I think I have a fairly low cost method for full implementation of the Audio Palette controls. Multiplexed CMOS switches.
Two 16 channel switches addresses 32 possible connections - so greater than 29 in several frequency and volume controls. The remainder of the controls can be addressed by a 16 channel and an 8 channel.
The Vishay DG506 (16) and DG408(8) can operate from +/-22V, and accept an audio signal equal to that voltage. They are simply addressed (so no microprocessors) and have an enable pin. So something like a center biassed toggle and some simple address decoding would advance the control one step either way by toggling. The on/off ratio is 85dB, and since they feed into the high impedance of unity gain buffers, variation of on resistance with voltage is not an issue.
Something like a ring of (low-brightness) LED's would indicate boost/cut setting or volume level.
The overall cost of all the multiplexers needed (48 x 16 and 16 x 8) including tax is UKP274. Plus the resistors of course, of which there are close to 500. Choosing the Dale RN60D or RN55D that Cello used adds a further UKP220. More sensible metal film resistors would be equally good at a fraction of the cost,
More expensive than rotary potentiometers for sure. But an order of magnitude cheaper than using real mechanical switches.
Two 16 channel switches addresses 32 possible connections - so greater than 29 in several frequency and volume controls. The remainder of the controls can be addressed by a 16 channel and an 8 channel.
The Vishay DG506 (16) and DG408(8) can operate from +/-22V, and accept an audio signal equal to that voltage. They are simply addressed (so no microprocessors) and have an enable pin. So something like a center biassed toggle and some simple address decoding would advance the control one step either way by toggling. The on/off ratio is 85dB, and since they feed into the high impedance of unity gain buffers, variation of on resistance with voltage is not an issue.
Something like a ring of (low-brightness) LED's would indicate boost/cut setting or volume level.
The overall cost of all the multiplexers needed (48 x 16 and 16 x 8) including tax is UKP274. Plus the resistors of course, of which there are close to 500. Choosing the Dale RN60D or RN55D that Cello used adds a further UKP220. More sensible metal film resistors would be equally good at a fraction of the cost,
More expensive than rotary potentiometers for sure. But an order of magnitude cheaper than using real mechanical switches.
Sawyers,
Sais this before but based on my own experience, I don't feel you HAVE TO USE stepped switches (or whatever). The dual Alpha pots (if you can still get them) work and "sound" just fine. Can get "repeatability" to around 1/2 dB which again is all I need. Surplus duals may be available if you look around - maybe guitar electronics suppliers.
Obtaining thru-hole parts today is CRAZY. Can't get many RN resistor values anymore and the prices for what remain is going thru the roof. I can envisage NO THRU-HOLE resistors in 3 to 5 years! Multiturn trimers are unobtainable in certain values! Fortunately, in my efforts to keep certain Cello designs going I've picked up a lot of the "original" semiconductors - otherwise forger them today.
Sais this before but based on my own experience, I don't feel you HAVE TO USE stepped switches (or whatever). The dual Alpha pots (if you can still get them) work and "sound" just fine. Can get "repeatability" to around 1/2 dB which again is all I need. Surplus duals may be available if you look around - maybe guitar electronics suppliers.
Obtaining thru-hole parts today is CRAZY. Can't get many RN resistor values anymore and the prices for what remain is going thru the roof. I can envisage NO THRU-HOLE resistors in 3 to 5 years! Multiturn trimers are unobtainable in certain values! Fortunately, in my efforts to keep certain Cello designs going I've picked up a lot of the "original" semiconductors - otherwise forger them today.
Indeed - you only have to look at the parts count at vendors for surface mount and through hole to see the inevitable trend. All of my professional designs in the last three years plus have been entirely surface mount. I have a large working distance binocular microscope and a Metcal GT90 with enough bits for fine work, but also larger ones for soldering the tab on power devices to copper area heatsinking.
The Viola stepped attenuator - a Colangelo rework of his Palette ones - used surface mount resistors. But surface mount metal film resistors are not cheap either.
The Viola stepped attenuator - a Colangelo rework of his Palette ones - used surface mount resistors. But surface mount metal film resistors are not cheap either.
They are available in assortment kits, on eBay, for low prices. Kits generally contain five or ten pieces each, of resistance 50, 100, 200, 500, 1K, 2K, 5K, 10K, 50K, 100K, 200K, 500K, 1MEG . Have a look.
