Earthing question.

[Diabolical Artificer]

Got you Joe, the only problem with your scheme that I can see is the connectors or more properly their resistance. AFAIK a mains earth has to have it's own very secure bolt onto any metal work. Fittings should be crimped and shakeproof washers etc used. Also any earth wire has to be able to handle the full fault current. I was also under the impression that earth has to be a dedicated earth connection, so, can't for instance share your HT ground. Could be wrong there.

Lets consider an actual probable fault condition... a HT wire comes loose in the amp and touches exposed metal work. The HT fuse should blow as current rises and or your RCD or whatever mains protection device should trip as it sees an imbalance in the circuit.

There's two fuses in the PSU, one for the PSU and one fuse on the high voltage secondary, this latter is more likely to blow as it's specific to the HT circuit. So in your case without a dedicated earth wire we should be ok here. But would the house protection gizmo, RCD or whatever trip? When I connect signal ground/power supply ground to any metal work I usually use only thin wire, if that's the case in Joe's amp might there be a chance of this wire burning out in a heavy current fault condition?

DIY amp building and any DIY gear building is a bit of a nightmare when it comes to protection I find. Take fuse ratings for example, what value do you choose? The quiescent current of the HT circuit is 200mA say, HT is 450v, max current is 500mA, so something like a 1A fuse should do the job right? Two issues here, if your using the fuse on the tfmr secondary, say 0v - 320v AC RMS your 240v rated fuse is no good strictly speaking and what about the big caps you may be using in the CLC filter? Inrush current might be over an amp, so a 1A slo blo?

In practice whilst first powering up the amp and setting/taking DC conditions I use a 2A or 3A then work my way down and after a soak test and a few weeks of use find the right value by a bit of trial and error, oh, and I forget about the voltage rating if it's not too far off 240v.

So, safety and protection of homebrew builds can take some thinking about, a separate power supply and amp/s does complicate the issue. Just had a thought, it's often an idea to have a look at commercial gear and see how they did it, I can't think of any commercial amplifier's with a separate PSU, but the Tektronix 555 scope might be worth a butchers,that uses a separate PSU.

Thanks again all, Andy.

[Sideband]

Originally Posted by Diabolical Artificer

I can't think of any commercial amplifier's with a separate PSU,

Thanks again all, Andy.

Just a thought Andy...try looking at Naim. They do an optional separate PSU....we've safety tested them!

[Radio Wrangler]

Originally Posted by Diabolical Artificer

Lets consider an actual probable fault condition... a HT wire comes loose in the amp and touches exposed metal work. The HT fuse should blow as current rises and or your RCD or whatever mains protection device should trip as it sees an imbalance in the circuit.

The RCD will probably NOT trip. Your HT is after an isolating transformer. HT negative may be earthed but there should be no connection between any part of the HT system and Line or Neutral in any amp with a mains transformer.

RCDs are simple little things, they ask if the Line current balances the Neutral current. Any discrepancy more than the specified current and they trip. They don't even need an earth connection.

David

[GrimJosef]

Originally Posted by Diabolical Artificer

... I can't think of any commercial amplifier's with a separate PSU ...

Some very large and esoteric valve hi-fi amps have separate power supplies, not least because once the whole lot gets to weigh more than about 40-50kg it can be more convenient to split it into separate lumps.

For a while there was a company called Border Patrol who would sell you a stand-alone after-market HT supply for your valve power amp. These replaced what was usually a CRC filter with an LC one and they tended to upgrade the mains transformer too. The new filter could lower the audio hum, albeit by lowering the 100Hz HT ripple rather than by any earthing improvement, and the over-specified transformer tended to give better voltage regulation, although any discussion of whether that's audible belongs in the audiophilia thread, not this one.

You also sometimes see separate power supplies with phono stages (used to apply the RIAA correction to the signal from magnetic cartridges, and then to amplify it up to line level). These deal with millivolt, or even sub-millivolt, input signals and the ability to move the mains transformer some distance away can help with hum induced in the very sensitive circuitry by magnetic flux leaking from it.

Cheers,

GJ

[Lucien Nunes]

There are two ways to combat the ill effects of earth loops. a) tackle the loop itself by breaking unnecessary connections, increasing their effective resistance at relevant frequencies and voltages etc. And b) tackle the sources of energy that are driving current around the loops and causing potential differences to appear in the signal path.

Let us suppose there is a single external loop formed in the following manner:
PSU mains lead > PSU > umbilical > amp > signal source > source mains lead. This is a common configuration that exists in principle even when the amp and PSU are combined. If you connect the PSU and source to the same mains block and physically position all the cables so as not to enclose an area in the loop, it will not pick up stray field from the environment. If you now have a problem with hum that is cured by breaking the loop (noting comments above about competing and cancelling hum sources) then the EMF driving the circulating current is coming from within the equipment and can probably be solved by altering the internal design, rather than by changing an already desirable arrangement of interconnections.

