Good stuff here ...
Valve Wizard - Grounding
Something I had not considered about can caps ...
Older amplifiers often used multi-section capacitors commonly known as can caps. These usually contain two or three capacitors with a single, shared negative connection. These were convenient at the time since they saved space and presumably money. The obvious problem with them, however, is that by having only one negative connection they force us to adopt a ground star that may not be convenient. The best advice would be not to use can capacitors at all; like noninsulating jack sockets they are an anachronism. However, for readers who still insist on using them, a reasonable compromise would be to use a single, dedicated capacitor for the reservoir, and a can-cap for later smoothing stages. Nevertheless, the ground connections should always be kept as short as possible.
Valve Wizard - Grounding
Something I had not considered about can caps ...
Older amplifiers often used multi-section capacitors commonly known as can caps. These usually contain two or three capacitors with a single, shared negative connection. These were convenient at the time since they saved space and presumably money. The obvious problem with them, however, is that by having only one negative connection they force us to adopt a ground star that may not be convenient. The best advice would be not to use can capacitors at all; like noninsulating jack sockets they are an anachronism. However, for readers who still insist on using them, a reasonable compromise would be to use a single, dedicated capacitor for the reservoir, and a can-cap for later smoothing stages. Nevertheless, the ground connections should always be kept as short as possible.
Star ground is convenient ... is kinda contradicts itself in the end by saying the ground connections should be as short as possible
Common or "ground" connections should have both low inductance (for low impedance at high frequencies),
and low resistance (for low impedance at low frequencies). The same also applies to power connections.
and low resistance (for low impedance at low frequencies). The same also applies to power connections.
Yes, that's easy to read. I like it.
Since I restuffed a can cap with three individual capacitors that are not connected to the can itself, I have the option of not connecting the ground lead from it to the chassis as described here:
As you can see, the negative terminals of the three new caps are not connected to the can. Instead, they are connected to the unused fourth terminal on the bottom of the cap. It was unused because this was a quad cap and now it's only a triple. So I can do whatever is best with that negative connection. According to the highlighted excerpt above, it seems that I would not connect it to the chassis and would use it as the common ground connection point for everything except the "safety" ground wire from the power cord, which screws to the chassis near where it enters on the back.
Too late. I already went down this path with the capacitor, the individual output tube bias kit add-on, insulated input jacks, and the "virtual center tap" resistors, which were an afterthought mod added by DynaKit that are just bolted to the chassis for their ground. One of those has a jumper back to the common ground point. The other doesn't (diagram posted earlier). Really, this isn't an ST-35 any more. Even the output tranformers aren't the same since they have been changed to 4/8 ohm outputs and the negative feedback has been moved to the 8 ohm output. So, given all of the changes, I'm calling it ST-35 series ii. 😉
Last edited:
Generally, the worst ground loop hum causes are:
Either the input stage ground connections (RCA input jack; volume pot; grid resistor Rg; and the cathode circuit (cathode, or the bottom of the self bias resistor and bypass cap; which becomes more complex with more parts if you apply global negative feedback there).
Insulate the RCA input jack from the chassis, and then connect all the above mentioned parts returns to a single point; that makes a Local ground loop (as good as it gets). Then, connect from there to the Central ground point of the amplifier.
Or, the other hum ground loop, the B+ circuit. The center tap (or negative of a bridge rectifier), needs to be tied very directly to the first filter cap (and to the second filter cap if the first cap is simply a voltage adjustment cap in front of a choke input filter). That is a local hum ground loop.
Then, connect from there to the Central ground point of the amplifier.
If possible, a 3 wire power mains, 3 wire IEC power cord, with ground should be used. Ground from there should connect to the central ground point of the amplifier
The central ground point of the amplifier should connect to the chassis (safety, and reduction of hum).
(most signal sources only use 2-wire power cords, so there is no direct 3-wire to 2 wire ground loop caused by the missing ground wire;
But there still can be common mode currents, and differential currents that can cause ground loops (often, reversing the 2-wire plug at the power mains outlet can fix a ground loop, but modern 2-wire plugs have a spade lug that is wide, so can not be reversed in a modern power mains wall outlet.
Poor B+ filtering will cause hum.
Any cathode resistor that is un-bypassed by a capacitor (some circuits can not, or should not be, bypassed) can cause hum from the AC filament supply. The filament supply should either be grounded by the center tap; or a matched set of resistor "pseudo" center tap; or at an elevated + voltage as needed (and that elevated voltage needs a bypass cap to ground).
