Hello all,
I'm planning out the umbilical cable to connect my 2-box 26 preamp chassis to its power supply chassis and I'd like to separate the grounds/returns as follows (I have 8 pins on the connectors I'm using):
HT
HT return
Left DC filament
Left filament return
Right DC filament
Right filament return
Signal reference ground
Safety ground
My question is...what should I call "signal reference ground" in the preamp box versus HT ground? I'm fairly certain about the final filter/decoupling cap (HT ground), gas regulator tubes (HT ground) input jacks (signal ground), grid leak (signal ground). How should I classify the 26 cathode grounds (signal or HT?), volume control ground (volume control is on the output of the preamp after the output transformers), output transformer shield, and output jack grounds (is signal signal, or is output signal different from input signal)?
I've been playing the various options over and over in my head and frankly it's starting to spin. Any help is greatly appreciated!
I'm planning out the umbilical cable to connect my 2-box 26 preamp chassis to its power supply chassis and I'd like to separate the grounds/returns as follows (I have 8 pins on the connectors I'm using):
HT
HT return
Left DC filament
Left filament return
Right DC filament
Right filament return
Signal reference ground
Safety ground
My question is...what should I call "signal reference ground" in the preamp box versus HT ground? I'm fairly certain about the final filter/decoupling cap (HT ground), gas regulator tubes (HT ground) input jacks (signal ground), grid leak (signal ground). How should I classify the 26 cathode grounds (signal or HT?), volume control ground (volume control is on the output of the preamp after the output transformers), output transformer shield, and output jack grounds (is signal signal, or is output signal different from input signal)?
I've been playing the various options over and over in my head and frankly it's starting to spin. Any help is greatly appreciated!
Last edited:
Think about where the currents go. For directly heated valves the filament circuit is unavoidably part of the signal circuit. For all valves HT ground is usually part of signal ground, although a separation of these could be possible. I would regard the cables in the umbilical as being merely extensions of the PSU, so things need to be grounded in the preamp and not in the PSU (apart from safety ground).
The idea is to make the single ground to chassis for everything in the PSU, at the star ground point where cap grounds and transformer CT meet.
I agree about the intermingling of the filament and signal in a DHT, another complication/puzzlement on the path to "perfect grounding".
I usually just lump the HT and signal together into one return, but a day of reading about grounding on the net and I come up with lots of creative (if not necessarily feasible) ideas; separating the noisier HT return, containing the ripple filtered out by the gas tubes, from the signal seems like a logical way to quiet things down.
In a completely opposite thought, I was also considering making the HT return the safety earth connection between the chassis as well. After all, the safety earth is only there in case of fault and doesn't do anything for the circuit until then, why not have it do double duty as the HT return and save a wire in the umbilical? My Supratek preamp does it this way. In case of a fault the fault current will still be sent directly to the chassis at the CT ground point of the PSU, so no harm there, right?
I agree about the intermingling of the filament and signal in a DHT, another complication/puzzlement on the path to "perfect grounding".
I usually just lump the HT and signal together into one return, but a day of reading about grounding on the net and I come up with lots of creative (if not necessarily feasible) ideas; separating the noisier HT return, containing the ripple filtered out by the gas tubes, from the signal seems like a logical way to quiet things down.
In a completely opposite thought, I was also considering making the HT return the safety earth connection between the chassis as well. After all, the safety earth is only there in case of fault and doesn't do anything for the circuit until then, why not have it do double duty as the HT return and save a wire in the umbilical? My Supratek preamp does it this way. In case of a fault the fault current will still be sent directly to the chassis at the CT ground point of the PSU, so no harm there, right?
Don't confuse the signal ground with the connection from signal ground to safety ground. If you use a star point then this should be in the preamp not the PSU. The transformer CT should not go to the star point even if they are in the same box. CT to cap -ve, then cap -ve to star point, is the right way to do it. If there are two caps (reservoir C1 and smoother C2) then -ves go CT-C1-C2-star. The CT-C1 connection must be in the PSU, with as small a loop as possible between this and the +ve connections from the rectifier. C2 could be in the preamp - up to you.
Keep signal connections out of the PSU. Don't send signals down the umbilical - it should just be an extension of the PSU.
Keep signal connections out of the PSU. Don't send signals down the umbilical - it should just be an extension of the PSU.
So DF, are you saying that the CT should NOT still be grounded to the PSU chassis? I've seen PSUs where the CT ground and all the individual PSU cap grounds come together at a star point in the PSU connected to the PSU chassis, signal and HT grounds come together at a star point in the preamp connected to the preamp chassis, and the two are connected by the umbilical.
Last edited:
The CT should be kept away from any star point, even in the PSU, as it carries charging pulses. As I said: CT-C1-C2-star is the way to do it. People often send charging pulses through their star point then wonder why they have buzz. You should have a connection to safety ground in the PSU but this should come from the 'clean' end of the PSU not the dirty end.
Having an umbilical makes things more complicated, as you have discovered.
Having an umbilical makes things more complicated, as you have discovered.
Well what do you know! I just rewired the power supply for my existing preamp so that the CT is no longer connected to chassis. Instead it is connected to the first cap (-) in a CLCLC, the first cap is connected to the second cap (-) and the second cap (-) is grounded to the chassis.
The preamp was quiet before, but I could still hear a faint buzz/hum with my ear to the speaker - NOT ANYMORE! It's dead silent! Maybe it's my imagination but the music sounds better, too!
Thanks DF96!
P.S: The 26 preamp I'm building will have an LCLC PS, I suppose CT will still go to the (-) of the first cap even though it's choke input, correct?
