Bobken said:
Bobken:
The 400 Ohm resistors were to lower the power dissipation in the regulators. Perhaps no necessary but a good precaution since the unit was designed to work from 90VAC to 135 VAC and the regulators were a little stressed at the max.
The ferrite beads were to reduce sensitivity to emi/rfi like the TV station problem I mentioned a while ago. I did not like them much either but they did help with RFI.
Tokyo Ko-on also makes (made?) the silver contact switches. They are solid silver, not silver over copper/brass and I'm sure that is part of the magic. One company, bought by Altec, used fine silver (the others use coin silver) to get even better performance. The good switches are not combined on a PCB- the connections will be unreliable so you are stuck wiring them
The reason I never used the two-gang switch was the challenge of soldering 29 Vishay resistors in such a confined space.
I went through the detent spring tweak on the Tech-labs switches many times.
You could make a stepper (noisy) or a servo motor to drive the switch but the mechanical design effort will get pretty heavy. The Bent solution looks good and addresses the concept well. The 23 step switch is a limitation, we used to get flak over the 29 step switch we used. But the 45 step switch is 4" on a side and doesn't fit. Also I have come to detest digital readouts for controls.
I would use relays in a new remote controlled design. And probably for a commercial design use the conventional miniature frame relays, which can be had with silver contacts that might sound better than gold. The classic complaint about this approach is the click noise of the relays. Levinson built a Jfet switch array to do the same. A huge amount of work and compromise to chase the noise away.
john curl said:
I have found that any unbalanced currents that end up returning through ground just invite non-stop problems. Since any current on a ground link creates a potential that is added to any ground reference I find it best to avoid the ground except for the specific input, output and feedback return. My supplies use ground as a reference voltage and not for dc current return. Fortunately Audio is AC w/o DC as long as everything is working right.
john curl said:
John,
I was not saying that you were wrong. I was just curious as to why you decided to use the dual bridge/virtual ground. Your use of it peaked my curiousity. I've always used the centertap referenced supply, your response was basic stuff I know.
Sorry if it came across differently.
Mike
1audio said:
Hi Demian,
Thanks for the explanation regarding the resistors across the DMC-10's regulators, and, although I haven't used that preamp for some years, it was always a favourite of mine and I have never parted with it.
Having now regularly used the prototype relay-switched attenuator I designed for a UK manufacturer around a year ago, I have appreciated some benefits from this device during longer-term domestic use.
Technically the absolutely constant 10k input impedance is an improvement over the Shallco 'L Pad' versions, and the output impedance is also reasonably similar across all settings except for slight 'steps' at -16dB & -32dB which don't concern me much.
More importantly, I find that with accurate 1dB steps down to -63dB, this is a considerable advantage over the more limited number of steps available with Shallcos. Perhaps with the 45 step Shallco you mentioned, this would be less of a problem but I now find that setting to 1dB is very useful for maximum enjoyment of various different sources, and I would probably now miss that benefit.
For some recordings, I would even like to have an intermediate setting of a 0.5dB step as one setting is marginally subdued for my tastes, but 1 dB higher output seems a little over the top!
This is clearly impractical, but I guess a 45 step device could be arranged with very close steps in the 'more-used' range, and at -32dB and beyond, maybe 3dB or similar steps would be adequate. This would doubtless be an ideal compromise for my personal listening, but for commercial designs one never knows for certain quite where the 'normal usage' region is likely to be, of course.
As John has implied, seemingly there isn't (yet!) an ideal solution for audio attenuation requirements, but some sort of compromise between the approaches we have touched on is probably about the best one can expect, and either approach betters any chip type attenuators I have experienced.
On the matter of Shallcos (and the similarly-constructed Blore Edwards) one of the most important issues here is the overall 'integrity' of the signal-path through these 'good-sounding' switches, by which I mean the least detrimental to the signal. The input/output connections are soldered directly to the rear of a small single solid piece of silver, the other end of which is the actual stationary contact. The wiper arrangements are similarly 'minimal', and there are no dissimilar metals riveted to each other nor convoluted signal-paths which one finds in most relays which I have inspected internally.
