I did a bunch of before-and-after measurements, just never got around to writing the article about it. One of these days...
Well, here are mine.
This is the electric output of my amp driving my mid-treble ESP panel, the worst imaginable amp load.
The THD curve (mostly south of -65dB which is like .05% total everything) reflects every component in the live measurement chain from my crummy headphone amp in my laptop through the entire measurement loop, DAC, pre-amp, 1982 amp, cheap Behringer mic mixer, etc. All these items are typically in a measurement chain, just mentioning it here so nobody thinks the already miniscule .05% THD is only from the DCX2496.
Fuzzy stuff south of Xover point typically measurement artifact. Please ignore.
DCX is doing EQ and Xover, as you can plainly see, and time delays. Using most of its brainpower
This is "before" mods and, as you can guess, I see no reason, so far, to try any mods.
This took me about 15 minutes to create. OK... who wants to post their curves before and after?
Ben
Attachments
Last edited:
I wish I had taken a bunch of measurements before upgrading my output stage, but this is an example of what can be done by replacing the main board 'lytics with OSCONs. Nothing else, just cap replacement. I have managed to improve things beyond this, but nothing that I can reliably document as "before/after" based on a single specific mod.
Attachments
Pictures always perfectly clear to the person who posted them.
What are we looking at? What does it tell us?
No before and after. Kind of sloppy way to make improvements, eh.
Ben
Last edited:
Where you see the original circuit fall down is distortion at higher frequencies. 1k usually looks fine. 5-10k, much less so.
If you need help interpreting the differences between the two, then maybe you shouldn't be stating things as being sloppy. There was improvement in cap upgrade, that's all you get, since your sarcasm is unwelcome.
If your DCX without mods does not need improvement, then don't bother posting in the DIY section of a DCX thread. After all, you haven't DIY'd anything.
There was no sarcasm what so ever. I stated that your pictures needed explanation and I stated my opinion that to proceed with mods without first establishing your baseline is sloppy. No innuendo at all.
BTW, I've been testing and modifying audio gear for more than half a century until 11:56 this morning, including testing in the world's largest anechoic chamber. So your gratuitous put-down is out of place.
Ooooops, I forgot to mention one mod I did to my DCX2496. I hacksawed off those awful rack-mount ears. Pity I didn't listen for sonic improvements like other folks.
Ben
Last edited:
Always great to get advice about removing the L-shaped wings from someone who hasn't done it.
I guess you've never seen what the sides look like with the wings missing? Have a peek.
Yup of course, anybody has got to take 'em off to hacksaw the rack-mount appendages. But then you've got to put 'em back to cover up! And I heartily recommend this mod to all who value nice looking equipment.
B.
Anyone who has removed the wings will know there is a big gaping hole left on the front end of the side covering. So the wings can not be left removed. I prefer honest comments to posturing.
I wish I could produce before and after pictures but I forgot to document the baseline condition.
Ben
You have been given ample about where you can find them. Don't make us do your homework!
What were the difference between the caps you replaced and the oscons, capacitance, tolerance, working voltage, package size and shape, operating temperature, manufacturer of the original caps etc.
I have seen similar (on volume products) between a prototype built with decent caps then production runs were purchasing get to choose the caps the main choice being value... Size and shape is something I got from an old analogue engineer who I use to work with, he would choose electrolytics based on size and shape as well as other parameters, looking for a large surface area where any power was involved and at other parasitic's such as inductance for his critical circuits, lead length, PTH or SMD shape etc. having an effect.
I presume the left plot is the before and the right the after... so some reduction at 2 and 3Khz mainly and a slight reduction in the noise floor.
Marce:
Your observations are correct; I knew the analysis wasn't too difficult, a simple before and after picture. The files are labeled, but they don't come through obviously on the post, so if you mouse over the image the name does show up in the browser status bar.
The improvement was arguably minor and subtle, but nonetheless was a measurable improvement. As SY notes, higher frequencies become even more obvious. At each step along the way of upgrading (digital stage, output stage), I have managed to obtain marked improvements in performance and reduced noise floor. It's actually a fun little device to mod.
My criteria for cap upgrades was to try and get a similar capacitance (sometimes larger, never smaller), with the constraint of exactly matching the lead spacing of the board. Annoyingly Behringer decided to pick very small lead spacing for the case size (there are two common standards) and this became the largest challenge.
