How to isolate pre-amp & power amp stages on same supply rail?

[Jasonmatthew911]

I'm asking this in case there is anyone here that can help me...I have another topic already open about it, but I decided to be more specific about what I need to do now, to make my LM380 guitar amp with 4580 pre-amp stage before it, work together of the same power supply...The LM380 has the recommended Staver V-7 Copper heatsink on it, and I also put a clip on the 4580 IC in case, since I am running on 18V, 1A Dunlop adapter...I already know for a fact that both of these work fine together on separate power supplies, but when I put them on the same supply, I get motorboating when my 500K Gain pot and 100K Volume pot are all the way up, everything works fine at lower volume levels....I also tried a few recommendations from my other thread, please refer to the thread that says "LM380 guitar amp has a pulsating sound at high volume level only, how can I fix?"....There I also mention the components I have on my power supply rail (I'm using vero/stripboards)...Another thing I did recently, was audio probe the circuit and at full levels on my pre-amp stage the signal path is clean all the way to the input pin of the LM380, the oscillation and motorboating starts right away at the output pin of the LM380...What does this mean exactly?...Is it the LM380 chip?...Is it being pushed to hard by the pre amp causing it to output noise at high levels?...I thought of trying the LM384 that is almost the same thing with higher ratings, to see if the motorboating was the same...The thing that I assume is that my Power supply rail isn't properly isolating the pre-amp from power amp stage....I was hoping that someone here would have a quick component solution to help isolate these 2 stages or maybe some experience with this kind of issue....I'd really like this circuit to work perfect, since I spent so much time on it and it actually sounds good before it starts motorboating...I made 2 separate guitar amps with the 4580 pre and LM380 amp, and they both do the exact same thing, so I'm trying to fix both....After trying many things, this issue is a mystery to me, and it's frustrating knowing they work perfectly together when on separate power supplies.

[merlinb]

Post a schematic. Use paragraphs. You will get more help that way.

[Jasonmatthew911]

Post a schematic. Use paragraphs. You will get more help that way.

You're right, sorry about that guys....I'm gonna try what you said about feeding the current hungry power stage before the pre-amp, hopefully you're right...I know the motorboating didn't go away when I tried feeding the 2 stages at the same time in parallel...If it doesn't fix the problem, I'll be sure to post my schematics...Thanks again.

[teemuk]

First, start with good ground return noding; make sure the ground current returns of the high current stages are not flowing in the lower current returns. Apply same proper noding schemes overall. It's a long story to discuss in a single post but I'm sure that a search should bring out a few threads about it. Try e.g. keyword "sewer ground", I know R.G. Keen uses that term and he has written a lot about this stuff.

The most basic decoupling style (RC filter) should stop most of motorboating issues and if the problem persists as severe one then you have other issues, such as the poor ground noding, or something else. Regulating the lower voltage / lower current rails is also a good idea. But you need to get the foundations right first and make sure the layout and noding won't cause issues.

[Jasonmatthew911]

Post a schematic. Use paragraphs. You will get more help that way.

Ok, I just Posted the Schematics in my original thread, "LM380 guitar amp has a pulsating sound at high volume only, how can I fix?".....I was thinking if it was a GND problem, but I really don't understand how to do the star grounding and avoid low current paths with the high current ones etc...I thought with GND that everything has to meet eventually at some point...Hopefully the mistake is in my schematics.

[R.G.]

I really don't understand how to do the star grounding and avoid low current paths with the high current ones etc...I thought with GND that everything has to meet eventually at some point...