No plating/cladding of one metal upon another has ever improved sonics in my experience, but such surface treatments are often needed for other reasons like wear or corrosion resistance. This being the case, a solid silver contact will sound better (in my opinion) and pure silver is likely to be better than an alloy like Sterilng Silver or coin silver, but pure silver is very soft (I know as I started my career nearly 50 yrs ago as a manufacturing jeweller/silversmith). Accordingly, whilst I guess pure silver contacts will 'sound' better, their longevity will inevitably be curtailed, but I have never made any listening comparisons between pure or alloyed-silver contacts.
It is always dangerous (at least on this Forum!) to generalise, but I believe that an alloy of metals where they are effectively intermixed throughout their composition will have less adverse affect on the sound than 'clad' metals, which have a harder surface layer of different metal applied to them, perhaps for wear-resistance. I believe that these dissimilar metal 'interfaces' are not good for the sound, most particularly when they are merely rivetted together as is often the case in relays. I have frequently seen signs of discoloration/corrosion when separating a used relay's harder metal switching contacts from the necessarily springy (usually copper alloy) strip which supports the moving contact, and this unwanted 'barrier' is not apparent with identical unused relays.
Clearly some adverse reaction causes this discolouration during usage, possibly through galvanic action or whatever with a current passing through these interfaces of different metals, and the resulting non-linear resistance is harmful to good sound.
I also initially found that the relay clicks were annoying and spent a lot of development time on minimising these as a result. However, the manufacturer involved wasn't at all bothered by this aspect and no-one else has commented adversely so far, and to be honest I no longer notice this at all myself. 'Normal' listeners don't usually spend much time going up and down the scale of attenuation during listening sessions, and I think any inital adverse reaction soon wanes with familiarity, and it only surfaces because it is something different from what they are accustomed to.
Regards,
Hi Bobken,
a Shallco fitted with all nude Vishay TX2352 resistors in a ladder configuration, is surely the creme de la creme for stepped attenuators. I plan for such an attenuator in my next balanced fully complementary stereo line amp powered by super shunt regulators for each of the four amps.
TX2352 in all its glory, now there is a new bulk metal foil resistor from both Vishay and Texas Coomponents, the VAR resp. the TX2575 which both are made with Z-foil instead of K-foil.
Have you maybe tested these new resistors?
Sigurd
a Shallco fitted with all nude Vishay TX2352 resistors in a ladder configuration, is surely the creme de la creme for stepped attenuators. I plan for such an attenuator in my next balanced fully complementary stereo line amp powered by super shunt regulators for each of the four amps.
TX2352 in all its glory, now there is a new bulk metal foil resistor from both Vishay and Texas Coomponents, the VAR resp. the TX2575 which both are made with Z-foil instead of K-foil.
Have you maybe tested these new resistors?
Sigurd
Bobken said:
Hi Sigurd,
No, regrettably I haven't yet tested these new naked Vishays although I am aware of their existance, of course.
My sonically 'best' Shallco-based versions were made with VSRJ Vishays (identical to S102s except for different lead-spacing), and the more-recent commercial relay attenuator used TX2352s.
Absolute accuracy with 1dB steps was a pre-requisite for this version, and where else can you obtain values like 81k9548 (the first value I just picked out of my bin) with 0.1% tol., other than in really vast quantities, if at all?
I will try the latest 'Nudes' at some future date, but unless things have changed recently there was a problem with minimum quantities of these latest resistors, and one attenuator of this (3-way stereo) design requires 26 different values which makes any trials a bit costly.
Perhaps fortunately in this instance, balanced connections were not required, but I am certain that your suggested 'recipe' would be very hard to better!
Regards,
No, regrettably I haven't yet tested these new naked Vishays although I am aware of their existance, of course.
My sonically 'best' Shallco-based versions were made with VSRJ Vishays (identical to S102s except for different lead-spacing), and the more-recent commercial relay attenuator used TX2352s.
Absolute accuracy with 1dB steps was a pre-requisite for this version, and where else can you obtain values like 81k9548 (the first value I just picked out of my bin) with 0.1% tol., other than in really vast quantities, if at all?
I will try the latest 'Nudes' at some future date, but unless things have changed recently there was a problem with minimum quantities of these latest resistors, and one attenuator of this (3-way stereo) design requires 26 different values which makes any trials a bit costly.
Perhaps fortunately in this instance, balanced connections were not required, but I am certain that your suggested 'recipe' would be very hard to better!
Regards,
There are several ways to connect two controls.
Obviously the gear thing works.
