I plan to build an passive attenuator to step down multiple identical audio sources (Class D amplifiers rated 3W@4-32 Ohms) to multiple sound card inputs +4dBu (MOTU 2408 MKII duplex sound card). Each source will go to a dedicated input channel on the MOTU sound card and thus it is my understanding this creates challenges due to summing of differential Class D voltages (compared to just a single source). The goal is to preserve sound quality as reasonably as possible and as tactile transducers rate to 12 Hz will be used, maintain source bass signals below 20 Hz. Also, the source signals will be summed in a VST host in various combinations (e.g., 3 sources may be summed to 1 output channel with another 2 sources into a different output channel). As electronic circuits are not my strength, I have multiple newbie questions:
1. The source volume can be adjusted. However, the listening volume will always be adjusted after the sound card. What level should the source volume be set at to maximize sound quality (S/N ratio/distortion)? I assume maximum volume, but perhaps when considering the interaction with the passive attenuator, a lower source output may be appropriate. Also, as the sources are Class D amps, perhaps there is an optimal volume level which has less distortion. It seems determining the volume level is important as it will change the attenuationl level needed and the component values (e.g., resistor Ohms rating).
2. What type of passive attenuator is ideal for this application? As bass/subsonic signals are desired, audio transformers are not ideal due to bass frequency loss and the cost of quality audio transformers which preserve bass signals is prohibitive with planned channel expansion (up to 40 mono sources). Thus, it seems that voltage dividers/"pads" such as L-Pads, Pi-pads, T-pads, etc. are the ideal solution considering compromises of cost vs. sound quality. However, in reading about the various voltage divider configurations, it is not obvious which voltage divider designs are suited for which applications, particularly this one.
3. Related to question #2 above, the impedance of the source and sound card are likely mismatched (I'm struggling to relate the amp Watts output to sound card dBu input). Some authors mention the voltage divider source and output impedances can be matched, but not for both impedances. Does this matter in an audio application? If so, then it seems some voltage divider arrangements are not appropriate and only designs in which the impedances are matched both directions should be used.
4. Some voltage dividers are unbalanced while others are balanced. Is either unbalanced or balanced preferred for audio applications?
5. Does the "balanced" aspect of a balanced voltage divider have any relation to "balanced" audio signals (e.g., XLR cables)? Edit: I forgot to mention that the Motu can receive a balanced (1/4" TRS plug) or unbalanced signal (1/4" TS plug).
6. The MOTU 2408 MKII sound card is DC coupled which may cause a DC offset. While I understand how the value of a DC blocking capacitor is determined [Xc(ohms) = 1/(2*3.14*f*C)], it is not clear where the capacitor should be place in a voltage divider. Does it matter if it is placed before or after R1 (or somewhere else)?
7. As the Class D amplifier sources have an unknown high switching frequency, should a low pass filter also be added to the voltage dividers? If so, it is not clear where the capacitor should go in a voltage divider. Does it matter if it is placed before or after R1 (or somewhere else)?
I try to learn as much as possible before posting, but as after many hours of studying, I'm not making much progress. It seems that much of audio involves rules of thumb known by those in the art and many educational articles do not directly related to audio or sound quality. Attached is the project with the areas of concern in red.
1. The source volume can be adjusted. However, the listening volume will always be adjusted after the sound card. What level should the source volume be set at to maximize sound quality (S/N ratio/distortion)? I assume maximum volume, but perhaps when considering the interaction with the passive attenuator, a lower source output may be appropriate. Also, as the sources are Class D amps, perhaps there is an optimal volume level which has less distortion. It seems determining the volume level is important as it will change the attenuationl level needed and the component values (e.g., resistor Ohms rating).
