Has anyone seen this front-end before?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
How? Sorry, I can't tell because of commercial interest.

Now I understand - you are looking for funding for your next amp construction. You are going to be a very rich man in the audio business! What about moving this to the Commercial Sector on this forum?

I have no idea how you implemented this in software, but spectra substraction is available in almost every DSP software package that I know, all free of charge. It's not rocket science, spectral components have to be substracted as vectors, so with both magnitude and phase.

Try the M-Audio 192 or the Xonar before you speak (not that I would expect from you a shadow of objectivity). And BTW, balanced inputs do look nice for advertising purposes, for measuring single ended power amps they are pretty much useless.

P.S. Averaging for 8 hours?
 
Last edited:
You stated
This is simply not true, they can very well supress ground currents influence at measuring (mains and harmonics) . No one talks that it can help to reach -160dB noise background..

Balancing is not necessary about "ground currents influence", but about common mode noise/artifacts.

If you can't reach better than -160dB floor in unbalanced configurations, then what's the point of specifying -160dB measurements.

Not sure what's the floor in Edmond's result, -210dB is suggested. That would be around 32 picovolts. Synchronous averaging (for 8 hours?) lowers the floor, but not the correlated common mode noise/artifacts.

My suggestion was to integrate and test the software with current (and much cheaper) sound card models. Expecting the diy potential customers to buy a L22 for the luxury of measuring ppb distortions (who cares about, anyway?) it's in my opinion not realistic.
 
L22

Hi BV,

Under the same conditions I got a slightly higher THD (-122 to -126dB), depending on which channel and which L22. Yes, I have two them, bought on ebay at roughly the same price as a new ASUS blah blah blah XTe thingy. Really 'snobbish', isn't it. :D
So you also noticed that near FS (in the yellow zone) distortion rises significantly. Any idea which IC in particular is responsible for it?
I think it's the INA137 or the MC33078 in the first place.

Cheers,
E.

BTW, what's your real name?
 
depending on which channel and which L22. Yes, I have two them, bought on ebay at roughly the same price as a new ASUS blah blah blah XTe thingy. Really 'snobbish', isn't it. :D?

Wow, you are a true EBay business man! Second hand L22's are running at about twice that money today, and are barely available. People who will have the chance to get a L22 for 100 squid are indeed a good market segment for your software product.

And BTW, MC33078 is a horrible opamp, get rid of it ASAP.
 
upgrading L22

Hi Ladislav,

So I think it is best to replace the chips step by step and check the performance after each step. As for the MC33078 (i.e the one just in front of the ADC), most of the time it is simply labeled '33078'. Or do you have another IC sitting there? Actually, I was a little amazed to find this mediocre chip on a L22. But, as I don't have reverse-engineered the card yet, I'm not sure if this chip sits in the signal path anyhow.
In the meantime I received a bunch of samples from ADI and TI. After getting (or making) a suitable desoldering tip*, I'll let you know about the results.

Cheers,
Edmond.

*Probably I'm making it myself, as buying one is too 'snobbish'. ;)
 
/OT

...
*Probably I'm making it myself, as buying one is too 'snobbish'. ;)

And so I did, made from a piece of 4/6mm copper oil tube and fitted on a 50 years soldering iron* (see 1st pic). Before modifying a L22, I will first practice on an old sound card (from Creative Labs of course, what else are they good for?:rolleyes:)

The picture was taken with my new camera, of course also a 'snobbish' one: same performance as a Leica M9 (though 1/7 of the price). So which one is it? Have a look at the 2nd pic. The connoisseurs among us will (and should) recognize it immediately.

Cheers,
E.

*This one is 20W and probably it is too hot for desoldering SO8 chips, so I will use a variac to lower the temperature.
 

Attachments

  • SO8-C.jpg
    SO8-C.jpg
    884.3 KB · Views: 424
  • handle-C.jpg
    handle-C.jpg
    57.8 KB · Views: 389
Modding a Lynx L22

Hi Ladislav,

I started with the ADC section. First, I replaced the MC33078 and INA137 by a LM4562 respectively AD8274. At 1kHz or 19kHz and 20dB or 3dB below FS, it had no significant effect on THD and noise. Only individual harmonics differ a few dB.
Next, I replaced the OPA2227's by LME49720's. At -20dB still no significant improvement. At -3dB and 1kHz dito. At 19kHz however, THD was 11dB lower. At least something has improved.

BTW1, you said "[t]here is no MC33078", so what else is sitting there? (I mean just in front of the ADC)
BTW2, the SO8 desoldering tip works fine and no variac needed, just pull out the mains plug when the tip is hot enough.

Cheers,
E.
 
Last edited:
Here is the result from my onboard Realtek audio, going through a feedforward EC buffer of my design. It's better than the Xonar DX. Interestingly, this comes from an Asus M5A99FX Pro R2.0 motherboard. Same manufacturer of the Asus.

5th harmonic I think is a software error cause by Audacity or the drivers.

Baudline is the software I'm using. It is incredible.
 

Attachments

  • Screenshot from 2013-05-03 16:12:37.png
    Screenshot from 2013-05-03 16:12:37.png
    338.1 KB · Views: 336
Sampling rate is 192KHz. Attached is a picture of the pure loopback test with a 5 inch cable.

5th harmonic was a lot worse when Audacity was at 96KHz. I switched it to 192KHz and it lowered significantly. The 6h harmonic wanders up and down regularly, which would seem to indicate sample conversion artifacts.

The Xonar was almost usable with the 5 inch cable but when I hooked it up to the few meters of cable to my bench, 2nd harmonic got significantly worse. However the left channel had H3 at -75db, which seemed to be a defect for that channel at both line in and output.

I'm afraid I don't want to show my EC buffer yet, especially since it should be capable of much less distortion, especially with the possibility of a complete function null. My previous post was a bit ill-conceived, sorry.
 

Attachments

  • Screenshot from 2013-05-03 17:04:13.png
    Screenshot from 2013-05-03 17:04:13.png
    352.4 KB · Views: 283
Distortion measurements

Hi keane,

I personally think that you should be interested in higher frequencies than 1KHz to exercise your EC buffer go why not 6Khz , 10Khz or for that matter 20Khz. Linearity at the low frequencies is not hard to achieve in my opinion and at the higher frequencies other things are revealed about the design.

Regards
Arthur
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.