| hotscot |
Hi all
I have apparently successfully wired my center tap transformer to my rectifier board from my L3875 kit.
I'm obtaining plus and minus 38.2 V after rectification and without any load.
I really want to proceed very carefully here and I'm seeking advice on earthing my system. Again I have been reading through the forums and sometimes the advice is conflicting and or confusing.
From what I understand I connect the Greem/Yellow earth wire from my AC cord to my chassis. From this point I run a wire to CHG (Chassis Ground Connection) location on the amp boards.
(I should also place a fuse in series between the AC cord and the transformer.)
Ok so far?
Alan |
|
|
| Sherman |
You've got it. The connection of the earth wire in your power cord to the chassis is often referred to as 'safety earth'.
One wire from that point to chassis ground on each amp board. Don't connect the chassis grounds of the two amp boards together then run a wire to your safety earth. Earth each board separately.
A fuse is definitely a must. For gainclones I generally use a 2 amp fuse, generally a 'Slo-Blo'. Put the fuse in line with the "hot" wire from the AC mains. Basically AC hot into chassis direct to the fuse, then from the fuse to the on/off switch then back to the rectifier board. |
|
|
| hotscot |
Basically AC hot into chassis direct to the fuse, then from the fuse to the on/off switch then back to the rectifier board. [/B][/QUOTE]
Just to clarify, do you mean AC to fuse to on/off switch then to transformer?
Thanks
Alan |
|
|
| Sherman |
Good catch!
The AC hot coming in from the wall socket goes first to the fuse, then from the fuse to the switch and then from the switch to the power transformer NOT the rectifier board like I stated first.
My fault for trying to multitask while typing. :xeye: |
|
|
| Nuuk |
| Alan, you many benefit from reading this ! ;) |
|
|
| AndrewT |
Hi,
sorry to disagree and throw in another option.
The mains incomer should go;- Live to fuse to switch, Neutral to switch, Earth to chassis safety earth.
I think we are all agreed on this first bit.
You can use insulated terminal strips to help with these first interconnections, but try to keep the wires short and any that go wandering (to find a switch) should be double insulated.
Keep the safety earth beside the mains incomer.
Make it permanent.
NEVER dismantle the safety earth from the chassis when carrying out maintenance and/or testing and/or modification if there is ANY chance someone might plug the unit into the mains socket outlet.
Any additional connections to the safety earth must not require dismantling of the safety earth to chassis connection.
Now on to the audio ground.
I do not recommend direct connection from audio ground to safety earth.
I do recommend an indirect connection consisting of one/some/all of the following forming a parallel network. Some experimentation here may prove valuable.
The network can include:-
10r power resistor.
power bridge diode wired in inverse parallel.
high frequency capacitor (10nF to 100nF ceramic).
Ground lift switch (common in PA gear but generally not needed in domestic. Closing this switch often causes injection of hum and buzz into the amplfier)
The RCA input grounds should be isolated from the chassis.
All the amplifier grounds should be isolated from the chassis. |
|
|
| Sherman |
| quote: | Originally posted by AndrewT
[B]Hi,
sorry to disagree and throw in another option.
The mains incomer should go;- Live to fuse to switch, Neutral to switch, Earth to chassis safety earth.
I think we are all agreed on this first bit.
... |
Andrew,
I agree a DPDT switch to switch both AC lines is the best setup.
About connecting signal ground to earth- it depends on where in the circuit we are. With chip amps it is pretty simple to have only one 'signal' ground that is floating. With tubes the concept of 'signal ground' is a bit more complex and there are areas where the signal ground will be earthed. |
|
|
| hotscot |
Guys...you'll understand now why I get confused.
If it makes it any easier I just want to make sure the box will be safe in the first instance. So I just need the simplest, safe, solution.
If there are different earthing configurations which affect sound quality I can explore those later as I learn more theory and practice.
(On a side point...in the UK we call the wires that come in on the ac mains cable Neutral, Live and Ground. In the US, where I am now, I guess it's Ground and what....Hot? For all practical purposes, being AC, do I just consider the two wires carrying the current as Hot relative to Ground?)
Regards
Alan |
|
|
| mrshow4u |
Hi HotScot, In the US if you have 2-wire, it's Hot and Neutral. Neutral connects to earth ground at the service box where power comes into the home.
