PCM179x pcb layout

Could someone from TI please comment on whatever they feel able to here??

Hi,

Regarding Q c) there seems to be a misubderstanding. The PCM179x Audio DACs are current output DACs. They don't produce a signal voltage, but a signal current.

It is the task of an transimpedance stage or I-V stage to generate a voltage from the DACs output current. The achievable output voltage basically depends on the design of the I-V converter. You can easily design converters that can put out enough voltage to feed a power buffer directly, that is capable of providing for 100W or more Power into a loudspeker load. The obvious difference between the 98 and the 94A is nearly double the current that the 94A can swing.  This in turn means nearly half the resistor value for the same value of output voltage and possibly a slight noise advantage. If  You design a discrete I-V converter, which I personally recommend due to better sound, say a current conveyor strucure feeding into a simple resistor, the lower current swing of the 1798 may be advantageous regarding THD, simplicity of circuit topology and possible choices of current conveyor transistors.

So the choice of DAC depends on Your application and I-V topology

regars

Chris

Thanks Chris. [i take it you mean Q4] I meant to include the op-amp I/V stage as described in the datasheet.

Updated question: Is the SNR improved for a '98 based DAC with op-amp I/V output (as per datasheet) when set up to output 4.5V, instead of 2V output? (like it is similarly using the  '94a)

As for the type of IV, I am looking to evaluate what is most suitable (and 98 vs 94a), so am trying an op-amp output first and may give other solutions (eg discrete) a go next. What is the purpose of the current conveyor (as opposed to just resistor) ?

Hi,

Yes, I meant Q4. Sorry for the typo.

In the 94´s Datasheet there are two circuits , Fig.24, resulting in 2Vrms output voltage and Fig.25, resulting in 4.5Vrms. A similar circuit structure You find in Fig.24 of the 98´s Datasheet., resulting in 2.1Vrms.

Now the similarity between all three circuits is the basic topology of a dual inverting OPamp stage functioning as IV-converter, followed by a differential-filter/differential-to-singlended-stage. Apart from the filter of the 98s circuit beeing 2nd order (not 3rd order as claimed) with a lower cutoff-frequency.

The filter of the 94a on the other hand, especially the one from Fig.24, are not lossless. They are 1st order with attenuation.

From the equations on page 19 and 18 resp  we can calculate the RMS output voltage of the IV-stages alone to:

Vrms =+- I(signal)pp x Rf/(2 x sqr2).

This voltage doubles due to the symmetrical nature and differential action of the differential-to-singleended-stage.

This results in 4.137Vrms for Fig. 24 and 4.523Vrms for Fig.25. The 98s IV-output ends up at 2.319Vrms.

If the 98 is supposed to generate 4.5Vrms of output voltage too, it´s Rfs need to be increased in value. For Rf = 1k6 the differential output becomes 4.525Vrms.

C1 and C2 should be made smaller then, approximately by the same factor that Rf is increased.

I´m not sure why Rf of Fig24. of the 94 is not made smaller in first place, since 2V could be achieved with Rfs of 360Ohms, or rather 390Ohms to accomodate for some losses of the filter stage. 390Ohms shouldn´t give issues here.

The beauty of the current output DACs is that the output voltage may at least in theory be as high or low as wished for. You´re not bound to the limits of the typical 5V supply. For linelevel tough 2Vrms are more than sufficient to drive most any power amplifier into clipping. There´s no advantage in going higher in output voltage if there´s no special reason asking for.

regards

Chris

Thanks again Chris.  I think I will stick with the 2v output then. I would like to use the same layout to evaluate 94a vs 98 (and some different OP-amps) [I understand i may have to use some different value components, because of the different current outputs]. Which differential-filter/differential-to-single-ended stage would you recommend for this?

Hi,

if it were for measurement purposes only, I´d opt for a steep filter as shown in the datasheets.

If it were intended for audio use I´d run the DAC on min 96kHz, but rather on 176.4 or 192kHz.

Then You may get rid of  filters as much as possible and leave just a little HF-taming bandwidth limiting.

The steep filters are only needed if the DACs are clocked at low speeds. But clocked fast and with the capabilities of the modern digital filters implemented into the DACs, steep filtering is a relic of the past that just costs on sonic quality.

You can see in the datasheets that the digital filter of the 1794A is clearly better in the stop-band while differences in the DAC-parts of the DACs are rather insignificant.

The NE5534 is a rather ´too slow´ OPamp for this application. It only works sufficiently in this application because of the large 2n2 and 2n7 caps in the feedback loop of the IV-converters, that slow down the rise time of the DACs output current to a value the NE5534 can handle.

If any, I´d prefer much faster OPamps here like the fully differential THS4130/4131 or the OPA1611/1612 or similar and even faster. Personally I´d opt for a current conveyor circuit running openloop out of sonic reasons.

The basic two problems with OPamps as IV-converters as I see it are:

a) fast current steps of the DAC output may drive the OPamps inputs into slew rate limits, thereby generating distortion due to the Opamp working under openloop conditions. So either one needs a hell-of-a-fast Opamp, or the di/dt of the signal current needs to be slowed down (well, it is shunted away from the IV-converters input by the feedback cap).

b) due to the low open loop bandwidth, the input impedance of the OPamp starts rising. It needed to be constant and low at least up to the DACs output clock frequency, which may be as high as several MHz.

I´d replace the IV-converters 820Ohms||2200pF by 390Ohms||220pF though. The first giving 4.5Vrms/88kHz filter, the second a 2.1Vrms/1.9MHz filter.

If You want singleended outputs You may choose the OPA1611for the differential-to-singleended filter stage also instead the of the NE5534. See Fig.35 of OPA1611/1612s datasheet.

The lossy 63kHz filter stage of 680R, 620R and 2.7nF I´d replace by a simple buffering 340kHz filter of 680R and 680pF (replace the 620R by shorts and omit with the 8200pF, replace the 330R by 680R and the 2700pF by 680pF).

Overall the original circuit generates 2.1Vrms/56kHz, my suggested circuit generates 2.1Vrms also with a bandwidth limit of 330kHz.

The NE5534 is specced for 600R minimum load impedance, which may be the reason for the 820R converter resistor. Otherwise I can see no good reason to generate a considerably larger signal voltage than required, which is then attanuated to its final value in the immediately following stage.

If You want balanced outputs You may use the fully differential THS4131 or OPA1632 in the second stage also (of course could You use just one of the outputs of the fully differential OPamps for singleended useage too, giving more flexibility of choices).

regards

Chris