the tone is in the iron... RightMark told me so

[Handcrafted Tone]

I can't believe this has happening... I think I ... nooooo!!!! ... I think I'm becoming a PC guy. DAMN YOU GATES!

My job bought me a new PC laptop. I still love my Mac G5 as a rock solid PT system, but I haven't used it for anything else in a month. Eagle runs slightly smoother, office suite is sweet, and I came across a stellar FREE audio measurement program called RightMark.

I measured a twin-servo preamp I built and the frequency response seemed pretty bad. This twin was ridiculous to start with, it had two fabio 2520 DOAs and a API2503 output transformer. In a design that is supposed to be clean, I used very colored components. I decided to run RightMark directly through the 2503 and found the culprit. If the color is in the iron, the 2053 has plenty of it.

DISCLAIMER - Unfortunately there's no firewire port on this PC. I haven't found a USB device with good converters, so I borrowed a crappy M-Audio Quattro from a friend. I included a direct loop of the converters as a base line to compare to.

That said, check out the frequency response of the transformers I had on my bench. We've got a 2503, a Cinemag CMOB-2S, and CMOB-2n.

http://www.handcraftedtone.com/xfer/RightMark/Comparison.htm

The picture I included is extremely exaggerated (there is only 1-2db of total frequency deviation in the 2503), but you get the idea.

Looks like the CMOB-2S isn't really that close to the 2503...

I've also measured a few mic input transformers. We should start a RightMark thread and share our results. I'll wait until I get a good :roll: USB interface.

 

[clintrubber]

I suspect your 2503-measurement is influenced by the thing driving the TX. Is it directly from the soundcard ? Has it got some muscle ?

IIRIC, the measurements CJ did on such iron showed that it was flat up to a few hunderd kHz at least. Don't recall the lows but that's also depending on the source impedance.

Regards,

Peter

 

[NewYorkDave]

Source and load impedances are very important when testing transformers. If those aren't specified, the results don't mean much.

 

[edanderson]

peter and NYD are very right. i would never try to drive the 2503 directly with a laptop, prosumer, or even a "professional" grade PC soundcard. the problem is that the outputs of most soundcards are designed to drive 10kohm loads. this will be a problem when testing transformers. especially a transformer like the 2503, rated as a 75ohm load.

a better solution in this case will be to swap the transformers in circuit. the 2520 has plenty enough current drive for the 2503. also, i would think that a THD sweep would be more revealing than a frequency sweep.

and yes, the cinemag CMOQ-2S has a similar winding structure to the 2503, but the inductances are totally different.

ed

 

[edanderson]

well, AP filed for a patent on their output stage, so:

http://www.pat2pdf.org/patents/pat4614914.pdf

i think that would probably be overkill for most of us. a high current output buffer, with a constant (vs freq) and known LOW output impedance would be the way to go. you could then build switchable networks on it do do higher impedance outputs. grounded or floating output is also a good option, as the performance of some output transformers is influenced by their grounding.

as for D/A and A/D converters, i wonder if any of the evaluation boards that are available have a way to disable or modify the filtering.

ed

 

[Guest]

[quote author="edanderson"]well, AP filed for a patent on their output stage, so:

http://www.pat2pdf.org/patents/pat4614914.pdf

i think that would probably be overkill for most of us. a high current output buffer, with a constant (vs freq) and known LOW output impedance would be the way to go. you could then build switchable networks on it do do higher impedance outputs. grounded or floating output is also a good option, as the performance of some output transformers is influenced by their grounding.

as for D/A and A/D converters, i wonder if any of the evaluation boards that are available have a way to disable or modify the filtering.

ed[/quote]

Ed, I don't know how somebody can patent what was used long time ago. Power amplifiers with negative impedance, DC motor drivers, etc... The same principle, the same base solution. May be somebody soon will patent method of connecting of conductors using melted metal?

 

[safe as milk]

hi

this would explain why i was so dissapointed with the cmoq2s
vs the 2503,
i asked cj about it, he said that the lower inductance of the cmoq2s
might explain why it seems so much brighter


cheers

 

[TomWaterman]

Can you do distortion plots, I think its the increase in low frequency THD that gives the 2503 its 'tone' or not?

Interesting.

-Tom

 

[Handcrafted Tone]

Thanks for your input guys, I'm new to audio measurements.

[quote author="clintrubber"]I suspect your 2503-measurement is influenced by the thing driving the TX. Is it directly from the soundcard ? Has it got some muscle?[/quote]

It is driven directly from the M-Audio Quattro. I noticed the frequency response of my twin-servo(with 2503) is near identical to the 2503 alone . I used pads to bring down the output of my sound card to mic level for the Twin.

If CJ is listening, how did you test your 2503?

[quote author="mediatechnology"]
I'll try and dig up that Jim Williams EDN series.
[/quote]

That sounds great. If you can find it, I'll post a little board design for it.

[quote author="TomWaterman"]Can you do distortion plots, I think its the increase in low frequency THD that gives the 2503 its 'tone' or not?
[/quote]

Sure

THD

IMD+N Swept Freq

IMD+N

I'm thinking of buying a P.O.S windows desktop PC so I can put a PCI card in it like this one - http://www.digitalaudio.com/DIGITALAUDIO/myarticles.asp?P=5210&S=75&PubID=4401#EmptyAnchor
Specs close to the LinxTwo, at half the price.

 

[clintrubber]

[quote author="Handcrafted Tone"]If CJ is listening, how did you test your 2503?
[/quote]
http://vacuumbrain.com/docs/Sub_Docs/api.html

but the pics don't show up here at the moment.