For 3-wire, Hot, Neutral and ground.
Colors (US):
Hot = Black
Neutral = White
Ground = Green |
|
|
| ash_dac |
| quote: | Originally posted by AndrewT
Hi,
The RCA input grounds should be isolated from the chassis.
All the amplifier grounds should be isolated from the chassis. |
If they wern't insulated would the RCA ground return (for a fault) current through any connected equipment ?
Note I also see numerous examples of mains powered amplifiers being tested on planks of wood. |
|
|
| Sherman |
| quote: | Originally posted by ash_dac
If they wern't insulated would the RCA ground return (for a fault) current through any connected equipment ?
Note I also see numerous examples of mains powered amplifiers being tested on planks of wood. |
If they aren't insulated from the chassis you have a very good to excellent chance of having hum in your amp. Unless your chassis is engergized in some way you shouldn't get damaging current through those components. You will get hum and noise. I always either use insulated RCAs or mount my RCAs on an insulated panel and attach that to the chassis.
Also both the speaker outputs both signal and ground should be completely isolated from the chassis.
I think that those two things in particular are what Andrew was referring to when he said signal ground shouldn't be connected to the chassis ground.
A plank of wood isn't a great insulator but it is usually sufficiently non-conductive that you shouldn't have problems in testing. |
|
|
| Nordic |
| Yep signal ground is often easily achieved by soldering a resistor between the two terminals on the input rca socket... the ground tab would be the signal ground star, from here you connect a THINNER wire to the power groundstar.... |
|
|
| hotscot |
Ok guys....so now I need to be careful about the thickness of the wires I'm using? :)
I was hoping this would be straightforward... |
|
|
| Nordic |
| only for the connection between the signal ground and powerground... i.e. you want more resitance between the ground and the input ground than you do between the ground and the speaker's ground return.... this will keep the nasty earth currents away from the clean input.... real good hum killer. |
|
|
| jaycee |
For doing the mains wiring in your gainclone, just go and get some 6A 3 core flex. Strip the outer insulation off so that you have the 3 inner wires free. Use that :)
Rod Elliot has a fab article on this - http://sound.westhost.com/psu-wiring.htm |
|
|
| AndrewT |
Hi,
ESP's grounding in that link is absolute balderdash! They will have to add that to their dictionary.
DO NOT connect the audio ground to the junction between the smoothing caps.
In particular DO NOT make MULTIPLE ground connections to a plate that has charging pulses across it. |
|
|
| jaycee |
Hmm.. I was looking at the pretty picture... he does not show PSU ground connected to chassis ground there...
I'm aware that a loop breaker should be used... something like two power diodes in inverse parallel.... my own gainclone doesn't connect power ground to chassis ground though. |
|
|
| Nordic |
If you insist on one of Rod's circuits (I love em)... then look at
Also if you your transformer has a ground tab, don't forget to connect it. |
|
|
| jaycee |
| That illustrates the ground loop isolator quite nicely :) For a gainclone though, I wouldn't go to the bother of using seperate bridges and capacitors - but that's just me. |
|
|
| AndrewT |
Hi,
Post18 diagram is just perfect.
All the audio grounds can meet at the junction right in the middle.
If ESP had only kept this good practice for that PSU link shown earlier. |
|
|
| ash_dac |
| quote: | Originally posted by Nordic
If you insist on one of Rod's circuits (I love em)... then look at
Also if you your transformer has a ground tab, don't forget to connect it. |
Hi,
 Naim Nap140
Looks like two bridge rectifiers as in the linked schematic ?
Green - ground
Red - positive rail
Grey - negative rail
White - output
Are the PSU cap' returning to the 'star' by separate paths in this design ? (Naim seem to have used two merged triangles with star at centre) |
|
|
| AndrewT |
Hi,
I see 5 separated grounds attached to the common between the smoothing capacitors carrying pulse charging currents.
Which nimcompoop copied the other?
I think ESP must have copied Naim's monstrosity.
The solution that almost saves this stupid grounding arrangement:-
drill a hole through the common between the caps. Insert a bolt and nut it tight through the PCB and common.
