yup, you can get plenty of attenuation permuations, but the problem is getting the precise 1dB steps i require. i have yet to derive an algorithm to calculate the minimum # of resistors for the 70dB attenuation range i want.
i agree the PGA2310 is not ideal, though there are some decent-sounding preamps that use it. i heard it's pretty musical for what it is. i will have to get my hands on some AD CMOS switches and try using that for shunting, instead of relays... could sound pretty good.
i agree the PGA2310 is not ideal, though there are some decent-sounding preamps that use it. i heard it's pretty musical for what it is. i will have to get my hands on some AD CMOS switches and try using that for shunting, instead of relays... could sound pretty good.
dorkus said:
I picked up some of those switches. They switch better then relays, but I see no difference when they are not switching. They also cost more. You can get 2 free samples from analog.com
I see no reason to compromise on a volume path, The pga2310 isn't simple by any means, and isn't very clean. You can try all you want by adding filters and cleaning it up, but nothing will beat the simplicity and preserve the signal path as good as resistors and a relays, except for maybe that patent-pending passlabs volume control by Wayne.
--
Brian
the primary advantage of the CMOS switches is you can pack a lot of them into a small area, much more practical if you need a lot of switches and minimizes the signal path length as well. an SMT switch combined with SMT tantalum resistors could make a very sweet attenuator circuit using no more space than a few relays alone. they're also easier to control w/digital logic, don't need any current amps as you might with most relays. a mechanical relay has better performance than a CMOS switch, but when used for shunting CMOS is quite good (Nelson Pass has said so himself). some people would argue that the magnetic field from a relay also adversely affects the sound, but i have yet to confirm it and it's still a minor factor compared to a CMOS.
What about on-resiStance when using CMOS switches..
The SSM2402 has appx 60ohms, but is reputed to have very stable and signal independent on resitanse, which is not the case with many other CMOS switches.....
Here in Scandinavia the SSM2402 retails for appx 10USD, maybe less if you have good connections.....but this is also the price of any decent quality relay....????
The SSM2402 has appx 60ohms, but is reputed to have very stable and signal independent on resitanse, which is not the case with many other CMOS switches.....
Here in Scandinavia the SSM2402 retails for appx 10USD, maybe less if you have good connections.....but this is also the price of any decent quality relay....????
It'may be possible to use a simple digital potentiometer in input. By example the DS1802 is a real digital potentiometer used by Audio Research last preamplifier, IMHO, it's not a bad argument...
The principal weakness of this componet is it maximum signal input, +-5V max rating.
The principal weakness of this componet is it maximum signal input, +-5V max rating.
the on-resistance of a CMOS switch is normally high but when used in a grounding or virtual ground configuration, where there is zero potential at the leg of it, it is actually very low - on the order of tenths of an ohm i believe. this is why they work well as shunt elements or into the inverting input of an opamp. it is not a cheap chip, but there are 4 switches to a chip which makes it comparable to or cheaper than relays, and the space savings are considerable.
the DS1802 is a nice chip, better than most in that it's passive, but it does not have the attenuation steps i desire, and i would rather not have an IC series element. using switchable shunt elements with fixed series resistor is the route i prefer.
the DS1802 is a nice chip, better than most in that it's passive, but it does not have the attenuation steps i desire, and i would rather not have an IC series element. using switchable shunt elements with fixed series resistor is the route i prefer.
OK! I think it's a good way...
I think too, relays (good parts, of course) will be very suitable in this case; with 7 relays by channel we have 128 steps (I know you know 🙂 ) and the serial resistor is very low, one more time , there is a weakness: the relay noise...
I think too, relays (good parts, of course) will be very suitable in this case; with 7 relays by channel we have 128 steps (I know you know 🙂 ) and the serial resistor is very low, one more time , there is a weakness: the relay noise...
again, the problem with doing relay combinations is calculating the optimum resistor values for precise 1dB steps. not all possible permuations may provide useful attenuation. however, with some thought, it may be possible to come up with values such that at least some combinations are usable. i insist on precision however, as the individual channels will be trimmed in 1dB increments and this trimming must remain consistent at all volume levels. assumming we want 60dB of control range and +/-6dB of trimming, that means we need 1dB steps from 0dB to -72dB.
