Convincing my boss that separate star grounding is trash versus GND plan.

[MBrahim]

He still believes in those 1980s kind of things and hard to convince him. and I'm getting tired, I even considered to resign.

[tggzzz]

Depends on the frequencies, currents/voltages, sensitivity and EMI/EMC considerations.

Without understanding those, it would be cargo-cult engineering.

[Roehrenonkel]

Hi,
 
existing design looks good for a tube-circuit.
It's high-impedance and sensible to ground-currents.
Don't resign, just make your own design and see if it's better.
Proof it to your boss (and us). ;->

Good luck

[Psi]

Compromise, keep the star ground but expand all traces so they take up the entire PCB like a ground plane.

[Berni]

Star grounds have their time and place too.

If you are doing sensitive microvolt level analog circuitry that shares a board with other higher power stuff, then a whole board ground flood fill is the absolutely worst thing you can do.

Filled internal ground planes is the best for high speed designs where giving the induced return currents the shortest path is the most important aspect.

The design shown looks more like a low frequency tube amplifier. There star grounds are sensible because with frequencies as low as for example audio, there electrons on PCBs pretty much follow DC circuit laws without any funky electromagnetic field effects or RF wizzardry.

As it often is in life the best is a compromise somewhere in between. This is called a sectioned ground fill with isolation cuts. You can flood the ground with copper to give you the nice shielding benefits, but then slice this fill into multiple smaller fills that connect together in strategic spots. This lets you control where the return currents flow so that you can still keep them out of the sensitive parts of the circuit.

[coppercone2]

I feel like you should try to learn from it, whenever I hear about some usual " 'time period' stuff" its a sign that someone is trying to be fashionable in engineering, its a term that might apply to teenagers dating. Unless its the auto router.

To me it seems like a bit of a joke, because you have tube sockets, then the tube, which has a significant wire length on it, and a contact resistance. The tube leads are like kovar wire (kinda like stainless, aka heater wire), its not even copper, so there is that too, so the ground connection is ALOT different, just from the part. I am not sure the ground plane is gonna change things too much in this situation.

With high voltage things, not having a ground plane is kind of nice too, because there is less area for high voltage to try to get out of. I.e. depending on how a design can fail, you could have something stupid, like a energized ground plane bringing HV to a wire that happens to be touching the side of the PCB away from any parts. In that case, the lack of conductor is a good thing. Think about what can happen if there is a broken wire and a short, then suddenly that 'poor ground' is actually increasing safety.

[mtwieg]

Star ground layout isn't "trash" in general IMO (so long as operating frequency is low). Main problem with it is that it doesn't scale well with circuit complexity (even in the simple layout you shared, there are some places which chain the ground). In a design limited to two copper layers, a star ground may work better than simply flooding empty space with GND.

For higher frequency circuits (>1MHz) I would never opt for a star ground, especially if a dedicated layer for GND is available.

[Sensorcat]

For higher frequency circuits (>1MHz) I would never opt for a star ground, especially if a dedicated layer for GND is available.

Since the OP talked about his boss, it is reasonable to assume that this is not a hobby project, but development of a commercial product. So they have to think about EMC/EMI, better sooner than later. And exactly this makes anything 'higher frequency circuits (>1MHz)', whether the band it operates with includes higher frequencies or not. This does not completely rule out star ground, but means that in any serious work, you must consider what happens with frequencies far outside your region of interest.

[CatalinaWOW]

Virtually any time you connect two or more pieces of gear you have connected star grounds.  You have dealt with them, and will be dealing with them whether you knew it or not.  So don't dismiss them as trash, figure out what problems you are trying to solve and come up with an appropriate topology.   

[coppercone2]

I am starting to become suspicious that star ground is what I should have done on this amplifier project I am working on, which seemed to work with similar parts in a 'bad' prototype that fails to work on a PCB. I think I need to get away from doing designs with methodologies to get rid of problems I did not know I had and stick with ones similar to what I saw worked. Amazing cheapskates volume #3 'drive your self crazy after you take random ass advice that some real EMC lab will laugh at'.

[floobydust]

OP, you don't know it all and quitting your job over this is a bit extreme. Fender used single-point grounding even in the 1950's it was known to do on point-point wiring.

A ground-pour is a poor choice with tube gear for at least two reasons:
One- people screw up the HV spacings and don't back off the pour enough. VOX put ground-pour around the HV transformer nodes, and they break down and carbon-track, burn up. Noob mistake.

Two - Tube gear is high impedance and the additional stray capacitance of the ground-pour affects HF response. Several pF on nodes with a few 100k or meg ohms rolls off the high-end or worse causes instability if this amplifier uses overall loop feedback.
If you are replicating vintage designs, then prove it - that you've met the old amp's performance. Check the frequency response you're new improved board accomplishes.

Star-grounding most important with the high current nodes and returns being grouped- like output tube cathodes, filter caps to the PSU. A preamp is mA only and it's not critical unless you have a lot of gain. Star-grounding falls apart at RF frequencies unless you look after that. Solid-state amps need the (high current) loudspeaker, PSU, filter caps at one node.

[uer166]

I have to say I'm with OP here: star grounding IS trash unless you have a very particular sub-optimal situation to deal with. It has nothing to do with tube circuits, rather space-constrained circuits where the fields happen to spread wide enough where you need to constrain it in presence of sensitive stuff like high-resolution ADCs.

It's rare enough where if your first instinct is to use a star ground, you've already fucked up substantially (either through cargo cult engineering or outdated concepts).

OP: I understand you completely, either deal with it as-is (provide the answer that is expected from you), or start sending resume to other places.

[Robert763]

As this appears to be a valve (tube) circuit star grounding is probably a good choice. Valves are inherently high impedance devices. Because of this stray capacitance matters. Ground planes are great in low impedance (logic) and constant impedance (RF) applications but not so much in high impedance circuits. As others have noted HV breakdown is an issue too as every pin or via has to passs through the ground plane. You need a new set of design rules for that. 
There is no "one size fits all" solution to gounding and signal returns.
There is nothing wrong with the current layout so just keep the boss happy.