Here's the problem: you're thinking in terms of voltages, not currents. All wires are ==> resistors <== just low value ones. When a current runs through a wire, it causes a voltage across the length of the wire it runs through. You simply have to know what grounds return current to the power supply (sewer grounds) and what grounds are used as a reference voltage. The wire to the input jack is a ==> reference voltage <== ground because the incoming signal voltage rides on top of whatever is there. The wire to a transistor's bias divider ground side is a ==> reference voltage <== because the signal voltage rides on top of it. The wire to the ground side of a power amp chip is a ==> sewer ground <== because all the current sent out/in to the load goes through it (in an AC sense). The V- pin of an opamp is a sewer ground because it powers the opamp. The ground going to the reference divider for an opamp is a reference voltage. Grounding a voltage reference at the V- of a power amp chip practically ensures feedback problems.

The whole point of star grounding is to make reference voltage grounds not share resistor/wires with the sewer ground return currents. That way the sewer currents can't muck things up.

[brett]

Hi
FYI
To do well, use separate lines to a star ground, and make them fat copper wires (household power cable is v cheap and fat and colou-coded - just strip out the conductors).
The worst scenario for a shared PS line is (i) when the power amp is further from the ground than the pre-amp, and (ii) the PS line has high resistance (like a PCB trace instead of a fat copper wire).
cheers

[Jasonmatthew911]

Hi
FYI
To do well, use separate lines to a star ground, and make them fat copper wires (household power cable is v cheap and fat and colou-coded - just strip out the conductors).
The worst scenario for a shared PS line is (i) when the power amp is further from the ground than the pre-amp, and (ii) the PS line has high resistance (like a PCB trace instead of a fat copper wire).
cheers

So basically you're saying, I should put my supply ground components of LM380 directly onto a thick copper wire, the input ground components on another thick copper wire, and then do the same with the 4580?...Then all those copper wires on separate paths should meet up at one point, but the power amp GND should be closer to the star GND?....Can they meet on a little copper plate separate from the chassis, or should they be meeting on the chassis?...Since my jacks are already grounding the metal chassis....

If I'm using stripboards for all of this, I shouldn't be using the copper traces for GND at all?...Also I'm using 22 gauge hook up wires for the GND going to my jacks and pots, is this ok?

[Jasonmatthew911]

[/quote]
Here's the problem: you're thinking in terms of voltages, not currents. All wires are ==> resistors <== just low value ones. When a current runs through a wire, it causes a voltage across the length of the wire it runs through. You simply have to know what grounds return current to the power supply (sewer grounds) and what grounds are used as a reference voltage. The wire to the input jack is a ==> reference voltage <== ground because the incoming signal voltage rides on top of whatever is there. The wire to a transistor's bias divider ground side is a ==> reference voltage <== because the signal voltage rides on top of it. The wire to the ground side of a power amp chip is a ==> sewer ground <== because all the current sent out/in to the load goes through it (in an AC sense). The V- pin of an opamp is a sewer ground because it powers the opamp. The ground going to the reference divider for an opamp is a reference voltage. Grounding a voltage reference at the V- of a power amp chip practically ensures feedback problems.

The whole point of star grounding is to make reference voltage grounds not share resistor/wires with the sewer ground return currents. That way the sewer currents can't muck things up.
[/quote]

Thanks R.G....I've been teaching myself about electronics, so I learn a lot as I go, thanks to the people that answer my questions...I think I'm understanding this a little better...So, I should put all my reference grounds on 1 path to the star GND and my sewer grounds on another path to star GND???...So these 2 paths will meet at the Star GND point, right?...Also, where should star GND be, if I have a steel plate chassis in the front of a wooden enclosure that gets grounded from the input jack and a separate steel plate on the back of the same wooden enclosure being grounded by headphone and speaker output jacks, this is also where my 2.1mm power jack (isolated from steel) is located as well???

[Jasonmatthew911]

I really don't understand how to do the star grounding and avoid low current paths with the high current ones etc...I thought with GND that everything has to meet eventually at some point...

Here's the problem: you're thinking in terms of voltages, not currents. All wires are ==> resistors <== just low value ones. When a current runs through a wire, it causes a voltage across the length of the wire it runs through. You simply have to know what grounds return current to the power supply (sewer grounds) and what grounds are used as a reference voltage. The wire to the input jack is a ==> reference voltage <== ground because the incoming signal voltage rides on top of whatever is there. The wire to a transistor's bias divider ground side is a ==> reference voltage <== because the signal voltage rides on top of it. The wire to the ground side of a power amp chip is a ==> sewer ground <== because all the current sent out/in to the load goes through it (in an AC sense). The V- pin of an opamp is a sewer ground because it powers the opamp. The ground going to the reference divider for an opamp is a reference voltage. Grounding a voltage reference at the V- of a power amp chip practically ensures feedback problems.

The whole point of star grounding is to make reference voltage grounds not share resistor/wires with the sewer ground return currents. That way the sewer currents can't muck things up.

Thanks R.G....I've been teaching myself about electronics, so I learn a lot as I go, thanks to the people that answer my questions...I think I'm understanding this a little better...So, I should put all my reference grounds on 1 path to the star GND and my sewer grounds on another path to star GND???...So these 2 paths will meet at the Star GND point, right?...Also, where should star GND be, if I have a steel plate chassis in the front of a wooden enclosure that gets grounded from the input jack and a separate steel plate on the back of the same wooden enclosure being grounded by headphone and speaker output jacks, this is also where my 2.1mm power jack (isolated from steel) is located as well???...Can I make a small copper plate cut out of 22AWG copper sheet to be my Star GND point?...This is what I'm still unclear of, since my input and output jacks are not isolated from the steel front and back plates of my wooden enclosure.

[R.G.]

So, I should put all my reference grounds on 1 path to the star GND and my sewer grounds on another path to star GND???

It's a matter of the currents. References generally don't source or sink much current. For instance, a JFET gate is referenced to ground through a 1M resistor, and its internal resistance is much higher than that. Same for a JFET input opamp. It's (+) input has an input impedance of many megohms, so it's fine being tied to a reference voltage through 1M, or more. Very, very little current flows through these grounds, so they could in theory be connected together, especially for the opamps which internally reject a lot of power supply noise. JFETs, on the other hand can't reject noise because their sources can be moved around because of source current through the source ground lead. It's probably better for the JFET to be referenced to its own source "ground" and have one wire from that go to star ground. The opamps can (usually!) share a reference voltage. What you DON'T want is any of these to be connected to, for instance, the ground return from an LFO that blinks an LED, or an LED switching circuit, because that would impart a voltage across its sewer-ground wire to anything that uses that same ground wire. So separating the grounds from anything that eats a lot more current than the input references is a good idea. Can they all be ganged together? It depends on how big the resistance of the wire is and how big the currents are. A classical mistake in building power amps is to bring the speaker ground return back to the power amp. That one has BIG currents, and must go directly to the main star point all by itself to avoid contaminating what it touches. The higher the current that flows, the more it needs its own wire.

...So these 2 paths will meet at the Star GND point, right?...

A "perfect" theoretical star ground has one wire to the star point from every component which connects to ground. It's impractical to wire that way, so usually, one ties all the grounds from one circuit section which does one function together. The grounds may contain a contaminating voltage, but it winds up as a feedback voltage which does not usually change function much. That means that there's a local star, perhaps after a two-device preamp section. These local stars have one wire each leading back to the main star point. It's less ideal, but massively better than just glomming them all together.

Yes, all grounds go back the main/master star point.

Also, where should star GND be,... This is what I'm still unclear of, since my input and output jacks are not isolated from the steel front and back plates of my wooden enclosure.

This gets near to meditation and religious revery sometimes. There is ideal perfection, which is guaranteed not to cause grounding problems. Then there is pretty good, and causes no problems, even if it's not perfect.