Be aware of the nonlinearity of these switches. They can be used in ways that negate this probem, but it usually means to design around the switch and it's peculiarities and not to use them instead of mechanical switch.
Probably more expensive than the 47 Position ELMA switches as well, once you take into account assembly cost and programming cost.
Seriously, at that point why no go digital?
Thor
Actually, they are, by the reel. Very cheap. I specifiy even in very cheap products for the Audio Path. Same for C0G capacitors which are pretty close to Teflon, subjectively and objectively.
Just buy a engineering set of MELF resistors and/or thin film resistors and C0G capacitors and you are set for DIY for a long time.
Thor
An engineering set of MELF and thin film E96 in 4 decades by the reel? A long time for me is running out at 67 - I'd never use a tiny fraction before I am pushing up daisies. I buy as needed in hundred off - that gives me enough stock for any projects (DIY or paid for) in practical quantities.
And going digital - well that is a skill I never (bothered or needed) to learn. In projects - like in the Mercury Imaging X-ray Spectrometer currently inbound to Mercury, I had a whole team (in Finland) doing the software, and a System Engineer (in the UK) to specify it. So in seriously big, and not so big, projects software is subcontractable, and therefore no reason to learn.
Even if I was responsible for a piece of commercial audio gear (and at one stage I was CTO of Wharfedale) I subcontracted any software.
Even if I was responsible for a piece of commercial audio gear (and at one stage I was CTO of Wharfedale) I subcontracted any software.
Could not agree more with Thorsten.
I use a lot of Beyschlag MELF, Susumu thin films, Panasonic ECHU, and Murata C0G these days.
Especially when dissipation is low.
And not at all difficult to had solder, if PCB designed accordingly.
Many advantages in PCB layout compared to through-holes.
https://www.diyaudio.com/community/...e-cen-iv-with-fixed-rails.389776/post-7208234
Patrick
I use a lot of Beyschlag MELF, Susumu thin films, Panasonic ECHU, and Murata C0G these days.
Especially when dissipation is low.
And not at all difficult to had solder, if PCB designed accordingly.
Many advantages in PCB layout compared to through-holes.
https://www.diyaudio.com/community/...e-cen-iv-with-fixed-rails.389776/post-7208234
Patrick
For DSP there is actually no need for doing software, just base the design on Analog Device DSPs and use SigmaStudio, which is a fully graphical block based programming tool.
I can fully understand and appreciate the nostalgia of this project like the Cello, and as you know I also spend too many hours trying to design a discrete opamp, but with a chip like ADAU1701 and it's bigger brothers and sisters, there is no reason not to go digital 😉
As for SMD, as Patrick says there is a great number of advantages, and today I get all PCBs made in China by JLCPCB, and get them to mount most parts (if they are in stock) and then hand solder the rest.
Yes you don't really get anything better than C0G/NP0 for caps (but they of course only comes in small values) and chip resistors are more or less1% as standard
I can fully understand and appreciate the nostalgia of this project like the Cello, and as you know I also spend too many hours trying to design a discrete opamp, but with a chip like ADAU1701 and it's bigger brothers and sisters, there is no reason not to go digital 😉
As for SMD, as Patrick says there is a great number of advantages, and today I get all PCBs made in China by JLCPCB, and get them to mount most parts (if they are in stock) and then hand solder the rest.
Yes you don't really get anything better than C0G/NP0 for caps (but they of course only comes in small values) and chip resistors are more or less1% as standard
Not a chance! Anyway that was three decades ago. Pass the walking frame....
Astonishingly for the date, we developed and launched a digital preamp. A-D, all functions in digital (volume, input select, tone controls etc) then D-A. I have a publicity photo somewhere; I'll look it out and post it. At that point 30 years ago Wharfedale owned Cambridge Audio and the Leak brand. So we used Cambridge Audio designers and a decidedly strange external designer (he produced the digital preamp prototype nailed on a wooden board!), and launched the (production engineered) digital preamp and monoblock power amps under the Leak brand name. There was a CD transport in development when Wharfedale imploded as a result of the 1990 recession. I beat a hasty retreat to Oxford Instruments.
The genius Cambridge Audio analog designer was a recently graduated Steve Sells, now chief engineer at Naim. The Naim Statement was his baby.
Last edited:
Engineering sets are usually 100pcs per value. They are often subsidised by the manufacturer if they are from the manufacturer.
Thor
I took an engineering set to be full reels, so 2,500 to 10,000 parts. 100 off is much more manageable, and comparable in number to my through hole stock.
