Suggestions needed for software, interface etc.

[Whoops]

Thank you Abbey

are there any affordable solutions that go above 100khz?

 

[abbey road d enfer]

Unfortunately no. So far, reasonably priced soundcards with 384k SR have only the DAC capable of that, the ADC is limited to 192.
The other day, I saw one that was capable of 384k (even 768k, IIRC) but it was about 4k€.
Actually measurements above 50k are just amplitude/phase, so they can be carried out with a different set of tools, typically a function generator, that can be had for about €100, and an oscilloscope ( I have a portable one that I paid €88 for on Ali).
There is usually no need for distortion measurements at >50k.

 

[Whoops]

I have an high end soundcard, it’s a Focusrite Red4Pre, I will make some measurements to check how it performs

probably I will ask you guys for some help on how to perform the tests with REW

 

[abbey road d enfer]

I have an high end soundcard, it’s a Focusrite Red4Pre, I will make some measurements to check how it performs

probably I will ask you guys for some help on how to perform the tests with REW

You need to send back the output into the input. You'll have to fiddle with the gain knob in order to avoid saturation.
Make sure the soundcard is not in loopback mode. That's in Preferences. That's also where you define the input and output used for measurement and sample rate.
Then the actual measurement is done from the Measure tag, where you define level and frequency limits. Make sure you have a recent version of REW, cause it allows extending measurements to 96kHz.
Then hit Start and it should produce the desired result. You should see a Bode graph (Amplitude & Phase). You may have to set the limits (there's a tag in the North-East of the screen).
I suggest you read the Help; I have read it twice, and even then I still have to ask for help on the forum. John Mulcahy is very helpful and answers quickly usually.

 

[Winston OBoogie]

Actually measurements above 50k are just amplitude/phase, so they can be carried out with a different set of tools, typically a function generator, that can be had for about €100, and an oscilloscope ( I have a portable one that I paid €88 for on Ali).
There is usually no need for distortion measurements at >50k.

Same. Except I use an old sine and square generator with levelled output good to 550kHz, and an ancient but well cared for Tektronix 465B.
Truth be told, the generator settles pretty quickly that, providing I sweep at a sane rate, I'm able to get a good handle on audio range frequency response this way before I even think about hooking up an interface.

 

[Tubetec]

Try different soundcard latency (buffer)settings to see if you can find a sweet spot with REW Abbey , certain values tend to work better than others and not nesesarily the lowest value. Ive never been able to get 32 bit float as an option in REW only via Asio so I assume your using the Windows driver ,

60khz with no wobbles and THD of 0.00013 % is still impressive enough though.
The SSL2/2+and the Audient are playing in the same ballpark for a little more money but I cant find a BW response for either beyond 20k . I did see a brand spanky new SSL2+ the other day with VST liscences intact , 150 euro , very tempted to bid .

 

[trobbins]

As indicated, even 96kHz bandwidth can be limiting for some hi-fi amps where roll-offs and resonances start significantly above 40-50kHz. Most stray parasitic resonances in valve amps I've come across show themselves coming out of the noise floor under 50kHz, but its nice to know there is some bandwidth margin to alert if there is a nascent issue (which can never typically be seen or appreciated on a scope).

Adding impedance measurement to your tool arsenal is likely to need up to 96kHz bandwidth to see and measure parameters after initial resonance.

When I looked at a few soundcards that had 384/768k capability, and that provided a bandwidth spec (a rarity) it seemed as if it was the opamps in the circuit path that were limiting bandwidth as I couldn't identify any soundcards with bandwidth exceeding about 140kHz.

 

[Whoops]

I think for 99% of my work going to 50khz is much more than enough, but I will try my most expensive soundcard to check it's limit.
I have also 2 other soundcards which are more affordable units. A Digidesign Mbox2 Pro and a Lexicon Lambda

 

[abbey road d enfer]

Same. Except I use an old sine and square generator with levelled output good to 550kHz, and an ancient but well cared for Tektronix 465B.
Truth be told, the generator settles pretty quickly that, providing I sweep at a sane rate, I'm able to get a good handle on audio range frequency response this way before I even think about hooking up an interface.

Of course, second-hand lab gear is an excellent solution.
I still use my old (1970's) LEA generator and THD meter, with a Hameg. It takes about half the space of my dining table, though, which is not small.

 

[abbey road d enfer]

Try different soundcard latency (buffer)settings to see if you can find a sweet spot with REW Abbey , certain values tend to work better than others and not nesesarily the lowest value. Ive never been able to get 32 bit float as an option in REW only via Asio so I assume your using the Windows driver ,

I use ASIO. Haven't tried but I've read somewhere that the Windows driver does not allow 192k.

The SSL2/2+and the Audient are playing in the same ballpark for a little more money but I cant find a BW response for either beyond 20k .

I still have to see a mfgr publishing frequency responses above 20k.

I did see a brand spanky new SSL2+ the other day with VST liscences intact , 150 euro , very tempted to bid .

If you do, tell us how it fares.

 

[abbey road d enfer]

When I looked at a few soundcards that had 384/768k capability, and that provided a bandwidth spec (a rarity) it seemed as if it was the opamps in the circuit path that were limiting bandwidth as I couldn't identify any soundcards with bandwidth exceeding about 140kHz.

Ah Ah! Which ones?

 

[trobbins]

Budget 192khz A/D

 

[abbey road d enfer]

Budget 192khz A/D

Are there any soundcards that use those chips?

 

[trobbins]

Sorry, I was referring to post #31 in that link.

 

[abbey road d enfer]

Sorry, I was referring to post #31 in that link.

Well, it hardly qualifies as a budget soundcard, does it?
Someone wants to plunge?

 

[trobbins]

My post was about finding anything >192kHz sampling. On review, what I saw as 140kHz bandwidth limit is what was reported earlier in that thread as the default ADC bandwidth, and likely not related to opamp limitations. I'm just pleased that there are present day product that can do 96kHz bandwidth that is somewhat entry level price (well at least the Scarlett is).

 

[abbey road d enfer]

My post was about finding anything >192kHz sampling. On review, what I saw as 140kHz bandwidth limit is what was reported earlier in that thread as the default ADC bandwidth, and likely not related to opamp limitations. I'm just pleased that there are present day product that can do 96kHz bandwidth that is somewhat entry level price (well at least the Scarlett is).

Indeed there are now several affordable soundcards that handle 192kHz sample rate. However many are restricted in actual banwidth. Of course none can achieve 96kHz BW.
It looks like the practical limit is ca. 0.32 SR.

 

[trobbins]

The -3dB bandwidth of my card is 90kHz loopback, and REW calibration flatlines that and adds a few kHz beyond in practice, so I'm not splitting hairs.

 

[abbey road d enfer]

The -3dB bandwidth of my card is 90kHz loopback, and REW calibration flatlines that and adds a few kHz beyond in practice, so I'm not splitting hairs.

Neither am I. It seems to be the exception.
So, what card is it?

 

[Whoops]

You need to send back the output into the input.

I have a question regarding the loop back,
Isn’t it possible than when doing that loop that the output circuitry of the soundcard (DAC and analog section) already limits the frequency response and that the input measures a lower kHz limite because the output can’t deliver that.

just to be sure shouldn’t the first tests of the input stage done with an external signal generator already known to be able to deliver the full sweep?

thanks