From that page:

PS: I ran this thing up to 1 mega hertz and it looked great!

Bye,

Peter

 

[Rossi]

Doesn't load here, either

You can set RMAA to different frequencies than 1000 Hz for THD measurements. Pretty useful for transformer measurements. In acoustics mode the RMAA makes a THD over frequency measurement. But that mode is meant for loudspeakers/room acoustics.

The inexpensive ESI Juli@ card is pretty nice for measurements. It's very linear, excellent THD and good driver software. PCI, though.

 

[Handcrafted Tone]

[quote author="Rossi"]The inexpensive ESI Juli@ card is pretty nice for measurements. It's very linear, excellent THD and good driver software. PCI, though.[/quote]

Thanks Rossi! Great specs for $140. Best price/performance I've seen so far. I haven't found a frequency plot yet, but 114db dynamic range is good enough for me.

I bit the bullet and bought a Compaq computer on ebay. 2ghz Pentium 4 + 18" LCD monitor for $300. No more USB crap. This one will sit next to my bench.

Are there any other free measurement programs I should know about? What about an accurate RTA?

 

[clintrubber]

[quote author="Handcrafted Tone"]Are there any other free measurement programs I should know about? What about an accurate RTA?[/quote]
There are a lot but that number reduces when apart from free they also need to be working, lo-hassle, of use & stable.
IIRIC there's a section about measuring software in a Meta.

Bye,

Peter

 

[BR]

Smaart Live has a beta version that you can download. It works pretty good to measure the freq response of line level stuff. You would still need some sort of buffer to run mic level and Z.

www.siasoft.com

 

[TomWaterman]

www.audiotester.de looks good but I haven't bought it yet.

Thanks for the distortion graphs - not as telling as I would have thought. I think we'd need to see THD % against frequency for differing input levels to find the differences.

Like the plots in the cinemag pdfs...

-Tom

 

[CJ]

I just run the signal generator straight into the xfmr via a fine adj pot.

50 ohm output from the generator.

 

[Handcrafted Tone]

Thanks CJ

I think I found a great sound card for cheap - the EMU 1212M

Check these specs - $150!!!!

Analog Line Inputs (2)
Type: servo-balanced, DC-coupled, low-noise input circuitry
A/D converter: AK5394A
Level (software selectable):
- Professional: +4dBu nominal, 20dBu max (balanced)
- Consumer: -10dBV nominal, 6dBV max (unbalanced)
Frequency Response (20Hz - 20kHz): +/- .05dB
Dynamic Range (1kHz, A-weighted): 120dB
Signal-to-Noise Ratio (A-weighted): 120dB
THD+N (1kHz at -1dBFS): -110dB (.0003%)
Stereo Crosstalk (1kHz at -1dBFS): < -115dB

Analog Line Outputs (2)
Type: Balanced, low-noise, 3-pole low-pass differential filter
D/A converter: CS4398
Level (software selectable):
- Professional: +4dBu nominal, 20dBu max (balanced)
- Consumer: -10dBV nominal, 6dBV max (unbalanced)
Frequency Response (20Hz - 20kHz): + 0.0/-.35dB,
Dynamic Range (1kHz, A-weighted): 120dB
Signal-to-Noise Ratio (A-weighted): 120dB
THD+N (1kHz at -1dBFS): -105dB (.0006%)
Stereo Crosstalk (1kHz at -1dBFS): < -115dB

Digital I/O
S/PDIF:
- 2 in/2 out coaxial (transformer coupled)
- 2 in/2 out optical (software switched at ADAT)
- AES/EBU or S/PDIF format (software selectable)
ADAT:
- 8 channels, 24-bit @ 44.1/48kHz
- 4 channels, 24-bit @ 96kHz (S-MUX compatible)
- 2 channels, 24-bit @ 192kHz
MIDI:
- 1 in, 1 out

Synchronization
Internal crystal sync at 44.1, 48, 96, 192kHz
External sample rate sync via
- ADAT (44.1 - 192kHz)
- S/PDIF (opt. or coax 44.1 - 96kHz)

Also includes a Firewire port

 

[Handcrafted Tone]

[quote author="mediatechnology"]Dave;

Do you know if anyone has posted the Audio Precision System One output stage? IIRC it used a tranny with feedback windings.

I agree that we need some kind of soundcard buffer to perform meaningful measurements of iron in particular. I've got the Jim Williams buffer series from EDN a few years ago I should scan.

Wayne[/quote]

Any luck?

 

[Rossi]

The Emus are great sounding cards. I have an 1820m (=1212m + more ins/outs). But they're no good for measuring, mostly due to the driver software and the overly complicated mixer. That's why I bought an ESI Juli@ just for measuring.

 

[clintrubber]

[quote author="Handcrafted Tone"]Thanks CJ

I think I found a great sound card for cheap - the EMU 1212M

Check these specs - $150!!!![/quote]

While I don't want to spoil the warm feeling that all those numbers may give, note that for soundcards more than a few times a tiny small line is added @ the specs that the given figures are the converter-specs.

So most likely measured on a near optimal test-PCB away from all nasty noises.

So the whole application (ICs on PCB in or near PC-hell) will never be as good. How good is difficult to tell since the exact environment a customer is using a soundcard in can't be predicted exactly.

(You may know all this already)


To get an impression how big the differences are: for instance dig up some online-SOS-reviews of soundcards. Martin Walker usually performs an in-PC-test when reviewing a soundcard, quoting some numbers vs the published specs.

But why... Simply enjoy that card ! :wink: