Hi!
I am designing a PCB with a LM1875 power amp, a linear power supply (transformer is not fitted on the board), and two regulators providing +/-15V for powering a preamplifier circuit on an other board.
I am gonna use a connector for signal input (from preamp PCB), and an other for +/-15V output (to preamp PCB).
What would be the best way to connect ground between those two PCBs please? One connector for signal and ground and the other one with +15V and -15V, or one with just signal and the other with +15V, GND, and -15V? Or maybe there is a better way?
I try to make the ground on the power amp PCB look as close as possible to a star ground.
The preamp board will have his own power filter capacitors, and decoupling capacitors on each ICs too.
Please apologize my bad english.
Thanks a lot for your help!
I am designing a PCB with a LM1875 power amp, a linear power supply (transformer is not fitted on the board), and two regulators providing +/-15V for powering a preamplifier circuit on an other board.
I am gonna use a connector for signal input (from preamp PCB), and an other for +/-15V output (to preamp PCB).
What would be the best way to connect ground between those two PCBs please? One connector for signal and ground and the other one with +15V and -15V, or one with just signal and the other with +15V, GND, and -15V? Or maybe there is a better way?
I try to make the ground on the power amp PCB look as close as possible to a star ground.
The preamp board will have his own power filter capacitors, and decoupling capacitors on each ICs too.
Please apologize my bad english.
Thanks a lot for your help!
Is the LM1875 pcb stereo or will you have two pcbs?
Also please read this:
https://www.diyaudio.com/community/attachments/wp_the_g_word-pdf.886214
Also please read this:
https://www.diyaudio.com/community/attachments/wp_the_g_word-pdf.886214
in any case, it is better to run different buses for power and input, because the bypass capacitors in the preamplifier will be connected to the power ground wire. In the preamplifier, the input and output of the signal ground will need to be done from the same point.
Hi! It is a mono amplifier, and only one PCB. I want to use it with instruments (guitar, synth, bass).
Thanks! So would you recommend that I connect the ground from the preamp output only to the reference resistor input of the LM1875, and then connect this point to the local ground of the preamp, connected to the power amp board by the power ground?
Thanks! Do you mean that I should connect all the bypass capacitors of the preamp to the power ground, and only them, and connect all the rest to the signal ground? Then connect signal GND and power GND in the power amp board only? Wouldn't this make the bypass capacitors noneffective due to long return pass between the load and the bypass caps?
a wire 20cm long is roughly equivalent to a track on a board 4cm long and 2mm wide.
if you still decide to make a full-fledged power connection with all buses, then a 10 ohm resistor will need to be inserted into the signal wire between the preamplifier and the power amplifier to open the resulting closed loop.
Is the preamp board already made? If not, what is the reason to use two boards? I assume the preamp has pots so it has to be mounted on the front panel?
Thanks for your answers!
I should have tell from the beginning, I already have a prototype of the power amp on perfboard, that is working well with the preamp. For now, the ground is connected between them only by the power connector. I would like to have a better looking power amp, and was wondering what would be the best way to do it.
Sorry, my English is not really good, I am not sure to understand. Do you mean if I have signal GND and power GND connected together on the power amp PCB, signal GND and power GND connected together on the preamp PCB too, and use a wire between the two power GND, and another wire between the two signal GND?
I should have tell from the beginning, I already have a prototype of the power amp on perfboard, that is working well with the preamp. For now, the ground is connected between them only by the power connector. I would like to have a better looking power amp, and was wondering what would be the best way to do it.
Thanks! I need to keep this in mind!
Yes, I already have a some preamp boards. They do have pots that need to be mounted on the front panel. But the goal is mainly to be able to use it with different preamps, or as a stand alone unit. I want to keep it as modular as possible.
OK
When you connect two boards powered from the same transformer windings, the problem is supply current flowing in the ground wire connecting the boards.
If the supplies for both boards are isolated from each other, this is less of a problem, but that isn't your case since you're deriving 15V from the LM1875 supply.
The worst is rectifier current pulses, which is why it is better to put all the rectifiers and regulators on the same board with proper layout, so rectifier current is confined in a tight loop between transformer, rectifiers and caps. And then take ground between the caps. I guess that's already what you're doing, so, fine, it's the most important to avoid hum.
So you have a preamp board and an amp board with 15V regs on it. The current flowing in the ground between these two will not be the supply current of the opamps, because they don't have a ground pin. It will be:
Assuming your regulators have low ripple and the preamp doesn't have too much capacitance (otherwise you can always add a resistor in series with the +/-15 lines)... If you do not use the differential amp topology in "the G word" for your amp board:
Basically the coax shield makes a nice thick low impedance ground connection between both boards.
If your preamp has blinking LEDs, or a microcontroller, or anything that dumps noisy current into ground, however, you will need a bit more care.
If you do use the diff amp, the coax becomes a shielded twisted pair (ie, balanced microphone cable, if you're a musician you must have some already) with shielded grounded at both ends, and one wire sensing ground on the preamp, connected to the negative input of the differential amp on the amp board. But it should not be necessary.
If you don't like to hear cellphones in your loudspeaker, the enclosure should be metal and all the RCAs should be grounded to it. This is much easier to do with a ground plane.
When you connect two boards powered from the same transformer windings, the problem is supply current flowing in the ground wire connecting the boards.
If the supplies for both boards are isolated from each other, this is less of a problem, but that isn't your case since you're deriving 15V from the LM1875 supply.
The worst is rectifier current pulses, which is why it is better to put all the rectifiers and regulators on the same board with proper layout, so rectifier current is confined in a tight loop between transformer, rectifiers and caps. And then take ground between the caps. I guess that's already what you're doing, so, fine, it's the most important to avoid hum.
So you have a preamp board and an amp board with 15V regs on it. The current flowing in the ground between these two will not be the supply current of the opamps, because they don't have a ground pin. It will be:
- Current through the decoupling caps on the preamp board due to whatever ripple is on the 15V supply. This is not signal related and can cause hum if the capacitors on the preamp board are too large and/or the 15V regs have a lot of ripple. Note decoupling caps on the preamp board between +15 and -15 do not count for this since they don't touch ground.
- Current dumped into ground by the various feedback networks and other analog circuits on the preamp board. This is related to signal, so it won't cause hum. It's not a problem.
Assuming your regulators have low ripple and the preamp doesn't have too much capacitance (otherwise you can always add a resistor in series with the +/-15 lines)... If you do not use the differential amp topology in "the G word" for your amp board:
Basically the coax shield makes a nice thick low impedance ground connection between both boards.
If your preamp has blinking LEDs, or a microcontroller, or anything that dumps noisy current into ground, however, you will need a bit more care.
If you do use the diff amp, the coax becomes a shielded twisted pair (ie, balanced microphone cable, if you're a musician you must have some already) with shielded grounded at both ends, and one wire sensing ground on the preamp, connected to the negative input of the differential amp on the amp board. But it should not be necessary.
If you don't like to hear cellphones in your loudspeaker, the enclosure should be metal and all the RCAs should be grounded to it. This is much easier to do with a ground plane.
Thanks a lot peufeu!
Yes, this is gonna be in a metal enclosure, with the input jack grounded to it.
To be sure to understand, are you saying that I should use a coax cable with one inner conductor for signal, and the shield connected at both ends carrying ground current ? Wouldn’t it be better to use a shielded cable with two inner conductors, with one for signal, another one for ground, and the shield connected at preamp output only, please ?
May I ask you to elaborate a little more on this please? That's the second time I read about too big filter capacitors on preamp causing hum, and I have never understand why.
Yes, that's what I'm doing 🙂
Yes, this is gonna be in a metal enclosure, with the input jack grounded to it.
To be sure to understand, are you saying that I should use a coax cable with one inner conductor for signal, and the shield connected at both ends carrying ground current ? Wouldn’t it be better to use a shielded cable with two inner conductors, with one for signal, another one for ground, and the shield connected at preamp output only, please ?
May I ask you to elaborate a little more on this please? That's the second time I read about too big filter capacitors on preamp causing hum, and I have never understand why.
Shields only work well when connected at both ends.
Also the shield of your coax is going to be a thick copper braid, much lower resistance than a wire in a 2-wire shielded cable, and low resistance is good.
