[AKSA]

Thanks Darren, Mike,

This is extremely interesting information......

Cheers,

Hugh

[stvnharr]

Hi Andy,

I got the ‘best sounding’ bit but my question as a non-engineer (and general ignoramus) was that I didn’t understand what needed to be bypassed.  Is the TDK not a good enough SA in its own right?  Would bypassing apply to other SAs?

Hi Lyn,

The TKD is fine on its own (it likely would not have survived as a product at that price otherwise) but in his never ending quest for "the best" Bob tried every mod he could think of. Primarily, he found that these bypass resistors improved dynamics.

Yes, the bypass mod could also be applied to other 'series' type stepped attenuators too, but you would need to experiment with various resistors to see which sounded best (to you). Originally I considered using some Caddocks (not cheap but much cheaper than the S102's) but Bob had tried them on the TKD and felt they sounded "crap".

Hi Darren,
I'm not at all interested in a new attenuator.  But I'm interested in learning about the "bypass" associated with the TKD.  From a post up above, I learned that its 40 steps in the 0.5 to 1.0 db area.  This helps explain much of the high cost.  But what are you doing with the bypass stuff? Are you making some kind of fancy fake shunt attenuator?  You don't have to divulge any proprietary secrets of course, but a couple words can explain a lot.

Steve

[LM]

I agree with Hugh’s thanks for all the interesting thoughts and options, more grist for the old mill.

Now we appear to have a couple of options for some relay based attenuators (AKSAphile the Dantimax RelVol1 and MikeC the Twisted Pear Joshua Tree).  Where would such items fit in sonic wise with the various series, shunt, ladder, bypassed etc SA’s that have been discussed so far?
 

[whubbard]

Well if Bob Crump was right the bypassed TKD would be better than the rest of the types of volume controls. I believe he didn't just try stepped attenuators, but tried everything available at the time. Therefore the bypassed TKD is superior to all. However, I’m not sure how another stepped attenuator would fit into the mix, but I would speculate a stepped attenuator is better than digital volume control.

Also, Steve, I'm not sure if this is what you are looking for, but the TKD is bypassed from input to output with resistors that are 10x-12x ohms the stepped attenuators. 10k stepped attenuator = 100k-120k resistors. That is the basics on the mod itself, although I'm not sure on what it technically speaking does.

-West

[stvnharr]

Also, Steve, I'm not sure if this is what you are looking for, but the TKD is bypassed from input to output with resistors that are 10x-12x ohms the stepped attenuators. 10k stepped attenuator = 100k-120k resistors. That is the basics on the mod itself, although I'm not sure on what it technically speaking does.

-West

West,
If I read the above correctly, then why would 4 resistors be required, rather than 1 per channel?
But really, since I don't want to do any of this, the answer doesn't matter.  I was just curious.

Steve

[AKSA]

Lynton mentioned 'grist for the mill', but I have one more to throw in here.

The sonic problems of potentiometers derive from the multiple contact phenomenon of the wiper, and they essentially rob resolution at very low levels.  The switched attentuator replaces a continuously variable carbon or plastic track with a ladder of resistors, where the only degradation is in the solder joints and the switching points.  This is known to be much 'cleaner', but there is still the problem of high cost, limited availability (or at least convenience of purchase), and highish wear rates, which knocks out the cheaper Asian imports.

The Rolls Royce (now there's a fading metaphor!) of switched attentuators is undoubtedly the Japanese TKD, which is veritably a scientific instrument, made like a Honda F1 engine cylinder head.  Just to hold it in your hand is a feeling of wonder, or at least for odd characters like myself who absolutely live for good engineering!!  But the cost is absurd, something you would only entertain for a $20K+ preamplifier.  That my GK1 is considered in this rarified company is a great flattery, and I'm pretty happy about it!


The Welborne Labs relay system is good, but I would suggest that the relay quality is the pivotal factor here, and they may be cheapies.  My favourite is the NAIS Aromat, with silver/palladium contacts, which is supplied in my Swift.  I've never heard any degradation of the signal using these relays, but I don't believe they are used in the Welborne kit product.

This third relay approach, wherein the complex, 24 position switch is replaced by several relays under microprocessor control driven from a rotary selector is pretty good, but it's electrically complex, and a whole project on its own to build.  I quite like it, and a Norwegian called Kolbrek (IIRC) created a marvellous design later added to by a clever young Viennese EE student, now graduated, with the interesting name of Timo Christ, which is a full on DIY project somewhere on the net.

However, since about 2002 when I designed the GK1 there have been a slowly growing list of microprocessor controlled ladder attentuators apperaing in the market with integrated opamp outputs for buffering and laser trimmed resistors.  The most famous is the PGA2310, originally a Burr Brown chip, now owned by Texas Instruments.  This product appeared in Oct 2001 but was revised in Sep 2004, and has stellar specification, namely -95dB to +31.5dB in 0.5dB steps (!!!), a distortion figure PLUS noise of just 0.0004%, a Zout of around 100R, a dynamic range of 120dB, and a unity gain bandwidth of 1.5MHz.  With a Zin of just 10K, two independent channels on the one chip, and noiseless switching, this is surely just the ticket for a premium quality high end system.