On the other side of the nut attach each ground on it's own earth tag and finally couple up with a second nut.
The modified version is almost as good as a separate audio ground.
edit: trying to decypher the PSU colours.
top red=right +ve
bottom red =left +ve
top black=right -ve
bottom black=left-ve
grounds starting at the top
1 green right power ground
2 green left power ground
3 turquoise possibly centre tap
4 right speaker return
5 left speaker return
yellow & grey appear to be transformer feeds to ~ & ~ on each bridge.
note no signal ground from amp PCB.
the two bridge diodes seem to be isolating the left and right halves not dual bridges as post18 |
|
|
| Nordic |
Another interesting one |
|
|
| AndrewT |
Hi Nordic,
are you drawing up these diagrams?
If so could you remove the audio ground from the smoothing capacitor common and attach it instead to the centre tap at or near the lower rectifier. |
|
|
| jaycee |
They're all diagrams from Rod Elliot's page.
The lower rectifier simply functions as the two diodes do on the previous diagram. The idea behind using the bridge is just to allow a higher fault current to pass.
I've always read it is best to put the star earth at the centre of the capacitor bank, and not near the transformer centre tap/rectifier as thats where all the charging pulses flow. My own Gainclone does pretty much this, and the only hum I get comes from the source. |
|
|
| AndrewT |
Hi,
this is the mistake that many knowledable builders make| quote: | | I've always read it is best to put the star earth at the centre of the capacitor bank, |
DO NOT put an audio ground on the common link between the smoothing capacitor banks.
Instead take a single wire from the smoothing cap common to an audio ground located elsewhere. Near the transformer centre tap is OK, IF the amplifier is also located nearby.
It might be better to put the audio ground near the amplifier/s.
The pulsed current loops run from transformer to rectifier back to transformer and from rectifier to capacitor back to rectifier. These loops must be kept short and compact (low loop area).
Go and re-read my comments on the Naim grounding in post22
and look at post 20 complimenting good practice. |
|
|
| Nuuk |
Andrew, the advice on properly grounding GC's has always been to have a separate signal ground and power ground, connected by a single piece of wire.
I've always used that arrangement and never had any issues with hum. :att'n:
I get the feeling that this thread has lost sight of the above advice and that's what is causing a bit of confusion! ;) |
|
|
| p robertson |
| according to my understanding and d.self the common ground is made on a short spur comming out of the connection between the pwr sppy caps,this is then connected to chassis along with seperate leads for spkrs earths and sig earth. It is quite obvious really.also mains earth then connects to chassis. |
|
|
| AndrewT |
Hi,
I don't agree| quote: | | It is quite obvious |
even you got this bit wrong| quote: | | this is then connected to chassis |
Do not directly connect safety earth to audio ground.
Instead use a disconnecting network. |
|
|
| AndrewT |
Hi Nuuk,| quote: | | the advice on properly grounding GC's has always been to have a separate signal ground and power ground, connected by a single piece of wire. |
I agree absolutely.
The problem seems to be:- how to assemble the PSU and connect the PSU to the outside world. |
|
|
| Nuuk |
| quote: | | The problem seems to be:- how to assemble the PSU and connect the PSU to the outside world. |
Do you mean what goes on the primary side of the transformer? I find a mains lead, fuse and switch is sufficient for say up to 500 VA, ie no soft-start required. Or are there other options? :confused: |
|
|
| AndrewT |
| No, I mean the secondary side. |
|
|
| kvholio |
How's this one ?
Regards,
Klaas |
|
|
| AndrewT |
Hi Kv,
do you want/need my approval? |
|
|
| kvholio |
No Andrew, this is as far as my reasoning concerning grounding goes. I hope it explains it clearly.
Best regards,
Klaas |
|
|
| Nuuk |
Kv, the bit that says 'keep this physically compact as possible" may deter people from putting the PSU in a separate housing.
In my experience, placing the transformer and bridge in a box, with up to a meter of cable connecting it to the amp circuit still results in a perfectly quiet GC.
For me the $64,000 question is, when housing the PSU separately, is it best to have the rectifier bridge with the transformer or in the amp case? :att'n: |
|
|
| kvholio |
| quote: | | For me the $64,000 question is, when housing the PSU separately, is it best to have the rectifier bridge with the transformer or in the amp case? |
Thats a tough one, Nuuk..
