I have been working on a passive pre. So far I've learned (again!) that the better the components the better the sound.
Now I'm thinking about putting the actual attenuator resistors in a very small box and plugging it directly in-line at the back of the power amp. This saves one pair of (very expensive) cables. The cable savings will buy all the Vishay bulk foil resistors one could dream of.
The problem of course is the switch (I used a Shallco, stereo, 43 posistion) They are too big and you couldn't reach it at the back of the amp anyway.
So... reed relays? Use a cheap switch at an arbitrary distance from the switched elements and a wall wart to power the relay coils.
A number of possible circuits come to mind. A simple example is an R2R ladder and 7 relays. This would get about 36 dB of range with steps of beter tham 1 dB.
Anyone have experience with reed relay contact sound quality?
Now I'm thinking about putting the actual attenuator resistors in a very small box and plugging it directly in-line at the back of the power amp. This saves one pair of (very expensive) cables. The cable savings will buy all the Vishay bulk foil resistors one could dream of.
The problem of course is the switch (I used a Shallco, stereo, 43 posistion) They are too big and you couldn't reach it at the back of the amp anyway.
So... reed relays? Use a cheap switch at an arbitrary distance from the switched elements and a wall wart to power the relay coils.
A number of possible circuits come to mind. A simple example is an R2R ladder and 7 relays. This would get about 36 dB of range with steps of beter tham 1 dB.
Anyone have experience with reed relay contact sound quality?
Should be OK, but use something better than a wall-wart. Those things can inject hum in anything!
Reed-relays are good because the contacts are totaly sealed, offer very low bounce and very low capacitance when open.
Why not a constant impedance network? Each section a certain attenuation in dB and you can put as much in series as you want. And you could match it to the input impedance of your amp.
Reed-relays are good because the contacts are totaly sealed, offer very low bounce and very low capacitance when open.
Why not a constant impedance network? Each section a certain attenuation in dB and you can put as much in series as you want. And you could match it to the input impedance of your amp.
Reed relays are good for the reasons mentioned, however consider the following if you are as obsessed as you must be to go this far- (Speaking from experience)
1) Reed relays use a magnetic material to pass the conductor since the armature is the contact in the design.
2) Emulating the switch with relays will use a lot of relays.
I would make a box at the back of the amp with long shafts for the switch. The mechanical problems to solve are easier than the electronic issues.
If you must have electronic control add a motor to the switch. you can defeat the detents so you can use a small motor.
The Shallco switch is about as good as you can get. It has solid silver contacts and should have at least two for each connection. Few relays approach that performance.
-Demian
1) Reed relays use a magnetic material to pass the conductor since the armature is the contact in the design.
2) Emulating the switch with relays will use a lot of relays.
I would make a box at the back of the amp with long shafts for the switch. The mechanical problems to solve are easier than the electronic issues.
If you must have electronic control add a motor to the switch. you can defeat the detents so you can use a small motor.
The Shallco switch is about as good as you can get. It has solid silver contacts and should have at least two for each connection. Few relays approach that performance.
-Demian
Several points:
1. Using a L pad at each position does allow either constant input Z or constant output Z but requires a 2 pole relay for each availble gain step, 20-30 steps, 20-30 relays per channel. Whereas the R-2R network does get you there with only 7 relays. Of course, now, multiple resistor junctions are in series with the signal and 7 relay contacts are all in parallel (through resistors). There are other minimal relay topologies, but the big question is the relays
2. It is precisely my concerns about the materials used in reed construction that prompted the post and the question about sound. And we haven't even begun to discuss contact materials; rhodium, gold, mercury wetted?.
3. A mechanical solution is in fact possible but would require a specific design for every brand of power amp and installation. The good switches require significant torque so the assembly would also need to be quite sturdy.
4. The little box in series with the amplifier input sort of looses it's appeal when it needs to hold that many large and heavy parts.
5. Relays allow for the possibility of a balance control without resorting to seperate volume knobs for right and left channel or some additional parts creating additional problems.
My audio buddy and I built a single one level fixed attenuator with 2 Vishay foil resistors, 1 male and 1 female Cardas connector and a little silver wire. We installed it directly into the back of his amp and connected his D to A with his exotic cables. Ta Da; pretty close to audio nirvana. Got me to thinking...
I have built a free standing conventional passive-pre and already own silver cables. But in building and using my design these questions came up.
1. Using a L pad at each position does allow either constant input Z or constant output Z but requires a 2 pole relay for each availble gain step, 20-30 steps, 20-30 relays per channel. Whereas the R-2R network does get you there with only 7 relays. Of course, now, multiple resistor junctions are in series with the signal and 7 relay contacts are all in parallel (through resistors). There are other minimal relay topologies, but the big question is the relays
2. It is precisely my concerns about the materials used in reed construction that prompted the post and the question about sound. And we haven't even begun to discuss contact materials; rhodium, gold, mercury wetted?.
3. A mechanical solution is in fact possible but would require a specific design for every brand of power amp and installation. The good switches require significant torque so the assembly would also need to be quite sturdy.
4. The little box in series with the amplifier input sort of looses it's appeal when it needs to hold that many large and heavy parts.
5. Relays allow for the possibility of a balance control without resorting to seperate volume knobs for right and left channel or some additional parts creating additional problems.
My audio buddy and I built a single one level fixed attenuator with 2 Vishay foil resistors, 1 male and 1 female Cardas connector and a little silver wire. We installed it directly into the back of his amp and connected his D to A with his exotic cables. Ta Da; pretty close to audio nirvana. Got me to thinking...
I have built a free standing conventional passive-pre and already own silver cables. But in building and using my design these questions came up.
Also keep in mind how the resistor+relay network is set up -- if you don't mind signal going through more than one resistor. You can use far less relays per total "step" desired.
Simple example:
2 relays, 2 resistors total in a series/parallel config (series relay with a resistor twice, put in parallel). Confused by my desription -- I never was good at writing
If the resisitors have value r1 and r2, you get 4 combos: 0, r1, r2, and r1*r2/(r1+r2).
Add a third relay/resistor in parallel and the number of permutations grows yet again.
If I recall by statistics theory correctly, it goes like 2 to the N, N = number of unique resistor values. So you could get the equiavelnt of 64 steps from 6 relays/resistors per channel.
Of course you're sending signal through multiple paths, resistors, solder joints, etc., which will not appeal to all. But rarely can you get something for nothing!
And don't ask how you would go about actually changing the volume on that puppy: 6 switches per channel
Simple example:
2 relays, 2 resistors total in a series/parallel config (series relay with a resistor twice, put in parallel). Confused by my desription -- I never was good at writing
If the resisitors have value r1 and r2, you get 4 combos: 0, r1, r2, and r1*r2/(r1+r2).
Add a third relay/resistor in parallel and the number of permutations grows yet again.
If I recall by statistics theory correctly, it goes like 2 to the N, N = number of unique resistor values. So you could get the equiavelnt of 64 steps from 6 relays/resistors per channel.
Of course you're sending signal through multiple paths, resistors, solder joints, etc., which will not appeal to all. But rarely can you get something for nothing!
And don't ask how you would go about actually changing the volume on that puppy: 6 switches per channel
That part's easy. Hook a series of diodes to each step of the master gain (cheap) rotary switch, route a diode to whichever relay coil needs to be closed for that step. This way you can activate any of the relays in any arbitrary sequence.
You could also use a rotary encoder as the gain selector. Connect it to a small ROM whose outputs select which relay is closed. No clocks are needed it can be a totally asynchronous design.
If you build a R-2R ladder the rotary encoder could drive the relays (with a transistor) directly. Rotary encoders cost a lot less than a nice Shallco silver rotary switch.
I plan some tests about series junctions on sound quality. I know that there are commercial stepped attenuators where many many resistors are in series and the signal is taken from the junctions in a sequential manner thereby emulating a pot almost exactly.
The tests arn't that hard, issues with break-in make the duration of each test longer than I would like
. So far no respondents have mentioned that they have built and accepted or discarded any of these ideas.
Further opinions are most welcome.
You could also use a rotary encoder as the gain selector. Connect it to a small ROM whose outputs select which relay is closed. No clocks are needed it can be a totally asynchronous design.
If you build a R-2R ladder the rotary encoder could drive the relays (with a transistor) directly. Rotary encoders cost a lot less than a nice Shallco silver rotary switch.
I plan some tests about series junctions on sound quality. I know that there are commercial stepped attenuators where many many resistors are in series and the signal is taken from the junctions in a sequential manner thereby emulating a pot almost exactly.
The tests arn't that hard, issues with break-in make the duration of each test longer than I would like
. So far no respondents have mentioned that they have built and accepted or discarded any of these ideas.Further opinions are most welcome.
Hi,
You can easily do better than that but there's always going to be some compromise.
What you're trying to do I did a long time ago already and I really pushed it way beyond the Chello stuff although I did limit the step number to 23 and the range to +/- 60dB.
A series type attenuator was used but instead of putting all the resistors in series, the values were simply summed as steps went up.
By doing so I only used a single resistor in the direct signal path and a shunt resistor at any given position bar the end stops.
You'll need to be able to buy precision resistors from the entire E96 series and IRCC I still had to solder two Rs in series somewhere...
An Elma switch was used to control the relays (I used mercury wetted relays as I believe they're the best) that were soldered by their metal backs on some PCB boards mounted as a stack (zig-zag style, 4*6 relays) so that it remained relatively compact.
