Stupid off subject question here. I was getting updates in my email whenever a new posting was happening and now they are not showing up. I am still showing that I am subscribed to the threads but the messages have stopped. What have I done. Perhaps it is that big capacitor I put on the CAT 5 cable to block spam, just kidding. How do I fix this, I want to keep up and I have to look to see if anything new has been posted?
Steven
Steven
Actually, and perhaps I've missed something here, but..
Relative to the wiper the 10K resistor has the thevenin equivalent resistance of a 2.5K resistor worst case whereas the 1K pot would be 250 ohms worst case. Johnson noise from thevenin equivalent source resistance at the pot wiper will be about 10dB worse for the 10K pot than for the 1K pot. Now if you wired a pair of ~560 ohm resistors across the pot from one end to wiper, and wiper to other end the situation would be different, the noise would be very close to what you would expect with a 1K pot.
Note that these comments wrt noise apply to the example mentioned in a previous post where a single 1K resistor shunts the 10K pot as well as in the case of an unshunted pot. In the case where a near ideal voltage source (any op-amp really) drives the pot and 1K resistor, the 1K resistor has NO effect on the noise seen at the wiper for the reasons stated above.
Now if one end of the resistor (edit) is connected to the wiper the situation will be different, but the taper resulting from such a connection would not be the most useful I could think of.
Last edited:
Still dodging the question
You didn't build the preamp and you aren't prepared to offer any justification for your assertion that I have modified the design and therefore the sound quality - so what is your point then? Should we just take your word for it?
Seems a bit strange to be talking about how to use cheap nasty carbon pots in a design aiming for the highest performance. Even if you can find one in the batch that has adequate mechanical alignment between the two wipers. You also need to check the power rating for this design (as stated in the article).
Even the Omeg pots (while conductive plastic and electrically better) are not up to scratch IMO as the "feel" is not smooth enough - they are all plastic and with all of the ones I have there is some uneven friction when you turn them. The centre detents (used on some of them) are not well defined either.
For a design you build yourself it's worth spending the money on good pots.
Even the Omeg pots (while conductive plastic and electrically better) are not up to scratch IMO as the "feel" is not smooth enough - they are all plastic and with all of the ones I have there is some uneven friction when you turn them. The centre detents (used on some of them) are not well defined either.
For a design you build yourself it's worth spending the money on good pots.
What steps would you use, 1db or less per step and what would the values have to be to do that? How close does the precision of the resistors have to be 1% or less?
If I were building one of these now I'd say that I'd want 0.5dB steps at the loud end, 1dB in the middle and 2dB at the quiet end. A 64dB range plus mute I'd find sufficient so long as individual preamp inputs were gain normalized to make up for differing output levels. 1% resistors are fine for the bigger steps but I'd go for 0.1% to make up the 0.5dB sized ones.
A long time ago I built a 128dB 1dB step attenuator using reed relays. Clicks were a constant niggle.
I have not seen the schematic, but I would caution against a switched attenuator and how you plan to configure the resistances since transition between positions could be full OLG. Check whether you need a make-before-break or a break-before-make switch (of course relays would need the appropriate code).
Last edited:
Good point Nico - with relays you need an arrangement that reverts to maximum attenuation between settings (ie during the brief intervals when the relays are all off including power up).
Besides it would be silly to use a stepped attenuator with the Baxandall active gain control since the whole idea behind it is to generate a log gain curve using a linear pot. With a stepped attenuator you are free to generate an accurate log curve using precision resistors, and this could be used with a simple shunt feedback arrangement to provide an active gain control.
Considering the previous accusations of modification for changing one capacitor type, going down this route would obviously be entirely unacceptable
Besides it would be silly to use a stepped attenuator with the Baxandall active gain control since the whole idea behind it is to generate a log gain curve using a linear pot. With a stepped attenuator you are free to generate an accurate log curve using precision resistors, and this could be used with a simple shunt feedback arrangement to provide an active gain control.
Considering the previous accusations of modification for changing one capacitor type, going down this route would obviously be entirely unacceptable
I've started another thread related to my construction and build of this new pre-amp design from Doug Self. It can be found here. I'll be posting pics as I go along, but it looks like it's going to be a time-consuming project. Feel free to share your experience if building this pre-amp or other relevant topic.
Absolutely correct. Using a stepped attenuator in the Baxandall control really would be missing the point.
I have, however, nothing against stepped attenuators as such. (I've just done a 43-step one for a client) There is the possibility of an active gain control that is quieter than the Baxandall configuration.
I would not try the relay method. The problems are many, not least what happens when your (presumably logarithmic) control code goes from 01111111 to 10000000. Relay timing is not that predictable- just think of the transients you're going to hear. It will be horrible.