Ideal is one single point to which all the ground points connect. It is best if this point is completely unconnected from the chassis. Once the circuit works with no connection to the chassis, a single wire is connected from the star point to the chassis. This forces no currents to flow through loops in the chassis, which then serves only as an electro magnetic shield. Again ideally, the only connection to chassis should be the ground sleeve of the input jack, and all others isolated from chassis. That's usually impractical, but it's the theoretical best. What is usually done is to connect all the jack grounds to chassis. This sets you up for hum and oscillation if anything goes wrong, but people get away with it often enough that it's become the standard because it's easy and mostly works, especially if no high current returns, like speaker returns flow across the chassis too. The AC power line safety ground wire must connect to chassis too, but it connects at its own point, not used for anything else.

The master star point has coming to it the wires from local-star collections of ground, and also exactly one wire to chassis, exactly one wire to the power supply ground, usually a filter cap terminal. That same filter cap terminal which connects to the star ground through one wire has connecting to it only the single wire to star ground and one wire from the rectifiers/transformer. That wire from the transformer/rectifiers carries the high current pulses in the rectifiers, and must not share any conductor in the rest of the grounding net. It's currents go in and out of the first filter cap, and can add hum that is impossible to get rid of. The single wire from the filter cap terminal to master ground carries (in theory!) only DC, and it certainly does not carry the rectifier charging currents. The wires to chassis and rectifiers are very special cases, as is the speaker return, because of the high currents. You can be sloppier with others until it gets you into trouble, as everyone usually does. Then they turn up here saying "My sadjfklsas oscillates!" or "My uyaureiop hums and I can't make it quit!". 

if I have a steel plate chassis in the front of a wooden enclosure that gets grounded from the input jack and a separate steel plate on the back of the same wooden enclosure being grounded by headphone and speaker output jacks, this is also where my 2.1mm power jack (isolated from steel) is located as well???...

Unless the currents are quite small, that's asking for trouble.
Ideal (I think... I have to ponder this stuff out too, and I make mistakes on it a lot) would be to take one ground wire from each isolated chassis section to the master star point. For the input side, that can be the wire on the ground side of the input jack. It would be better if the output jacks were isolated from chassis and carried their own wires back to the master star, or all but one are isolated from chassis. The power jack should be isolated from the chassis, and wires taken back to the master star point. This is a common disaster in making a univibe/neovibe, because the AC in cannot be connected to ground, so using an uninsulated power jack causes massive hum.

Can I make a small copper plate cut out of 22AWG copper sheet to be my Star GND point?...

Yes. However, it really ought to be isolated from chassis, and have one and only one wire to chassis from it.

[Jasonmatthew911]

So, I should put all my reference grounds on 1 path to the star GND and my sewer grounds on another path to star GND???

It's a matter of the currents. References generally don't source or sink much current. For instance, a JFET gate is referenced to ground through a 1M resistor, and its internal resistance is much higher than that. Same for a JFET input opamp. It's (+) input has an input impedance of many megohms, so it's fine being tied to a reference voltage through 1M, or more. Very, very little current flows through these grounds, so they could in theory be connected together, especially for the opamps which internally reject a lot of power supply noise. JFETs, on the other hand can't reject noise because their sources can be moved around because of source current through the source ground lead. It's probably better for the JFET to be referenced to its own source "ground" and have one wire from that go to star ground. The opamps can (usually!) share a reference voltage. What you DON'T want is any of these to be connected to, for instance, the ground return from an LFO that blinks an LED, or an LED switching circuit, because that would impart a voltage across its sewer-ground wire to anything that uses that same ground wire. So separating the grounds from anything that eats a lot more current than the input references is a good idea. Can they all be ganged together? It depends on how big the resistance of the wire is and how big the currents are. A classical mistake in building power amps is to bring the speaker ground return back to the power amp. That one has BIG currents, and must go directly to the main star point all by itself to avoid contaminating what it touches. The higher the current that flows, the more it needs its own wire.

...So these 2 paths will meet at the Star GND point, right?...