Hum should be "attacked" and minimized Before applying negative feedback. The better an amplifier is Before negative feedback, the better it should be if negative feedback is used (in terms of hum reduction; this is not a treatise on whether to use, or not use, negative feedback).
Negative feedback is a little like Penicillin, if you wait until after the amplifier is struggling to fix it by the use of negative feedback, it is like waiting 15 days of infection before you take your Penicillin.
Just a few hints, that work very well for me.
Most of my power amplifiers (mono-blocks) have less than 100uV of hum and noise (< 100uV; < 0.1mV).
Either the input stage ground connections (RCA input jack; volume pot; grid resistor Rg; and the cathode circuit (cathode, or the bottom of the self bias resistor and bypass cap; which becomes more complex with more parts if you apply global negative feedback there).
Insulate the RCA input jack from the chassis, and then connect all the above mentioned parts returns to a single point; that makes a Local ground loop (as good as it gets). Then, connect from there to the Central ground point of the amplifier.
Or, the other hum ground loop, the B+ circuit. The center tap (or negative of a bridge rectifier), needs to be tied very directly to the first filter cap (and to the second filter cap if the first cap is simply a voltage adjustment cap in front of a choke input filter). That is a local hum ground loop.
Then, connect from there to the Central ground point of the amplifier.
If possible, a 3 wire power mains, 3 wire IEC power cord, with ground should be used. Ground from there should connect to the central ground point of the amplifier
The central ground point of the amplifier should connect to the chassis (safety, and reduction of hum).
(most signal sources only use 2-wire power cords, so there is no direct 3-wire to 2 wire ground loop caused by the missing ground wire;
But there still can be common mode currents, and differential currents that can cause ground loops (often, reversing the 2-wire plug at the power mains outlet can fix a ground loop, but modern 2-wire plugs have a spade lug that is wide, so can not be reversed in a modern power mains wall outlet.
Poor B+ filtering will cause hum.
Any cathode resistor that is un-bypassed by a capacitor (some circuits can not, or should not be, bypassed) can cause hum from the AC filament supply. The filament supply should either be grounded by the center tap; or a matched set of resistor "pseudo" center tap; or at an elevated + voltage as needed (and that elevated voltage needs a bypass cap to ground).
Hum should be "attacked" and minimized Before applying negative feedback. The better an amplifier is Before negative feedback, the better it should be if negative feedback is used (in terms of hum reduction; this is not a treatise on whether to use, or not use, negative feedback).
Negative feedback is a little like Penicillin, if you wait until after the amplifier is struggling to fix it by the use of negative feedback, it is like waiting 15 days of infection before you take your Penicillin.
Just a few hints, that work very well for me.
Most of my power amplifiers (mono-blocks) have less than 100uV of hum and noise (< 100uV; < 0.1mV).
If I am reading those two items correctly, then it seems that:
- Every ground wire gets connected to a single point, which would be my power supply cap negative, not to the chassis.
- That single point then gets connected to the chassis with one wire to the point where the saftey ground from the power cord attaches to the chassis, as close as possible to the entry point.
Because there are so many ground connections to that 2-terminal strip, I would just solder a heavy piece of bare copper wire between the two terminals and use that as my point for soldering all 12 wires I suppose.
Yes, I drew all of that, and it's all to exact size in my graphics program. It's easier to draw and adjust the drawing rather than wire it wrong and adjust the wiring! 😀
Whenever this project is finished, I'll print out the diagrams for reference. The amplifier will last a very long time, and if anyone ever needs to troubleshoot it in the future (including me), they would be valuable to have.
Last edited:
If I am understanding your method, it sounds like you are using the famous Star Grounding technique.
But that means you are Not using Local ground loops. (Local input circuit ground loop; Local B+ ground loop for examples).
Of course, it is much easier to create local ground loops if you use point to point wiring, but often difficult if not impossible if you use PCBs.
It sounds like you are using 3-wire mains power and cord, and are connecting the mains ground to the chassis.
You also need to connect the Star Ground to the chassis.
Safety First!
I remember the brand new output transformer that had a primary short to the transformer laminations and end bells. Ouch!
Prevent the "Surviving Spouse Syndrome"!