PGA2310 sounds ok but it is not great because it has internal opamps which limit its sound quality. it's actually an ok sounding chip, the editor of the magazine i write for reviewed the McCormack preamp which i believe uses it and said it sounded nice, but not true high-end transparent. our ideal attenuator should be entirely passive, leaving the active circuitry at our discretion. the Wolfson attenuator is passive, but it still has a lot of circuitry in series with the signal. PGA2310 is good if you want something flexible and don't need the very highest level lof transparency - its sound is equivalent to a decent-quality IC opamp.
OK, thank for your clear answer, I mean want you want ...
In reality, I never hear PGA and I never compare it with alps potentiometer . I am in late concerning Wolfson attenuator, is it an patent concerning attenutor, doesn't it ?
😕
In reality, I never hear PGA and I never compare it with alps potentiometer . I am in late concerning Wolfson attenuator, is it an patent concerning attenutor, doesn't it ?
😕
a good potentiometer like Alps Black Beauty will always sound more transparent than an IC like the PGA2310, particularly if you use the Alps as a shunt element only. passive is always cleaner than active, at least when the active is as complex as the PGA. however, it is not a truly valid comparison as the PGA has an output buffer, so the IC by itself constitutes a complete preamp... the Alps is passive so you'd need some sort of active stage with at least 12dB gain to be comparable. i believe Alps followed by a good IC like the AD8610 would be better than the PGA; if followed by a good discrete stage, even better. however, this would not give you remote control, unless you got a motorized pot, which does not have the precision of the PGA.
i heard the latest Rowland preamp, which uses the PGA, and it sounded fairly good. i think if you want something quick and dirty it's a very nice chip - it has decent output drive and some voltage gain, so it's essentially a preamp on a chip. keep in mind you'll need a microcontroller to implement it properly, but it's definitely a convenient way to get a decent-sounding digitally controlled preamp.
i heard the latest Rowland preamp, which uses the PGA, and it sounded fairly good. i think if you want something quick and dirty it's a very nice chip - it has decent output drive and some voltage gain, so it's essentially a preamp on a chip. keep in mind you'll need a microcontroller to implement it properly, but it's definitely a convenient way to get a decent-sounding digitally controlled preamp.
I know what you mean and I really like it, 🙂 it have been mine too.
Concerning the remote control, I let a note in the appropriate thread but I think I can, may be, help you concerning the µC prog.
Concerning the remote control, I let a note in the appropriate thread but I think I can, may be, help you concerning the µC prog.
I really like relay noise. I t sounds really high tech to me.Those little clicks only happen when you adjust the thing, and the sound isn't coming from the speakers. For me the noise is a plus. (I think Nelson agrees with me so there!!......) At the least I doubt most people will be bothered by it. The Pass gear has this "feature"
Variac said:
Agreed, I love it when you are sitting across the room, and change the volume with the remote, and you hear a faint ticking from the relays. I like the feedback. If it really bothered you, you could dampen the inside of the chassis that they are situated in, and I am sure that they wouldn't be too loud.
--
Brian
DAC attenuator
FYI i just got in for review an interesting multichannel preamp from MSB, their GOLD Multiple Volume Control (MVC). basically an 8 channel volume control/gain stage using a "DAC-based" attenuator. not sure exactly how it's implemented but i'll take a look inside later. yeah, it's ful of IC opamps and other nasties, but it actually sounds really, really, really good... extremely transparent in fact, with great speed and articulation. i'll compare it to my borbely JFET buffer and passive preamp soon but all things considered this is an impressive-sounding piece, and it has the attenuation flexibility i've been looking for. more info here:
http://www.msbtech.com/
FYI i just got in for review an interesting multichannel preamp from MSB, their GOLD Multiple Volume Control (MVC). basically an 8 channel volume control/gain stage using a "DAC-based" attenuator. not sure exactly how it's implemented but i'll take a look inside later. yeah, it's ful of IC opamps and other nasties, but it actually sounds really, really, really good... extremely transparent in fact, with great speed and articulation. i'll compare it to my borbely JFET buffer and passive preamp soon but all things considered this is an impressive-sounding piece, and it has the attenuation flexibility i've been looking for. more info here:
http://www.msbtech.com/
I'm deciding on whether to use a top notch motorised pot or a relay based attenuator. No silicon for me. The balance has to be between cost and performance but my budget is quite healthy.
Another compromise is between topology and size, the layout will affect perfomance, therefore all relays should be of the SIL variety (I don't know how reeds compare to regular relays - having never used them). This is to keep the signal and ground paths as linear and short as possible.
