Dual Pultec Build Ground Noise

[RoadrunnerOZ]

Can I send dc into the rectifiers on the boards for testing?

If you put the DC from your new rectifier board to the DC output points of the on-board rectifiers on the two boards there will be no conduction of any current backwards, so no issue there - unless there is any diode or diodes breaking down in the on-board rectifier(s). If you put the DC from your new rectifier board to the AC input points of the on-board rectifiers on the two boards, you will get DC passed through the diodes, at the correct polarity, with a voltage drop across each diode of less than 1V - also no issue unless there is a breaking down diode or diodes in any of those rectifiers. That’s why it’s best to just remove them if you’re unsure of their integrity. This all comes down to needing an oscilloscope.

 

[ccaudle]

we’ll just go ahead and remove the onboard rectifiers

That sounds like making random changes without any theory of operation to test. If you do that at the same time as moving the transformer mounting location that is two simultaneous changes without any way to distinguish which (if either) made a difference.

To track down problems like this you have to both be systematic in testing, and have a grasp of fundamentals of current flow to understand (at least somewhat) the effects you see to know what a reasonable next step would be.

Here is what I would do to try to track down this problem.
1. Remove the spaghetti mess of connections from the different PCBs to the chassis. There should be 1 connection from the power supply return to the chassis, ideally right at the filter cap(s). The PCB does not appear to have an easy connection point there, so don't sweat exactly where it connects.
Since there are separated power supplies for each buffer PCB that really means one connection from each SRPP board.
I cannot see the connections on the bottom side of the PCB, making only one connection to chassis may involve using insulated stand-offs if the circuit gnd node surrounds the mounting holes.
Connect the cable shields on input and output connectors directly to the chassis and only there.

2. Assuming that did not solve the noise problem, measure the noise voltage at the output (i.e. between pins 2 and 3 of an output connector), then disconnect the powered monitors and check the noise voltage again. That is making the assumption that the voltage is high enough and your voltmeter is sensitive enough to easily see if the noise is present just from the AC voltage measurement.
If removing the connection to the monitors gets rid of the noise then figure out where the rule about cable shields connect only to chassis was broken and fix that. If the problem is in the monitors (i.e. the monitors have a pin 1 problem) then most likely you will be able to hear the problem, but not measure it at the output pins. You might be able to bandage your monitors with a modified cable if that is the case.

3. Assuming disconnecting the monitors did not solve the noise problem, strip the problem back to the most basic configuration possible.
You cannot have an output without the SRPP boards, so disconnect everything except power to the SRPP boards and pins 2 and 3 on the output connectors to the SRPP boards.

The diagnosis splits at this point. If the noise it not present, then start connecting pieces back one at a time, one connection at a time. So connect a twisted pair from the output of one EQ board to the input of the buffer board and check the noise. Connect the second buffer board and check the noise, then connect a twisted pair from the secondary of the input transformer to the input of one EQ board and check the noise, etc.

If the noise is still present with the only connections being SRPP boards with power input, and the output connections to the output XLR connectors, you will really need to find a friend with an o'scope so you can see what is going on in the buffer circuits. That circuit is just about as simple as you can get for a tube buffer stage, just one step up from a cathode follower, with nothing connected on the input side there should only be a slight amount of white noise.

If there is still noise, power only board A (which you said was quiet before), then only board B (did you ever try only board B, or just A alone, then A+B together?), then A and B together again. Is the noise really the same from both boards, or just similar? Do the heater voltages change at either board between one powered or both powered? They are regulated, so should not change at all, should be 6.3V always.
How much does the HT voltage change at each board between each board powered singly then together? The HT voltages are not regulated, so I would expect the voltage to drop at least a little when both boards are powered compared to only one.
Is the voltage across the 1K cathode to plate resistor and the 470 ohm cathode resistor the same on both SRPP boards? They should be close if both tubes are operating properly.

Hopefully somewhere along that path a change makes enough improvement to give a clue what is going on.

 

[Janalex]

Ok thanks. Let’s just dissect point 1 above for clarity. The pic below shows 3 ground points on an srpp board. They all have continuity. Should I connect one ground wire to any one of those since they are the most available grounds on the board and then connect to my point on the chassis? Then connect the in/out shields to the same point on the chassis? Do the same for the other board and use the same chassis location? Also should the AC ground be connected to this chassis location? After that is clear I’ll ask what to do about grounding the filter board.