2. What type of passive attenuator is ideal for this application? As bass/subsonic signals are desired, audio transformers are not ideal due to bass frequency loss and the cost of quality audio transformers which preserve bass signals is prohibitive with planned channel expansion (up to 40 mono sources). Thus, it seems that voltage dividers/"pads" such as L-Pads, Pi-pads, T-pads, etc. are the ideal solution considering compromises of cost vs. sound quality. However, in reading about the various voltage divider configurations, it is not obvious which voltage divider designs are suited for which applications, particularly this one.
3. Related to question #2 above, the impedance of the source and sound card are likely mismatched (I'm struggling to relate the amp Watts output to sound card dBu input). Some authors mention the voltage divider source and output impedances can be matched, but not for both impedances. Does this matter in an audio application? If so, then it seems some voltage divider arrangements are not appropriate and only designs in which the impedances are matched both directions should be used.
4. Some voltage dividers are unbalanced while others are balanced. Is either unbalanced or balanced preferred for audio applications?
5. Does the "balanced" aspect of a balanced voltage divider have any relation to "balanced" audio signals (e.g., XLR cables)? Edit: I forgot to mention that the Motu can receive a balanced (1/4" TRS plug) or unbalanced signal (1/4" TS plug).
6. The MOTU 2408 MKII sound card is DC coupled which may cause a DC offset. While I understand how the value of a DC blocking capacitor is determined [Xc(ohms) = 1/(2*3.14*f*C)], it is not clear where the capacitor should be place in a voltage divider. Does it matter if it is placed before or after R1 (or somewhere else)?
7. As the Class D amplifier sources have an unknown high switching frequency, should a low pass filter also be added to the voltage dividers? If so, it is not clear where the capacitor should go in a voltage divider. Does it matter if it is placed before or after R1 (or somewhere else)?
I try to learn as much as possible before posting, but as after many hours of studying, I'm not making much progress. It seems that much of audio involves rules of thumb known by those in the art and many educational articles do not directly related to audio or sound quality. Attached is the project with the areas of concern in red.
Attachments
Last edited:
The main volume is before the class D amps? And why coming from amps rather than a pre amp or wherever volume is controlled? (What's the signal chain before the amps?) Do you have a spec sheet for the amps themselves?
Edit: I re-read your post and see the main volume adjustment is after sound card. I'm still curious on why it needs to go through the initial amp at all.
Edit: I re-read your post and see the main volume adjustment is after sound card. I'm still curious on why it needs to go through the initial amp at all.
Last edited:
1. Simple trimmer is enough. It seems that this will need to be adjusted just once, to match the signal sources level. If you identify lowest signal source, you do not need to put anything there, just attenuate higher signal sources.
7. No, this is separate issue. This is addressed by proper amplifier construction.
Besides, if the amplifier generates higher fr hash above audible range, probably less than your dac, how is any filter in your passive attenuator going to affect it? Makes ni sense.
Besides, if the amplifier generates higher fr hash above audible range, probably less than your dac, how is any filter in your passive attenuator going to affect it? Makes ni sense.
The source is propriety and the manufacturer will not provide any specifications other than what I have already provided.
Yes simple trimmer is all you need, something x10 smaller than the input impedance it will need to drive. Maybe 1k pot with audio taper?
But make sure you put that in parallel will a 4-32 Ohm load. The Class D amplifier needs to see that on its output. By putting a large value pot in parallel with the load, the amplifier basically just sees the load. But make sure the pot has a much larger value than the load.
No need to impedance match or mess around with L-pads. You just want to transfer voltage to the input of the sound card.
This makes sense. However, I don't know how "proper" the source Class D amplifier was constructed. The only manual I can find for the MOTU 2408 MKII is in German, but it seems the max frequency response is 20kHz, typical of MOTU's other pro audio gear of the same vintage.
OK, understood. Without spec sheet or details, not possible to recommend ideal volume setting to minimize distortion. @boyfarrell is spot on re: paralleling a 4-32 ohm load since amp is class D. Load may affect frequency response and THD+N. 3W at 4 ohm is 3.5V RMS, at 32 ohm it's 9.8V RMS. Some experimentation may be necessary to get optimal performance.