So, why not just use a ladder attenuator, and get the benefit of bulk buying for both relays and resistors. It may not be as perfect as a proper a fully switched resistor pairs attenuator but is easier to implement control etc..
I am torn between 1dB steps or 3dB, maily in an attempt to keep the signal paths short. What if there was a primary ladder stage with 8dB steps (say upto -64dB), followed by a buffer, then another ladder to implement 1dB steps (upto -7db) giving a total range from 0 to -71dB. Now my question is would the extra buffer do more harm than the extra long chain to give 1dB steps (cost of course will be better). Of course I intend the buffer to be a very high quality discrete design.
Unfortunately I cannot make my mind up and the proof through prototyping can only be done using the final PCBs so the cost hit will be full.
Any guidance or guesses as to the correct path?
Another compromise is between topology and size, the layout will affect perfomance, therefore all relays should be of the SIL variety (I don't know how reeds compare to regular relays - having never used them). This is to keep the signal and ground paths as linear and short as possible.
So, why not just use a ladder attenuator, and get the benefit of bulk buying for both relays and resistors. It may not be as perfect as a proper a fully switched resistor pairs attenuator but is easier to implement control etc..
I am torn between 1dB steps or 3dB, maily in an attempt to keep the signal paths short. What if there was a primary ladder stage with 8dB steps (say upto -64dB), followed by a buffer, then another ladder to implement 1dB steps (upto -7db) giving a total range from 0 to -71dB. Now my question is would the extra buffer do more harm than the extra long chain to give 1dB steps (cost of course will be better). Of course I intend the buffer to be a very high quality discrete design.
Unfortunately I cannot make my mind up and the proof through prototyping can only be done using the final PCBs so the cost hit will be full.
Any guidance or guesses as to the correct path?
MJ said:
I am also looking into building a great volume control with relays . I have the control worked out for it (microcontroller), but I haven't decided on the best way to use the relays for attenuation. My roommate and I are working together on this, and our first prototype is based on the Pass Labs Aleph 1.7 relay circuit:
http://www.passlabs.com/pdf/aleph/apserv17.pdf
It seems like it doesn't have enough attenuation, and the steps are on a linear scale, as to say that it takes 3-4 db steps at -48db and when it gets around 0, it changes in .1 dB steps.
I have resources to get pcbs done here for myself at college for prototyping for no cost, so I am planning on trying out a few different designs, then if I find one that works great, I can get real pcbs made.
As for the 1db steps, how many relays were you looking to use. Any input into this is appreciated.
--
Brian
I looked at the aleph circuit, but was not too impressed, I wanted a log circuit so as to ease the control side. I could use a PIC or other Micro-Controlled but it is just a big head ache, as to get true 1dB steps down to -72dB the attenuator stage actually needs to be at least 12bits not 8bits as in the aleph and that does not help.
With the twin ladders there would only be 16 relays and 32 resistors per channel so cost is reduced. A pure ladder would require 72 relays and 144 resisitors per cannel. The control logic is simple up/down counting with no need for linear to log lookup. Zipper noise is virtually eliminated by making the relays arrangement make before break - this can be done using a single RC chain parallel with each relay.
The extra cost is in the buffer stage, use something like the Borberly buffer or a simple emitter follower buffer with very high quality components (or for experimentation a simple opamp buffer using an OPA627 would be superb).
For control I intend to use a simple optical encoder. The control can be setup with simple discrete logic, same as for the source switching or use a uP. Remote control can then be simply applied as well.
With the twin ladders there would only be 16 relays and 32 resistors per channel so cost is reduced. A pure ladder would require 72 relays and 144 resisitors per cannel. The control logic is simple up/down counting with no need for linear to log lookup. Zipper noise is virtually eliminated by making the relays arrangement make before break - this can be done using a single RC chain parallel with each relay.
The extra cost is in the buffer stage, use something like the Borberly buffer or a simple emitter follower buffer with very high quality components (or for experimentation a simple opamp buffer using an OPA627 would be superb).
For control I intend to use a simple optical encoder. The control can be setup with simple discrete logic, same as for the source switching or use a uP. Remote control can then be simply applied as well.
MJ said:
Is this the dual ladder that you are speaking of?
http://www.diyaudio.com/forums/showthread.php?postid=19490#post19490
ftorres made his with 16 relays per channel.
It might be interesting to make it possible to control with just an optical decoder. I will look into this.
--
Brian
- Status
- Not open for further replies.