A "perfect" theoretical star ground has one wire to the star point from every component which connects to ground. It's impractical to wire that way, so usually, one ties all the grounds from one circuit section which does one function together. The grounds may contain a contaminating voltage, but it winds up as a feedback voltage which does not usually change function much. That means that there's a local star, perhaps after a two-device preamp section. These local stars have one wire each leading back to the main star point. It's less ideal, but massively better than just glomming them all together.

Yes, all grounds go back the main/master star point.

Also, where should star GND be,... This is what I'm still unclear of, since my input and output jacks are not isolated from the steel front and back plates of my wooden enclosure.

This gets near to meditation and religious revery sometimes. There is ideal perfection, which is guaranteed not to cause grounding problems. Then there is pretty good, and causes no problems, even if it's not perfect.

Ideal is one single point to which all the ground points connect. It is best if this point is completely unconnected from the chassis. Once the circuit works with no connection to the chassis, a single wire is connected from the star point to the chassis. This forces no currents to flow through loops in the chassis, which then serves only as an electro magnetic shield. Again ideally, the only connection to chassis should be the ground sleeve of the input jack, and all others isolated from chassis. That's usually impractical, but it's the theoretical best. What is usually done is to connect all the jack grounds to chassis. This sets you up for hum and oscillation if anything goes wrong, but people get away with it often enough that it's become the standard because it's easy and mostly works, especially if no high current returns, like speaker returns flow across the chassis too. The AC power line safety ground wire must connect to chassis too, but it connects at its own point, not used for anything else.

The master star point has coming to it the wires from local-star collections of ground, and also exactly one wire to chassis, exactly one wire to the power supply ground, usually a filter cap terminal. That same filter cap terminal which connects to the star ground through one wire has connecting to it only the single wire to star ground and one wire from the rectifiers/transformer. That wire from the transformer/rectifiers carries the high current pulses in the rectifiers, and must not share any conductor in the rest of the grounding net. It's currents go in and out of the first filter cap, and can add hum that is impossible to get rid of. The single wire from the filter cap terminal to master ground carries (in theory!) only DC, and it certainly does not carry the rectifier charging currents. The wires to chassis and rectifiers are very special cases, as is the speaker return, because of the high currents. You can be sloppier with others until it gets you into trouble, as everyone usually does. Then they turn up here saying "My sadjfklsas oscillates!" or "My uyaureiop hums and I can't make it quit!". 

if I have a steel plate chassis in the front of a wooden enclosure that gets grounded from the input jack and a separate steel plate on the back of the same wooden enclosure being grounded by headphone and speaker output jacks, this is also where my 2.1mm power jack (isolated from steel) is located as well???...

Unless the currents are quite small, that's asking for trouble.
Ideal (I think... I have to ponder this stuff out too, and I make mistakes on it a lot) would be to take one ground wire from each isolated chassis section to the master star point. For the input side, that can be the wire on the ground side of the input jack. It would be better if the output jacks were isolated from chassis and carried their own wires back to the master star, or all but one are isolated from chassis. The power jack should be isolated from the chassis, and wires taken back to the master star point. This is a common disaster in making a univibe/neovibe, because the AC in cannot be connected to ground, so using an uninsulated power jack causes massive hum.

Can I make a small copper plate cut out of 22AWG copper sheet to be my Star GND point?...

Yes. However, it really ought to be isolated from chassis, and have one and only one wire to chassis from it.

I originally tried the circuit without the chassis, jacks and pots in the air...So is it really necessary to isolate the headphone and speaker output jacks from chassis?...Since my jacks are the metal switchcraft jacks, this will be kind of hard...Can't I have my back steel plate chassis being grounded by the speaker output jack to star GND, and from the speaker output GND will connect to headphone jack GND, so only 1 wire would be grounding the back steel chassis, since the DC jack is already isolated from chassis and will connect directly to star GND as well???...Why can I GND the front chassis with the input jack to star ground and not the output jack with back chassis, which is actually a separate piece of metal, so it is isolated from the front steel chassis....Also, should I be using very thick solid core copper wire for my paths to Star GND?...Is 18AWG enough or do I need thicker than that, or would stranded wires be alright?...Thicker wire will make it harder for me to connect to my stripboards etc......

