Hello fellow DIYers !
I'm in the very last stages of my MM/MC preamp clone and I would like to validate the earting/grounding scheme before the PCBs get sent to the fab house. My design handbook states that the central point of a star-grounding arrangement for the whole system should be right at the most sensitive input stage of the amplifying chain to reduce the length of its negative terminal to ground, including the mains ground. That means the phono stage input in my case.
This build is kind of a "scrapbox challenge", re-purposing a TV set-top box steel chassis and a power transformer with EI-lams. A challenge as far as avoiding induced noise from the environement and mains ground but I know it can be done as I have a Rotel MM/MC preamp here that is built just like that and is dead silent. I use the same precautions Rotel did : shielding the power transformer and having the power supply as far as possible to the sensitive circuitry. I won't even consider a separate power supply chassis for this project.
See diagram below. The main elements in the chassis are positionned at their correct place and about to scale.
Key points:
- The two PCBs at top right of the chassis drawing are identical amplifier circuits for each channel; It is a discrete BJT fully symmetrical bipolar design which helps in reducing common mode noise sensitivity.
- Since mains power is fed by transformer and as the output is capacitor-coupled, there are four "grounds" in the system: Mains earth, power supply ground, input ground and output ground.
- Input and output grounds are NOT joined on the PCB but rather their RCA sockets are tied together using a piece of bare copper wire. This is duplicated for both channels.
- Likewise, power is fed to the PCBs using two distinct sockets and their common ground is not tied on the board but routed to the power supply star point (screw terminal at bottom right). This power supply ground star point is then tied to both input/output RCA sockets pairs, from which point a wire goes to the main ground post.
- All AC power wiring is twisted. I may also twist the signal input/output wiring (not shown on diagram).
- The power transformer is rotated at 45° to the chassis corners for maybe a bit of magnetic field cancellation by reflection... Not sure if that's such a big thing but it is recommended in my handbook and made the layout cleaner anyways.
- The power transformer secondary has no center tap so the power supply ground reference point is set with two 1% metal film resistors (middle left).
- Two strips of copper-clad phenolic board act as a non-magnetic shield to the AC power and regulator sections. Each strip is grounded separately directly to the chassis' integral ground screw.
- Power supply and preamplifier PCBs sit on 10mm nylon standoffs with nylon screws to reduce possibility of noise coupling. If push comes to shove more copper-clad board pieces can be added below and over the boards to completely surround them with non-magnetic shielding.
- There is a manual link between the chassis ground screw and the preamp ground post. The preferred mode of operation would be to have the chassis ground separated from the preamp star ground, with the power amplifier ground screw tied to the preamp as the whole system link to mains earth. However if the preamp is used standalone and connected to a "floating" input such as a battery powered laptop for example, the ground link is shunted and the preamp mains becomes the path to mains earth.
- The signal path is sketched on the rightmost board. The parts layout is quite packed and derived from the original amplifier, which was on single-side PCB and used a lot of 0-ohm resistor links across the board. I get away with those by using a 4-layer stackup, taking care to cross superimposed traces at 90° and avoiding close paralleled traces.
Sorry for the very long-winded post, this is my first phono preamp build and I wanted to be as precise as possible. I think I have covered all the bases, does that makes any sense to you? Are there obvious No-No-s?
As always, thanks in advance for any insights!
- Joris
I'm in the very last stages of my MM/MC preamp clone and I would like to validate the earting/grounding scheme before the PCBs get sent to the fab house. My design handbook states that the central point of a star-grounding arrangement for the whole system should be right at the most sensitive input stage of the amplifying chain to reduce the length of its negative terminal to ground, including the mains ground. That means the phono stage input in my case.
This build is kind of a "scrapbox challenge", re-purposing a TV set-top box steel chassis and a power transformer with EI-lams. A challenge as far as avoiding induced noise from the environement and mains ground but I know it can be done as I have a Rotel MM/MC preamp here that is built just like that and is dead silent. I use the same precautions Rotel did : shielding the power transformer and having the power supply as far as possible to the sensitive circuitry. I won't even consider a separate power supply chassis for this project.
See diagram below. The main elements in the chassis are positionned at their correct place and about to scale.
Key points:
- The two PCBs at top right of the chassis drawing are identical amplifier circuits for each channel; It is a discrete BJT fully symmetrical bipolar design which helps in reducing common mode noise sensitivity.
- Since mains power is fed by transformer and as the output is capacitor-coupled, there are four "grounds" in the system: Mains earth, power supply ground, input ground and output ground.