First of all, characterise the hum as this can give a clue to its origin. Is it pure 50Hz, significantly 100Hz, full of higher harmonics or what? If the loop path through the PSU from mains lead to umbilical passes through some leakage flux from a mains transformer, it is likely to produce fairly clean 50Hz tone. If OTOH it is being coupled from wiring in the reservoir charging path, it might be rich with harmonics and buzzy.

Scrutinise the physical layout of the DC ground and AC earth paths within the PSU chassis. The mains input earth should be connected to the points that require earthing using dedicated termination points as you describe, i.e. don't use mechanical fixing screws to double-up as earth terminals. The transformer frame and interwinding screen are likely sources of primary-side earth faults and need solid connection back to the mains lead as much as the exposed metal panels. However, the routing of interconnections to achieve this and the locations of the terminations to the panels, can be arranged to condense any loops. This is crucial in the avoidance of linking any leakage flux into the ground pathway between the mains earth and the PSU output. The best point to connect the mains earth to the PSU chassis might be near the output connector or the star point.

There is a separate question of whether the amp chassis needs to be connected to mains earth. The HT and heater supplies could be made electrically separate from the mains, meaning that the amp itself cannot realistically become live with mains and needs neither class I (earthed) or II (insulated) construction. Whether or not the secondaries of the mains transformers are considered electrically separate (i.e. whether it counts as an isolating transformer with more than just basic insulation) depends on its internal construction and unless that can be verified by approvals and tests, it is not wise to assume that the PSU outputs are electrically separate. This means that the secondary DC 0V connection needs to be earthed. Again, this can advantageously be done at the star point.

That brings us finally to your bolted-on, green-and-yellow flying wire. I don't think this is necessary. For sure, the amp 0V line needs to be solidly connected to the PSU's HT 0V line as far as the secondary circuits are concerned. Failure of this connection could see the amp chassis and signal inputs fly up to HT volts, which I have experienced when an inter-chassis connection broke on the bench. It was not pleasant - I threw the receiver chassis into the air and injured my hand in the process - but could have been much worse. This critical 0V rail continuity could be achieved using multiple redundant contacts in the umbilical connector, linking the amp star point to the PSU star point, and this is what I have used when building separate amp / PSU chassis*. I certainly would not use a separate single-pole connector between the two for the main HT (as seems to exist) in case that was connected in the absence of the 0V connection resulting in HT on the chassis. But none of this relates to mains earthing. If the 0V star point connects to the mains earth in the PSU, and the wiring in the PSU is constructed so that no single fault will bring one of the other conductors (e.g. bias feed) independently into contact with the mains, then the amp chassis does not need a mains earth as it cannot independently become live with mains.

* Not for any phoolish reason but to spread the weight and size so that each part is carryable.

I have spent much time correcting hummy systems in studios etc including with valve gear. It can be rewarding when you get the exact configuration and also proof of why that configuration works and the original one doesn't. Sometimes there are so many parameters you never get that proof, only a quiet system.

[GrimJosef]

Very thorough Lucien !

I'd add just one thing - I can well believe that a comfortably spec'd mains transformer tends just to leak flux at 50Hz. But if the transformer is working hard, with low duty-cycle current pulses into the reservoir cap, then there's a risk of core saturation at the current peaks. It seems that flux leaks more badly when the core saturates, so the flux waveform can have higher harmonics in it too.

I've quite often seen a hum spectrum with prominent peaks at 50Hz, 150Hz and 250Hz, with the 150Hz peak commonly being the largest. I've found it worth checking whether this is coming from the transformer itself, either by rotating the tranny to see if the hum can be reduced or at least by separating it from the chassis if that happens to be steel.

Cheers,

GJ

[Lucien Nunes]

Good point, I should better have said mainly low order harmonics rather than 'clean'. There was a garbled edit there too that axed a reference to the SMPSU. This can produce all manner of nasty stuff both common-mode and differential. If it is Class II FE, i.e it has a functional earth connection on its input and that is common to the DC output negative (typically found on larger supplies >30W), even a very smooth, well regulated output can ride on aggressive mains harmonics.

[Terry_VK5TM]

A couple of sentences from the Mullard audio amplifier book:

Firstly concerning earthing:

In this amplifier, the chassis is insulated electrically from the internal components and wiring except for a single connection, by way of C21*, between the chassis and the busbar** return line of the amplifier. (This connection is required for reasons of stability.)