Other chips have since appeared in the market, I believe.

More grist for the mill?   

Ciao,

Hugh

[jules]

Grist!! That's evil Hugh 

So at $9.95 for a PGA2310, what's the catch and where does the cost come in?

jules

[whubbard]

The Rolls Royce (now there's a fading metaphor!) of switched attentuators is undoubtedly the Japanese TKD, which is veritably a scientific instrument, made like a Honda F1 engine cylinder head.  Just to hold it in your hand is a feeling of wonder, or at least for odd characters like myself who absolutely live for good engineering!!  But the cost is absurd, something you would only entertain for a $20K+ preamplifier.  That my GK1 is considered in this rarified company is a great flattery, and I'm pretty happy about it!

Hugh,

My personal thinking is that the main component in any preamp is the pot or the attenuator, and then the selector switch. The reason for my thinking this is that those are the only two parts I believe serve a purpose. However, the Gk-1 is special, and it adds tubes. I too purchased 7308s, so at this point the only thing that I have really left alone is your circuit, and from what I hear...it deserves all these amazing parts. I, in my mind, have created the ultimate preamp (TKD + DACT) and then the your wonderful tube circuit makes it everything I could ask for. It is all 'clean and pure' and then your GK-1 adds life and makes it special. Even the best circuit sounds bad with bad parts, so if you take the best circuit and add the best parts...

That is why I got the TKD, DACT, 7308, and Platinums (and soon Solens).

-West

[LM]

As an electronic solution then, I sure hope a voice control circuit could be added.  Asking my GK1 to set –40db would sure save a lot of effort in looking for the misplaced remote.

On a more serious note Hugh, is this the sort of circuitry underlying the Volume control on my Arcam AVR and perhaps for other brands that have moved on from the old motorised ALPS Blue Velvet or similar.  It certainly has a lovely feel and has a full logarithmic range display.  If so, are any such complete chip based volume control assemblies available commercially and at reasonable cost for GK1 fitment.

Edit:  Answering one of my own questions, Arcam appear to use the LM1972 from National Semiconductor.  Still haven't found any 'commercial' package or kit though but will keep looking.

[RonR]

Google turned up this one:



http://electronics.dantimax.dk/Kits/Selectors_-_attenuators/index.html - ChipVol1

They have a PGA2310 chip attenuator, a relay attenuator and motorised pots, with controllers with or without displays and remote control.

Cheers,

Ron

[MikeC]

The Welborne Labs relay system is good, but I would suggest that the relay quality is the pivotal factor here, and they may be cheapies.

The input switching on the Welborne kit is by relays. The attenuator has a series resistor in conjunction with an IC shunt attenuator. I'm not sure what the IC is as all markings have been scrubbed, but the total range is only 36dB (60 steps of 0.6dB each). One of my issues with the kit is the limited range - volume setting invariably not low enough on some highly compressed recordings, and can't get high enough with really dynamic recordings. As I said, I will be changing this at some stage, I just need to find something that will work within a casing optimised for remote switching and volume control. A rotary attenuator and/or input selector will require a completely new preamp case, so the TKD is disqualified, unless I can work out a stepper motor drive and display driver for it.

There are some kits becoming available that look like they may fit the bill for me. I really like the look of the Bent resistor (not Tx) kit, but the control board will not fit my case, and it is a bit pricey. The re-designed Twisted Pear should also work, and the Dantimax kits mentioned warrant some investigation.

I think I smell a new upgrade kit from Hugh here 

Mike

[LM]

They have a PGA2310 chip attenuator, a relay attenuator and motorised pots, with controllers with or without displays and remote control.

I missed the 2310 based kit at this site earlier so thanks Ron.  I must admit the appealing bit here (to me) is the fine adjustment provided by the 1db steps over the whole usable attenuation range.  No worries about being caught up in the relatively course steps of the DACT or similar in the -38db attenuation or greater region, even if such fine steps are of little practical use below about -60/70db,

Now the only problem I personally have is whether I should try going that way or go with a more straightforward DACT etc. install.  I feel I would have no problems assembling/soldering such a kit but would need advice in ensuring that I had ordered all of the other ancillary kit/components etc to make the package work, particularly with options like remote control.

[AKSA]

WOW!!       Thanks Ron, this is a pearler from Dantimax, and inexpensive, too!!

It's beautifully engineered from the pictures, and when you realise there are several hundred lines of machine code in the software, it's an absolute bargain!

Thanks West, very nice of you.....  you are very positive, I appreciate it.

Lyn, we shall speak tomorrow, but in truth you need no more than 1dB steps, and while a range of 40dB is acceptable, 50dB is better, as Mike has suggested. 

Not sure I can do one of these cheaper......  I like the Dantimax, I really do.   

Cheers,

Hugh

[AKSAphile]

Well...PGA 2310 is of course one of the best "digital" volume attenuators available today. It has many advantages over mechanical volume controls such as very high precision, crosstalk, almost unlimited life, but...as any other thing in EE the PGA 2310 has also some disadvantages compared to normal potentiometers.