The most advanced relay volume volume control I am aware of is in the Cambridge Audio Azur 840E (and I must declare an interest here, because I designed it) and that uses some really rather subtle and complex techniques to solve the transient problem completely. However, I don't think it would be right to divulge them here.
Last edited:
Please tell, curious minds would love to know, just kidding I know you can't and wouldn't do that. I only brought up the stepped attenuation as it looked like there was a problem with sourcing the 1K linear pots required for the build. I am not one to suggest anything here beyond what color of wire jacket to use in the build..... Just watching and trying to learn.
Thanks for sharing your expertise once again Mr Self.
I actually managed to make a passive "L" segment version that works without any noticeable signal interuption with relay switching of 7 "bits" in series, where each bit represents a binary attenuation segment - ie the first segment is 1dB, the second 2dB, third 4dB etc until 7th at 64dB. The SPDT relay contacts simultaneously short out the series R and open the parallel R when the relay is on so that with power removed attenuation is at maximum. This gives 128dB range, though I'm only using 99dB in my preamp as it's enough and I only have two 7 segment LED digits on the display...
When I first tested it, there were indeed some scary transients when changing steps that involved multiple relays switching. However I managed to solve the problem quickly and easily using a method you suggested in your speaker protection article in EW&WW years ago - using a zener in series with the diode across each relay coil to speed up switch off. I was very surprised to find that this solved the problem completely. I guess this isn't a universal solution but it works with two different "telecom" signal relay types I tried. I guess also if the relay network was part of the feedback network in an active gain stage the switching interval would be far more critical.
Incidently, my preamp that uses this relay attenuator is my favourite for sound quality. It has a single non-inverting gain stage after the attenuator using a discrete opamp that is based on the Figure 3.33 circuit in your Small Signal Audio Design book, but with the addition of a current mirror in the input stage and some tweaking of bias currents etc.
So I just wanted to offer my thanks for your ideas and make it clear that I remain a big fan of your writing and indeed the sound quality of some of your published circuits. I'm kind of embarrassed about not being happy with the sound of my Precision Pre '96 - I still don't know why this is but it remains so. Anyway I sincerely hope that my discussion of it hasn't worried you, despite it apparently stiring up some aggression in certain camps
If there are any intrepid designers out there who'd like to forge new pathways in audio attenuators, ADI's X-amp might provide inspiration for getting away from the relay/CMOS/JFET switch style of volume control.
http://www.analog.com/library/analogDialogue/archives/36-01/xAmp/xAmp.pdf
I've been using AD603 and AD605 which implement this technique - the distortion's not as low as using a pot or switched attenuator, the noise certainly is higher because of the fixed gain element, but a discrete solution might improve on these drawbacks.
http://www.analog.com/library/analogDialogue/archives/36-01/xAmp/xAmp.pdf
I've been using AD603 and AD605 which implement this technique - the distortion's not as low as using a pot or switched attenuator, the noise certainly is higher because of the fixed gain element, but a discrete solution might improve on these drawbacks.
http://www.analog.com/library/analogDialogue/archives/36-01/xAmp/xAmp.pdf
Mmmm...500MHz bandwidth?..almost enough for high-end audio, you might think.
Seriously though, how would you rate AD60X series chips, their successors and competitors in consumer applications? As the chips have advantages with physical size and isolation from other sources of EMI, do you think there's a future for them at this extreme end of audio? Perhaps you are even dangling an interesting idea or two for that discrete approach?
A big problem for audio purists and those just wanting to follow traditional construction techniques, is that whilst great advances are made at the silicon end, the majority of interfaces for users have become tedious button-poking or multi-turn rotary encoded operations. Simple pots are now either large, emphatic and expensive boutique parts, scarce industrial specialties or sad little bits of carbon paper in a 10mm square sandwich of diecast metal and plastic that may look functional but where performance and precision is a travesty. Surely, it doesn't take much to outperform these when, as we increasingly find, suitable parts are no longer viable.
Mmmm...500MHz bandwidth?..almost enough for high-end audio, you might think.
Seriously though, how would you rate AD60X series chips, their successors and competitors in consumer applications? As the chips have advantages with physical size and isolation from other sources of EMI, do you think there's a future for them at this extreme end of audio? Perhaps you are even dangling an interesting idea or two for that discrete approach?
A big problem for audio purists and those just wanting to follow traditional construction techniques, is that whilst great advances are made at the silicon end, the majority of interfaces for users have become tedious button-poking or multi-turn rotary encoded operations. Simple pots are now either large, emphatic and expensive boutique parts, scarce industrial specialties or sad little bits of carbon paper in a 10mm square sandwich of diecast metal and plastic that may look functional but where performance and precision is a travesty. Surely, it doesn't take much to outperform these when, as we increasingly find, suitable parts are no longer viable.