Yes each board was tested individually and working well. I’m just looking for any redundant connections that are causing issues when both connected.

 

[Janalex]

I understand that I am not to connect anything to xlr pin 1 except chassis however i can use one of the three ground points on the srpp board to connect to chassis, no? Or should I leave this completely empty and find another point on the srpp board to ground?

 

[RoadrunnerOZ]

I understand that I am not to connect anything to xlr pin 1 except chassis however i can use one of the three ground points on the srpp board to connect to chassis, no? Or should I leave this completely empty and find another point on the srpp board to ground?

SRPP board is ground connected to the filter board by the audio cables isn’t it?
Seeing as how the SRPP board is the active board I would only have the one ground from that to star ground and let it provide the ground for the filter board via the audio cabling. The ground for the SRPP should be the one closest to the power supply.

 

[ccaudle]

Should I connect one ground wire to any one of those since they are the most available grounds on the board and then connect to my point on the chassis?

That should work. The ideal location would be closer to the power supply section, but if you are strict about only using a single reference connection it won't matter very much.

Then connect the in/out shields to the same point on the chassis?

No, the shields should connect to the case as close as possible to where the connector contacts the case. Think of the idealized case being cable shields consisting of the twisted pair signal wires surroundec by solid metal tubing welded to the chassis where the cables come in, and try to emulate that as much as practical. Unfortunately standard XLR connectors don't provide a way to connect the shield all the way around and contact directly to the chassis, but that would be ideal from an RFI standpoint (like high frequency BNC or SMA connectors which bolt directly to the chassis). For power line frequencies a low resistance connection from pin 1 to the closest chassis point you can get to (or the shell tab on the connector) works fine.

Also should the AC ground be connected to this chassis location?

It does not have to be directly connected there, but usually they should be close together. Typically the AC protective earth connection connects close to the power entry, so that the resistance is as low as possible and so that there is not a cumbersome wire running across part of the chassis. To reiterate an earlier post: that protection scheme relies on the chassis being truly conductive, because the point of the protective earth conductor is that if a power wire breaks inside the chassis and comes into contact with the metalwork, the protective earth connection should be low enough resistance that it keeps the voltage from rising to a dangerous level, and conducts well enough that the current trips the breaker.
Usually the AC line input goes directly to the power supply, so it should not be much problem for the power supply ground to connect to the chassis right beside the safety/protective earth connection.

This is how AES standard 48 diagrams the recommended connections:



In this context the two wires going in on the left would be the signal hot and return wires from pins 2 and 3 of the input connector into the input transformer, the two wires going out to the right would be the hot and return signals from the output transformer to pins 2 and 3 of the output connector, and the "signal circuitry" box would encompass both the filter PCB and the SRPP buffer PCB.

I’ll ask what to do about grounding the filter board.

Return path should follow signal. It appears that on the SRPP board the return pin is right in between the out and in pins ( it appears that the naming is from the standpoint of the filter board, the in pin on the SRPP board is not for signal coming in to SRPP, but should be connected to the filter board input, in pin comes from input transformer and goes to the filter board, the connection to the out pin would come from the filter board to the input of the SRPP buffer, right?), so it should be easy to have three wires twisted together carrying signal and reference to and from the filter board, or two twisted pairs, with one pair going in/gnd to the filter board, and the other pair carrying out/gnd back to the SRPP board, with the gnd wires in each pair meeting up at that middle pin on the connector.

As an aside, I notice that you have used the term "continuity" several times. This is an important concept to grasp:

3 ground points on an srpp board. They all have continuity

If you look at the drawing of the layout, the connections to the pins labeled "GND" are maybe 0.01"/a few mm wide, and several inches long.
Let's say for argument sake that the width is 0.01", and 3" long. That is about 0.2 Ohms resistance at DC (so at 60Hz will be basically the same). If there is 1mA of current running through that trace the induced voltage (Ohms law, v=i*r) is only 0.2mV. That doesn't seem like much, and it isn't in an absolute sense, but that is -70dBu. So if you let even 1mA of AC power related current flow through just 3" of your circuit ground, you are already limited to a -70dBu noise floor. Those PCBs look longer than 3" to me, and every doubling or resistance or doubling of current is going to increase the voltage by 6dB. Hopefully you can see that just thinking about connections in terms of "continuity" and making connections from the circuit reference (aka "ground") just willy-nilly isn't really the precision needed when you want to get to -80dBu and lower noise floors.

That also circles around to something I suggested earlier, but I guess I should clarify: I suggested measuring the voltage between pins 2 and 3 of the output in lieu of just listening, but that assumes you have a voltmeter which can measure accurately down in the tenths of mV range. That may not be a good assumption, I forget that not everyone has been collecting test equipment for many years, so I just wanted to call that out.
The original suggestion about disconnecting the active monitors was just about being thorough in eliminating every point that power line related currents can come in, so using a battery powered active speaker or battery powered headphone amp would be another way of accomplishing the same thing, and would probably be more sensitive (i.e. using your ears to pick out low levels of hum) than relying on an inexpensive multimeter with indeterminate accuracy down in that low voltage range.

 

[Janalex]

Eureka! I would say right now it’s absolutely fixed meaning the ratty ground issue is gone just with the wiring above. The channel furthest from the transformer is silent and the channel closer to the transformer has ever so slight hum but this I’m now certain is the proximity probably to the ac rails crossing over more so than the transformer itself. Roadrunner stated I could test without removing the onboard rectifiers and just see if using my external rectifier board may make a slight difference. I mean it’s built already. Sounds like a plan?

 

[ccaudle]

the channel closer to the transformer has ever so slight hum

Sounds like magnetic field from the transformer affecting that channel.

I’m now certain is the proximity probably to the ac rails crossing over more so than the transformer itself

Why are you certain? There is an easy experiment to do to check: make a long twisted pair and instead of routing the power under or beside the closer channel, run the long pair straight up from the transformer a foot or two, then over and down to the second channel. If the noise goes away, then probably is due to the AC running underneath or beside. If the noise doesn't change, then put the original wiring back, unmount the transformer and move it outside of the chassis and check again.

see if using my external rectifier board may make a slight difference.

If you are careful and systematic about that it may be educational. There are several ways to create unintentional differences that aren't directly related to using a different rectifier, such as changing the routing of the wiring, or rotating the transformer to get easier access to the wires (magnetic leakage pattern is usually not symmetrical, so rotating the transformer can change hum pickup).
My educated guess is that either there is no difference, or the noise becomes more annoying because the wiring carrying power to the second (farther) channel now has a lot of high frequency content rather than just 60Hz (or 50Hz, I don't remember if you said you were in North America or not).

 

[RoadrunnerOZ]

Eureka! I would say right now it’s absolutely fixed meaning the ratty ground issue is gone just with the wiring above. The channel furthest from the transformer is silent and the channel closer to the transformer has ever so slight hum but this I’m now certain is the proximity probably to the ac rails crossing over more so than the transformer itself. Roadrunner stated I could test without removing the onboard rectifiers and just see if using my external rectifier board may make a slight difference. I mean it’s built already. Sounds like a plan?

I would remote the transformer first before anything, you may not need the new rectifier - if the hum in the first channel is due to proximity of the transformer you could always construct a metal shield around the transformer to minimise this.
Try first by moving the transformer outside the case then secondly bolting the transformer to the outside of the chassis (use the same mounting hole) with flying leads to the terminals inside - this will test to see if proximity is a problem, or you are getting induced noise from the transformer radiating into the chassis or both.
To alleviate the AC induced noise from your wiring to the second board you could always run your AC using Star Quad shielded cable - this may help reduce the amount of radiated energy when passing cables near the first circuit board.

 

[Janalex]

Can I alligator clip the transformer in from outside the chassis? It was difficult with its short leads to connect to both boards. It must be connected to both boards for a proper test.

 

[RoadrunnerOZ]

Sure - it’s just a test. Remember to try both transformer loose and also with it bolted to the outside of the box in the same place.

 

[Janalex]

What does the bolting change? Interestingly at this moment the transformer was inside chassis but not bolted

 

[RoadrunnerOZ]

What does the bolting change? Interestingly at this moment the transformer was inside chassis but not bolted

The transformer may induce ground currents in the casework. Hence testing with it hard up against the case and well away from the case to see if there’s a difference.
How is the transformer mounted inside the case?
Thiese tests are to eliminate all possible causes.

 

[Janalex]

It was bolted on with a single bolt. I have not bolted it back since the new wiring. I suppose now it’s possible some noise returns when bolted.

 

[RoadrunnerOZ]

It was bolted on with a single bolt. I have not bolted it back since the new wiring. I suppose now it’s possible some noise returns when bolted.