We shouldn`t forget that PGA2310 is not a passive volume control, but an active, since it has opamp output buffer to lower output impedance. This is not a major concern, but the question is how good is actually this buffer opamp? BurrBrown (TI) never described in their datasheets which opamp they use. Some folks are sure that PGA 2310 has OPA2134 or OPA2604 on the output. These opamps are very good, but not the best to today standards. 

Few weeks ago I spend a lot of time upgrading my CD player output stage, testing and comparing various opamps from OPA2604, OPA2134 to the latest National LM4562. All the people that were with me during listening tests, confirm my opinion that LM4562 is SEVERAL times better than his BurrBrown competition. It sound is more like to  high class discrete design, not opamp. When I put opa 2134 /2604 back to chain I quickly realize that these chips are not totally transparent and they compress sound. I now threw away my opa2134 preamp gain stage and I simply use just 10k Alps pot between CD and AKSA and music just flows naturally. (with LF this should be probably much more evident )

I have not tested PGA 2310 yet, but with these listening experiences, I am afraid that PGA 2310 cannot be the purest volume control available. It has a lot of pros, but I think that for high end reproduction, where purity is essential, nothing can beat stepped or relay volume attenuator.


BTW: One of the most popular audiophile integrated amplifier - Lavardin, still uses blue ALPS... aa

[AKSA]

Thanks Andrej,

I can agree with all your points, but, as you say, the pros and cons of these devices must be weighed up.

I see it this way.  The spread of distortions introduced by premium quality electronics in the front end are below about 0.001%, and if we use tubes, the distortions they introduce in open loop are all low order, and high, around 1% minimum, and more likely two or three percent.  The Cathode Follower used on the GK1 introduces these levels, all H2 and H3, nothing higher, and these colorations, which are responsible for the 'sound' of the GK1 (detailed yet warm and engaging), dwarf the tiny but objectionable distortions of the preceding solid state stage.

We are reaching a point in the technology where people will not buy a product without a remote, and this requirement forces either a motor driven pot (expensive and flawed) or a digital pot in an IC (cheap and flawed)!

In the early nineties ABS was made available on some automobiles as sold as a safety feature.  Now, it is mandatory, and I see preamps going the same way - if they survive in their present format, that is.

I'm surprised the Lavardin uses the blue ALPS, I would have expected the black version, or even a DACT!!

I would like to see a Taiwanese company produce an all smd, laser soldered attentuator, using teflon substrate.  It should be durable, self lubricating, and sonically wonderful.  The problem is one of quantity and ROI.  It's not a difficult design problem.

Thanks for the contribution.  You give an EEs perspective, very interesting and informative.

Cheers,

Hugh

[AKSAphile]

Thank you Hugh for nice comments...


I 100% agree that in case of GK1 preamplifier, PGA 2310 should perfectly sit between solid state and output valve unity gain stage. PGA`s harmonic distortion is several times lower that typical valve and you got remote too. Ideally PGA should sit directly on GK1 board eliminating several signal wires. I would call this musical or joy solution.


In the case of passive preamp, I would rather put relay volume control there and you can have it remote controlled too...


About Lavardin...I was also surprised about use of standard blue Alps option, but here is the picture...

[LM]

Very interesting thoughts AKSAphile.

On another tack though, does anyone know where else one might obtain a TKD attenuator other than M Percy?

Going right back to the opening post, the 20KOhm version is no longer in stock at that site but I’m sure it has to be available somewhere and at a broadly equivalent price.  If so, I could optionally go the TKD way (perhaps at first without the Vishays and then update later with more budget).  But if I did so, I would ideally like to get the 20KOhm version.  Whilst the forum could undoubtedly help me choose resistors and setup etc to optimise use the 10KOhm version, I would prefer the simpler route.  This would apply particularly if I had to experiment with optimising the system gain.  At least I can always get 20KOhm and 25KOhm options from Goldpoint and DACT without any extra considerations applying and Goldpoint has some good tables on its site dealing with the resistor values needed for such adjustments.

[DSK]

It has been a long time but I seem to recall that the TKD was available directly from TKD in Japan.

Also, I recall that www.thlaudio.com had them.

It is the 25K version that I use with 250K Vishay S102's.

[AKSA]

Thanks Darren,

Great information!!  There are many pots offered by TKD, but for those interested in chasing it up, it's this model:

2P65S TKD Step Attenuator 
Round, stereo, Best rotation 315°, 25 x 25mm, 19.5 mm(L), 
Ohmic law type A( log)
Impedance: 10K, 20K, 50K, 100K, 250K & 500K
Installation triangle hole(center to center): 28 mm
Shaft: 6 mm dia.

For best results on the GK1 choose the 20K version.

Cheers,

Hugh

[LM]

Thanks also Darren.

Guess from looking at the TKD pdf, a 2P65CS20KOhm would be the complete description.  Interestingly, the S (or blank) is for 'click' or not and it seems to me a click would be useful.  Just wondering what you had Darren.  No price list at those sites so an enquiry is will be needed to ascertain the damage.