I normally try to increase the voltage rating if space allows, but in this case options were limited due to space. Increase capacitance if possible, but don't go crazy and double it. Sometimes you create more problems by arbitrarily tossing in extra capacitance (startup sequence, impedance mismatch, etc). Since I had a board schematic, and knew if the particular cap was running on the 5V regulated rail, I had no concerns about putting in a 10V OSCON when the original might have been 16V. Use smart engineering when you were forced to make a decision.
The bulk capacitors fed directly from the off board power supply were tricky, as the package diameter was a little larger, and they fit very tight. But I was able to meet the voltage rating and get a little extra capacitance out of them. Since I am still running the stock power supply (gasp) I prioritized the board level caps to get my performance squeezed out. Happy to see it did help a little.
I can't recall the mfg of the caps; but they were some off-brand that you don't find at a Newark, Mouser, Digikey, or Allied. It felt good to get them out. OSCONs were either Panasonic or Nichicon (now discontinued models) that were based on what I could actually locate in stock from Mouser. Standard radial through hole packages.
The biggest improvement I have implemented on the device is the use of Jensen output transformers JT-11-DMPC directly connected to the DAC output. No resistors, no capacitors, just xfmr isolation and filtering. Levels are now consumer grade, and noise floor is fantastic. Distortion about as low as can be obtained from the DAC; it's nearly identical to AKM's app note on the DAC.
Your observations are correct; I knew the analysis wasn't too difficult, a simple before and after picture. The files are labeled, but they don't come through obviously on the post, so if you mouse over the image the name does show up in the browser status bar.
The improvement was arguably minor and subtle, but nonetheless was a measurable improvement. As SY notes, higher frequencies become even more obvious. At each step along the way of upgrading (digital stage, output stage), I have managed to obtain marked improvements in performance and reduced noise floor. It's actually a fun little device to mod.
My criteria for cap upgrades was to try and get a similar capacitance (sometimes larger, never smaller), with the constraint of exactly matching the lead spacing of the board. Annoyingly Behringer decided to pick very small lead spacing for the case size (there are two common standards) and this became the largest challenge.
I normally try to increase the voltage rating if space allows, but in this case options were limited due to space. Increase capacitance if possible, but don't go crazy and double it. Sometimes you create more problems by arbitrarily tossing in extra capacitance (startup sequence, impedance mismatch, etc). Since I had a board schematic, and knew if the particular cap was running on the 5V regulated rail, I had no concerns about putting in a 10V OSCON when the original might have been 16V. Use smart engineering when you were forced to make a decision.
The bulk capacitors fed directly from the off board power supply were tricky, as the package diameter was a little larger, and they fit very tight. But I was able to meet the voltage rating and get a little extra capacitance out of them. Since I am still running the stock power supply (gasp) I prioritized the board level caps to get my performance squeezed out. Happy to see it did help a little.
I can't recall the mfg of the caps; but they were some off-brand that you don't find at a Newark, Mouser, Digikey, or Allied. It felt good to get them out. OSCONs were either Panasonic or Nichicon (now discontinued models) that were based on what I could actually locate in stock from Mouser. Standard radial through hole packages.
The biggest improvement I have implemented on the device is the use of Jensen output transformers JT-11-DMPC directly connected to the DAC output. No resistors, no capacitors, just xfmr isolation and filtering. Levels are now consumer grade, and noise floor is fantastic. Distortion about as low as can be obtained from the DAC; it's nearly identical to AKM's app note on the DAC.
Cheers.
There are some suppliers such as Anglia that provide bulk components for very high volume, then you can also go direct to the far east... I have had some fun with capacitors and assemblies such as 35,000 snap fit caps bought cheep by a purchasing dept. that didn't snap into the boards and raised up during selective solder causing lots of re-work, but they did save a few pence per capacitor....
There are some suppliers such as Anglia that provide bulk components for very high volume, then you can also go direct to the far east... I have had some fun with capacitors and assemblies such as 35,000 snap fit caps bought cheep by a purchasing dept. that didn't snap into the boards and raised up during selective solder causing lots of re-work, but they did save a few pence per capacitor....
There are some interesting things available from the Far East. 😀