- Input and output grounds are NOT joined on the PCB but rather their RCA sockets are tied together using a piece of bare copper wire. This is duplicated for both channels.
- Likewise, power is fed to the PCBs using two distinct sockets and their common ground is not tied on the board but routed to the power supply star point (screw terminal at bottom right). This power supply ground star point is then tied to both input/output RCA sockets pairs, from which point a wire goes to the main ground post.
- All AC power wiring is twisted. I may also twist the signal input/output wiring (not shown on diagram).
- The power transformer is rotated at 45° to the chassis corners for maybe a bit of magnetic field cancellation by reflection... Not sure if that's such a big thing but it is recommended in my handbook and made the layout cleaner anyways.
- The power transformer secondary has no center tap so the power supply ground reference point is set with two 1% metal film resistors (middle left).
- Two strips of copper-clad phenolic board act as a non-magnetic shield to the AC power and regulator sections. Each strip is grounded separately directly to the chassis' integral ground screw.
- Power supply and preamplifier PCBs sit on 10mm nylon standoffs with nylon screws to reduce possibility of noise coupling. If push comes to shove more copper-clad board pieces can be added below and over the boards to completely surround them with non-magnetic shielding.
- There is a manual link between the chassis ground screw and the preamp ground post. The preferred mode of operation would be to have the chassis ground separated from the preamp star ground, with the power amplifier ground screw tied to the preamp as the whole system link to mains earth. However if the preamp is used standalone and connected to a "floating" input such as a battery powered laptop for example, the ground link is shunted and the preamp mains becomes the path to mains earth.
- The signal path is sketched on the rightmost board. The parts layout is quite packed and derived from the original amplifier, which was on single-side PCB and used a lot of 0-ohm resistor links across the board. I get away with those by using a 4-layer stackup, taking care to cross superimposed traces at 90° and avoiding close paralleled traces.
Sorry for the very long-winded post, this is my first phono preamp build and I wanted to be as precise as possible. I think I have covered all the bases, does that makes any sense to you? Are there obvious No-No-s?
As always, thanks in advance for any insights!
- Joris
Lack of ceneter tap could be a problem. The result is equivalent as if there is a center tap but connected via resistor half the value of used divider resistors. Perhaps it's better to put this divider after the rectifying bridge.
Transformers with central tap are not too expensive, is it worth the unsecurity of the final result?
Transformers with central tap are not too expensive, is it worth the unsecurity of the final result?
On a second thought, keeping divider before the bridge would reduce riple on filter caps but asimetry during charging might be more pronounced. So pick your poison 🙂.
Thanks chip_mk for your reply.
Perhaps. However the regulators can probably handle that?
You want to make the wires between xformer, rectifier and first filter caps as short as possible. I would move your entire regulator board to be next to xformer or put the rectifiers and caps with the transformer. Connect mains earth to chassis and then connect circuit earth to chassis by the input sockets - a wire to mains earth may have more resistance and dertainly more inductance than the chassis.
Yes they will - to some point. Asimetry is in the + and - voltage preceding regulators. If + rail sources 10mA more than - rail sinks, and you use 2k divider resistors, than it would cause 10V decrease on + rail before regulator and 10V increase on - rail. It may become insufficient for proper operation of + regulator.
The power transformer is rated 1A and the circuit draws 45mA per rail, so I can probably spare some current on the divider resistors. 2x 470R would give 32mA current through the center tap divider, would that "absorb" variations in circuit current draw simmetry? Remember the design is fully differential, so current draw will largely be equal between rails.
Agreed that would be best, since there is about 1W lost in the two 470R center tap divider resistors, if need be I'll buy one but for the sake of experiment let's say I keep the current one. I believe this is an interesting exercise.
Agreed that would be best, since there is about 1W lost in the two 470R center tap divider resistors, if need be I'll buy one but for the sake of experiment let's say I keep the current one. I believe this is an interesting exercise.
It might work but since I can't recall seeing other such solutions in a wild, it's yet to be proven as success.
Half wave rectifier also comes to mind as alternative that makes up for missing central tap.
Half wave rectifier also comes to mind as alternative that makes up for missing central tap.
Speaking of grounding, I see 4 gnd connections on each board. But it's not clear if they're interconnected on the board, and how are they used.
IMO the resistor bridge is interesting but should be fine.
Earth - I think you simply need to follow the schematic, better that star grounding IMO.
Star ground is a reference, but the output can wiggle it, and it's used for the input! Horror!