If there is ripple on the supply voltage, that is some AC on top of your DC, then the caps on the preamp board are going to shunt this AC to ground, which creates a current in your ground, and therefore a voltage across the impedance of the ground wire. It is desirable to have caps to filter the power supply, but all they do is dump the noise as current into ground wherever they are actually connected to ground. So the big caps have to be on the power supply board. Or if there is only one board, they should be at the point of power entry.
There was one case on EE.SE where the guy had made a motor control board, and he put the power entry connector on the left side and the motor connectors on the right side. Then his ADC current sense was noisy. If he put big caps near the motor controllers, then the ripple current from the SMPS would go through the whole board to get to the caps. And if he put the caps near the power entry connector, then the ripple current from the PWM motor controllers would also go through the whole board to get to the cap. Screwed by wrong connector placement... should have put them all on the same side.
Actually if you are installing these in a metal chassis and all of the boards are commonly grounded to the case , you should only connect the shields on one end only, you should never have main power supply ground currents flowing through your signal shielding.
If you connect both ends of the shielding to ground on both ends to the boards and the boards are then also each grounded to the case then you create a ground loop with the shielding and the case through the to boards that you are bridging a signal to.
The shield will do nothing for the noise and the ground loop inside the case is more susceptible to noise and rf coming off of the line and also emitted from the transformer or directly up the ground side as well, this does happen I have tested it and have had such situations, especially with switching supply and such.
I learned this a long long time ago setting up many multi point patch bays for mine and others recording equipment and on stuff I have built, I have had my bouts with ground loops and all it takes is one lousy ground connect in the Wrong place, and, many times have to use ground lifters because of such badly practiced techniques that can set the whole system at a Hummmmmmmm !!
I had to Re-do some things and this was one of them besides adding caps from all the signal jack grounds to the chassis in my Fostex 454 mixer so that I could hook it to my CB (high power of course) and have some fun with my echo's and effects with it.
Only then would it work and it was a lot of simple modding that the engineers that designed the thing obviously knew nothing about !! :/
I had the same issue when connect my two Mackie boards together from separate rooms, I spent hours on the phone with mackie, to no avail, in the end years later it was a ground loop and all I had to do was lift one end of the shield feeding the other board, it had to be cut in order for the system to be dead quiet while all everything was wide open.
I was told it doesn't matter which end you leave un grounded just as long as they are all situated the same way if you have a choice.
My preference is to ground only the side of the cable that is the input to a device as it is an antenna.
I don't think it really matters at all but it just makes sense too me coming from the RF world where sometimes I only use an open coax cable as a probe to probe circuits with my scope, it works well and would not work very good if it were reversed.
FWIW 😉
jer 🙂
If you connect both ends of the shielding to ground on both ends to the boards and the boards are then also each grounded to the case then you create a ground loop with the shielding and the case through the to boards that you are bridging a signal to.
The shield will do nothing for the noise and the ground loop inside the case is more susceptible to noise and rf coming off of the line and also emitted from the transformer or directly up the ground side as well, this does happen I have tested it and have had such situations, especially with switching supply and such.
I learned this a long long time ago setting up many multi point patch bays for mine and others recording equipment and on stuff I have built, I have had my bouts with ground loops and all it takes is one lousy ground connect in the Wrong place, and, many times have to use ground lifters because of such badly practiced techniques that can set the whole system at a Hummmmmmmm !!
I had to Re-do some things and this was one of them besides adding caps from all the signal jack grounds to the chassis in my Fostex 454 mixer so that I could hook it to my CB (high power of course) and have some fun with my echo's and effects with it.
Only then would it work and it was a lot of simple modding that the engineers that designed the thing obviously knew nothing about !! :/
I had the same issue when connect my two Mackie boards together from separate rooms, I spent hours on the phone with mackie, to no avail, in the end years later it was a ground loop and all I had to do was lift one end of the shield feeding the other board, it had to be cut in order for the system to be dead quiet while all everything was wide open.
I was told it doesn't matter which end you leave un grounded just as long as they are all situated the same way if you have a choice.
My preference is to ground only the side of the cable that is the input to a device as it is an antenna.
I don't think it really matters at all but it just makes sense too me coming from the RF world where sometimes I only use an open coax cable as a probe to probe circuits with my scope, it works well and would not work very good if it were reversed.