Let me try (this is my reasoning)
The bridge-rectifier is noisy part, so you want to keep it as far away from the amplifier-circuitry as possible.
By moving the bridge-rectifier to the separate psu-case, the R-component between rectifier and psu-capacity will increase, effectively creating an RC-filter.
This will make the psu less noisy.
Best regards,
Klaas |
|
|
| Nuuk |
| quote: | The bridge-rectifier is noisy part, so you want to keep it as far away from the amplifier-circuitry as possible.
By moving the bridge-rectifier to the separate psu-case, the R-component between rectifier and psu-capacity will increase, effectively creating an RC-filter.
This will make the psu less noisy. |
That's the reasoning that I have used so far. Plus, to me at least, it seems preferable to have DC in those umbilicals than AC! :att'n: |
|
|
| kvholio |
| quote: | | it seems preferable to have DC in those umbilicals than AC! |
Perhaps you can elaborate on that ?
I don't see any problem with AC in those umbilicals.
The only problem i see is this AC entering the separate amplifier-case.
Best regards,
Klaas |
|
|
| Nuuk |
| quote: | | Perhaps you can elaborate on that ? |
It's based on nothing more than a hunch! I thought AC may affect anything around it more than DC so with the interconnects etc passing close to the PSU umbilicals, I prefer to to stay with DC. |
|
|
| filholder |
| To me that sound right as well, AC gives far more eletro-magnetic interferance than dc doesnt it. Else we would be able to have DC transformers. |
|
|
| Nordic |
My Approach has always been using a seperate transformer from the amp which also contains the rest of the PSU...I.e. the power cables running into my GC's box is carrying stepped down AC.
Mains earth I terminate on the bolt on the transformer's case, from there I run a trhee-core wire with ends connected to the amp case from where its split with witha network of two 8A diodes, a 10R resistor and .1uF cap alll in series, nothing from the amp connects anywhere before this network... PS make the resistor a nice fire retardent one.... it burns to a crisp when something goes wrong on the amp in my experience...
mains---Transformer case ---- amp case ---- network---pgs/sgs/speaker returns...
The above describes the mains ground wire from where it is plugged into the wall onward. |
|
|
| BrianDonegan |
I think what is missing from kvholio's diagram is secondary-side fuses. These would be what blows in the event of a sort to the chassis, rather then blowing up the bridge(s) and possibly melting the transformer secondaries, which may have a much lower current rating than the bridge. I think without these you risking electrocution, until the bridge burns up, when you risk fire.
If you use a seperate chassis for trafo/bridges, you shoudl have fusing at it's output as well as the input of the amp box. |
|
|
| AndrewT |
Hi Kv,
I would keep to your compact philosophy.
i.e. keep the transformer and rectifier and first stage smoothing together.
Then run DC carrying cable to larger than normal decoupling to form second stage smoothing.
The short (low inductance) cable/track from decoupling to amplifier supports the high frequency demands.
The low frequency demand is still fed from the main smoothing capacitors.
Your grounding diagram needs a little refining. |
|
|
| Nordic |
| I have to admit that my transformer is about 6 feet from the amp, so i just went with what was easiest at the time... |
|
|
| BrianDonegan |
Here is another option. I mention it, as no piece of consumer audio equiptment I have owned (until recently) has had a three prong plug (no earth safety ground).
In this picture, mains hot goes to fuse, then power switch, the torroid. Nuetral goes to torroid. Again, no ground.
The chassis is "grounded" to the torroid's center tap. Both secondary rails are fused.
 |
|
|
| kvholio |
I'm not prepared to discuss the possibility of ac-mains powered equipment in a metal case
without a connection from the case to safety-earth.
I think it's inherently unsafe to do so.
We all know it's theoreticallly possible to construct it that way,
and we probably all have practical examples at home.
In my opinion it's not doable from a diy- point of view.
Also i don't see what there's to be gained from 'physically stretching' the psu by dividing it into 2 cases
(assuming a gainclone with only smoothing-caps on the amplifier-pcb).
Only in case of 2-stage smoothing (like AndrewT suggests) i can see some possible merit in putting the psu in a separate case.
