Coulomb said:
I wouldn't agree with you about 3dB changes only. It may be what they say in books, but it's not like that in reality.
Anyway, this is a serious product, so minimum resolution would be required. I would think that 1dB would be a nice step, especially if this could be shown on display. The current setup has steps as low as 0.1 dB, which is good if you calibrate your 6 channel system, but for 2 channel everyday listening is simply not convenient. There are 19 steps on my transformers, and I can assure you that this is enough also. Unless someone gets complain from wife and has to be more carefull with volume settings.
Anthony,
is not true that uman changes can discern volume changes less than 3 dB. Our sensibility to volume changes it's strictly connected to the frequency and to the level( see Fletcher and Munson curves). Around 3,5 Khz, the most sensitive frequency for human ear, we can discern volume changes less then 3 dB.
Regards,
Antonio
Grataku, posso avere il piacere di scambiare due parole con te in privato, scrivimi mi farebbe molto piacere, grazie.
is not true that uman changes can discern volume changes less than 3 dB. Our sensibility to volume changes it's strictly connected to the frequency and to the level( see Fletcher and Munson curves). Around 3,5 Khz, the most sensitive frequency for human ear, we can discern volume changes less then 3 dB.
Regards,
Antonio
Grataku, posso avere il piacere di scambiare due parole con te in privato, scrivimi mi farebbe molto piacere, grazie.
Peter,
You are certainly correct about the funny jumping in impedance.
The resistors in parallel (shunt) all cause nice changes in impedance but the series(or pass) resistors cause some complications.
The C++ program I wrote tried to do two things.
1.) have evenly spaced steps in dB.
2.) given the choice between two closely spaced gains.
choose the one with the lower impedance.
unfortunately one of the limitations I found was the smallest resistor that I can get in 0.1% is not very small. This makes the
impedance go very high when I needed very large attenuation.
Also to keep the even dB steps, you start getting stange jumps
in impedance.
When I made all of the curves of the attenuation. they are 3 nasty asymptotic curves. Ideally for minimum impedance
you would like to be on only one of the curves, in the linear section of the curve. but that would require that the user
has the gain of amp/preamp such that that attenuation range falls perfectly into his/her acceptable listening volume range.
But if anyone has a better idea of how to improve the sound, please let me know.
It would certainly be simple to have fewer volume steps.
but I originally thought the user would like the extra levels?
Thanks,
-Craig Beiferman
You are certainly correct about the funny jumping in impedance.
The resistors in parallel (shunt) all cause nice changes in impedance but the series(or pass) resistors cause some complications.
The C++ program I wrote tried to do two things.
1.) have evenly spaced steps in dB.
2.) given the choice between two closely spaced gains.
choose the one with the lower impedance.
unfortunately one of the limitations I found was the smallest resistor that I can get in 0.1% is not very small. This makes the
impedance go very high when I needed very large attenuation.
Also to keep the even dB steps, you start getting stange jumps
in impedance.
When I made all of the curves of the attenuation. they are 3 nasty asymptotic curves. Ideally for minimum impedance
you would like to be on only one of the curves, in the linear section of the curve. but that would require that the user
has the gain of amp/preamp such that that attenuation range falls perfectly into his/her acceptable listening volume range.
But if anyone has a better idea of how to improve the sound, please let me know.
It would certainly be simple to have fewer volume steps.
but I originally thought the user would like the extra levels?
Thanks,
-Craig Beiferman
It is normal that such an elaborate project as creating 255 steps volume attenuator won't be completely perfect right from the beginning and only end users can actually comment on the real life performance.
So far I didn't listen to it yet, just connected signal generator and true dB meter to see how accurate the steps were. I tested the board which I bought preassembled (one of the specials) and it measured pretty well. All the steps were not exactly equal, but in reasonable range (I'm talking here 0.3dB, 0.2dB and 0.1dB variety). Then I connected the ohm meter and checked how the impedance changes. What I noticed is that it was always switching with ea. step (like indeed the program was choosing the resistors for best step values). So, the imp was changing between 3k and 6k with each step in a certain range. Well, it's not really something to be concerned about, but if there is a choice that it can be more stable by sacrificing number of steps, I would rather go for the latter. The thing is, it shouldn't be changing that much in some preferrable listening volume settings. Let's say your listening range is between -10dB and -20db. You do adjustements to volume by changing shunt resistors, but at least series resistors shouldn't be changing constantly in that range, as I believe this somehow may affect the sound. Of course it is hard to avoid when you move to the next listening range, but than again it is partl of the way things work.
The biggest problems started when I tried to put some other randomly chosen series resistors. I got lost and completely couldn't figure out what to do.
