[bg4533]

I am about to order some parts to mod my EMU 0404. Below is a link to a picture of what I want to do. I only care about improving the analog output of this card.

Picture

Color Codes:
Yellow: 10uF/16V caps, replace w/ Panasonic FC 10uF/25V (16V unavailable)
Red: 100uF/25V caps, replace w/ Panasonic FC 100uF/25V
Pink: 47uF/16V caps, replace w/ Panasonic FC 47uF/16V
Green: Output caps and buffers, remove and jumper
Orange: Opamp, replace with LT1364, bypass cap, Bias to class A with ~1.5mA CRD

Removing buffers:
The datasheet lists the pinout of these opamps as:
1. A Output
2. A- Input
3. A+ Input
4. V-
5. B+ Input
6. B- Input
7. B Output
8. V+
I obviously have to bridge A output to one (or both) A input, but which? The same goes for B.

Bypass caps:
I think I am just going to use a Panasonic ECQP Polypropylene here. I have not yet decided on a size. Should I bypass from rail-rail or rail-ground? Rail to rail will be easier. What are the benefits of each method (I want to know this for future projects mainly). If I go rail to ground can I use the silver box by the buffers as ground? Anywhere better for ground?

CRDs:
I have been unable to find these on Digikey. Anyone know any part numbers?

Capacitors:
I want to keep this a fairly cheap mod, so I am going to stick with Panasonic FCs for now. Are there any other caps on this board I should upgrade or are there any that I am upgrading that don't pertain to analog output? Provided the caps fit, could there be any problems with using a larger capacitance cap? Prices are about the same, so I might as well use larger ones if it won't hurt.

Thanks for the help. After I finish I can put together a modding guide with pictures if people want. The cost for all the mods should be ~$15.

 

[llmobll]

WOw, someone has finally thrown together a little "how to." I anxiously wait to see the final result. Just out of curiousity are you planning on doing everything at once? If not what do you plan on doing first. All the Caps? Or removing the whole output stage?


Come to think of it why are you taking out the buffers?

 

[bg4533]

I was planning on doing things in 3 stages. In stage 1, I would do the free mods (remove buffers and ouput caps). Stage 2 would be replacing the opamp. Stage 3 would be replacing the caps. I will briefly test the card and write down my observations after each stage.

I have heard the buffers aren't necessary. The buffers are actually the same opamp as the main opamp. According to Glassman the buffers help keep distortion down a tiny bit. Removing them apparently improves the sound a bit. Ask Glassman for a better explanation.

 

[gaboo]

Quote:

Originally Posted by bg4533 I obviously have to bridge A output to one (or both) A input, but which?

If those really are non-inverting buffers (can't tell from the pic), the you bridge A+ input to A output. The same goes for B.

Here is the datasheet for the DAC used on 0404. You may find page 24 interesting. It is most likely what E-MU engineers have implemented.

 

[S_Dedalus]

Quote:

Originally Posted by gaboo If those really are non-inverting buffers (can't tell from the pic), the you bridge A+ input to A output. The same goes for B. Here is the datasheet for the DAC used on 0404. You may find page 24 interesting. It is most likely what E-MU engineers have implemented.

The + and - inputs are already bridged on the board, you just bridge the output to one of them.

 

[S_Dedalus]

C10 (red) probably doesn't need to be replaced, it looks like its part of the digital section.

 

[bg4533]

Quote:

Originally Posted by S_Dedalus C10 (red) probably doesn't need to be replaced, it looks like its part of the digital section.

Thanks. It is actually C103, but the 3 isn't visible in that pic. I wasn't really sure about that cap. It looks like it connects to the DIP8 chip in the bottom left and the E-MU chip in the upper right.

 

[Glassman]

the cirquit in the AK4395 datasheet indeed isn't what's used on 0404 but it's very interesting, why? E-MU engineers decided on using noninverting unity gain buffers to beef up the voltage outputs, because they found them weak and thus limiting measurable performance a bit.. but look at the AKM's suggested cirquit - they don't use any buffers of that kind and on top of that use ridiculously low resistor values, that means pretty high loads on the DAC's outputs.. hmm, that's really fun..

you don't need C103, C31 replaced, but try to find out if there are any caps connected to the VR1 voltage regulator and try to replace them too, it's the analog 5V for ADC and DAC.. I see just the 10u tantalum there.. C75 and the other yellow marked cap nearby could be higher value.. like 47uF/16V (purple