FWIW 😉
jer 🙂
Last edited:
Thanks geraldfryjr !
The boards are not grounded to the case. The only chassis ground connection is at the preamp input jack.
Wouldn't the shield connected at both ends be efficient against magnetic field for only higher frequency than audio?
One would thinks so, but once you have a ground loop created all frequency's get through as more some than others as well but effectively the shielding does nothing at this point and you are relying on the case as the shield, if the ground loop is on the inside of the case then you risk picking up the EMI coming off of the magnetic fields that surround the transformer, most circuits my not have a high enough gain an input impedance for it to matter, but Phono and Tape preamps do and this is where you will find a huge difference.
IMHO Ground loop issues are mysterious and the worst to track down ans I can only help to give you my best advice as to try and avoid them completely from the git go, however I don't have enough space to type out all of the scenarios I have seen so far, all the boards have to be ground for power yes but it is best not to run power currents through the shields.
Infact shielding inside a metal case does little usually, except for high level RF and it causes unneeded ground loops if not done correctly, there are other techniques as well as I mentioned, as I saw inside my B&K CB RF signal generator those boards are triple shield on the inside from one another, the boxes are all layered in three triers, all with only one ground point to the outside world and a cap on that box shield to the main case to the outside world in order to reduces the TTL digital noise inside the thing and produces a super clean and super Quiet Modulated 2uv rms RF signal at its output for receiver alignment.
I took that thing apart only one time and I will never forget it, it all made sense as to what I had done earlier with my mixer.
With just a few boards it may not be that big of an issue but if they were stacked up as in a mixer with many channels then shielding and routing is a big deal.
However my Mackie 32-8 is made up of a bunch of boards and there is No Shielded Cabling in the Whole Mess and has a noise floor of better than -118 tp -124db to spec as measured by my stuff.
And ground loops have been my biggest nightmare outside of its very own case. 😉
jer 🙂
P.S. As it turns out for me, some equipment you can just hook them up with an unbalanced shielded cable and all is Good...
And Some, Will just not work and you have to go around snipping ground's somewhere to make it all noiseless again. :/
IMHO Ground loop issues are mysterious and the worst to track down ans I can only help to give you my best advice as to try and avoid them completely from the git go, however I don't have enough space to type out all of the scenarios I have seen so far, all the boards have to be ground for power yes but it is best not to run power currents through the shields.
Infact shielding inside a metal case does little usually, except for high level RF and it causes unneeded ground loops if not done correctly, there are other techniques as well as I mentioned, as I saw inside my B&K CB RF signal generator those boards are triple shield on the inside from one another, the boxes are all layered in three triers, all with only one ground point to the outside world and a cap on that box shield to the main case to the outside world in order to reduces the TTL digital noise inside the thing and produces a super clean and super Quiet Modulated 2uv rms RF signal at its output for receiver alignment.
I took that thing apart only one time and I will never forget it, it all made sense as to what I had done earlier with my mixer.
With just a few boards it may not be that big of an issue but if they were stacked up as in a mixer with many channels then shielding and routing is a big deal.
However my Mackie 32-8 is made up of a bunch of boards and there is No Shielded Cabling in the Whole Mess and has a noise floor of better than -118 tp -124db to spec as measured by my stuff.
And ground loops have been my biggest nightmare outside of its very own case. 😉
jer 🙂
P.S. As it turns out for me, some equipment you can just hook them up with an unbalanced shielded cable and all is Good...
And Some, Will just not work and you have to go around snipping ground's somewhere to make it all noiseless again. :/
Last edited:
Hi, sorry for my late answer !
Yes, the case is shielded, and high frequencies are filtered at the input.
I can hear the noise from the transformer coming in the spring reverb circuit (and I am still struggle with it...), but as you say other circuits don't have enough gain to cause noticeable issues. It is totally possible that the shielded cable is unnecessary.
Yes, the case is shielded, and high frequencies are filtered at the input.
I can hear the noise from the transformer coming in the spring reverb circuit (and I am still struggle with it...), but as you say other circuits don't have enough gain to cause noticeable issues. It is totally possible that the shielded cable is unnecessary.