DCX is by no means bad, it's is actually very good piece of equipment. Even unmodded DCX is very transparent in the signal path of a hi-resolution system. Modding DCX besides op-amps and (someway) reduced hiss might be totally waste of time.
The half that goes away when switching to AES/EBU will only be reduced in a modded DCX that changes the input buffer, like maybe to an OPA1632, or maybe if the DCX has an upgraded power supply. Personally, I'd bet on the OPA making a bigger difference.
Is the input buffer right in the next to the power supply in the PSU(dip8 op-amp looking thing)? I will change that also when I change PSU's capasitors. OPA1632 is the best one available? Or do you mean the op-amps in the I/O board?
edit: I am going to change all the 17 op-amp of the DCX to LM4562s, also input op-amps.
Last edited:
Gain Structure
OK, so now we know that part of the hiss is coming from the analog input and part from elsewhere. As Michael says, once you get the levels set properly, you shouldn't get any hiss. This is known as "Proper Gain Structure."
Here is one method of setting it up properly:
I have to run now - but will continue later.
* If you don't know how to measure this, I can help.
OK, so now we know that part of the hiss is coming from the analog input and part from elsewhere. As Michael says, once you get the levels set properly, you shouldn't get any hiss. This is known as "Proper Gain Structure."
Here is one method of setting it up properly:
- Choose a piece of dynamic music, about the most dynamic you usually listen to. A track with at least an average to peak value of 16dB is needed.*
- Set all gains in the DCX to 0dB (or at nominal, if you use different gains for each section)
- Play this track at the loudest level you think you would normally listen to. Check the input and output VU meters. Note what they say.
- Where are the peaks on the inputs? They should be just hitting the yellow LEDs. If they are not, go to next step.
- Reduce the gains as much as possible on the outputs. -15dB. Now bring up your preamp volume until you hit the yellows on the input VU meters - or - you get to the same high acoustic level as before.
- If the track is now hitting the inputs and lighting the yellows LEDs, your input level is good. If it's still below that, you'll need more than 15dB attenuation on the DCX outputs.
- If you are not at the loud music level from before, bring up the gains on the outputs until you hit the same volume. Note this setting and note where the peaks are on the output VU meters.
- If you succeed in obtaining a strong level on the inputs and a "loud enough" volume with the outputs attenuated by the DCX menu, then you will know that you need the same amount of attenuation as found in the DCX output menus.
- If -15dB on the outputs is not enough to give you a strong level on the outputs and not blow up your speakers, you'll need more than 15dB attenuation between the DCX and your amps.
I have to run now - but will continue later.
* If you don't know how to measure this, I can help.
Optimising the Gain Structure
Legis
The best results with the DCX have generally been acheived by using a digital source in to the DCX so that it is running flat out and then attenuating the outputs using a six channel volume control.
You are on the other hand using the DCX as a conventional active crossover.
In order to optimise the gain structure, you need to drive the input harder (to pro audio levels) and then attenuate the DCX outputs into your amps.
Simple resistive divider networks on your signal leads between the DCX and your power amps will give you a more optimal gain structure, and reduce the hiss. I think that in this case this would give you the performance gains that you are after with least effort / cost.
Legis
The best results with the DCX have generally been acheived by using a digital source in to the DCX so that it is running flat out and then attenuating the outputs using a six channel volume control.
You are on the other hand using the DCX as a conventional active crossover.
In order to optimise the gain structure, you need to drive the input harder (to pro audio levels) and then attenuate the DCX outputs into your amps.
Simple resistive divider networks on your signal leads between the DCX and your power amps will give you a more optimal gain structure, and reduce the hiss. I think that in this case this would give you the performance gains that you are after with least effort / cost.
OPA1632 is a replacement for the arrangement of JRC4580's and other components used in the input section between the A/B/C inputs and the ADC's. It requires circuit redesign to swap in.
Jan's I/O board uses the OPA1632.
Cheers,
Paul
What I have read, changing all the 17 I/O board's the op-amps will reduce the hiss/noise significantly. If the changing won't help at all, I'll have to look for those attenuator to be put after DCX.
My main plan was to use DCX with analog inputs as I have read that it actually performs better (excluding louder hiss) with them than digital inputs. Analog inputs are also more convenient than digital inputs since I can control the volume with my preamps (especially with my tube preamp
).
Yep, I read that OPA1632s perform well, but, as you said, they don't suit well in the original desing as they run too hot. LM4562s should work fine based on what I have read. Do you have experience of how much changing the input and output op-amps to OPA 1632/LM4562 or equivalent will reduce the hiss?
What I have read, changing all the 17 I/O board's the op-amps will reduce the hiss/noise significantly. If the changing won't help at all, I'll have to look for those attenuator to be put after DCX.
My main plan was to use DCX with analog inputs as I have read that it actually performs better (excluding louder hiss) with them than digital inputs. Analog inputs are also more convenient than digital inputs since I can control the volume with my preamps (especially with my tube preamp
Maybe, but that's a bit brute force. There are other circuit optimizations that can be done besides just swapping different op amps into the existing topology. For instance, switching from the existing input buffer to the OPA1632 reduces the input path from six op amp stages to the one OPA1632, simplifies the addition of the 2.5V offset, and allows for DC coupling.
