Hello.
I am designing a board to add on to my DG-1000. This board uses a THAT1646 to create a balanced output. The problem is the 1646 comes with a gain of 2, which would cause the audio signal to clip at the rails. The audio signal is 24Vpp coming out of the IOut of the PCM1702 DAC, as calculated by the ±1.2mA current output and the 10k resistor of the inverting opamp following it:
V = 2.4mA x 10k = 24Vpp
I'm wondering if it would be better for SNR to either reduce the feedback resistor of the inverting opamp from 10k to 5k or use an otherwise unused section of an opamp to reduce gain by half before the 1646?
Thanks in advance for any help.
I am designing a board to add on to my DG-1000. This board uses a THAT1646 to create a balanced output. The problem is the 1646 comes with a gain of 2, which would cause the audio signal to clip at the rails. The audio signal is 24Vpp coming out of the IOut of the PCM1702 DAC, as calculated by the ±1.2mA current output and the 10k resistor of the inverting opamp following it:
V = 2.4mA x 10k = 24Vpp
I'm wondering if it would be better for SNR to either reduce the feedback resistor of the inverting opamp from 10k to 5k or use an otherwise unused section of an opamp to reduce gain by half before the 1646?
Thanks in advance for any help.
You need only about 1.2V RMS for line-level audio and besides I'm not able to see a 5V DAC give 24Vpp !!
Well, that's what the IOut of the dac is for - it doesn't reallt matter what voltage is supplied to the DAC, the current output can be "converted" into voltage based on the analog supply voltage.
EDIT: The opamp in the IOut path has a 68p in parallel to the 10k feedback resisitor - not sure exactly how this effects the Vpp. Pretty sure its for filtering out extraneous noise.
EDIT: The opamp in the IOut path has a 68p in parallel to the 10k feedback resisitor - not sure exactly how this effects the Vpp. Pretty sure its for filtering out extraneous noise.
Okay so there clearly is a lot more going on - I asked my college lecturer and I looked at the schematic with him and he told me there's a lot of filtering going on which would affect the Vpp output of the DAC - which i should have thought of beforehand just looking at it. Furthermore he suggested the frequency the DAC operates at could further decrease the output Vpp. Not sure how that applies here but it is worth considering. I don't know an awful lot about this implementation. I will hook it up to one of my college's oscilloscope and see what's going on.
You can’t arbitrarily utilize whatever value I/V resistor you wish. DAC distortion increases significantly above a few hundred mV amplitude at the current output pins of the PCM17xx family, as I recall. Conversely, the lower the I/V resistor value, the lower the signal voltage amplitude there, and therefore the lower the DAC distortion in passive I/V mode. Which, by the way, is not a mode recommended by T.I.
SNR will theoretically degrade (to a inconsequential degree for audio, IMHO) whether you reduce the signal at the DAC current output, or at an op-amp output. So, I would disregard that concern. For benefit of legitimate THD concerns, I suggest that you reduce the signal amplitude at the DAC’s I/V resistor. 100mV, or less is a good target, but Audio Research utilized the PCM1702 with passive I/V in their tube DACs, and which featured a 200mV signal amplitude at the I/V resistor, if I remember accurately.
SNR will theoretically degrade (to a inconsequential degree for audio, IMHO) whether you reduce the signal at the DAC current output, or at an op-amp output. So, I would disregard that concern. For benefit of legitimate THD concerns, I suggest that you reduce the signal amplitude at the DAC’s I/V resistor. 100mV, or less is a good target, but Audio Research utilized the PCM1702 with passive I/V in their tube DACs, and which featured a 200mV signal amplitude at the I/V resistor, if I remember accurately.
Last edited:
Thanks for your response, Ken.
How should I go about doing this in the DG-1000? The I/V resistor in the inverting opamp signal path is the first thing the IOut of the DAC 'sees'.
I’m guessing here, without the schematic, but you likely can simply replace the existing I/V (transimpedance) op-amp‘s feedback resistor with one half it’s value, or alternatively, solder a second I/V resistor of the same value in parallel with the existing I/V resistor. There probably is also a small capacitor in parallel with the I/V feedback resistor which would need doubling in value to maintain the same low-pass filtering effect. Considering this filter is 1st order, however, it won’t make a huge difference if you leave it as it is.
https://yamahamusicians.com/forum/download/file.php?id=9163&sid=b2166b10c2b2420680f75deb23feadac
It's on the AD/DA board where the area of interest is. Down at the bottom of the schematic.
I was thinking the same, but it seems to be part of higher order filters down the line.
All this gain stuff might not even be a problem, though. I'll actually scope the output to see what Vpp exists on the output. It's hard to believe it actually goes to 24Vpp! I think I'm missing a lot on how this DAC current out works.
You are correct about the output voltage being 24Vpp. The unit appears intended for professional application, so I guess the engineers had their reasons for making it so large. Anyhow, you, indeed, may easily reduce the fullscale output voltage to meet your requirements by appropriately reducing the value of that 10K feedback resistor.
You probably are aware that the de facto fullscale output voltage of most commercial DACs is, 2VRMS, or 2.83Vpk. So, for example below, with the PCM1702.
You probably are aware that the de facto fullscale output voltage of most commercial DACs is, 2VRMS, or 2.83Vpk. So, for example below, with the PCM1702.
Seems absolutely bizarre to me because at full scale that will clip the hell out of what comes next in the signal chain. How does the input on DAC scale the output? It wouldn't make sense if it was a constant 24Vpp signal on the output because then you would have literally no dynamic range lol.
As a separate note, would changing R234 (10k) not also imply you have to change R235-237 (all 2.2k)? They're all in the feedback path of the inverting opamp on IOut, it's just that they're not drawn in the typical way on the schematic i.e. they're visually placed on the output pin.
When you halve R234, you could double C212 to keep the time constant the same and hence keep the shape of the filter response as is.
R235 ... R237 are part of the following filter stage (a third-order Sallen and Key low-pass filter stage). I don't see any reason to change those.
R235 ... R237 are part of the following filter stage (a third-order Sallen and Key low-pass filter stage). I don't see any reason to change those.