HiFi Haven Group Project - Building The "World's Best Tube Phono Preamp"

Still working on the ground buss. First pic is a rough mock-up of a couple of grounding schemes. You can se that unlike our fearless leader's build, my power supply caps are at the opposite end of the chassis from the input jacks.
1) Should I ground the buss close to the input jacks (black text) or near the four big caps in the power supply compartment (white text)?
2) Should I leave the LED's grounded midway along the ground buss as in the second picture or split them off and bring them back to the D3A cap as shown in the white dashed lines of the first pic?
3) In general, should the noisier components be grounded nearest the chassis ground point of the buss or further away?
4) Other ideas welcome as always?

IMG_20191119_193247 copy 2.jpg


6thAttempy.jpg
 
Yes, the earth connection for the signal ground should be between the input jacks.

You could remove your center divider and drill it out for the caps, then have the cap terminals available for convenient connection inside the audio portion of the chassis.

You definitely do not want to run those long grounds back to the power supply caps, that's what the ground buss is there for.
 
If I may offer a method that seems to work for me over the years is to think about the earth /ground/0v as 3 separate areas.

First you have the audio reference point for the audio which is usually a point between stage 1 and 2 of a channel, closer to stage 1. I usually use a small square of pcb stuck to the chassis or on a small insulated standoff. All the components for that audio channel go to this and that includes the stage decoupling caps which should be nearby too. So this is an area that you can solder a number of components and wires to that is compact hence a square or rectangle of pcb close to the chassis for screening purposes. Forget about audio in and out for now. Pt 2 in a mo.
 
The idea of the above is to keep the audio signal unambiguous in terms of its reference i.e. there's not a mix of audio and other PSU signals mixed in wiring going here and there.

Now there's the power distribution points which are the final capacitors in the power supply, after the umbilical entering the audio chassis. In a one chassis build, just imagine there's an umbilical. In mine, this is just one cap with a common 0V tag but more commonly, this is two capacitors so stick them nearby to the audio and keep the connection between them close and join the 0V connections together. Form this point feed each audio ground and the HT feeds to the stages come from the positive tag.

Going back further is the power supply wiring itself. Make the first cap in the power supply the main reference point for 0V and take everything in the power supply back to it. Then take a wire from that point to the power distribution 0V. This is like the 0V wire in the umbilical.

Finally there is the ground reference point. I don't use a metal chassis but do use PCB to mount the audio circuit on so same thing. The ground post for the input can be a good point to ground the chassis so take the earth ground wire to this point and also a wire from the power distribution capacitors 0V point. If using floating PCB in a chassis, then a wire from them to the ground point is also required.

I treat the in and outs as extensions of the interconnects so don't think of them in terms of any other signal or grounding. Wire from the in to the input grid resistor and audio ground which I hope are close together and the output from the audio ground and output coupling cap to the output socket. Given the size of caps these days, those points might be far apart but at that signal level now, it won't make much difference. The input is the one we want to keep close and tight. I twist these wires.

I wire the heaters in a similar manner but as there are more wires coming off one power distribution capacitor, I fashion a multi tag strip out of copper sheet and solder that to the capacitor tag thereby giving it multiple tags. Each valve heater is taken from there via a twisted wire. I like Paul's suggestion of ceramic cap decoupling right at each valve socket. As that's a dirty connection at the junction of the two caps, take it directly to the chassis at that point; say a tag under the nut of the valve socket. I wouldn't daisy chain them one to the other as these are high gm valves and that could act like an aerial. Keep all the wiring unambiguous in its function.

I credit the above ideas to John Camille in an article he wrote some place. It made a lot of sense even though I think he was applying it to a power amp.

I realise there are bus bar schemes as well which I confess not to understand. Hope these ideas don't throw a spanner in the works.
 
Still working on the ground buss. First pic is a rough mock-up of a couple of grounding schemes. You can se that unlike our fearless leader's build, my power supply caps are at the opposite end of the chassis from the input jacks.
1) Should I ground the buss close to the input jacks (black text) or near the four big caps in the power supply compartment (white text)?
2) Should I leave the LED's grounded midway along the ground buss as in the second picture or split them off and bring them back to the D3A cap as shown in the white dashed lines of the first pic?
3) In general, should the noisier components be grounded nearest the chassis ground point of the buss or further away?
4) Other ideas welcome as always?

View attachment 17979


View attachment 17980

With your layout, I would run the ground buss from the PS caps over to the 5687 tubes, then up between the tubes to the middle of the D3a tubes. All of the ground connections would tie into the buss. As for the input jack connections, I (like Izzy) view that wiring as an extension of the interconnects. I use some short wires, twisted together, running from the jacks to the audio circuitry. The ground wire separates and is tied to the buss; the signal wire goes on to the gridstopper R on the D3a.

The main difference between your description and my build is that I placed the final PS caps close to the tubes they are feeding so the ground buss connections are short and direct. This seemed like a good idea but I don’t know how important it may be.

I do feel strongly that signal and ground connections should be made using component leads wherever possible. In other words, avoid jumper wires. Tag strips are ok since the electrical connections can still be made with component leads soldered directly to each other.