Another way that has very little backlash is to use a ribbon of metal, usually copper or stainless that wraps around two flat surface "gears". One merely has the knob connected to one shaft - and the other is not connected except via the ribbon. A bit of a slack take up slot for the "follower" gear on the mount for the control it is on and you don't have to be ultra precise about the distances or tolerances of the ribbon.
Fwiw, that is a method that I have seen used in USA military gear, as well as in ham rigs like the Heathkit Apache (behind the front panel).
As far as remote control? If you can handle a stepwise "Up-Down" means, then you want to find out how the old "clicker" type TV remote controls worked, and duplicate the basic methodology, but updated. Those were used on "turret" type TV tuners. Bet those might make nifty audio switches, parallel the contacts for very low contact resistance? Good up to UHF??
The other way is to use a stepper motor. Remove the detents and you can use a very small stepper with a rotary encoder. But I don't want any uprocs in my audio gear, I don't think I do anyhow... ?
_-_-bear
PS. glad someone here can afford bins of Vishays, naked be they or not!
PPS. Almost forgot, on many of these high end Shallco or similar switches, the shaft is a 0.250" stainless steel rod with a few holes fitted for tapered pins. One can, and I have removed these rods, machined up new longer shafts, and ganged two separate attenuators one behind the other... in fact Shallco, Daven and other manufacturers have examples of units made with a shaft sticking out the back, precisely for this purpose. A flex or non flex shaft coupling does the job here.
Obviously the gear thing works.
Another way that has very little backlash is to use a ribbon of metal, usually copper or stainless that wraps around two flat surface "gears". One merely has the knob connected to one shaft - and the other is not connected except via the ribbon. A bit of a slack take up slot for the "follower" gear on the mount for the control it is on and you don't have to be ultra precise about the distances or tolerances of the ribbon.
Fwiw, that is a method that I have seen used in USA military gear, as well as in ham rigs like the Heathkit Apache (behind the front panel).
As far as remote control? If you can handle a stepwise "Up-Down" means, then you want to find out how the old "clicker" type TV remote controls worked, and duplicate the basic methodology, but updated. Those were used on "turret" type TV tuners. Bet those might make nifty audio switches, parallel the contacts for very low contact resistance? Good up to UHF??
The other way is to use a stepper motor. Remove the detents and you can use a very small stepper with a rotary encoder. But I don't want any uprocs in my audio gear, I don't think I do anyhow... ?
_-_-bear
PS. glad someone here can afford bins of Vishays, naked be they or not!
PPS. Almost forgot, on many of these high end Shallco or similar switches, the shaft is a 0.250" stainless steel rod with a few holes fitted for tapered pins. One can, and I have removed these rods, machined up new longer shafts, and ganged two separate attenuators one behind the other... in fact Shallco, Daven and other manufacturers have examples of units made with a shaft sticking out the back, precisely for this purpose. A flex or non flex shaft coupling does the job here.
Once again, I am impressed by the latest inputs on making SOTA volume controls for audio equipment. Many here have gone even farther than we did with the BLOWTORCH, as we used modified TKD pots that are pretty darn good, certainly expensive, but still seem a slight compromise to the pure Shallco, hand wired switch. It is refreshing that so many, located around the world, can still hear and appreciate the difference.
This is why I think that this thread is important.
This is why I think that this thread is important.
Bobken,
the TX2352s can be bought in one-quantaty, and I bought some TX2575 a month ago to test them out, but as you say the TX2575 are usually sold in larger quantaties. 25 pcs minimum is the info I got, but maybe that have changed as I could buy just a handfull.
I have this huge bespoke solid brass 70mm diam volume knob which turns an ELMA switch like a varm knife cuts in butter. That know might turn a Shallco in a smooth way.
Regarding remote controlling Shallcos. That is what Ayre has done in their new preamp. Using a belt drive. Very elegant!
One nice thing about realis in remote controlled attenuators, is that one can use only a few relais and a few resistors to get a lot of different volume levels. Drawback is of course many relais and many resistors and many joints in the signal path. The budget sets the limit here.
Some relais make a lot of noise in my experience. However, this can be dampened a bit with a good acoustic absorber. How the attenuation is reached (how many relais are needed to switch for each setting) also affects the noise.
I think that Nelson Pas abandoned relay attenuators as his customers did not fancy the noise.
