[sapphari]

This seems like a super basic topic but after googling and searching here I haven't really found the answer.

I'm using a digital out from a head unit, which is a first for me (I've only used analog out previously). My first question is, assuming EQ and all that in the HU are flat, do I need to check to see at what volume the HU stops putting out a clean signal? If so, how is that done with a digital signal?

Second question is how do you set input gains on a DSP when using a digital input?

 

[Slave2myXJ]

A digital signal will not clip unless the source signal has already done so, such as if you're using an external ADC, which you're not.

Since the digital communication spec dictates what 0 to 100% means, there is no ambiguity between input and output, thus no need for gain adjustment...

The problem with a digital connection is that you lose granularity as volume is reduced, much like encoding an MP3 at a lower bit rate... If you're going to use a digital connection, keep the volume at max on your source and use the DSP to adjust overall volume.

Edit: I'll add that you should always leave your source at the highest level it can support without distorting, regardless of whether it's analog or digital, to maximize the signal, and then regulate the volume (voltage in an analog system, range in a digital system) as close to the amplifier as you can, generally the DSP or EQ, maintaining the highest possible signal strength throughout the system.

Keeping the signal strength at the maximum possible value keeps the effect of any stray noise low, and then with minimal gain, we amplify the noise very little. Gain = amplification. THE WORST POSSIBLE THING YOU CAN DO is set your gain with the clip light, or with a multimeter, and then turn your volume down to achieve a normal listening level. In doing so, you're reducing the signal strength while the noise signal is staying the same.

I highlighted distorted. A digital source won't distort since it's just 1s and 0s, but an analog source certainly will long before it clips. Clipping occurs when amplifying a signal and there's simply no more range available. Instead of being a smooth sin curve, it runs out of steam and flattens off at the top, causing rapid acceleration/deceleration of the voice coil. Clip lights on amplifiers are actually very accurate for determining whether their amplification is causing clipping of the signal; it's a basic input to output comparison with a common IC.

Distortion on the other hand, will happen long before clipping as the transistors providing the regulation becomes less and less "linear". Distortion can only be determined by comparing waveforms since it's a much more "nuanced" changing of the signal than a clip. Clipping was much more of an issue back when a 25w x 2 amplifier was the size of a dinner plate and cost as much as two car payments. Now a days, a person who actually listens to their music rarely has to worry about clipping if they're doing things properly.

So, long story short, with a digital connection, you should be able to keep it at maximum volume for maximum signal strength, and the modulate volume downstream. With an analog connection, keep it at a nice high level, where there's no distortion, and by extension, no chance for clipping at this point in the signal path. Always modulate volume at the last reasonable point in the signal path to keep the effect of stray noise as low as possible.

 

[Anu2g]

I agree with everything that @Slave2myXJ is saying.

I, personally, have elected to go with a bit of a trade-off. I'm using a Stinger Heigh10 head unit with optical out to a MiniDSP 8x12 DL. I am using the Stinger Heigh10 to control the volume, which means the digital signal gets compressed at lower volumes as @Slave2myXJ has pointed out. However, I've tuned it in such a way that I'm generally in the top 25% of the volume range of my head unit, so the compression should be minimal.

The trade-off is that I am gaining the ability to use my steering wheel volume controls (not the DSP remote) in exchange for a bit of compression. But the amount of compression I get, I cannot notice. When I demo'd at the last meet, I demo'd at 35 out of 40 on the head unit.

But again, I agree with the prior post that the best possible sound would be by setting my HU to 40/40 and using the DSP remote to control the volume (analog outputs to the amps).

Hope that helps.

 

[Anu2g]

Also, to specifically answer how I set gains, I had set my HU to 35/40, left the DSP outputs at 0dB, and then adjusted amp gains accordingly. No need to measure clipping on the HU. This is assuming you want to use the HU to control volume. If you go with @Slave2myXJ 's approach, then you should set it to 40/40.

 

[Slave2myXJ]

I absolutely agree with what you're doing. Convenience always wins out and you are still maximizing signal strength.. It just boils my blood when I hear of people setting gains with a calculator and a multimeter...

 

[sapphari]

Thank you both for your replies, both are super helpful. Good to know that the digital out on the HU should be clean up to 100%.

My DSP will also be the CDSP 8x12 DL, so @Anu2g I can relate (and probably have some more questions for you). The remote for the CDSP is huge, 4" long! I don't have a great spot for that anywhere in my console area, so I may see how things go using the HU to control volume (and try to stay in the upper 25% like you do). I saw this article recently that points out that this digital signal attenuation might not actually degrade or compress the data, as long as upsampling is taking place. I'm not 100% sure this applies here though since he's talking about DAC, and we are not.

Digital volume controls – PS Audio

www.psaudio.com

 

[sapphari]

Also, to specifically answer how I set gains, I had set my HU to 35/40, left the DSP outputs at 0dB, and then adjusted amp gains accordingly. No need to measure clipping on the HU. This is assuming you want to use the HU to control volume. If you go with @Slave2myXJ 's approach, then you should set it to 40/40.

Please let me know if this is how you are doing things:

1. You leave the input gains on the CDSP at 0dB
2. You leave the gains in the routing section and mixer section at 0dB too
3. You use a "straight through" (no Dirac applied) setup, so the gains in the DL tab are all 0dB too (??)
4. Then you set gains on your amps

What do you use as the input source in this process? Something like a -5dB 1kHz or 40Hz sine wave, or...?

That's how I was thinking about doing it. Then I'd go back and apply Dirac. In this case I think the Dirac target curve needs to not have anything greater than 0dB. I also need to think about how to set my remote bass knob during this process, since I definitely want to be able to adjust bass level on the fly.

I think gain setting and CDSP might need its own thread!

 

[Anu2g]

Please let me know if this is how you are doing things:

1. You leave the input gains on the CDSP at 0dB
2. You leave the gains in the routing section and mixer section at 0dB too
3. You use a "straight through" (no Dirac applied) setup, so the gains in the DL tab are all 0dB too (??)
4. Then you set gains on your amps

What do you use as the input source in this process? Something like a -5dB 1kHz or 40Hz sine wave, or...?

That's how I was thinking about doing it. Then I'd go back and apply Dirac. In this case I think the Dirac target curve needs to not have anything greater than 0dB. I also need to think about how to set my remote bass knob during this process, since I definitely want to be able to adjust bass level on the fly.

I think gain setting and CDSP might need its own thread!

There are a ton of different ways to set gain; I don't think my way is better than how others do it, nor do I feel that strongly about my particular approach, but I'll outline it here anyways in case it's helpful:

1. Set amp gains at minimums
2. Leave input/mixer/output gains on DSP at 0dB
3. Set your Dirac settings to 7 channels (assuming active 3 way plus sub)
4. Run Dirac
5. Only take the first measurement
6. On the filter design screen, draw (or import) your target curve
7. Export the filters to your DSP
8. Look at the Dirac tab, specifically at the gain adjustment that was performed for each of your drivers
9. Looking at the gain adjustments that Dirac performed, you can kind of get a baseline for what the difference in gains needs to be between each channel. Assuming your amps only have gain adjustment per pair of channels, i'd look at whichever channel played with lower volume (i.e. right tweet, as opposed to left tweet)
10. Now I adjust the amp gains accordingly, both between pairs of drivers, but also I adjust them all up a bit to make sure I get enough volume

It's up to you if you want to repeat a ton of times. I don't. Another approach i've done was to adjust the amp gain on the tweeters up to the point that I heard some floor noise out of my tweeters, then adjust it down a notch so that there's no noise; then run Dirac, and figure out how much to adjust up the amp gains on the other drivers.

Does that make sense?

It may be of note that my main amp is an MMATS HiFi-6150d, which puts out 150w RMS to each channel, and I've not gotten to a point where amp clipping is an issue.

 

[Theslaking]

The remote for the CDSP is huge, 4" long!

I probably I have one of the half-sized, older, less capable ones laying in a box of it would help with your install.

 

[sapphari]

There are a ton of different ways to set gain; I don't think my way is better than how others do it, nor do I feel that strongly about my particular approach, but I'll outline it here anyways in case it's helpful:

1. Set amp gains at minimums
2. Leave input/mixer/output gains on DSP at 0dB
3. Set your Dirac settings to 7 channels (assuming active 3 way plus sub)
4. Run Dirac
5. Only take the first measurement
6. On the filter design screen, draw (or import) your target curve
7. Export the filters to your DSP
8. Look at the Dirac tab, specifically at the gain adjustment that was performed for each of your drivers
9. Looking at the gain adjustments that Dirac performed, you can kind of get a baseline for what the difference in gains needs to be between each channel. Assuming your amps only have gain adjustment per pair of channels, i'd look at whichever channel played with lower volume (i.e. right tweet, as opposed to left tweet)
10. Now I adjust the amp gains accordingly, both between pairs of drivers, but also I adjust them all up a bit to make sure I get enough volume

It's up to you if you want to repeat a ton of times. I don't. Another approach i've done was to adjust the amp gain on the tweeters up to the point that I heard some floor noise out of my tweeters, then adjust it down a notch so that there's no noise; then run Dirac, and figure out how much to adjust up the amp gains on the other drivers.

Does that make sense?

It may be of note that my main amp is an MMATS HiFi-6150d, which puts out 150w RMS to each channel, and I've not gotten to a point where amp clipping is an issue.