The resistors were than soldered directly to the relays PCB mount
pins using 6% silver loaded solder.
Once everything worked as planned, the lot was sprayed with PTFE and potted in epoxy to prevent oxidation.
The resistors used were 0.1% Vishay bulkfoils considered to be the best available in those days, lead lengths were kept to the minimum manageable.
All other wiring was bare 6N silver, .6mm diameter, sheethed with a losely fitting translucent PTFE tubing just to avoid shorts.
This is about all I can remember for the moment....
Oh, yes it beat everything else hands down in those days especially the high one off prototype costs.
Cheers,
You can easily do better than that but there's always going to be some compromise.
What you're trying to do I did a long time ago already and I really pushed it way beyond the Chello stuff although I did limit the step number to 23 and the range to +/- 60dB.
A series type attenuator was used but instead of putting all the resistors in series, the values were simply summed as steps went up.
By doing so I only used a single resistor in the direct signal path and a shunt resistor at any given position bar the end stops.
You'll need to be able to buy precision resistors from the entire E96 series and IRCC I still had to solder two Rs in series somewhere...
An Elma switch was used to control the relays (I used mercury wetted relays as I believe they're the best) that were soldered by their metal backs on some PCB boards mounted as a stack (zig-zag style, 4*6 relays) so that it remained relatively compact.
The resistors were than soldered directly to the relays PCB mount
pins using 6% silver loaded solder.
Once everything worked as planned, the lot was sprayed with PTFE and potted in epoxy to prevent oxidation.
The resistors used were 0.1% Vishay bulkfoils considered to be the best available in those days, lead lengths were kept to the minimum manageable.
All other wiring was bare 6N silver, .6mm diameter, sheethed with a losely fitting translucent PTFE tubing just to avoid shorts.
This is about all I can remember for the moment....
Oh, yes it beat everything else hands down in those days especially the high one off prototype costs.
Cheers,
Thanks Frank.
I was reasonably sure that the relay idea could be made to work.
My current passive pre has a single L pad pair with one resistor switched in parallel in the shunt leg for all normal listening levels. A single resistor is switched accross the series element for low levels of attenuation. I used molded Vishay foils for the 2 fixed resistors and the conformal Vishay foil for the parallel element of the 10 most likely listening volumes. Holcos everywhere else. I bought a cheap commercial passive pre used, gutted it and used the case, knobs and front panel, looks nice and professional.
I used 4 nines silver and a Shallco silver contact rotary switch. I still have some stridency which I think I have traced to the RCA jacks. New ones arrive tommorow and some Cardas 5 nines silver wire for good measure.
I can not evaluate megabuck pre-amps in my system because I can't afford them anyway. Testing to date shows the passive design is quite superior to a Conrad Johnson mid price pre (PV12L) and to a Classe Audio DR3(?).
Much better resolution in all frequencies especially the bass, faster attack and better harmonic decays - I'm sold! And the price beats all but the least expensive used pre-amps out there. In reading reviews from others about passive pre amps many felt that dynamics were compressed, I didn't find that, quite the opposite. The new pre makes cable evaluations a cinch, the differences are very easy to spot now.
All in all, a worthwile project leading me to wonder what else can be done.
Thanks again for sharing your experience.
I was reasonably sure that the relay idea could be made to work.
My current passive pre has a single L pad pair with one resistor switched in parallel in the shunt leg for all normal listening levels. A single resistor is switched accross the series element for low levels of attenuation. I used molded Vishay foils for the 2 fixed resistors and the conformal Vishay foil for the parallel element of the 10 most likely listening volumes. Holcos everywhere else. I bought a cheap commercial passive pre used, gutted it and used the case, knobs and front panel, looks nice and professional.
I used 4 nines silver and a Shallco silver contact rotary switch. I still have some stridency which I think I have traced to the RCA jacks. New ones arrive tommorow and some Cardas 5 nines silver wire for good measure.
I can not evaluate megabuck pre-amps in my system because I can't afford them anyway. Testing to date shows the passive design is quite superior to a Conrad Johnson mid price pre (PV12L) and to a Classe Audio DR3(?).
Much better resolution in all frequencies especially the bass, faster attack and better harmonic decays - I'm sold! And the price beats all but the least expensive used pre-amps out there. In reading reviews from others about passive pre amps many felt that dynamics were compressed, I didn't find that, quite the opposite. The new pre makes cable evaluations a cinch, the differences are very easy to spot now.
All in all, a worthwile project leading me to wonder what else can be done.
Thanks again for sharing your experience.
There probably isn't any such thing as "normal" switch contacts.