Similarly, the output side can be simplified and DC coupled as well. If you're looking at such wholesale changes as swapping out all the 4580's, I'd really consider swapping in Jan's I/O board instead.
Cheers,
Paul
Yep, Jan's mods are quite pricey though. Are these I/O board kits good? Selectronic : Vente par correspondance de composants électroniques
You are right, maybe I should have ordered just two of these kits (if they are any good). But this way I did it I saved the price of the other kit.
Interesting comment on the "too hot." Do you have more details on this?
The LM4562 idles at 10mA = .3W.
The OPA1632 idles at 14mA = .42W.
A bit hotter, yes, but not excessive for what I'd expect to be able to radiate from an SOIC-8 package.
FWIW, the NJM4580 idles at 6mA = .18W ... so the LM4562's are already almost twice as hot as the stock IC's... and there are a lot more of them on the board. One OPA1632 replaces three NJM4580's, so final power draw is about the same.
It was said that OPA1632 runs (too) hot if used in +/- 15 volts: Print Page - DCX2496 audio mod (Post by: jdbakker on March 10, 2009, 03:19:00 pm)
Last edited:
Optimising the gain structure
Legis
It may be possible to reduce the noise floor of the DCX but please also concider that this is a digital unit, and running it at low signal levels will restrict the resolution of the unit.
The aftermarket modifications for the DCX usually result in a much reduced voltage level for full scale digital input. Replacing the op amps would not optimise this aspect of the DCX performance.
Legis
It may be possible to reduce the noise floor of the DCX but please also concider that this is a digital unit, and running it at low signal levels will restrict the resolution of the unit.
The aftermarket modifications for the DCX usually result in a much reduced voltage level for full scale digital input. Replacing the op amps would not optimise this aspect of the DCX performance.
Have I understood it right that the hot and cold is floating on the output(within reason)? (relative to ground)
Could I then just take hot and cold plugging it directly into single ended amp. Hot to + and then grounding cold at the amp side.
Amp will be a simple follower with a gain of little under 1. Driving 100dB speaker.
I think the gain structure should be fine for my use, probably ending up controlling volume at the input. Ether digital or analogue in depending on final setup.
And quick PA related question. Will there be any ill effects if the DCX is powered on and off by removing external power. The amps will receive the same treatment (Yamaha P7000s and P5000s) by the same master switch.
Could I then just take hot and cold plugging it directly into single ended amp. Hot to + and then grounding cold at the amp side.
Amp will be a simple follower with a gain of little under 1. Driving 100dB speaker.
I think the gain structure should be fine for my use, probably ending up controlling volume at the input. Ether digital or analogue in depending on final setup.
And quick PA related question. Will there be any ill effects if the DCX is powered on and off by removing external power. The amps will receive the same treatment (Yamaha P7000s and P5000s) by the same master switch.
Do you mean that it will lose some of it's resolution if LM4562s are used? It would meand that LM4562s would drop the gain compared to stock op-amps, do they? Both have quite the same maximum output voltage swing (this measures the maximal gain?). I think the lower overall noise of LM4562 is the reason for lower hiss, not that it would actually reduce the gain?
If I turn the output's gain -7,5dB from DCX, doesn't that mean that DCX will always have 7,5 db greater dynamic range than the music signal will ever have. Does the "extra" (compared to signal) resolution and DR benefit somehow?
Last edited:
No. What he's saying is that most output mods also include lower gain in the output stage, which directly results in reduced noise from within the DCX and increased use of the digital dynamic range of the DACs - because you're more likely to drive the DCX near to its maximum signal level.
On the other hand, if you have background noise from your source and/or preamp, having to "crank it up", to make up for the reduced gain of the DCX after an off-the-shelf mod, may negate some or all of your benefit.
If all you're doing in swapping JRC4580's for LM4562's, the gain will be identical to stock.
On the other hand, if you have background noise from your source and/or preamp, having to "crank it up", to make up for the reduced gain of the DCX after an off-the-shelf mod, may negate some or all of your benefit.
If all you're doing in swapping JRC4580's for LM4562's, the gain will be identical to stock.
I've done it plenty and never had any problem at all. AFAIK, all the power switch does is cut the AC anyway.
EDIT. I checked. The power switch just cuts the hot wire of AC power. Nothing fancy. Pulling the plug or using a power switch on a remote panel would do the same thing. No worries.
Neither of those units have a digital input, so are less than useful for some of us. All the filters all these units use are FIR as far as I'm aware, so should be no difference in the digital domain, the analogue I/O are reasonably easily fixed by DIYers and they don't need to withstand the physical rigours of pro use at home, I don't see any benefit to either of them.
assuming 10Vpk from the DCX and using that through a unity gain buffer to feed 100dB/W @ 1m speaker will give ~111dBpk @ 1m. That is loud, if your speakers are capable of going that loud.
The system has sufficient gain.
This is the way to set up the DCX, set the maximum output of the DCX so that the final signal coming from the speakers is not quite clipped.
Then turn down the volume for normal listening levels, preferably after the DCX.