Thanks for spelling it out. Two questions though. Step 5, why only take 1 measurement? Do you go back after this whole process and run the full 9 measurements? And in step 10, how do you know how much to increase the gains on the amp? It this just a bit of intuition (and informed by the results from step 8/9)? Or is there something more empirical happening here?

 

[Anu2g]

Thanks for spelling it out. Two questions though. Step 5, why only take 1 measurement? Do you go back after this whole process and run the full 9 measurements? And in step 10, how do you know how much to increase the gains on the amp? It this just a bit of intuition (and informed by the results from step 8/9)? Or is there something more empirical happening here?

My pleasure!

You could take all measurements, but we're just doing this measurement for setting amp gains, so I think one is enough. Once gains are set, then start over with the measurements and complete all 9.

And yea, just using intuition and guess/check. Not using any fancy science, keeping in mind that Dirac will also update the DSP gains accordingly.

 

[bertholomey]

Good stuff Anu - good way to use the DSP as a tool to get a desired result.

I have always been challenged with this question - never truly understood gain setting though I have seen it done multiple times. The idea of setting gain by ear never made sense (crank it till you hear distortion, and then back off a bit).

Those who have set my gains typically have used an O-Scope to determine ‘max-unclipped’ signal.

When I had the Helix - set everything at 0 - no EQ, levels at 0. Then use the Autosound 2000, disc 104 to get various tracks at various frequencies specific to individual drivers - at either -5db or -10db. Adjust gain on the amp until the O-Scope shows a good sine wave - not a flat top.

Then we would tune the system - adjusting levels, making significant cuts to EQ - which sometimes left me with very little headroom - near the top of the volume knob, and not enough output. We were using a lot of EQ cuts on those tunes.

A friend stated that in his research of how engineers set up sound systems for large venues (concerts, clubs), pre-EQ is done for each speaker near field, that is stored, then the speakers are placed in the venue and EQ for the room is done, then gain is verified / adjusted to max unclipped to compensate for the reduction of signal through EQ (this is from a faulty memory, but I think I have stated it correctly).

So we started to do it this way in the car - gains at zero, tune the car with levels / EQ applied, then use the disc 104 to set gains. More output was achieved......but the problem I have had and haven’t solved - what if you want to do another tune on another preset? In the concert / club analogy - the pre-EQ (near field of each speaker) doesn’t change - that is stored.....they zero the gains, put the speakers in another venue, EQ to that room, then maximize the gains on the amp.

But if I want to do another tune, my gains are set based on the tune I did before - which may or may not had a lot of level difference in the DSP and a lot EQ or very little compared to the new tune. This is my current situation.

Yours or Ryan’s method where Dirac is employed (for those of us with the MiniDSP DL) could be very effective for those have easy access to the gains on their amps and want one good tune (not a different tune per preset - comp, drive, fun, genre-specific). Zero everything, run Dirac (which will make correction for EQ / levels), then maximize the gain for that tune. And potentially through this straightforward method, one could do this for each ground-up tune that is done if desired. Unfortunately, my gains are a real pill to get to and to adjust. My installer recently adjusted mine - had a tune on the car, used disc 104, set everything according to the disc, and I get decent volume - but I’m always wondering if it is ‘optimal’.


Sent from my iPhone using Tapatalk Pro

 

[sapphari]

I hear you about the 'wondering if it's optimal' part. It's a bit of a puzzle when the DSP has 3 stages of gain, and they are channel specific. Then gains are changed again through EQ. And in the situation when gains on the amp control more than 1 channel... and a lot of us have remote gain on the sub stage... there is a lot to consider! I suppose one solution is just to have a lot of power on tap, so that even if your amp is not optimally gain matched to its input, you'll still have sufficient volume. Luckily this is the boat I'm in.

At some point I may create a new thread for gains and the CSDP 8x12 DL since it's a different topic than this thread and specific to this piece of gear.

 

[hella356]

I also have a Heigh10 & DL via optical. I was lucky enough to have a good spot in my dash for the DL controller. I keep the HU volume at max & use the DL for volume. Bonus is that I prefer that knob over the slower acting HU knob. I just left my amp gains the same as they were from my previous HU.

 

[sapphari]

Just adding this here for anyone reading who is interested in the topic of setting gains with a DSP. I never did start a thread on this specifically for the CDSP, but there is a useful thread on the general topic (specifically post #21):

The whole gains thing is making me lose my hair.

gain setting is done for the amps, not drivers, with the drivers disconnected... disconnect all drivers turn HU up to clipping, then down one (on many HUs sub output will have its own clipping point to, at the max clean volume, so check it next before moving on), then turn dsp up to clipping...

www.diymobileaudio.com