* C21 in this particular circuit is 0.05μF.
** busbar is the 0v line

Secondly on hum prevention:

To prevent hum pick-up, there are no isolating capacitors in the input circuit. For this reason great care must be taken when connecting auxiliary equipment to the amplifier.

This last one is obviously going to be a problem when connecting the likes of modern stuff to the amp.

[joebog1]

So, just to explain a little more of my build ( without stealing Andy's thread ).

My transformers are toroids. THey are wound on 350 watt cores but run at a tad over 200 watts at full load. They have an interwinding screen ( copper tape ) and an external flux band made from mumetal. The transformers have extra insulation between primary and seconday.
Andy has a similar transformer designed for a preamp.

There are four fuses in my amp. One on each channel primary mains of the transformer, and one each on the HT secondary side after the rectifier. ( 6 amp fast diodes arranged in a bridge).

The amplifier has delayed startup, the circuit of which I posted here for advice a few years ago.
Basically a capacitor charges up till it trips a large relay and applies full HT to the circuit around 2 seconds after switchon. The mains switch is just a tiddler rotary switch, but all it does is switch on a 30 amp contactor type relay, which in turn applies mains through a large wirewound resistor to the mains traffos and the delay. All primary wiring is at max a couple of inches long and laced into a loom that contains mains wiring only, and follows the corner of the steel chassis.

The main thing is ( at least I think ) there is only one earth in the amp and power supply, all connected together with heavy guage wire. The only "weak " link would be the two umbilicals which are constructed of 21 /.2 mmsqr wire, and effectively are in parallel and precisly the same length.

The connectors are standard circular aircraft connectors very similar to Andy's, with contact ratings of 500 volts and 6 amps.

I am trying to anotate a drawing in block form of whats in my head, but I will post that in a new post( old that already exists, thread ), and after I find a program that will let me do that NEATLY !!.

Joe
p.s. Sorry Andy for taking so much space on your thread.

[Ed_Dinning]

Hi Joe, that sounds like you have covered all the bases!

Ed

[Diabolical Artificer]

I'll have a look at NAIM, and GJ can you name some of these amps with separate PSU's please?

Thanks Lucien, a lot of good info there, I unpacked a lot of it but will have to through and have a re read not just of your post but the others too, there's a lot to take in but ifI understand correctly the mains earth is there to stop any metal becoming live and not there to stop the metal work from achieving a DC potential. If I also don't need the flying separate earth leads then that solves the problem to a great extent.

However in visiting this topic it's revealed a few things about my power supply that are less than optimal and whilst I have a pretty good intuitive feel for the whole ground/hum/layout field there's quite a few gaps in my knowledge.

No worries Joe, though different our amps share a lot in common, any sharing of info is welcome.

Andy.

[RichardGM]

I have no advice to offer on the black art of avoiding hum, but perhaps some comments on the safety aspect may be worthwhile. Apologies to anyone for whom this is egg-sucking. My credentials are that for a while I was chairman of the BSI committee dealing with the general standard for medical electrical equipment, BS EN 60601-1, and convenor of the IEC working group responsible for the electrical safety section of that standard.

The purpose of protective earthing of accessible conductive parts is to address a possible fault condition when they become connected to the line side of the mains supply. If and when that happens, the protective earth connection allows a large fault current to flow, taking out the fuse or circuit breaker on the supply circuit. With modern installations, even a small current down the earth wire (and therefore an imbalance between the currents in the line and neutral wires) will trip a residual-current device (RCD), but not all installations have those. My own house acquired a new consumer unit, replacing the old fuse box, only a few years ago and some of my friends' houses still have fuses.

In the OP's system, everything downstream from the PSU is (presumably) separated from mains. The question to be addressed is whether any possible fault condition, either within the amps or between the mains and output circuits in the PSU, could make any accessible parts "hazardous live" (in the terminology of some other standards but not the 60601 series). If all the metal gubbins is connected to the zero-volt line, whether or not that is connected to the incoming mains earth, is there any way that a fault could make anything "hazardous live" or would that just short-circuit the PSU and blow a fuse?

[Radio Wrangler]

Originally Posted by GrimJosef

Very thorough Lucien !

But if the transformer is working hard, with low duty-cycle current pulses into the reservoir cap, then there's a risk of core saturation at the current peaks. It seems that flux leaks more badly when the core saturates, so the flux waveform can have higher harmonics in it too.

Quite right.

The winding produces a high level of magnetomotive force, scaled to the peak current of the reservoir charging peaks. But the saturated core is no longer magnetically effective and no longer routes the flux from the coil back round to its other pole. So the radiated flux outside the core increases.

This only happens at the time the core is saturated, so the flux radiated is modulated with the harmonic rich pulse function.