What relais did you use?
Sigurd
the TX2352s can be bought in one-quantaty, and I bought some TX2575 a month ago to test them out, but as you say the TX2575 are usually sold in larger quantaties. 25 pcs minimum is the info I got, but maybe that have changed as I could buy just a handfull.
I have this huge bespoke solid brass 70mm diam volume knob which turns an ELMA switch like a varm knife cuts in butter. That know might turn a Shallco in a smooth way.
Regarding remote controlling Shallcos. That is what Ayre has done in their new preamp. Using a belt drive. Very elegant!
One nice thing about realis in remote controlled attenuators, is that one can use only a few relais and a few resistors to get a lot of different volume levels. Drawback is of course many relais and many resistors and many joints in the signal path. The budget sets the limit here.
Some relais make a lot of noise in my experience. However, this can be dampened a bit with a good acoustic absorber. How the attenuation is reached (how many relais are needed to switch for each setting) also affects the noise.
I think that Nelson Pas abandoned relay attenuators as his customers did not fancy the noise.
What relais did you use?
Sigurd
Bobken said:
john curl said:
Sigurd Ruschkow said:
Hi Sigurd,
I will try the latest 'Nudes' as already mentioned, but with 26 values required for one attenuator, 26 x 25 x approx $7 (for TX2352s) does amount to a substantial outlay for further experimentation, and even if they will make me a "handful" of each value, it won't be cheap as I guess the 2575s are rather more costly per piece, too.
The company I did this design for are quite satisfied with the existing results, so they are unlikely to help out financially here either, and I still have some supplies of the original TX2352s to hand.
I am reminded of being restricted by an NDA which precludes me from divulging the relay I finally ended up using here, but I found several of the miniature types (20mm x 10mm footprint) which were 'acceptable'. I would not go for anything smaller like sub-miniature (nor low-profile) as with smaller packages the coils will inevitably be physically closer to the contacts and signal paths, which I don't consider to be good for the best (untainted) sound.
Different makers use alternative internal mechanical arrangements/layouts and I look for ones which have the greatest distance between the coils and signal areas. One high-isolation type was particularly good in this respect (although more mechanically noisy), and don't be tempted to use a poorly-filtered PS for energising these coils. It may not be intuitive, as there should not (usually) be any direct electrical connection between the coil's circuits and the signal areas, but I have experienced unwanted effects when using a relatively poor supply which would otherwise be perfectly acceptable for energising relays. These problems were fully eradicated when I went over to a well-smoothed DC source, so there is clearly some kind of 'bleedthrough' possible in relays, perhaps due to an internally radiated field from the coils. It goes without saying that any coil-energising PS source and grounds should be entirely separated from the signal PS and grounds, for similar reasons.
There is little to beat making certain there is adequate distance between these regions to ensure there is no possible 'contamination', and I have seen other circuits where for easy PCB layout the tracks from the coils have run between, and far too close to IMO, the tracks and solder-pads carrying signals, so care in PCB layout is important here too. A lot of good achieved elsewhere in the circuit can be very easily countered by careless PCB layouts, unfortunately, and all of these details are important if you wish to achieve the finest possible sonic results.
Also, it is vital to go for bifurcated contacts, which I believe most suitable relays will probably have anyway, for the following reason.
Due to customary tolerances in manufacture (nothing is perfect in life, nor in audio!) with every bifurcated-contact relay I have inspected internally, one contact will 'lead' the other, albeit by a very small margin. The effect of this is that the contacts will make and break at different times (maybe only by a few milli-seconds, but this helps here) which is an 'unintentional bonus' to us perfectionists, which I am very happy to take advantage of.
What happens in practice is that the leading contact will always take the brunt of the electrical/thermal shocks as it mates first, and (even better, so this is a real "double-whammy") it also breaks last, and this is usully when the contact deteroration is most likely to take place. So, whatever spark-erosion or other deterioration might occur will be heavily biased towards (if not completely - as in most cases) this one contact, leaving its 'partner' unscathed (or certainly less adversely affected) during its usable lifetime. The one remaining 'good' contact can then still pass the delicate audio signals with the least harm, and inspecting used relays with bifurcated contacts I have seen this one-sided deterioration result several times.
Apologies to John for hogging his thread, but I hope that the above is of some value/interest to others.
Regards,
- Status
- Not open for further replies.