Cosmos APU a notch+LNA $70 to outperform APx555b for $30000

[IVX]

This is not a fake Viking but another supplier of precision resistors NYS RJM75P0207 https://www.diyaudio.com/community/...perform-apx555b-for-30000.386001/post-7027277
I did buy that from the factory, and all APU use the RJM75P0207 in these positions(diff amp, if your notch gain is 0db there are just 4pcs of the same resistors in action i.e. that circuit isn't sensitive to the resistor nonlinearity due to mutual compensation. +6db position may affect the H3 performance and recommended for THD+N tests). I did individual testing of H2 and H3 performance for the first 70+ units, after I found the performance is consistent(-150-160db, mostly -150-155db), I stopped testing that parameters.

 

[IVX]

Last month I reported the same issue (high H3) about my Cosmos APU to IVX in PM. Mine also has MELF207 resistors that are not according IVX's post. Although I doubt the issue is with the resistors.

Your result(H3 -121db) was nearly 10x times worse than matulo_g -140db, and the source was your DAC, not Victor's OSC, hence, this is not the same case at least. H3 -120db in the diff amp with 4x680ohm is okay with 0402 thick-film Yageo.

 

[IVX]

PS: I tried the publicly published circuit of Victor's OSC(unbalanced one), and found that 2.5-2.8Vrms has a noticeably higher H3 vs < = 2Vrms.
Probably the current version isn't similar to what I tried but perhaps there is an invertor to get a negative output to be balanced, in other words, I recommend trying 2+2=4Vrms balanced output to see if H3 will be -150db or lower.

 

[bohrok2610]

Your result(H3 -121db) was nearly 10x times worse than matulo_g -140db, and the source was your DAC, not Victor's OSC, hence, this is not the same case at least. H3 -120db in the diff amp with 4x680ohm is okay with 0402 thick-film Yageo.

Without APU my DAC measured H3 -138dB. So is nearly 10x times worse (17dB) to be expected with APU?

 

[matulo_g]

PS: I tried the publicly published circuit of Victor's OSC(unbalanced one), and found that 2.5-2.8Vrms has a noticeably higher H3 vs < = 2Vrms.
Probably the current version isn't similar to what I tried but perhaps there is an invertor to get a negative output to be balanced, in other words, I recommend trying 2+2=4Vrms balanced output to see if H3 will be -150db or lower.

When I lower osc output H3 is going down, but it is not solution for my problem...
My osc has best THD at full output and H2 and H3 are less than -160 dB - confirmed by my passive notch (-52 dB) with APU LNA used.

 

[IVX]

matulo_g, I see, so when you reduce the output level of the OSC your passive notch shows the same -160db of H3?
If so could you please show me REW FFT for 5.5V, 5V and 4.5V after APU? I did test 70+ APUs with 5Vrms input sine @ notch gain 0db.
Also, I want to quote here what I replied to bohrok2610 in my PM(he didn't reply back, perhaps I wasn't polite enough), I hope it can help.

How did you make sure your real H3 is lower? I recommend cutting 9038 harmonics with 1kHz LPF 2ord, H3 may be reduced to 15db or about. Another option is "Victor's OSC" analog gen.
Did you get increased H3 on the first day you got the APU or that happened later? If later, I tend to suspect some input opamp IC3 OPA1612 needs to be replaced. As I see, the H2 isn't too high so I believe the input diodes-clamps(OVP) are fine. Just in case pls measure the input 0603 10ohm resistors nearly to IC3. I do not suspect passive parts at all because they are 100-250V(MLCC) and very long resistors with practically zero VCR. If you got the H3 increased right the first day, I could suspect the output opamp IC1 RT6863S only because it is Chinese and in theory, it may be not so reliable as Ti's opamps. I use that opamp due to its input stage symmetry, which let APU see the H2 at -150-160db. You can replace that for OPA1612 and get H2 -140-150db, but H3 for sure will be -150-160db. I tried to find from 5 pcs OPA1612 the best symmetrical one and got H2 -155db. Literally, each OPA2156 gives H3 -160db or so but very bad input symmetry makes H2 -135-140db. All db referred to 1kHz test. Pls let me know what you'll find finally, I should know that to improve the production quality. Any opamps etc I'll send you if needed(but probably take that from the mouser-digikey is 10x faster way).