Fit a bigger reservoir capacitor and it discharges less between peaks, so the time between when the rectifiers switch on and the top of the peak charging where they switch off is reduced significantly. The amplifier takes the same average current and so the same discharge in coulombs per cycle, but the transformer and rectifier have to replace that charge in a significantly reduced amount of time... so the current is in shorter, larger pulses. I2R losses are scaled up more flux is leaked etc etc. This is where choke input filters have an advantage, the snag is the regulation goes lousy if the current taken falls below a critical level and the rectifiers start peak charging instead of mean charging.

You also get wire clatter in the transformer and magnetostriction in the core, so the frankenstein movie sound effects are increased.

There is a good bit of art in designing a sociable power supply.

David

[Lucien Nunes]

In the OP's system, everything downstream from the PSU is (presumably) separated from mains

A problem with some one-off valve equipment builds is that the transformer cannot be relied upon to provide electrical separation because it is not a new item or made to any recognised standards e.g. EN61558-2. I don't know the spec of the transformer in use here. In transformers without an externally connectable interwinding screen, in particular, there are failure modes in which the secondary supplies could become hazardous live if left floating. There is also the functional problem of in-bound capacitive leakage and interference; therefore it is conventional to earth the 0V rail in the PSU although possibly via loop-breaking network e.g anti-parallel diodes shunted by a capacitor and resistor.

either within the amps or between the mains and output circuits in the PSU, could make any accessible parts "hazardous live" (in the terminology of some other standards but not the 60601 series).

Within the amps, not with respect to the LV AC supply; all supplies are derived from the PSU and no mains is present. However unlike most modern external PSUs that produce ES1 / SELV outputs, a valve amp PSU is an ES3 source which requires consideration of single-fault touch-voltage conditions at the amp chassis. Therefore, one way or another, things all end up connected together and the fun begins.

[Diabolical Artificer]

Thanks Richard, that clarify's matters.The transformer secondaries I wound myself and is over specced as regards the windings and core in that the core is 1KVA but the PSU itself supplies around 500VA. I used this core in case I wanted to add more windings later. The wire gauge used is generous so winding loses will be lower. The pri/sec insulation was two layers of mylar and tested to 10kv, as is insulation between HV winding's. Good insulation but not good pri/sec C but there should be no chance of the amps metalwork becoming live.

One of the things I wish I'd have done is fit a screen, I didn't have any suitable copper tape at the time and wanted to get the job done. I'd have also re-wound the primary, but still, there's next time, everytime I wind a tfmr I learn more.

[GrimJosef]

Originally Posted by Diabolical Artificer

... GJ can you name some of these amps with separate PSU's please ...

The big beasts are rare and exotic Andy. I've seen them at the odd audio show but never worked on one, so I don't have records of them. But for what it's worth here are links to a few units of this type, mostly a bit more lightweight:

Thomas Mayer makes a range of 2-chassis amps (or 4-chassis ones if they're monoblocks). Here are some 300B ones of his https://www.vinylsavor.com/en/products/300b/.

Dared made these monoblocks with a common separate power supply about 15 years ago https://www.goodsound.com/equipment/...ini_vp300b.htm.

Audiozen make a hybrid amp (valve front end, solid-state for the power stages) with the power supplies in a separate box https://www.audiozenaudio.com/embrac...ing-amplifier/

I'm not sure if the Cayin HA-300 is still in production. It's nominally a headphone amp (!) despite using 300B power valves and substantial output transformers. Again the power supply is separate https://www.audioemotion.co.uk/cayin...uk-33345-p.asp.

I could spend more time looking and I imagine I could find more, but none of them will be on show in a hi-fi shop near you, or me.

Cheers,

GJ

[Ed_Dinning]

Hi Andy, some of the larger amps had their circuits "lifted" from the GEC book on Audio Amps, types with outputs up to 1100W !

Ed

[RichardGM]

[QUOTE=Lucien Nunes;1468988]

However unlike most modern external PSUs that produce ES1 / SELV outputs, a valve amp PSU is an ES3 source which requires consideration of single-fault touch-voltage conditions at the amp chassis. Therefore, one way or another, things all end up connected together and the fun begins.

Absolutely, hence the need to analyse the configuration of the various circuits and see what would be the result of each foreseeable fault condition. When we were developing the third edition of IEC 60601-1, in one of the early meetings we decided that actually telling people to do that would be better than the approach that had been used in the previous editions, which had been to try to list all the places where separation is needed (which can be numerous when some bits are intended for connection to patients), and how good the separation needs to be at each place.

[Diabolical Artificer]

Thanks GJ.

Hi Ed, I've got the GEC "Approach to audio amps" I'll have a look.

ES3? will have a look, and give some deeper thought to this PSU.

Andy.