Hi
I'm trying to understand the use of transformers in the SPDIF interface. According to my knowledge the transformers are used to avoid ground loops. But why is this so important for digital signals compared to analog signals? For short distances(<1 m) few uses transformers for analog signals, but it seems to be the standard for the SPDIF interface. Is it because of the high frequency components of the digital signal or is the digital circuitry at the ADC or DAC end especially noisy?
I'm trying to understand the use of transformers in the SPDIF interface. According to my knowledge the transformers are used to avoid ground loops. But why is this so important for digital signals compared to analog signals? For short distances(<1 m) few uses transformers for analog signals, but it seems to be the standard for the SPDIF interface. Is it because of the high frequency components of the digital signal or is the digital circuitry at the ADC or DAC end especially noisy?
There's no fundamental difference. The large change in bandwidth used motivates quite different implementations.
Dean Banerjee's book on PLLs (free download from TI) is a good reference for understanding basic behavior of S/PDIF receiver cores. The short version of the answer is AC variation between the transmit and receive grounds introduces jitter. This is rejected by a lowpass in PLL which corners anywhere from mHz to kHz. Additional rejection can be obtained through the highpass provided by an RF transformer. 100Hz - 10kHz are typical PLL corner frequencies and a few tens to a couple hundred kHz typical trafo corners.
THAT's design support materials (also free downloads) are a good reference for understanding CMRR requirements for maintaining analog audio signal integrity over differences in ground potential---look at the InGenius in particular. There's also a larger literature on instrumentation amplifier CMRR to consult; same problem, different application spaces. The basic answer is analog audio requires large and expensive transformers. So alternate implementations are favoured.
Dean Banerjee's book on PLLs (free download from TI) is a good reference for understanding basic behavior of S/PDIF receiver cores. The short version of the answer is AC variation between the transmit and receive grounds introduces jitter. This is rejected by a lowpass in PLL which corners anywhere from mHz to kHz. Additional rejection can be obtained through the highpass provided by an RF transformer. 100Hz - 10kHz are typical PLL corner frequencies and a few tens to a couple hundred kHz typical trafo corners.
THAT's design support materials (also free downloads) are a good reference for understanding CMRR requirements for maintaining analog audio signal integrity over differences in ground potential---look at the InGenius in particular. There's also a larger literature on instrumentation amplifier CMRR to consult; same problem, different application spaces. The basic answer is analog audio requires large and expensive transformers. So alternate implementations are favoured.
Gunders,
Based on your somewhat tentative acknowledgment, I suspect that twest820's knowledgable engineer's explanation may have been a bit more technical than what you were hoping for. My hobbyist oriented addition to what twest820 wrote is that all unbalanced signal interfaces are susceptible to transferring ground loop noise existing between components. This is true whether that unbalanced interface carries a narrowband analog audio signal, or a wideband digital signal. While transformers are well known for blocking mid and low frequency ground noise across an balanced signal interface, they also are effective in blocking noise across an unbalanced interface, just not quite as effectively.
Consumer application S/PDIF is an unbalanced digital signal interface. As twest820 says, noisy grounds can contribute to the manifestation of digital clock jitter. Since transformers help block ground noise transfer they help block the manifestation of ground loop noise induced clock jitter.
The reason you see the common utilization of S/PDIF transformers, yet relaively rare use of analog audio interface transformers, is because quality analog audio transformers are much, much more costly to make. The reasons they are so costly to make is because audio signals are far lower in frequency than are S/PDIF signals, and because analog audio signals must be passed with very low distortion. These two requirements dictate a much larger transformer core, which may need be constructed from costly metals or metallurgical processing. Neither of those requirements apply to S/PDIF transformers.
Based on your somewhat tentative acknowledgment, I suspect that twest820's knowledgable engineer's explanation may have been a bit more technical than what you were hoping for. My hobbyist oriented addition to what twest820 wrote is that all unbalanced signal interfaces are susceptible to transferring ground loop noise existing between components. This is true whether that unbalanced interface carries a narrowband analog audio signal, or a wideband digital signal. While transformers are well known for blocking mid and low frequency ground noise across an balanced signal interface, they also are effective in blocking noise across an unbalanced interface, just not quite as effectively.
Consumer application S/PDIF is an unbalanced digital signal interface. As twest820 says, noisy grounds can contribute to the manifestation of digital clock jitter. Since transformers help block ground noise transfer they help block the manifestation of ground loop noise induced clock jitter.
The reason you see the common utilization of S/PDIF transformers, yet relaively rare use of analog audio interface transformers, is because quality analog audio transformers are much, much more costly to make. The reasons they are so costly to make is because audio signals are far lower in frequency than are S/PDIF signals, and because analog audio signals must be passed with very low distortion. These two requirements dictate a much larger transformer core, which may need be constructed from costly metals or metallurgical processing. Neither of those requirements apply to S/PDIF transformers.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.