The preamp was quiet before, but I could still hear a faint buzz/hum with my ear to the speaker - NOT ANYMORE! It's dead silent! Maybe it's my imagination but the music sounds better, too!
Thanks DF96!
P.S: The 26 preamp I'm building will have an LCLC PS, I suppose CT will still go to the (-) of the first cap even though it's choke input, correct?
Your conclusion is correct but the inductance is worse than the resistance, because it creates voltage differences that are proportional to the time-rate-of-change of the current, rather than merely the amplitude of the current. So even extremely-small-amplitude currents can create large voltage differences, if they are quickly changing.
Conductors have both resistance and self-inductance and both the inductance and resistance can be reduced in the same way, i.e. by using more parallel conductors in place of a single one.
For the case at hand, maybe the most-important thing will be to make sure that the input-signal ground-reference point, in each preamp, is not connected to anything that goes back to the power supply, if that connection would have ANY other type of current flowing in it. So it seems like, for example, you would also NOT want to connect all of the star grounds from all of the preamps to some single wire that goes back to the power supply. That might make all of the star grounds' voltages bounce, which would make all of the input signal voltages bounce.
Also, shouldn't you ideally have a separate umbilical for each preamp? Mixing all of the power rails and ground returns just seems like it's asking for trouble. Every unit's currents will affect the voltages at every other unit.
What size are the wires in this umbilical?
This "standard practice" is the main reason why people believe that rectifiers and power cables always "sound". When it is connected properly (like DF96 already described) neither rectifiers, nor power cables "sound".
The "star point" must be as close as possible to the output of the filter, i.e. to the last capacitor.
OK, maybe there's a little confusion here. There are two separate preamps, each with its own dedicated power supply. One is my existing Supratek Chenin preamp which I have upgraded to a CLCLC PS from CLC, converted to use 2C22 tubes in place of 6SN7, converted to CCS plate load with diode bias, and various other tweaks. It came with a grounded CT so I left it that way and just changed it last night on DF96's recommendation. It also uses the HT return (which is the same place where the signal grounds terminate) as the safety ground link between boxes (no separate safety ground wire in the original umbilical, or in the new one I made for it).
The second preamp is a 26-based unit I am building. My original question was about the umbilical for it and the best way to separate the grounds. The wire I plan to use is 15.5 gauge copper litz for a few reasons: I have some, it's flexible, of sufficient gauge, teflon insulated so easier to solder without melting the insulation, and based on your comments above the multiple parallel conductors should be an advantage for reduced inductance and resistance. It's what I used to wire up the Supratek power supply and umbilical as well.
So Gootee, can you recommend a better way to separate the returns/grounds than this? Keep in mind I only have 8 pins on the connectors I'm using:
HT
HT Return
Left Filament
Left Filament power return from Coleman filament supply board
Right Filament
Right Filament power return from Coleman filament supply board
Remote control power
Remote control return
In my original plan, star point in the preamp chassis would have the HT and signal returns going to it (ground of gas regulator tubes, cathode ground of 26, input grounds, output grounds, grid leak, volume control ground, output transformer electrostatic shield). Filaments are also necessarily grounded via the cathode of the 26 since it's a DHT. Filaments use the Rod Coleman filament supply boards right at each 26.
I could remove the remote power and return from the umbilical (9VDC - could use a wall wart, for example) and run that separately into the preamp if having two more open pins would be an advantage.
I'm thinking that at least the signal grounds could be sent to a separate star point in the preamp chassis from the HT grounds, but that wouldn't really affect how the umbilical is wired since I don't want to send signal grounds down the umbilical to the PS.
I'm open for suggestions...
In most circuits HT ground and signal ground are the same thing, so you can't separate them. If you don't want a single star then you could use a bus ground.
In grounding arrangements most wires have two functions:
1. establish a reference potential
2. carry a current
These two functions are in direct conflict, unless you ensure that each wire does one and only one purpose. A little pure DC on a wire does little harm, but where is there any pure DC in an amplifier? You can't perfectly separate these functions but you can get sufficiently close.
In grounding arrangements most wires have two functions:
1. establish a reference potential
2. carry a current
These two functions are in direct conflict, unless you ensure that each wire does one and only one purpose. A little pure DC on a wire does little harm, but where is there any pure DC in an amplifier? You can't perfectly separate these functions but you can get sufficiently close.
I'm thinking a bus ground in the preamp box is the way to go. I have some nice 8 gauge copper wire I could use for that, and wire the signal grounds further "upstream" from the other HT grounds such as the gas regulators, basically in order of noisiness from quietest to noisiest to PS, and ground to the preamp chassis at the quiet end.
One last question, regarding shielding.
If I use a braided copper shield over the umbilical, should I ground it at both ends or just the PS end? I've seen arguments for both in my reading on the subject. If I ground it to both chassis it forms the most complete shield, and avoids becoming an antenna for RF, but might that create a ground loop problem if the HT ground is also grounded to both chassis? Can I avoid grounding HT to chassis in the PS and just rely on the shield to provide a chassis connection between PSU and preamp (where everything is grounded to chassis)?
One last question, regarding shielding.
If I use a braided copper shield over the umbilical, should I ground it at both ends or just the PS end? I've seen arguments for both in my reading on the subject. If I ground it to both chassis it forms the most complete shield, and avoids becoming an antenna for RF, but might that create a ground loop problem if the HT ground is also grounded to both chassis? Can I avoid grounding HT to chassis in the PS and just rely on the shield to provide a chassis connection between PSU and preamp (where everything is grounded to chassis)?
It can be of any length, if currents that flow through them do not cause hum or undesired feedback.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.