So far I didn't listen to it yet, just connected signal generator and true dB meter to see how accurate the steps were. I tested the board which I bought preassembled (one of the specials) and it measured pretty well. All the steps were not exactly equal, but in reasonable range (I'm talking here 0.3dB, 0.2dB and 0.1dB variety). Then I connected the ohm meter and checked how the impedance changes. What I noticed is that it was always switching with ea. step (like indeed the program was choosing the resistors for best step values). So, the imp was changing between 3k and 6k with each step in a certain range. Well, it's not really something to be concerned about, but if there is a choice that it can be more stable by sacrificing number of steps, I would rather go for the latter. The thing is, it shouldn't be changing that much in some preferrable listening volume settings. Let's say your listening range is between -10dB and -20db. You do adjustements to volume by changing shunt resistors, but at least series resistors shouldn't be changing constantly in that range, as I believe this somehow may affect the sound. Of course it is hard to avoid when you move to the next listening range, but than again it is partl of the way things work.
The biggest problems started when I tried to put some other randomly chosen series resistors. I got lost and completely couldn't figure out what to do.
Some things can be done, others cannot be done. One can build circuits with the following characteristics:
You can have constant input impedance
You can have constant output impedance
But you can't have both at the same time
You can have many levels, but any constraints such as the above reduces number of available levels.
My theory is that you want to look at energy transfer - how much of the signal is lost as heat, and how much is available for driving the next stage. That is one of the things that the SHM is optimized for while trying to keep parts count down.
Since I like few parts, many volume steps, shunting between phases, balanced signals etc. I believe the current iteration of SHM is just about as good as you can get from a theory standpoint. How well it will work in practice depends on how sensitive your circuitry is to changes in input impedance. The dipchip guys have another product offerings for those who prefer other alternatives.
Peter, may I suggest you measure the input impedance of a standard pot, say 100K driving a 5K load and consider how the input impedance and output impedance of that setup impacts sound. Also, consider how much of the signal is lost as heat when you use a standard pot to drive a signal at near maximum attenuation.
As you see, there will always be tradeoffs and outliers. It is quite possible to utilize excellent engineering the wrong way.
Petter
You can have constant input impedance
You can have constant output impedance
But you can't have both at the same time
You can have many levels, but any constraints such as the above reduces number of available levels.
My theory is that you want to look at energy transfer - how much of the signal is lost as heat, and how much is available for driving the next stage. That is one of the things that the SHM is optimized for while trying to keep parts count down.
Since I like few parts, many volume steps, shunting between phases, balanced signals etc. I believe the current iteration of SHM is just about as good as you can get from a theory standpoint. How well it will work in practice depends on how sensitive your circuitry is to changes in input impedance. The dipchip guys have another product offerings for those who prefer other alternatives.
Peter, may I suggest you measure the input impedance of a standard pot, say 100K driving a 5K load
and consider how the input impedance and output impedance of that setup impacts sound. Also, consider how much of the signal is lost as heat when you use a standard pot to drive a signal at near maximum attenuation.As you see, there will always be tradeoffs and outliers. It is quite possible to utilize excellent engineering the wrong way.
Petter
Hi Anthony,
You could short the output (X2-1 and X2-2) and measure the resistance across the input.
The values should change as you change volume.
Also, if the board is working, the volume on the display will change.
You could also put 5V across the input X1-3 and X1-2.
Now measure the voltage across the outputs.
Repeat for the other channel.
Dale
You could short the output (X2-1 and X2-2) and measure the resistance across the input.
The values should change as you change volume.
Also, if the board is working, the volume on the display will change.
You could also put 5V across the input X1-3 and X1-2.
Now measure the voltage across the outputs.
Repeat for the other channel.
Dale
Petter,
May I suggest you measure your SHM board and see what's the output impedance at 255 volume position and let's say 200 position.
On my board it was 50k and 2K approx,
Since I want to use my boards at the input of a GC, the volume control sets the input impedance. My goal is to have it at least 20K and never more than 50K. I don't know what load this volume pot will drive, but its output is connected straight to the + input on LM3875 chip. No other resistors are used except for two other setting the gain. I know about the other board, with constant resistance, but I can't force myself to use 10 resistors in series at the input of my amps.
I don't care for constant impedance, I care for the impedance in a reasonable range and the similar resistance in subsequent steps, although it can change gradually. But my comfort range on a pot is last quater of the complete turn and in this range I want my impedance reasonably steady. This can be achieved pretty easy.
May I suggest you measure your SHM board and see what's the output impedance at 255 volume position and let's say 200 position.
On my board it was 50k and 2K approx,
Since I want to use my boards at the input of a GC, the volume control sets the input impedance. My goal is to have it at least 20K and never more than 50K. I don't know what load this volume pot will drive, but its output is connected straight to the + input on LM3875 chip. No other resistors are used except for two other setting the gain. I know about the other board, with constant resistance, but I can't force myself to use 10 resistors in series at the input of my amps.