I posted a photo of the inside of my audio chassis in the other D3a phono thread, but I am posting it again here to illustrate the component lead connection approach. It looks messy, especially in a 2-dimensional photo, but the connections are secure and reliable. Even though it has never left my listening room, I wouldn’t have any concerns if I had to ship it cross country.

4046A102-569F-4B4C-B114-0B759FAFA95A.jpeg
 
Last edited:
I would run the ground buss from the PS caps over to the 5687 tubes, then up between the tubes to the middle of the D3a tubes. All of the ground connections would tie into the buss. As for the input jack connections, I also view that wiring as an extension of the interconnects. I use some short wires, twisted together, running from the jacks to the audio circuitry. The ground wire separates and is tied to the buss; the signal wire goes on the gridstopper R on the D3a.

The main difference between your description and my build is that I placed the final PS caps close to the tubes they are feeding so the ground buss connections are short and direct. This seemed like a good idea but I don’t know how important it may be.

I do feel strongly that signal and ground connections should be made using component leads wherever possible. In other words, avoid jumper wires. Tag strips are ok since the electrical connections can still be made with component leads soldered directly to each other.

I posted a photo of the inside of my audio chassis in the other D3a phono thread, but I am posting it again here to illustrate the component lead connection approach. It looks messy, especially in a 2-dimensional photo, but the connections are secure and reliable. Even though it has never left my listening room, I wouldn’t have any concerns if I had to ship it cross country.

Ah the problems of words vs pictures ;)

I do have the final PSU caps local to the stage. I call them the stage decoupling caps. I cannot recommend this highly enough as did Paul also and is like Salectric's layout. It keeps the AC/audio loop as close and tight to that stage as possible.

Where we differ is I wire each channel to its own audio reference point; a star ground if you like. This keeps each channel's reference very tight and local. I understand this isn't possible for you Salectric as you have a common tag on those Wkz caps and you wire it with a common bus between those final caps. I then only common them up at the power distribution caps. This all assumes I can read your picture accurately.

Reference the layout in post #522. What concerns me is the distance between what should be tight common reference points. The cathode LEDs and the grid resistor should go to the exact same point. Sure there can be a small distance between stage 1 and 2 but not too far. Close to stage 2 common point, should be the RIAA EQ connection to ground as it's the reference for stage 2's input. So to tighten this up, wire the cathode of 5687 to the other end of that LED string and put the grid resistor above as we look at the pic and attach it to the same point as the 0V connection of the LED string. Salectric's layout is like this.

Some of this might sound fussy but it's worth it. I've built this thing many times and more than once with the exact same parts but a better layout in line with the suggestions here and it does sound better so worth the pain and effort now.
 
Yes, the earth connection for the signal ground should be between the input jacks.

You could remove your center divider and drill it out for the caps, then have the cap terminals available for convenient connection inside the audio portion of the chassis.

You definitely do not want to run those long grounds back to the power supply caps, that's what the ground buss is there for.

Thanks Paul. I've already drilled the top plate for the caps so I'll probably leave them in the position shown. I'll tie the buss earth to the bolt for the TT ground post as they'll both be right there at the same point.
I'm thinking of tying that LED for the D3A to the buss down by the other 4 LEDs so those cathode LEDs are the last (or first) thing on the buss even though it is a bit away from the D3A cap where Sal tied his LEDs.
 
I would run the ground buss from the PS caps over to the 5687 tubes, then up between the tubes to the middle of the D3a tubes. All of the ground connections would tie into the buss. As for the input jack connections, I also view that wiring as an extension of the interconnects. I use some short wires, twisted together, running from the jacks to the audio circuitry. The ground wire separates and is tied to the buss; the signal wire goes on the gridstopper R on the D3a.

The main difference between your description and my build is that I placed the final PS caps close to the tubes they are feeding so the ground buss connections are short and direct. This seemed like a good idea but I don’t know how important it may be.

I do feel strongly that signal and ground connections should be made using component leads wherever possible. In other words, avoid jumper wires. Tag strips are ok since the electrical connections can still be made with component leads soldered directly to each other.

I posted a photo of the inside of my audio chassis in the other D3a phono thread, but I am posting it again here to illustrate the component lead connection approach. It looks messy, especially in a 2-dimensional photo, but the connections are secure and reliable. Even though it has never left my listening room, I wouldn’t have any concerns if I had to ship it cross country.

I understand about the component leads and am working hard to accommodate that concept. I'll probably need to extend the output cap leads but am hoping that is it. I find your photo helpful and how it sounds is the criteria for success, not how it shows up in a photo. I'm much more handsome in person than pictures of me for example!
 
@izzy wizzy . Thanks so much for that clear and complete view of grounding!! I may be too far down my path to revise the whole grounding plan but will incorporate what i can.

