[ginger]

A few comments only to fuel your discussion.

Interconnects GENERALLY have shield connected BOTH ends. That is signal currents flow from the source down the centre wire of the shielded cable to the "recipient" (the Amp in this case) and then back through the shield to the source. You need a full loop for it to work. When you break the shield of the interconnect the return path is via the mains earth wire which will have other return currents and noise etc on it.

When you have BOTH return paths it is possible that you can experience trouble with hum etc. because you have an earth loop. When this occurs you need to either break one of these return paths OR make some circuitry changes to make one of the paths the "preferred" path. For electricial signals that means the lower resistance path.

That is precisely the function of the 10R resistors at the input of the AKSA Amps. Return Current via the mains earth will see at least 10 Ohms more resistance than the return current via the interconnect shield. This is the BEST method for coping with these sorts of problems, its better noise wise than breaking an interconnect shield and SAFER than breaking a mains earth.

Occassionally we have a situation where we have no option to break one of the return paths and this can be done by connecting the interconnect shield one end ONLY, OR breaking the mains earth return path.

The second option is fraught with danger unless you know what you are doing. Remember the mains earth is a SAFETY earth. Its there to present a preferred path for FAULT currents - preferred to the path thru' you that is - we hope.

Breaking a mains earth return loop involves separating the Signal 0V and hence the Power Supply 0V from chassis earth and generally requires floating the entire amplifier electronics. Doing this SAFELY is not trivial and is NOT RECOMMENDED for home experimenters.

So what usually happens - the interconnect shield return is broken instead. The amp input signal then becomes subject to ground noise.

So whats the answer? - no one answer for every situation, the "Balanced" Input was an attempt to address just these sorts of issues, with a balanced input the return currents are a dedicate return wire and not on a 0V referenced wire so most of the problems go away.

Using shielded cable for interconnects with the  shield connected only one end works very well within a single chassis where multiple boards all have their 0V points tied back to a single "Star" 0V point.

If you have found it necessary to do this between your source and amplifier (separate equipment) might I suggest its probably due to another problem - that is you've introduced a second problem to cure the first. A NOT uncommon thing to see in electronics.

Cheers,
Ginger

[andyr]

A few comments only to fuel your discussion.

Interconnects GENERALLY have shield connected BOTH ends. That is signal currents flow from the source down the centre wire of the shielded cable to the "recipient" (the Amp in this case) and then back through the shield to the source. You need a full loop for it to work. When you break the shield of the interconnect the return path is via the mains earth wire which will have other return currents and noise etc on it.

 ...

 Ginger, there is another topology - for those of us who believe solid-core wire is the only way to fly (ie. no multi-stranded wires ... which shields are).

This is ... to have an IC made up of 2 wires, inside a braided shield.  This is what I have done.  The signal and return wires form the loop for the signal to flow from source to amp.  The shield is simply there to be a shield.  As such it has to be grounded - at one end only, otherwise it becomes part of the signal loop.

Hence my Q: should the shield be connected to the RCA barrel (together with the 'return' part of the signal loop) at the source end or at the amp end ... where the 10nF bypass cap attached to the RCA earth lug is??

To an amateur like me, it seems more logical for the shield to be earthed at the amp end, so any noise picked up along the length of the IC is straight-away shorted to the chassis.  However, to an electrical engineer like yourself, maybe there is a good reason to earth it at the source end.

As an amateur, I is a tad confused here, hence my Q!!

Regards,

Andy

[ginger]

Andyr,
Yep - there is the connection method you describe. In this case the shield is a "pure shield" and doesn't carry signal return currents and the shield MUST be connected one end ONLY. In the philosophy of connecting the shield to the "quiet" end - this connection should be the source end, not the amplifier end.

In such a case the caps shunting the signal return at the amplifier input are NOT as important BUT still may be worth while. The theory is still to shunt any RF Hash etc from the signal return side to chassis and keep it out of the amplifier box.

I've always been a bit sus about these sort of cables. My view is that the cable capacitance, that is the capacitance between the signal wire and the signal return wire plus the shield (connected together one end) MUST be higher than for an "ordinary" shielded cable and you MUST therefore suffer additional top end (high frequency) signal loss. All depends on cable quality, dielectric type (in particular) etc. of-course.

Cheers,
Ginger

[andyr]

Ginger,

Thanks very much for your explanation.  I shall make sure my IC shields are grounded at the source end.

Now, I made up a pair of 10.5m long shielded ICs as I described (bcoz my preamp & t/table are at one end of my 8m-long room and the AKSAs and speakers are up the other end).  I measured the following:

*  Capacitance of these ICs is 490pF.

*  Inductance is 0.014mH.  (Of course, I had to connect the central RCA conductor to the barrel at the other end, in order to measure this ... so I don't know whether this reading is for a 10.5m length or a 21m length!!)

Are these considered "high" values??

Regards,

Andy

[ginger]

Andyr,
490pF is NOT high for a 10.5m cable however its is getting towards the limit of what most middle of the range CD players etc. will drive happily.

How does this work??
If you want to pass audio signals up to say 30kHz you can calculate the capacitive reactance as 1/2pi.freq.C
ie for 490pF and 30kHz the reactance = 10.8 KOhms

This reactance is across (in parallel with) the Input Impedance of the Receiver Equipment (amp, preamp or whatever) - which for the AKSA is 47KOhm.

The 47K in parallel with the 10.8K gives about 8.8KOhms (roughly speaking - these actually combine as vectors. 8.8KOhms is a good value to use for a WORST CASE calculation)

For you to achieve the 30kHz response we chose above then the Output Impedance of the Source equipment driving the cable MUST be less than 8.8KOhms.

Good gear will manage that - some cheaper stuff with Output impedances of 10KOhms or more will struggle and might ONLY be able to manage 25kHz or less.

More cable means more capacitance and the problem gets worse.

This is where a line amp can be very usefull. Line amps are designed to drive long cables (lots of capacitance) - designed how, well the most important specification for a line amp design is Output Impedance. A good line amp will have an Output impedance of say 600 Ohms. A 600 Ohm Output Impedance Line Amp could drive your 490pF of cable capacitance at frequencies up to 500kHz (541.3kHz to be exact).

Or conversely could drive 8,842 pF of cable capacitance to 30kHz - anyone for a 200 metre cable???

Cheers,
Ginger

[andyr]

Thanks, Ginger.

My current (very old) Naim preamp which I use just as a line-stage has quite a low output impedance, so that explains why I hadn't heard any bad effects with my long ICs.

And my new GK-1 (when I get it finished!!) is only 120 ohms, so it won't have any trouble either.

Regards,

Andy

[David Ellis]

Gents,

My AKSA PCBs are gathering dust, but I have not lost interest.

I have a question regarding the cost of Black Gate capacitors. 

I read through this string and found some folks paying $10-$12 for a 50v 100uf Black Gate STD series capacitors for C3,C4,C10,C12 .  This seems strange    .  There is at least 1 guy in the USA selling them for $2.88.  Am I missing something here  ?  Are you gents using the bigger and more expensive N-Series caps?  My hunch is "no" since they are slightly too big for the PCB.

Also, for the aksa 55 power supply, I'm thinking of rolling the dice and trying the 35v Black Gate electrolytics.  They cost $30 each, and are significantly less expensive than the 50v variety ($92 each).  I pushed a Black Gate voltage limit in my CD player (due to size constraints) in a few places with no "boom" yet.  But... what happens when electrolytics are pushed too hard?

Dave

[ginger]

David,
With 25-0-25 V AC power trannies you end up with about +/- 36V rails so the 35V BG for the power supply is real risky unless you settle for a 22-0-22 or 20 -0 - 20 or similar power tranny to get the rails down a bit. That will reduce the output power a bit BUT I'm told that it also improves the sound. Anyway the 55W spec is into 8 Ohms - it delivers more like 90W into 4 Ohms so you can afford to ditch a bit of power unless driving maggie panels or such.

BG are expensive here in OZ because we are such a small market that we are treated like the "rest of the world" by most manufacturers/suppliers. I have found for example that the best price/service for ordering BG and similar is to order them from Parts Connexion in Canada. Thus they go right around the world before ending up on my door in Oz.

Rubycon ZL do very well on the Amp board power supplies but I found that you can screw the last bit of detail out of it by going to BG Stds - BUT as you point out they do NOT fit on the PCB conveniently and so this mod is for experience constructors who are happy to relocate adjacent parts to make space. My view is that it is the stunning noise performance of the BG that is at work here. Yes they sound good BUT the overall impression is more detail - that is a "veil" has been lifted allowing detail which was always there but masked by low level broadband noise to show through.