The correct saying is "Keep the Local Ground Loops wiring as short as possible".
That is one reason why star grounding sometimes is not the optimum way to do things.
I will be real interested on how much hum you get at the 8 Ohm tap, with an 8 Ohm non-inductive load connected:
With no signal source connected
With a signal source connected, but no signal playing from it (a CD player at Stop, for example).
Please let us know when you get the amplifiers up and running.
But that means you are Not using Local ground loops. (Local input circuit ground loop; Local B+ ground loop for examples).
Of course, it is much easier to create local ground loops if you use point to point wiring, but often difficult if not impossible if you use PCBs.
It sounds like you are using 3-wire mains power and cord, and are connecting the mains ground to the chassis.
You also need to connect the Star Ground to the chassis.
Safety First!
I remember the brand new output transformer that had a primary short to the transformer laminations and end bells. Ouch!
Prevent the "Surviving Spouse Syndrome"!
The correct saying is "Keep the Local Ground Loops wiring as short as possible".
That is one reason why star grounding sometimes is not the optimum way to do things.
I will be real interested on how much hum you get at the 8 Ohm tap, with an 8 Ohm non-inductive load connected:
With no signal source connected
With a signal source connected, but no signal playing from it (a CD player at Stop, for example).
Please let us know when you get the amplifiers up and running.
Last edited:
I'm trying based on the references provided earlier. There are so many mods to this amplifier already, why not add one more easy one. Adding a 2-terminal strip and a few extra ground wires is no big deal. I have to ground the insulated input jacks anyway.
Yes. 120 VAC, 3-wire, United States, with ground wire straight to the chassis exactly at the entry point to ensure safety.
I don't see any way to shorten any ground wires without using the chassis for multiple ground points. The longest wires are the ones to the isolated input jacks and the two PCBs, which can't run very many places. The directions say to ground the jacks to the chassis at the center terminal of the front 5-terminal strip and the PCBs to the can of the quad cap, which is no shorter than my diagram. The ST-35 chassis is so small, most of the leads except those and the ones to the speaker binding posts are no more than about 3 inches at most. None of them are any longer than the directions for the kit call for. The added 2-terminal strip would be only 1 inch away from the usual ground point of the can of the multi-cap.
The chassis decals will arrive tomorrow. All white lettering on an all black chassis with all black components including the cap can and choke and all screw heads. Of course, the PCBs will still be brown. Can't change that.
Last edited:
Never Get Old,
I was re-editing my Post # 87, when you replied in your Post # 88.
You might want to go back and look at Post # 87, because I changed it.
Happy building, and Happy listening!
I was re-editing my Post # 87, when you replied in your Post # 88.
You might want to go back and look at Post # 87, because I changed it.
Happy building, and Happy listening!
The little Chinese EL84 amp kit that I built a few months ago is absolutely dead silent with my ear directly on 91dB efficient speakers, which surprised me. It takes a slightly different approach to "star" grounding, which required some rather long ground wires running from front to back. Every ground wire, including the ground from the AC power cord, goes to one single point on the chassis with a whole lot of round lugs on one bolt. I messed up one connection and had to remove that chassis bolt to correct it, and getting all of those lugs back through that one bolt was a real headache.
You probably don't want to run two "ground" lines (per channel) from the input jacks and the PCB separately to chassis, and instead run a short "ground" wire between them and a single wire to chassis; minimizes loop area.
I don't see the "ground" from cathode bias to chassis so it must be the obvious connections.
Best Practice calls for the PE/safety ground connection to have its own dedicated chassis connection - its own screw, lockwasher, nut - but I'm not certain whether or not this is NEC code. It's easy to imagine the value of this if the mechanical connection to chassis were to fail.
All good fortune,
Chris
I don't see the "ground" from cathode bias to chassis so it must be the obvious connections.
Best Practice calls for the PE/safety ground connection to have its own dedicated chassis connection - its own screw, lockwasher, nut - but I'm not certain whether or not this is NEC code. It's easy to imagine the value of this if the mechanical connection to chassis were to fail.
All good fortune,
Chris
Got it. Thanks.