So what are we building here ? i thought the thread-starter was going for a one-box solution.
Best regards,
Klaas |
|
|
| kvholio |
| quote: | | Your grounding diagram needs a little refining. |
I would appreciate suggestions.
best regards,
Klaas |
|
|
| AndrewT |
Hi Kv,
the link from upper cap to lower cap carries the impulsive charging current.
Similarly the link between the transformer secondaries (the centre tap).
The link from centre tap to capacitor 0V (zero volt or PSU common) carries no current when the amp is idling ( it carries a small current if the voltage amplifier uses different currents from the +ve and -ve rails). It carries current when the amplifier puts current into the load. This output current follows the signal and so is less impulsive than the charging pulses and for low output is very small. For large outputs it is very large but then a tiny bit of voltage difference is unlikely to be heard.
I suggest you connect centre tap to PSU common as shown but move the other two connections.
The disconnecting network, from the safety earth could be connected to the centre tap.
The speaker return and the decoupling returns could also be connected to the centre tap.
All the low current audio grounds should be brought together and then linked to the centre tap.
This is nearly the same as you have, the main differences are moving everything across to the centre tap and separately connecting the large current returns to the 0v ground. Some might say this is no significant difference and I have to admit I have not been able to measure any difference. However many amplifiers have multiple smoothing capacitors. With multiple caps there is no single 0V point whereas the centre tap is almost a single point.
Show the box of low current audio grounds as star connected rather than a string. In your diagram only the middle connection needs to be relocated up to the star point.
Delete the reference to a resistor in the note "offset between powersupply-ground and audio ground ( length of wire, bolt, or trace on pcb)"
I like the way you show the broken line box around the charging circuit. This could have a note added to say "keep the loop area of the two charging circuits as small as possible" |
|
|
| kvholio |
AndrewT, thank you for taking the time to write such an extensive reply.
If i understand correctly, you want to use the transformer's center-tap as ps-ground.
I see your reasoning behind it , but isn't there a practical issue to consider here also ?
A lot of times the transformer (and thus the center-tap) is located far away from the audio-circuit for practicality-reasons.
Using the center-tap as ps-gnd would in this case mean a longer ground-connection from the audio-gnd to ps-gnd, OR longer connections from the amplifier-circuitry to audio-gnd.
I think this longer connection/connections would be more sensitive to hum/noise pickup.
Using a bolt-configuration my suggestion would be as follows:
Connect center-tap at bolt FIRST
Then connect speaker-gnd at bolt
Then connect audio-gnd at bolt (or a piece of wire to create more offset from audio-gnd to ps-gnd)
Connect 2nd-stage smoothing/decoupling on opposite side of 0v-common
Like this:
Best regards,
Klaas |
|
|
| CarlosT |
| Can someone please post the recommended resistor and cap values, capacities and types for the optimum mains ground decoupling circuit shown above? That was very interesting stuff. I'm an extreme newbie and I engaged AndrewT in a battle of wits on a related thread which I was not prepared to win :D |
|
|
| AndrewT |
Hi,
I would swap second stage smoothing with the centre tap.
The current from centre tap to PSU 0v is the connection likely to have the worst charging pulses and as a result the worst buzzing.
Keep that current away from the signal side of the bolted connections.
I would take the Thiel/Zobel return to the decoupling / speaker gap.
You have shown spacers. I think it will do no harm to remove most of them. I use none, just solder tags. Experimentation should be very easy since all is bolted.
The idea you have shown is very similar to what JE Sugden did on the elderly P128 but the decoupling and signal and RCA input all share a wire to a remote part of the PCB. There is a small residual hum and buzz. I intend separating the grounds at the PCB to try and improve this and show that an audio ground can successfully share the bolt with the PSU 0v if done properly. But I have a lot of cutting and patching to do on the PCB so this part of my research is still some weeks away. |
|
|
| AndrewT |
Hi Carlos,
your search button must be broken.
Try c=10nF to 220nF
R =1r0 to 10r
25A bridge
Thermistor = no idea. |
|
|
| kvholio |
| quote: | | I would swap second stage smoothing with the centre tap. |
Agreed.
To me the most important thing is, your transformer center-tap is your powersupply-ground.NOT your common connection between your smoothing-caps.