For an "over the top" mod I run 2 100uF/50V BG N "Super E" connected for the bootstrap cap BUT the 100uF BG Std is almost as good. BG N on the Diff Amp supply bypass was something I tried but did'nt like and went back to a BG Std. This mod really requires some thought as to mounting and how to tack one cap each side of the PCB.

I'm just about to complete a Hugh's "bog standard" AKSA55N+ to stack up against my "over the top, supermodified" 55N+ and so will be revisiting all of my "valhalla" mods over the coming weeks as I add them one at a time.

AS for what happens when you blow an electrolytic cap - well I've never blown a BG but have cerainly blown a few of the ordinary type electrolytics in my time. Results vary from gass escape from the vent hole which just stinks a bit (and the cap goes short) to having them really explode making a mess eveywhere which is very difficult to clean up.

Cheers,
Ian

[David Ellis]

Ian,

Thanks much for your extremely throrough response  .  It would appear that you are on par with Hugh regarding his continued enthusiasm and support for this amp  .  It's indeed a very good product.

Your comments about Black Gate capacitors in Australia make very good sense.  I do know that Jeff at Sonicraft will ship parts overseas, but the shipping cost will be the same as everyone else.  http://www.soniccraft.com/black_gate_capacitors.htm Also, I must admit that I have not compared Jeff's prices on Black Gate capacitors to other suppliers.  I have purchased loudspeaker capacitors and inductors from Jeff for many years and simply followed this "habit" by purchasing the Black Gate capacitors from Jeff too.  I have learned to completely trust Jeff at Soniccraft and didn't compare his prices.  Other folks may wish to compare prices and it might be a good idea.

It is comforting to know that electrolytic capacitors will physically explode .  I was worried that I might not be able to troubleshoot a blown 35V Black Gate, but I can troubleshoot this very easily.  20 Years ago I was a Patriot Missile System Technician.  The Radar power supply filters blew quite frequently, but I wasn't keen to hifi back then and didn't pay attention to what type of capacitor was in those power supply filters.  I think that I'll roll the dice at 35v Black Gates  .  I clearly rememeber the smell of a blown electrolytic cap and the associated black grunge  .

I can control the impedance of the speaker's used with this amp and the low dcr will be 6 ohms.  The nominal dcr is a fair 8 ohms.  As Hugh mentions, this is uncommon.  Most commercial stuff is nominally 4 ohms.  Hence, I don't see a need for reducing the transformer output voltages.

[stvnharr]

Dave,
Jeff, Michael Percy, and Partsconnexion all have the same prices for Black Gates.
I've popped a few electrolytics in my time and they just go pop, takes about a second to go pop.  It teaches you to get in the habit of having a few spares handy.

Steve

[David Ellis]

Jeff, Michael Percy, and Partsconnexion all have the same prices for Black Gates.

I kinda' thought this was the case. 

Those guys in Australia have a rough time in this regard.  They pay significantly more for capacitors.

I recently stuffed @ 30 smallish Black Gates in my CD player.  The total cost was probably $75.  For $75 this was a very worthile upgrade.  However, if those capacitors cost 3-4 x the cost... ouch.

[multibit16]

Sorry to drag up an old thread but I've recently been trying out some various caps in my Aksa amp, after recently fixing a blown output stage (thanks to Hugh for the advice btw;)  )  thought I'd do a bit of cap rolling since its in bits.

For my PSU's I use a pair of 24v toriodal transformers, discrete rectifiers made from MBR20200 and 4-pole T-network capacitors.
Electrolytic caps on the amp pcb was a mixture of Rubycon ZA/ZLH,  bootstrap was a Hitano.
I never really got on with Blackgate caps even after running in so stuck with the ZA/ZLH's.
Anyway looking through my boxes of components I found some Elna Stargets, enough to replace all the caps on the pcb so gave them a try, I wasn't expecting much tbh as I've tried the Stargets in other gear and wasn't too impressed but this time was quite surprised tbh!  I don't know if its because these go well with the t-nets in the psu but must say I'm quite pleased.
Anybody else tried the Stargets?  I have tried both the Silmics and Cerafines before in my amp but they didn't sound anywhere as good as this:)

[multibit16]

Now just fitted BC136's for the main decoupling