R/E/P Community

R/E/P => R/E/P Archives => Bruno Putzeys (Designer) - Dave Hecht (Master Tech) => Topic started by: andshesbuyingastairway on May 07, 2007, 07:43:01 PM

Title: 'uncommon' converter circuitry
Post by: andshesbuyingastairway on May 07, 2007, 07:43:01 PM

i'm familiar with a few "uncommon" so to speak ADC such as say the universal audio 2192 that i heard uses the 990 op-amp or lavry gold which i also was told that uses hybrid discrete components with its monolithic ones (to draw a little more current and stabilize the temperature more sufficiently i presume, maybe if one is lucky squeezing a few more bits of resolution out, etc.)

does anyone else know of any other converters that use hybrid IC or discrete op-amps?

also does anyone know of any converters that use high sample rates (above 192 say DXD or DSD level) coupled with high oversample ratios, (above 128 and between 1024) or is there even such a thing as i'm referring to yet. the only oversample rates i've seen for such device is 64fs-128fs.

Title: Re: 'uncommon' converter circuitry
Post by: bruno putzeys on May 08, 2007, 03:38:55 AM

Is your question about the use of discrete buffers to drive standard A/D converter chips or about converters designed from the ground up without using A/D chips?

Title: Re: 'uncommon' converter circuitry
Post by: andshesbuyingastairway on May 08, 2007, 05:22:03 PM

in relation to the op-amps, yeah discrete transistors for instance driving monolithic op-amps.  

if by A/D chip you are referring to the analog and digital elements contained on the same monolithic mixed IC, i wasn't aware of any superior converter companies that i've been looking at thus far utilizing such design.  in the converter designs i've been looking at there have been possibly only monolithic op-amps and monolithic digital chips 'discretely' placed on the same PCB.  

Title: Re: 'uncommon' converter circuitry
Post by: Andy Peters on May 08, 2007, 07:18:29 PM

andshesbuyingastairway wrote on Tue, 08 May 2007 14:22

in relation to the op-amps, yeah discrete transistors for instance driving monolithic op-amps.   if by A/D chip you are referring to the analog and digital elements contained on the same monolithic mixed IC, i wasn't aware of any superior converter companies that i've been looking at thus far utilizing such design.  in the converter designs i've been looking at there have been possibly only monolithic op-amps and monolithic digital chips 'discretely' placed on the same PCB.

Your questions are remarkably confusing.

-a

Title: Re: 'uncommon' converter circuitry
Post by: bruno putzeys on May 09, 2007, 02:49:47 AM

Indeed. Anyhow, the creativity of designers in devising alternate input buffers / output filters other than those proposed in the IC data sheets has been unlimited. ADC's have been driven by transformer coupled tube front-ends, DACs with discrete op amps in the output filters are positively common. The only thing that's rare is converters that build the actual conversion stage out of loose parts. On the A/D front only four readily spring to mind (Lavry, dCS, EMM, Grimm Audio).

Title: Re: 'uncommon' converter circuitry
Post by: Larrchild on May 09, 2007, 03:53:09 AM

Bruno, does thermal memory distortion have the potential to adversely affect the digital domain?

And if so, would getting the converter off of a single common substrate have advantages?

I'm sure it doesn't in terms of propagation delay and parasitics, but could "spread out" be better?

Title: Re: 'uncommon' converter circuitry
Post by: bruno putzeys on May 09, 2007, 05:23:37 AM

Separating a converter into two dice improves performance of the analogue section. The digital section only crunches numbers so as long as it doesn't choke on a bit it'll produce the same results regardless of any cohabiting analogue circuitry. Historically, separating the two was the only way that the analogue circuitry could be shielded from rubbish produced by the digital part (substrate noise), but these days chip designers know how to make the analogue circuit highly immune.

The tricks used to get rid of 1/f noise (alternating the inverting/noninverting sides each clock cycle) are equally effective at combating errors caused by thermal drift. (Note that the thermal distortion thread mixes two effects. One is DC shift, the other is bias shift)

Title: Re: 'uncommon' converter circuitry
Post by: zmix on May 09, 2007, 10:42:31 AM

Andy Peters wrote on Tue, 08 May 2007 19:18

andshesbuyingastairway wrote on Tue, 08 May 2007 14:22 in relation to the op-amps, yeah discrete transistors for instance driving monolithic op-amps.   if by A/D chip you are referring to the analog and digital elements contained on the same monolithic mixed IC, i wasn't aware of any superior converter companies that i've been looking at thus far utilizing such design.  in the converter designs i've been looking at there have been possibly only monolithic op-amps and monolithic digital chips 'discretely' placed on the same PCB.   Your questions are remarkably confusing. -a

And it makes me wonder: "Is 'andshesbuyingastairway' possibly a relative of 'maxdimario'?"

Title: Re: 'uncommon' converter circuitry
Post by: andshesbuyingastairway on May 09, 2007, 02:17:26 PM

thanks for your reply bruno, i haven't looked into the actual circuitry designs of many of the converters yet so i wasn't exactly aware of this, but can you tell me if that sort of practice is so common, what converter companies actually do couple transformer with tubes or use discrete op-amps in the analog sections?

p.s. no man i've actually never been on PSW before, i've been on recording.org but haven't had as good of results and answers as i've had thus far here.  

Title: Re: 'uncommon' converter circuitry
Post by: John Roberts {JR} on May 15, 2007, 04:31:19 PM

I would expect we have probably passed the point where discrete design can can approach what can be done on integrated circuits, and it will likely just get better, or worse depending on your POV.  When I see some of the linearity specs on modern opamps, it's not a why bother issue, but more like why set yourself up for failure (perhaps a personal problem)?

Sure there's always room for improvement. The paradox of digital is at best it's only an approximation, but a damn fine approximation. Putting a high impedance tube in the same chassis with anything digital sounds suspect to me, but the customer is always right, and there must be customers buying that stuff...or it wouldn't exist.  

JR

Title: Re: 'uncommon' converter circuitry
Post by: Larrchild on May 15, 2007, 04:41:51 PM

The only pure solution seems to be a condenser capsule and polarizing voltage attached to the A/D chip's input and singing really loud.