Some synthesisers and rack gear I have serviced have the transformer mounted on an independent isolated mounting frame with rubber insulating feet and with no ground connection. This is to prevent mechanical noise coupling plus ensuring no induction from the windings to the case. This is more common with the non toroidal transformers.

 

[Janalex]

Ok so did some testing. Transformer bolted vs not bolted to inside of chassis no change. Transformer laying horizontal a foot away from the chassis unchanged. Mini external bridge rectifier HT rail wired in no change. We are talking about every so slight noticeable hum that the unit is powered. It possible after I remove all these alligator clips and wire properly it might even get better. I don't see this as causing any issue in terms of tracking or mixing but if it does Ill revisit it. Want to experience it in a real world scenario. The control room is quite silent and the smallest things can be heard.

The last thing I will be testing is if any noise appears after I connect in the red twisted pair seen in my initial pics that goes to the unbalanced input on the srpp boards. So what was sold to us was 2 active bypass board. We extended a short jumper from the 6.3V heater on each tube to the hot position on the unbalanced input on each board. This was used to get 6.3vdc to power the relays and the LEDs so you can see two red wires connected there on each board and running around the chassis to the power switch LED, bypass PCBs, and bypass LEDs. It is low voltage and nicely twisted so hoping I can still use this. The alternative would be to wire in a switch to each EQ board where is shown in the schematic and dump the LEDs and bypass boards. We shall see.

I double checked and the XLR chassis shells are jumped to pin 1. I was bypassing these for testing.

 

[Janalex]

I connected the jumper between 6.3V pin on the tube and the unbalanced input hot. I have included a pic here. We broke the tracing as part of this mod as shown with the small red line. However I tried it both ways with and without jumper as quickly as I could and I'm pretty certain this jumper is adding noise. There is certainly continuity between many new points on the board with this jumper in place. Just connecting the (-) side of the DC to ground at the unbalanced input is likely producing some ground loop. Can anyone confirm just looking at the pic. If so, I'm gonna go ahead and purchase the Don Audio transformer with the extra 5v secondary unless there is another way to harvest 6v dc without introducing noise. Will have to rectify it to DC with a small board in the corner but should be noise free.

 

[RoadrunnerOZ]

No wonder you’re getting noise. Your Heater DC is running all round the chassis acting like an antenna - it would appear it’s got one twisted pair running parallel to the mains wiring to the power switch LED.
MThen also that the two SRPP board grounds and + would appear to join at the bypass switch via your two twisted red pairs meeting up there and joining. This would create an instant ground loop.
Is that correct? Why do you need two sets of DC wiring to come to the same place?
Because from what I see in your first pics visually tracing those twisted red wires that seems to be the case.
If so, you need to separate the whole LED and other items network from your audio boards and have a separate DC supply. You can always use the 12V AC from the transformer and use your new rectifier board and add a voltage regulator and filter capacitors. Just use a 6V fixed regulator (7806) with a 4700uF 25V cap + on the input pin 1 where your rectifier + connects, pin 2, cap - and regulator - create ground, pin 3 is out, connect across this a 10uF 16V cap +, cap - to ground

 

[Janalex]

Yes just to be clear the entire LED bypass network was disconnected during all testing. This was definitely a ground loop somewhere in the mess of wiring before. Now that it has largely been resolved I'm introducing the LEDs and bypass boards back before closing up. Thank you for confirming that getting the DC voltage from the tube pin is not a good idea. I didn't think to use the 12V secondary already there. Therefore the wiring would be as follows:?

12V secondaries to input on external rectifier board ---> separate wires coming out of that same input into Board A SRPP ( so a branch point at the input of the rectifier board). Then branch off again at the Board A 12V input to Board B...I cant get more than 3 wires in the same terminal block so would this work?

On the rectifier board I have a bridge rectifier feeding a voltage regulator a smoothing capacitor like in this pic? Or do I need more capacitance after the rectifier to get actual DC first?

 

[RoadrunnerOZ]

I just did an edit to my previous post if you refresh the page it should show - I have put bhigher cap values than your circuit pic shows - you could probably use as low as 1000uF 25V on the input to the 7806. The best idea is take the AC from the transformer onto a 4 way terminal block straight from the transformer, have jumpers at the back to create two pairs and then use that to feed your boards from one pair and your new rectifier from the other.