) for instance.. also, you don't need any cap with more than 16V capability, so 220uF/16V is a better idea than 100uF/25V.. just be shure they will fit..

as far as the bypass caps go, I suggest rail-to-rail bypass for simplicity.. finding ground there that wouldn't disturb the signal itself ain't easy.. 10-100nF polypropylene looks pretty reasonable..

and you of course can try any other opamp instead lt1364 if you want to.. ad8066 for instance would be great too.. try to google opamp's sonic characteristics and dedice on which would suit you the best, they will all work..

also you might find smt resistors connected after the four green marked caps to ground, directly after the caps, you should remove them too if you can.. if you're not up to it, please take detailed shot of that place after you remove the green caps.. thx..

 

[tschanrm]

I just removed the 4 caps and the buffer opamps, and replaced the opamp with lt1364 biasing it to class a with two 1.1 ma crd's. Doing all this at once gave a n improvement over the sound. Guitars felt much faster, and lower bass notes seem more refined. Cheers to that

My only issue is that when I went to test my card with RMAA I get the following results at 44khz, 24 bit:

Freq: +.08, -.02 (20k-40)
Noise: -111.8
Dynamic Range: - 111.7
THD %: - .0008
Int.mod % .0020
Stereo Crosstalk -112.4

Before doing all these mods I had the following for 44khz, 24

Freq: +.09, -.06 (20k-40)
Noise: -112.2
Dynamic Range: - 112.1
THD %: - .0005
Int.mod % .0018
Stereo Crosstalk -111.5

Looking at S_Dedalus' RMAA, he got noise and dynamic range closer to what I had before the mod. I just did these mods less than an hour ago, should I wait and retest, or is there any part of my soldering that I should look over? Would a less than clean solder connection cause my dynamic range to lower? Any body with any idea? Thanks.

Edit: When I removed the buffer opamps, I bridged a- and a+ inputs both to the output of a, and did the same for b. I did this because bridging the output of a to just -a yielded no sound. Just more information for those doing this mod.

edit 2: well since i got 2 lt1364's just in case I trashed one trying to bias it to class a. Figured I'd try the other one to see if results got ay better. results were better, by about half a db for dynamic range, and a couple for noise level.

 

[S_Dedalus]

Well, I did my measurements at 24/48, so they aren't really comparable to yours. Let the opamp burn in, maybe things will stabilize. I would say, though, that a 1dB difference in noise at the levels you are talking about is inconsequential. Also, look at the graphs, I noticed that humps that were there before the mod were gone, and also look at the frequency response, on mine it extends quite a bit further at both ends before rolling off.
The LT1364 is a whole lot faster than the JRC2068, I believe thats why details and dynamics are better, but they also have a little bit more noise than most high end opamps, I don't hear the noise, but I do hear the difference in sound.

 

[tschanrm]

Yeah, I think I wait a while before retesting the card in RMAA. Right now 24/48 is nearly identical to 24/44, very little diference in any measurement. But you're right though I don't think I'd ever notice the 1 db difference in results, just wish I knew why there was a discrepency between our two scores.

 

[lan]

The computers are different so the scores are different. I even venture to say the power source is different. I have measured different RMAA figures plugged in different outlets. I wouldn't worry about small differences. I'm glad to hear you found improvements.

 

[bg4533]

My 0404 will be modded by next weekend. I will do some RMAA tests before and after the different mods I do.

 

[deadkenny64]

Quote:

Originally Posted by bg4533 My 0404 will be modded by next weekend. I will do some RMAA tests before and after the different mods I do.

I for one would love to see an "after mod" pic or pics. I know just enough to be dangerous but pictures help immensely.

I keep telling myself not to risk it but after successful mods of my motherboard, video card and dvd player, I'm feeling bold.

 

[S_Dedalus]

It looks like the yellow cap below the buffers is a filter cap for the digital section of the dac, and the two yellow caps to the right just above the dac are filter caps for the 5.0 volt regulator and the analog section of the dac.
I'm guessing to improve the filtering, replace these with larger caps.
here's the dac datasheet
http://www.asahi-kasei.co.jp/akm/en/...395/ek4395.pdf

EDIT: The yellow cap furthest right probably shouldn't be increased, as it is the same value recommended, but the middle one looks like it could be increased.