One question HaHaHa. I don't understand the statement "Reference the layout in post #522. What concerns me is the distance between what should be tight common reference points. The cathode LEDs and the grid resistor should go to the exact same point. Sure there can be a small distance between stage 1 and 2 but not too far. Close to stage 2 common point, should be the RIAA EQ connection to ground as it's the reference for stage 2's input. So to tighten this up, wire the cathode of 5687 to the other end of that LED string and put the grid resistor above as we look at the pic and attach it to the same point as the 0V connection of the LED string."

Is not the grid resistor the Rinken 300Ω? I thought it runs between the input and pin 2. I don't see where it goes to ground and how to tie it closer to the LED string(s).

great to have you here.
 
@izzy wizzy . Thanks so much for that clear and complete view of grounding!! I may be too far down my path to revise the whole grounding plan but will incorporate what i can.

One question HaHaHa. I don't understand the statement "Reference the layout in post #522. What concerns me is the distance between what should be tight common reference points. The cathode LEDs and the grid resistor should go to the exact same point. Sure there can be a small distance between stage 1 and 2 but not too far. Close to stage 2 common point, should be the RIAA EQ connection to ground as it's the reference for stage 2's input. So to tighten this up, wire the cathode of 5687 to the other end of that LED string and put the grid resistor above as we look at the pic and attach it to the same point as the 0V connection of the LED string."

Is not the grid resistor the Rinken 300Ω? I thought it runs between the input and pin 2. I don't see where it goes to ground and how to tie it closer to the LED string(s).

great to have you here.

Sorry, I was a bit vague. In the grid circuit, there's the grid stopper that goes to the actual grid of the valve and the grid resistor that the input is developed across which goes from the grid stopper to 0V. That resistor and the cathode LEDs should go to the exact same point. For the first stage there's a small gap and in the 2nd stage there's a big gap hence my suggestion of rewiring the LED string and putting the grid resistor "above" the valve. The local decoupling capacitor 0V should go to that point too IMO. And I would keep the distance between those stage 1 and 2 points very close together on the bus with the EQ connection in between.

I'm not so sure you can't revise things as you haven't wired anything or screwed anything down yet except the caps if I understand you right. I realise no one wants their new build to look like a swiss cheese but those big caps are part of the audio circuit so should be close to the audio circuit especially as they seem to have a common lug and the long wire to them is shared between channels if I think I've got the picture right so that wire will contain the audio and power supply noise for both channels all mixed up together. Salectric gets away with this due to his layout positioning and very short bus between them, Honestly, I can't advise how to wire it with them way over there. Maybe someone else can.
 
Well, it will be easy to attach components to the buss in closer fashion as you suggest.
I wonder if I split the buss lengthwise so-to-speak and use one set of cap grounds for one channel and one for the other? They would both tie together at the input jacks. I'll have to sketch it out.
 
let's assume I can tighten up the troublesome ground points regarding the RIAA, grid stoppers and LED chains, would either of these schemes be preferable to the single buss scheme I started with?
The first one with the dual ground buss bars running fore and aft might, in my little mind, reduce the amount of noise each buss is carrying.
The second one accomplishes the same while keeping similar tubes are tied together.

IMG_20191120_191109 copy.jpg


IMG_20191120_191919.jpg
 
You're flirting with at least one loop in the bottom photo. I would just use a single buss. I do really like that you are using tinned ground buss. Whenever I see untinned copper in a build it hurts my eyes.
 
Maybe pictures might help in solving this step by step. Here's the circuit as I understand it.

IMG_20191121_082915379_compress38.jpg

Let's go with an idea of separate buses for now. Very short ones that only run between the two valves at most. Even shorter if you can. A short diversion here for a moment. Wire isn't always just wire but has an impedance. Sure not very much but we're dealing with very small signals, large voltages and very sensitive valve in the D3a plus single ended circuits which have very little ability to reject noise so this kind of thing makes a difference especially as 0v is an audio reference and power supply point. Pt2 in a mo.
 
The preference would have been for C1 and C2 to be local but they're not so let's see what can be done.

C1 not only decouples power supply noise to 0V but also joins the top of R1 to 0V so it is part of the anode circuit ; the signal is developed across R1 and the D3a in parallel which is why it would be best local but let's deal with that in a mo.

C2 isn't so much of an issue as the CCS isolates the 5687 as it is a very high impedance. The signal is therefore only developed, mostly, across the 5687 locally with very little audio in the CCS.

So as C1 is going to be remote and C2 isn't doing much in terms of audio, C1 and C2 could be in the same package with the common wire taken to where C1 connects to the bus.

The input socket gets wired as a twisted pair to the D3a grid resistor; no grounding of input socket. The output gets wired as a twisted pair to the socket with the final resistor mounted across it.

However, I now wonder why have you got 8 caps in there in 4 packages? I must have missed something so when I know that, we can sort out an arrangement of the power distribution.

The above is they way I would do it. There are other build methods using long bus bars, tinned of course ;) with clean end and dirty end signals and so on. It's just that no one has been able to explain it in a way I understand how it works. There always seems to be unanswered questions in my head but then maybe that's my limitation.
 
I agree that C2 isn't critical. The CCS impedance is high enough that the AC signal current loop is primarily from the plate through the coupling cap, grid leak resistor, then back up through the biasing LED.
 
Back
Top