So design the circuit in a line, solder the input ground to the input side, and the output ground to the output side, and tie that to the chassis ground, perhaps with a parallel of:
1. X2 cap 100nF etc (AC ground)
2. 120 Ohm resistor (DC ground)
3. AC pins of a bridge rectifier (Safety override ground for signal ground).
Remember you are not amplifying the input signal, you are amplifying the difference between the input signat and ground - i.e. what the transistor/FET/Tube can 'see'.
That's my 2c anyway 🙂
Earth - I think you simply need to follow the schematic, better that star grounding IMO.
Star ground is a reference, but the output can wiggle it, and it's used for the input! Horror!
So design the circuit in a line, solder the input ground to the input side, and the output ground to the output side, and tie that to the chassis ground, perhaps with a parallel of:
1. X2 cap 100nF etc (AC ground)
2. 120 Ohm resistor (DC ground)
3. AC pins of a bridge rectifier (Safety override ground for signal ground).
Remember you are not amplifying the input signal, you are amplifying the difference between the input signat and ground - i.e. what the transistor/FET/Tube can 'see'.
That's my 2c anyway 🙂
Let's consider current path of the output signal. Starts from the nearest decoupling cap on the board, through the output stage (transistor), down the core of the output interconnect cable (capacitive), through input impedance of the line stage (preamp), back to the board (through screen of the cable). Instead going straight to the destination and that is the mentioned capacitor, the current should first go to the star grounding point, then to the terminal block, then to the board, and finally to through PCB trace back to the sourcing capacitor. This last part of the current path is considerable loop with L and can induce noise from stray magnetic flux.
I would keep input and output RCA isolated from chassis. They should meet at the point where FB resistor is grounded. Power GND from decoupling capacitors meets signal GND in the same point. Note resistors in + and - lines coming from PS board. This ensures transient currents are completely sourced from the local decoupling capacitors. Like this:
Chassis grounding can remain as is. But there are other options to experiment with if needed.
SomeJoe, BTW I like your avatar, reminds me of 50's, when DIY revolution started 🙂
I would keep input and output RCA isolated from chassis. They should meet at the point where FB resistor is grounded. Power GND from decoupling capacitors meets signal GND in the same point. Note resistors in + and - lines coming from PS board. This ensures transient currents are completely sourced from the local decoupling capacitors. Like this:
Chassis grounding can remain as is. But there are other options to experiment with if needed.
SomeJoe, BTW I like your avatar, reminds me of 50's, when DIY revolution started 🙂
my suggestion is build it without the mains psu, and run it with batteries, if its ok then add the mains psu
This image is indeed inspired from 1940-50's clip art and represents J. R. "Bob" Dobbs, the "messiah" of the Church of the SubGenius, a parody religion created in the 'wild 70s 😎. I took it directly from Wikipedia, if you want to know more and laugh a bit : Wikipedia article
I'm using it everywhere, so if you see an account on the interwebs with this avatar it's probably me!
Regarding the power supply decoupling RC network, I don't have one per se but instead each of the two amplifying stages (for each channel) has a Darlington current buffer on each rail as so (T63/T64/R147/C55 at right):
In the original amplifier this was connected to a bipolar regulator output rails, l reproduced that configuration. Should be enough no?
Es-tu Québécois?
Let me summarize. You're design strictly follows star ground topology. That is textbook approach and is valid.
My concern is when the star becomes very large introduced inductance 'decouples' different parts from the central point at HF. Tighter the star, higher the frequency at which this becomes significant.
One way to improve it is to split the star to two smaller stars on each board, which I usually do in my projects. The basic idea is explained in the diagram in my previous post.
If you want to implement it you will have to make some changes in the PCB routing and probably add some parts (bigger local decoupling elcos, serial resistors... etc).
Still, original proposal is not faulty and might work.
Let me summarize. You're design strictly follows star ground topology. That is textbook approach and is valid.
My concern is when the star becomes very large introduced inductance 'decouples' different parts from the central point at HF. Tighter the star, higher the frequency at which this becomes significant.
One way to improve it is to split the star to two smaller stars on each board, which I usually do in my projects. The basic idea is explained in the diagram in my previous post.
If you want to implement it you will have to make some changes in the PCB routing and probably add some parts (bigger local decoupling elcos, serial resistors... etc).
Still, original proposal is not faulty and might work.
En effet, qu'est-ce qui m'a trahi ? lol
I guess a small board with the serial resistors and decoupling caps can be easily added at the power supply local star ground like in your drawing, thanks for that suggestion. Then join with the rest of the signal grounds at the RCA input sockets.