 

[IVX]

Without APU my DAC measured H3 -138dB. So is nearly 10x times worse (17dB) to be expected with APU?

Well, it depends on how did you make sure your real H3 is -138db?
I don't try to avoid returning your money if you'll ask about it. I just offering some way to find&solve the issue quickly.

 

[bohrok2610]

he didn't reply back, perhaps I wasn't polite enough

It is not about being polite enough. You have not really suggested any fixes but just possible part swaps for me to try out. I don't have any suitable OPA1612s laying around (I use only SON-8 package). Also IMO tinkering with something I have just bought as a product is not first on my todo list.

What is the H3 level to be expected from Cosmos APU?

 

[IVX]

if -138db is real APU should give -138+/-2db as well. Of course, if H3 levels DUT vs APU match let's say at -150db, and the phase of both H2 magically match as well, then we may expect -144db as a result.

 

[matulo_g]

matulo_g, I see, so when you reduce the output level of the OSC your passive notch shows the same -160db of H3?
If so could you please show me REW FFT for 5.5V, 5V and 4.5V after APU? I did test 70+ APUs with 5Vrms input sine @ notch gain 0db.

For 5,5 Vrms REW FFT is in my post #277 here:
https://www.diyaudio.com/community/...perform-apx555b-for-30000.386001/post-7344669

For lower values I can try it later this weekend, but for more than 4,5 Vrms H3 is starting to rise more than declared on your web. For the lower values noise of the osc/passive notch is dominating.

Martin

 

[1audio]

WRT Victor's oscillator- He carefully selects the components and hand tweaks each unit. They are about as good at the circuit can be. I have noticed the distortion shift at higher outputs but need to make sure its the oscillator and not the analysis circuit.

I'm interested in the distortion residual you are measuring on the resistors. I can get to -170 HD3 at 10 KHz on the Radiometer CLT-1. Which parts should I be looking at? Is there a small volume supplier (for testing I only need 5-10).

 

[IVX]

matulo_g, please confirm that you use the APU with 0db gain switch, not 6db right?

 

[matulo_g]

Yes, 0 dB gain - with 5,5 Vrms and +6 dB gain APU clips (heavy distortion due to the overdriving).

 

[IVX]

matulo_g, I'll prepare the low THD setup(down to -160-170db, usually I calibrate Cosmos ADC with that source) and test a few APUs to post the result here.

 

[IVX]

matulo_g, sorry for the delay but 2 of my kids got the flu(maybe some covid mod as usual), and don't go to school. I'm busy with them so far.
My wife Sunny is busy with our younger daughter. So I need a few days more.

 

[matulo_g]

IVX, no problem, we are not in hurry.
Family first.

 

[matulo_g]

@IVX

Here is Victor osc balanced output > APU Notch 1 kHz 0 dB gain> Cosmos ADC (2,7 V) in REW with SW notch compensation

for 5,0 Vrms:

for 4,5 Vrms:

for 4,0 Vrms:

Martin

 

[diyralf]

I don't see any unusual behavior here. The higher the level, the more the op amp comes out of its class A region.

 

[matulo_g]

The H3 is up to the 10 dB higher than IVX shows on his web:

 

[G-K]

@matulo_g IVX shows the data he measured with his DAC and low pass filter and not with Victors oscillator, so you cant compare his measurements with your measurements.
We dont know your measurements of your victors oscillator without APU, so its unknown if your harmonics are coming from your oscillator or from your APU
We dont know too your FFT and sampling rate data.
I can see interferences lower 1 kHz. Is it a power supply problem or a shielding problem?
Here are some measuring results from my Victors oscillator and APU