I should've just stuck to making pedals, but I want to know more...Thanks for your time R.G.

[R.G.]

I originally tried the circuit without the chassis, jacks and pots in the air...So is it really necessary to isolate the headphone and speaker output jacks from chassis?...Since my jacks are the metal switchcraft jacks, this will be kind of hard...Can't I have my back steel plate chassis being grounded by the speaker output jack to star GND, and from the speaker output GND will connect to headphone jack GND, so only 1 wire would be grounding the back steel chassis, since the DC jack is already isolated from chassis and will connect directly to star GND as well???

Maybe. That's the problem with grounding. Star grounding can be shown to be a sufficient answer in all cases. It cannot be shown that something else can't be made to work, or that the gain of the circuit is so low that it won't matter. In your case, with only outputs on the back panel (speaker and headphone) and no inputs, you may be fine.

I guess another way to think of it is that star grounding is sufficient in all cases. However, it may not be necessary in all cases. There may be simplifications which work as well, or 90...95...98% as well, depending on what is where and how big the currents are. If you can afford to dink with it for a couple of weeks to get it right with a custom ground buss or chassis grounding, you may not have to have star grounding. But if it simply has to work perfectly first time, every time, star grounding is a way to get that.

However, do worry about safety grounding. For safety reasons, you really should connect the front and back panels with a thick braid of wire or metal that will conduct lots of current - the safety test is 25A last I remember - so that if the AC power line gets loose and touches either one, it will pop the fuse and not the connecting wire. The safety guys think it's better to not be dead than not hear hum.  Sometimes this makes hum removal harder.

...Why can I GND the front chassis with the input jack to star ground and not the output jack with back chassis, which is actually a separate piece of metal, so it is isolated from the front steel chassis....

That's because the chassis really should be continuous for safety ground/shielding reasons, and the Rules say, only ground the chassis at one point. People have been getting away with grounding things everywhere to the chassis for decades. And it is that: getting away with it. Current prefers to run where the impedance is lowest, just like people prefer to walk on paths instead of cross country through the bushes. It's easier. So if the chassis has the inputs on one side so any chassis current flows to the ground point in the middle and does not combine with current flow from the speaker outputs on the back and get shifted around, you get away with it.

So: can you use the jacks to ground the chassis plates? Sure. To the degree that they isolate high currents and sewer ground returns from inputs and references, you will get away with it. Do be sure to worry about that safety ground current thing, though. That means you may have to use 25A capable wires to your speaker jack and input jack which ground the chassis, and use star washers under the jack nuts to ensure a bite into the chassis metal. Otherwise, there is a potential (sorry...  ) shock hazard.

Also, should I be using very thick solid core copper wire for my paths to Star GND?...Is 18AWG enough or do I need thicker than that, or would stranded wires be alright?...Thicker wire will make it harder for me to connect to my stripboards etc......

Actually, the wire can be as thin as can be guaranteed to not get too hot to make it easier to wire. The nice thing about star grounding is that the wire size matters less, because you're not trying to keep voltage drop in the wires low by simply glomming on more copper. For things in the range of a few ma, use thin stranded wire. For speaker wire, use 22ga to 18ga. The extra resistive drop does not cause problems because only that particular circuit sees it. For AC power wiring, go massively for safety in current carrying and insulation voltage.

I should've just stuck to making pedals, but I want to know more...Thanks for your time R.G.

Actually no, you should not have. You're in the uncomfortable position of having conceived something that needs made that  you don't ...yet... know how to do. I personally have adopted that as a life style for my entire adult life. I know how to do more cra... er, stuff  than you'd believe. I'm a firm believer in the quip that a monstrous mind is a toy forever. Finding oneself in the position of needing to learn to get the next whatsit designed and working is a delight.