I don't care for constant impedance, I care for the impedance in a reasonable range and the similar resistance in subsequent steps, although it can change gradually. But my comfort range on a pot is last quater of the complete turn and in this range I want my impedance reasonably steady. This can be achieved pretty easy.
Peter Daniel said:Petter,
May I suggest you measure your SHM board and see what's the output impedance at 255 volume position and let's say 200 position.
On my board it was 50k and 2K approx,
The original suggestion was that "only top 2 dogs can play" where one would limit the settings so that one would toggle only between the the two available lowest settings of pass resistors. This will of course limit the number of steps.
How about this: If you can convince the dipchip guys to have a top dog, top 2 dogs and top 3 dogs setup the user can choose between tradeoffs on the display ... So you will trade the number of volume settings and the linearity between them for smaller jumps in impedance.
Petter
I wasn't aware of the top dog issue, but after playing with the volume and measuring the way it works, I don't think that this is such a good approach. I noticed that the system is switching constantly between series resistors with every click of the knob. If it happens that there is a big difference in those resistors value, it may actually affect subjective performance (in other words the way the system sounds). I would rather see the resistors being changed in certain operating ranges, where they simply improve the total impedance and provide more gradual and more stable impedance values.I don't even know if this has any big impact on the performance, but in my opinion seems like more elegant and more purist approach.
One thing I know for sure, there will be difference in sound between one series resistor and 3 series resistors in parallel. If your control interface is switching between all parallel combination of those resistors in 10 subsequent steps, covering 2dB range, how good really is it? I guess it's a question of priorities
One thing I know for sure, there will be difference in sound between one series resistor and 3 series resistors in parallel. If your control interface is switching between all parallel combination of those resistors in 10 subsequent steps, covering 2dB range, how good really is it? I guess it's a question of priorities
Peter Daniel said:Congratulations on a job well done. Putting my blah,blah aside for a moment, I was very nicely surprised with how quiet the relays operate, aren't they?
What resistance value did you choose?
Hello Peter, I used the 5 K kit, and at first I was concerned the relays were not switching they are so quiet.
I think functionally this is the best DIY project yet, lots of parts, connectors and PCB's
Mind you I start building a Jukka DAC this weekend that looks promising as well. It was a little difficult getting all the parts but it should make a nice addition to my DAC stable.
My CDPRO chassis is coming along, I will be adding a lid next weekend that opens with Gas struts. I hope Brian manages to get everything working soon.
Here is a link for the Jukkka DAC.
http://koti.mbnet.fi/siliconf/JukkaTolonen/projects/jt-dac_no.3/jt-dac_no.3.html
Optional tube output.
http://koti.mbnet.fi/siliconf/JukkaTolonen/projects/tubedac/tubedac.html
Anthony
Assembly Question :
How do you hook up the volume for balanced operation?
One board for L and one for R ? Or one for + L/R and one for - L/R? Does it make a difference?
Also not understanding how to hook up a second pair of input boards, in series, for 8 stereo-inputs....and if I should remove the HT-Bypass relays on the board not used for output?
Arne K
How do you hook up the volume for balanced operation?
One board for L and one for R ? Or one for + L/R and one for - L/R? Does it make a difference?
Also not understanding how to hook up a second pair of input boards, in series, for 8 stereo-inputs....and if I should remove the HT-Bypass relays on the board not used for output?
Arne K
Arne K,
Download the IS1 manual from DipChip site to see how to wire more than 4 inputs. (Basically, you need to use a jumper on JP1-1 or JP1-2.) As for the HT Bypass, there is a photo of a section of the IS1 board in the manual that shows how to place jumper wires in lieu of the relays. This is how I use my IS1s: no output relays.
As for the SHM, I believe between the manual (section marked "Important") and the schematic pdf for this board you can see how to hook up a balanced connection. Essentially, only one SHM is needed for a given balanced, channel: (-) signal in/out is marked on SHM as is (+) signal in/out; ground/shield is the other input. For strictly single ended use, I shorted the (-) input to ground.
Regards, Robert
Download the IS1 manual from DipChip site to see how to wire more than 4 inputs. (Basically, you need to use a jumper on JP1-1 or JP1-2.) As for the HT Bypass, there is a photo of a section of the IS1 board in the manual that shows how to place jumper wires in lieu of the relays. This is how I use my IS1s: no output relays.
As for the SHM, I believe between the manual (section marked "Important") and the schematic pdf for this board you can see how to hook up a balanced connection. Essentially, only one SHM is needed for a given balanced, channel: (-) signal in/out is marked on SHM as is (+) signal in/out; ground/shield is the other input. For strictly single ended use, I shorted the (-) input to ground.
Regards, Robert
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.