Both relays and manual switches contain various kinds of metals and contact materials. Usually contacts are further plated with another metal to aid life. Common contact or contact plating materials are: Gold, Palladium, Rhodium, silver and mercury films. Any of these can be found in both switches and relays.
Silver may be best followed by mercury (danger, this is an opinion) silver may be more common in manual switches since surface films that inhibit performance are easily removed with a little rubbing caused by wiping the contact. This rubbing force is easily applied by a human hand when activating the switch. This is harder to do in relays especially tiny ones. Reed relays keep the contacts inside a hermetically sealed glass envolope filed with inert atmosphere. This significantly reduces the formation of oxides or other films
So get data sheets. Assume that cheap switches at Radio Shack are suspect and experiment or buy your switches from a reliable audio source. !Blatant Plug! I have always found Michael Percy audio to have good stuff and reliable opinions.
Both relays and manual switches contain various kinds of metals and contact materials. Usually contacts are further plated with another metal to aid life. Common contact or contact plating materials are: Gold, Palladium, Rhodium, silver and mercury films. Any of these can be found in both switches and relays.
Silver may be best followed by mercury (danger, this is an opinion
) silver may be more common in manual switches since surface films that inhibit performance are easily removed with a little rubbing caused by wiping the contact. This rubbing force is easily applied by a human hand when activating the switch. This is harder to do in relays especially tiny ones. Reed relays keep the contacts inside a hermetically sealed glass envolope filed with inert atmosphere. This significantly reduces the formation of oxides or other filmsSo get data sheets. Assume that cheap switches at Radio Shack are suspect and experiment or buy your switches from a reliable audio source. !Blatant Plug! I have always found Michael Percy audio to have good stuff and reliable opinions.
Hi,
I use reed relays in my TVC preamp with very good results.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=23086
Regards
I use reed relays in my TVC preamp with very good results.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=23086
Regards
Relay Contacts
Here are some links to details on relay contacts:
http://www.agastat.com/app_pdfs/13c3236.pdf
http://www.cougarelectronics.com/pdf/reed_relays.pdf
http://www.cotorelay.com/Contact_Resistance___Dynamics.pdf
http://zone.ni.com/devzone/conceptd...86256DD400569EB7?opendocument&node=DZ52331_US
I nwas looking for a tutorial I ran into years ago that discussed the physics of small signal contacts. The fundamental issue is that there needs to be enough energy to weld the contacts to pass the signal. Some materials do it better than others, and at very low levels getting the contact to weld is the challenge. (Not to be confused with welding the contacts with lots of power)
Mercury relays are good but difficult to use.
The third reference points out some of the problems with reed relays.
I still think a mechanical solution is better. Make a base to put the amp on with a box at the back. 1/2" should be enough. Use gears at the back to drive the switches. Most of the hardware should be in the Allied or Mouser catalog. Berg has useful hardware for this as well including differentials so you can have vol/bal controls on the front and seperate pots inside. Stax did this in their preamps.
-Demian
Here are some links to details on relay contacts:
http://www.agastat.com/app_pdfs/13c3236.pdf
http://www.cougarelectronics.com/pdf/reed_relays.pdf
http://www.cotorelay.com/Contact_Resistance___Dynamics.pdf
http://zone.ni.com/devzone/conceptd...86256DD400569EB7?opendocument&node=DZ52331_US
I nwas looking for a tutorial I ran into years ago that discussed the physics of small signal contacts. The fundamental issue is that there needs to be enough energy to weld the contacts to pass the signal. Some materials do it better than others, and at very low levels getting the contact to weld is the challenge. (Not to be confused with welding the contacts with lots of power)
Mercury relays are good but difficult to use.
The third reference points out some of the problems with reed relays.
I still think a mechanical solution is better. Make a base to put the amp on with a box at the back. 1/2" should be enough. Use gears at the back to drive the switches. Most of the hardware should be in the Allied or Mouser catalog. Berg has useful hardware for this as well including differentials so you can have vol/bal controls on the front and seperate pots inside. Stax did this in their preamps.
-Demian
Thanks,
but the link is a little bit superficial, sorry. That is just an overview sheet which I have seen already. But I can't find anywhere a hint that the contacts are gold-plated! Especially not for the SPDT ones. Or am I blind? Which type do you mean?
I am looking for such thing almost two month! Gold-plated reed relays could be the solution (mercury wetted isn't RoHS-conform and ugly).
Burmester 035
Regards
Thomas
but the link is a little bit superficial, sorry. That is just an overview sheet which I have seen already. But I can't find anywhere a hint that the contacts are gold-plated! Especially not for the SPDT ones. Or am I blind? Which type do you mean?
I am looking for such thing almost two month! Gold-plated reed relays could be the solution (mercury wetted isn't RoHS-conform and ugly).
Burmester 035
Regards
Thomas
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.