Klaas |
|
|
| CarlosT |
| Thanks, AndrewT...you Scots are cranky but ultimately affable :D |
|
|
| falcott |
Hi there,
is it recommended that
a) the grounding should be positioned in the centre of the amp chassis, or that
b) the grounding should be centred between the 2 amp boards (in the case of a stereo amp) if they are off-centre of the chassis?
Cheers! |
|
|
| john_ellis |
Hi all
In my view, using insulated RCA phonos at the inputs, and insulated output and ground return sockets for the speaker power seems to work without needing an earth disconnector.
With this approach all the power return grounds and input grounds are brought (with separate wires, twisted pairs for the power etc) to a common earth point on the chassis. I usually make this the 0V from the power caps also, since although there is a high current noise from the caps, the resistance will be lowest at this point. Which makes the power ripple voltage as small as it can be.
Prof. Cherry used to recommend bringing the output ground return to the chassis and using chassis-connected phonos for the inputs. I have not found this to work when a preamp is connected which is also earthed. I have not tried disconnector schemes in that setup yet.
Anyone else tried Cherry's approach?
cheers
John |
|
|
| CarlosT |
| Great ideas...especially the centric ground thing. |
|
|
| AndrewT |
Hi John,| quote: | | In my view, using insulated RCA phonos at the inputs, and insulated output and ground return sockets for the speaker power seems to work without needing an earth disconnector. |
Can you give more detail?
It would be nice to know how to make this work.
From what you have described, I think I have tried this and failed to eliminate all noise from the output.
I wonder if the reason my version failed is that the impedance of the mains supply earth is lower than the amplfier impedance. So then the spikey low impedance contaminates the higher amplifier impedance.
Your observation that a mis-wired source will cause hum is very valid. Two mis-wired units causes the hum loop.
That becomes the main downfall after you eliminate the buzzes. |
|
|
| AndrewT |
Hi,| quote: | Originally posted by john_ellis
Prof. Cherry used to recommend bringing the output ground return to the chassis and using chassis-connected phonos for the inputs. I have not found this to work when a preamp is connected which is also earthed. I have not tried disconnector schemes in that setup yet.
Anyone else tried Cherry's approach? |
I have read some of Dr Cherry's WW article's .
I really liked his treatment of Thiel Network. And use it.
I would still like to try his nested feedback loops, but some others think his theory is flawed, leading to poor sound quality/oscillation on awkward loads. |
|
|
| AndrewT |
Hi,| quote: | Originally posted by falcott
a) the grounding should be positioned in the centre of the amp chassis, or that
b) the grounding should be centred between the 2 amp boards (in the case of a stereo amp) if they are off-centre of the chassis?
|
I think that precise location does not matter.
Since it is floating "chassis" is irrelevant.
I would be tempted to keep it between the pair of amp modules but close to the PSU.
However, I believe a much more important issue is keeping the LOOP AREA of the PSU cables VERY SMALL. Those are the cables forming the centre tap, the two AC cables to the rectifier, the two +v & -ve cables to the PCB. But some PCBs have well separated PSU & Power ground input connections. The designer is either very clever or has not recognised the relevance of small loop area and the impulsive magnetic fileds coming from single cables. |
|
|
| CarlosT |
Obviously twisting grounds together (PG- and PG+) helps? Does twisting any other specific set of wires help?
Luckily I got some cool twist pliers from my motorcyle trackday tool box :D |
|
|
| AndrewT |
Hi,
when two cables carry the same current but in opposite directions then close coupling them will attenuate the magnetic field. Twisting achieves an extra measure of attenuation.
So, any pair that carry equal and opposite currents can be twisted.
Transformer psuedo centre tap.
Transformer AC to rectifier.
Rectifier to smoothing caps.
Loudspeaker to PCB, but don't connect the return. It goes to audio ground.
smoothing caps +ve & -ve AND 0v to PCB.
note: that last one NEEDS the 0v wire to make a threesome. The currents in the threesome cancel but not in any pair from the three.
A warning on the Loudspeaker suggestion.
The single return passing near the PCB on it's way to the audio ground may be worse than the line led to speaker terminal and then return from terminal led straight to audio ground. |
|
|
|
