makes no difference at all, unless somewhere else in the amp you use the chassis for a signal ground. An input jack, for instance.

The tube that the input jack feeds has its cathode and probably its grid referenced to the FFC by the tube's power wiring. The jack's shield contact rides the chassis voltage. The difference between the chassis and the FFC negative is now applied right to the grid-cathode voltage and gets amplified.

Depending on the exact size and routing of the conductors, the induced voltage may be big enough to be a problem, or may be below the noise floor. Many amps do get away with doing this. So - yeah it may be fine. Depends on the particulars of the amp, chassis, transformer, wire gauge, etc. I always mess up complicated things, so I look for ways to "just do it this way" and not have to hunt for OK-enough setups.

So I'm clear, when you talk about the PT CT, you're talking about the HT secondary, not the heater CT? Or both?

How about this layout? HT CT, heater CT, and filter cap ground go to chassis star point. Not the exact same amp, but a DR504 clone
https://ceriatone.com/wp-content/upload ... ec2021.pdf

[R.G.]

So I'm clear, when you talk about the PT CT, you're talking about the HT secondary, not the heater CT? Or both?

Yes, just the HT secondary.

How about this layout? HT CT, heater CT, and filter cap ground go to chassis star point. Not the exact same amp, but a DR504 clone
https://ceriatone.com/wp-content/upload ... ec2021.pdf

The first issue is that it uses chassis grounding. This raises the issue that certain points on the chassis are quieter than others, depending on what currents flow between which points.
Chassis ground can be quiet enough if done well. It's just complicated to get it done well, because there are so many places where hum and noise can get in.

In this case, the DC power and chassis can be measured from that one chassis star point where they join with the HT CT. I'll call that the power star point.

The rectifier pulses flow through the wire to the capacitor stack. That means that the wire has pulsed voltages across it. In this amp, the signal circuits tie their signal and DC grounds to the chassis, so the capacitor-wire voltages get added to the series capacitor voltages and appear on B+. The wire effectively adds its resistance to the ESR of the capacitor stack. So far, so good; rectifier pulse noise appears on somewhere other than signal ground reference.

If the signal ground references are not to chassis, but moved to the actual filter cap negative, leaving power and chassis ground where they are, then the rectifier pulses appear as an offset of signal ground to power ground, and you get the rectifier pulses injected into grids and cathodes. What matters so far is which end of that wire the signal grounds are connected to. If there are no other voltages between wherever on the chassis the signal grounds connect and the power star point, it's quiet.

But there are other voltages across the chassis. The chassis carries power return for the power tube cathodes. The path between the power tube cathode connection point on the chassis and the power star point will have a voltage across it. The signal tubes have both their power and signal grounds connected to a chassis star point that's not the power star point, and is not the same point as the power tube cathode connection to chassis. The signal tube decoupling caps ("32 +16") are connected to the signal tube ground star point, so any currents they "bypass" flow through the chassis on the same path as the signal tube power ground and signal ground. These caps carry whatever fraction of the rectifier pulses that were on the B+, plus any local signal decoupling currents. The currents are smaller than the first filter cap rectifier pulses, but still there. Whether they get heard or not depends on the gain of the signal tube cascade. As a side note, the single ground path on the chassis returning to the power star point carries the sum of voltages generated by the signal tubes' operation, and amounts to a small amount of local feedback in just the signal tubes. A setup like this could oscillate from this feedback. It depends on the amount of gain in the signal tube path.

Low frequency currents follow the path of least resistance. So the power tube cathode currents flow as directly as they can back to the power star point. So do the signal tube return currents. If these paths happen to share some of the chassis, then the cathode currents wobble the signal tubes' ground around. This is another and bigger amplitude feedback path to the signal tubes, opening the gate for oscillation.

[FourT6and2]

So I'm clear, when you talk about the PT CT, you're talking about the HT secondary, not the heater CT? Or both?

Yes, just the HT secondary.

How about this layout? HT CT, heater CT, and filter cap ground go to chassis star point. Not the exact same amp, but a DR504 clone
https://ceriatone.com/wp-content/upload ... ec2021.pdf

The first issue is that it uses chassis grounding. This raises the issue that certain points on the chassis are quieter than others, depending on what currents flow between which points.
Chassis ground can be quiet enough if done well. It's just complicated to get it done well, because there are so many places where hum and noise can get in.

In this case, the DC power and chassis can be measured from that one chassis star point where they join with the HT CT. I'll call that the power star point.

The rectifier pulses flow through the wire to the capacitor stack. That means that the wire has pulsed voltages across it. In this amp, the signal circuits tie their signal and DC grounds to the chassis, so the capacitor-wire voltages get added to the series capacitor voltages and appear on B+. The wire effectively adds its resistance to the ESR of the capacitor stack. So far, so good; rectifier pulse noise appears on somewhere other than signal ground reference.

If the signal ground references are not to chassis, but moved to the actual filter cap negative, leaving power and chassis ground where they are, then the rectifier pulses appear as an offset of signal ground to power ground, and you get the rectifier pulses injected into grids and cathodes. What matters so far is which end of that wire the signal grounds are connected to. If there are no other voltages between wherever on the chassis the signal grounds connect and the power star point, it's quiet.

But there are other voltages across the chassis. The chassis carries power return for the power tube cathodes. The path between the power tube cathode connection point on the chassis and the power star point will have a voltage across it. The signal tubes have both their power and signal grounds connected to a chassis star point that's not the power star point, and is not the same point as the power tube cathode connection to chassis. The signal tube decoupling caps ("32 +16") are connected to the signal tube ground star point, so any currents they "bypass" flow through the chassis on the same path as the signal tube power ground and signal ground. These caps carry whatever fraction of the rectifier pulses that were on the B+, plus any local signal decoupling currents. The currents are smaller than the first filter cap rectifier pulses, but still there. Whether they get heard or not depends on the gain of the signal tube cascade. As a side note, the single ground path on the chassis returning to the power star point carries the sum of voltages generated by the signal tubes' operation, and amounts to a small amount of local feedback in just the signal tubes. A setup like this could oscillate from this feedback. It depends on the amount of gain in the signal tube path.

Low frequency currents follow the path of least resistance. So the power tube cathode currents flow as directly as they can back to the power star point. So do the signal tube return currents. If these paths happen to share some of the chassis, then the cathode currents wobble the signal tubes' ground around. This is another and bigger amplitude feedback path to the signal tubes, opening the gate for oscillation.

You lost me a bit haha. I'm not an engineer.

Here's the game plan:

1. Move PT Heater CT to filter cap neg.
2. Move PT HV CT to filter cap neg.
3. Single ground wire from filter cap neg. to chassis ground
4. Power tube cathodes will stay where they are (grounded to their individual mounting lugs like in 99.99% of all other amps).
5. Preamp star ground will remain where it is by V1/input/gain control.

There is still another chassis star ground point over by the NFB/bias circuit. The filter caps are currently connected to that point, along with the NFB low-pass, and bias circuit. I will need to lift the filter caps from this point so they can ground to number 1/2/3 above.

Hopefully this reduces the noise floor some. It's not perfect, but I can't really rewire this entire amp without completely removing all the boards and desoldering everything.

Oh... also, the SMPS for the DC heaters has its own chassis ground over by the power/standby switches. What about that?
Same thing for the OT. It's grounded to chassis at the output jacks.