I think it complies with that. All I have to do is put a lock washer on the one lug/tab where the safety ground from the power cord bolts on to the chassis. The fact that the tab has one other lead soldered to it shouldn't violate the rule. In any case, I'm not overly concerned with the code. I'm not producing a product for resale. The little Chinese amp kit pictured above doesn't follow that strict standard because all grounds go to one bolt with no lock washer. I suppose there is the possibility that decades from now it could theoretically loosen itself. I never considered that. I may have put a lock washer on the whole stack. I don't remember. I do know that I tightened the heck out of it and used no-oxid electrical grease between each lug (no grease under the screw head because that's the last place you want a slippery surface).
I'll adjust the diagram to accomodate these suggestions tomorrow. Thanks!
Last edited:
Between cap negative and the 2 ground terminal strip you should put a very thick wire , which is called bus ground wire .
Yes I will. I thought about using a heavy, bare copper solid wire to connect the two terminals and then run that same wire to the cap negative. It gives plenty of area to solder all of the wires. I'll adjust the diagram to accomodate the suggestion tomorrow. Thanks!
No grommet on that hole with wires going in and out in that picture of the Chinese amp.
I have found it useful to consider '(safety) earth' and 'ground (reference)' as being two distinct things. They only combine because it is essential to tie the ground to the chassis in order to capture the anomalies that will trigger the blowing of the mains fuse and potentially save a fire or electric shock.
The rule AIUI for the safety earth in Europe are that it should be as close to the inlet as possible, and a dedicated M4 bolt with star washers and no other connections. There is no current flowing to safety earth, so no concern about a ground loop (so one less long wire on your diagram).
Is there no volume control?
I have found it useful to consider '(safety) earth' and 'ground (reference)' as being two distinct things. They only combine because it is essential to tie the ground to the chassis in order to capture the anomalies that will trigger the blowing of the mains fuse and potentially save a fire or electric shock.
The rule AIUI for the safety earth in Europe are that it should be as close to the inlet as possible, and a dedicated M4 bolt with star washers and no other connections. There is no current flowing to safety earth, so no concern about a ground loop (so one less long wire on your diagram).
Is there no volume control?
Last edited:
All holes have grommets. I just didn't draw them (yet).
I got that part. It's about 1/4" from the inlet to the chassis ground.
I guess technically I might "violate" that rule by having one other wire soldered to that ring terminal. Seems trivial to me. I'm not making a product with liability from retail sales or government fines. I doubt that it would start any fires or shock anyone because of one additional solder point on one tab. With a lock washer on it, it will never come loose. I could also put Loc-Tite on that bolt to be 100% sure, but at some point "safe enough" is just that. The world has gone insane with fear of things that are a one in a billion chance. I always rode a bicycle with no helmet! Horror by today's standards! I could have died! 🙄
No. The ST-35 is just a power amplifier. Schematic is in post #1.
Last edited:
Helmets are arguably the wrong analogy for electrical safety. Nobody gives RatsAssOne about your skunky butt or mine, but there are other people and pets in the world. We're old and cranky, but I'd argue that should make us more rather than less aware of potential safety issues.
If this example, imagine the voltages appearing between PE and chassis, or signal ground (and therefor other connected electronics) and properly Earthed stuff, at failure of a chassis connection. Two is one and one is none.
All good fortune,
Chris
If this example, imagine the voltages appearing between PE and chassis, or signal ground (and therefor other connected electronics) and properly Earthed stuff, at failure of a chassis connection. Two is one and one is none.
All good fortune,
Chris
My point was that I think some people have a slightly higher risk tolerance than others. Nevertheless, if something is easily accomplished, I might as well do it, and I'm always interested in "best practices."
Based on the information provided yesterday, I have adjusted the wiring diagram as follows:
Based on the information provided yesterday, I have adjusted the wiring diagram as follows:
- Moved cable clamp for power cord to its own screw to avoid any possibility of interference with or loosening of the chassis ground screw.
- Used two ring terminals at the chassis ground point instead of one ring terminal with two wires, although I don't know why this is needed.
- Crimped connection at the ring terminals since solder alone is not considered valid.
- Added star lock washer to the chassis ground screw.
- Grounded PCBs to input jack lugs instead of running 2 wires per channel all the way to the circuit ground point.
- Added heavy bare copper wire "bus bar" to facilitate connecting all of the circuit ground wires.
Last edited:
Solder the wire from PCB to star ground somewhere at the middle of the long trace , not using the input jacks ... should be a little star there too . Like it is now all the current will flow through the input GND trace . A small filter cap could be added there , lets say 10uF + 100nF