Forgive my lack of knowledge, but why use a pot? The source already has adjustable output (volume) control. I failed to mention in the OP that the volume level of each source will be identical as the sources are all identical circuits, they just contain different sound files (all the same manufacturer). I see the value of using a pot in the design stage to determine the attenuation required. However, when there are 40 channels, that means 40 pots which will never be adjusted once initially set. So space becomes somewhat of a concern. It seems a voltage divider would take up less space.
You can get tiny trim pots, I like the ones from Bourns https://www.bourns.com/products/tri...ers/trimpot-trimming-potentiometers-multiturn
It would not be too unusual to see these in audio electronics, for infrequent use applications like biasing or setting levels.
If you can get away with a simple resistor divider that’s fine.
It would not be too unusual to see these in audio electronics, for infrequent use applications like biasing or setting levels.
If you can get away with a simple resistor divider that’s fine.
Yes, if there was only one input, there would be less concern. However, I'm not sure if pro audio sound card inputs are isolated so that the power from one source does not feedback into another source. In other words, sound cards are intended to accept low level signals (e.g., line level, mics, etc.), not power Class D amplier outputs.
That makes sense. In effect, the sound card would only be seeing a low level signal. Are parallel resistor(s) sufficient to simulate the 4-32 Ohm load?
The signal that goes to the sound card is taken across the potentiometer. So you adjust this to line level.
I don’t quite follow everything you wrote about feedback.
But it did make be think a bit more. Some class D amplifiers need floating output - without ground connection. Connecting to the input of the sound card might short it out. Obviously the is no way to know without a data sheet.
If you want it to be bullet proof you could use an audio isolation transformer after the voltage divider and before the input to the sound card. In this case maybe I would use a 1:1 (600 Ohm or 1k Ohm). They tend to be a bit expensive for good frequency response.
Yes total fine to do that. Use the smallest number you can get away with and do check the final value of the combination; tolerances with systematic errors can add quickly if you use a lot parts.
This makes sense now. Thank-you!
Yes, this is what is concerning. MOTU won't provide a data sheet. They will not even provide a PDF as this sound card is legacy.
I was hoping there would be a bulletproof means without audio transformers due to the cost and space. Space is not a huge concern, but it would be nice to fit a PCB board (or two) in a small 19" AV rack with the sound cards. This is an interesting approach to place transformer after a voltage divider. The other approaches I had found used a step-down transformer instead (and were only one source). I will look for 1:1 audio transformers as they may be less expensive.
Some classD amps can not have negative signal connected to ground.
Some can not have two negative speaker outputs, like left and right connected to each other.
This may present problem when all plugged into the same soundard with only one ground.
Those isolation transformers may be a necessity.
Some can not have two negative speaker outputs, like left and right connected to each other.
This may present problem when all plugged into the same soundard with only one ground.
Those isolation transformers may be a necessity.
Thank-you for explaining the concern better than I could.
I still do not understand what are you trying to achieve, if just mix many signals, or individually select...why classD as signal source, why not just operational amp or simple buffer. Since i do not understand the purpose, i can not help anymore. Cheers.
The purpose is to listen to muliple sources of audio. Once the sources are in the digital domain, they can be selected/deselected, mixed, manipulated, etc. indefinitely via VST plugins. Then output for listening. The souce is a Class D amplifier because that is the form factor it is commecially available in. The source normally drives small speakers (e.g., mobile phone size). My goal is to listen via headphones instead and be able to manipulate the sounds as desired. As I already have the source and sound cards, I'm simply trying to find the best way to send the sources' signals from amplifier to sound card. Thank-you for your input and Cheers!
I don’t understand either. Guess OP has his reasons.
I’d say it’s pretty unlikely the inputs of the sound card are transformer coupled. So if output of class D needs to be floating or connected with care… not good.
10 transformers are fairly inexpensive https://www.aliexpress.com/item/32827431499.html no idea if the frequency response is any good. Ask around maybe somebody here has measured them.