No need.
Many excellent sounding sounding software EQ's exist for Studio's.
Setting up a 6-way parametric (say MAXX Audio Renaissance 6-Way parametric) to act a like a Palette is not hard.
Thor
I've been (because like Everest, it is there) working out how the input attenuator works. Before publishing the resistor chain values, the way it works is shown in the attachment.
In Burwen's original design on the left, he shows in the schematic a 2k switched attenuator with a 71.5 ohm resistor from the bottom of the attenuator to ground. Calculation shows that the maximum attenuation is -29dB, consistent with the Audio Palette. The problem with this arrangement is that it calls for a number of sub-1 ohm resistors..
The Audio Palette uses a different arrangement, shown in the block diagram that Charles published a while back. The resistor chain is split into two sections, with the centre tap grounded with a 402R resistor, and the bottom of the chain through a 1k15 resistor to ground. The centre tap is at an attenuation of -14.5dB (half way to -29dB). With some fiddling around with values, the upper chain is 1k625 and the bottom chain 6k25. It is easy to show that the combined arrangement has a terminal resistance of 2k (the same value as Burwen's prototype). But the cunning use of 405 ohms from centre tap to ground means that the minimum resistor value is 23R2. It also correctly produces a maximum attenuation of -29dB.
In Burwen's original design on the left, he shows in the schematic a 2k switched attenuator with a 71.5 ohm resistor from the bottom of the attenuator to ground. Calculation shows that the maximum attenuation is -29dB, consistent with the Audio Palette. The problem with this arrangement is that it calls for a number of sub-1 ohm resistors..
The Audio Palette uses a different arrangement, shown in the block diagram that Charles published a while back. The resistor chain is split into two sections, with the centre tap grounded with a 402R resistor, and the bottom of the chain through a 1k15 resistor to ground. The centre tap is at an attenuation of -14.5dB (half way to -29dB). With some fiddling around with values, the upper chain is 1k625 and the bottom chain 6k25. It is easy to show that the combined arrangement has a terminal resistance of 2k (the same value as Burwen's prototype). But the cunning use of 405 ohms from centre tap to ground means that the minimum resistor value is 23R2. It also correctly produces a maximum attenuation of -29dB.
Attachments
And this is the resistor values. This is not the official values; however it meets the criterion of half dB steps all the way to -29dB with a maximum error of +/- 0.01dB - assuming all the resistors have perfect values (so zero tolerance).
To me that is an "analogue preamp" as it is analogue in and out.
A "digital preamp" should have only digital in & out... Semantics I l know.
Unless using hideously expensive SOTA multibit ADC (SAR ADC using PCM63 as DAC) & DAC (PCM63) with high levels of oversampling I have my doubts on the overall sound quality.
A modern digital preamp would actually make some sense now. App on Tablet or Phone for Palette type EQ. Together with digital input Class D Amplifiers that have programmable gain & PSU this could offer a lot of useful functions.
Real physiologically correct volume control.
Auto Level adjustment of incoming audio to (say) -20LUFS for 85dB SPL.
Peak expansion for overly compressed audio.
Low level expansion for overly compressed audio.
Remastering EQ (digital Palette).
Speaker/Room EQ.
Spacial correction for Speakers and Headphones.
DAC and Headphone Amplifier may be included.
Thor
There is certainly a whole range of possibilities Thor. My good friend and mentor Gordon Edge (RIP) challenged thinking about technology, business and spin out companies. He was the founder of Cambridge Audio in the 1960's - but was very active in pushing for exactly the technology you are talking about - but thirty odd years ago - and entirely digital environment. The killer is always at the end of the chain - the loudspeaker/headphone - which is always analog. If a truly "digital loudspeaker" were possible, that would close the whole digital chain.
Alas the technology was not available at that time to implement Gordon's vision, and indeed even the Wharfedale digital preamp in the very early 90's was the bleeding edge of what was possible - analog in and out and a digital filling to the analog sandwich.
Indeed this is the way that current product offerings work, for example https://www.minidsp.com/products/minidsp-in-a-box/minidsp-2x4-hd .
Alas the technology was not available at that time to implement Gordon's vision, and indeed even the Wharfedale digital preamp in the very early 90's was the bleeding edge of what was possible - analog in and out and a digital filling to the analog sandwich.
Indeed this is the way that current product offerings work, for example https://www.minidsp.com/products/minidsp-in-a-box/minidsp-2x4-hd .