What is the reason, that RF ICs usually don't have output impedance of 50ohm? As you said, with lower frequencies and with frequencies more than 6ghz, impedance is matched in die, why can't manufacturers do the same with RF ICs?
I don't know the answer definitively. This is something I have seen on MCUs or ASICs with an RF interface, not always on RFICs like amplifiers, attenuators, etc... The moral of the story is to check the datasheet and make sure the appropriate termination is applied.
Did you already or you will make a video about front end input for analog signals and how we can specify the cut-off frequency of rc low pass filter and select the value of capacitor that the pin can drive ? I'm starting my career in Hardware Design as a Mixed Signal Board Designer and need knowing these topics and also if you have made a video about tips and tricks in Mixed Signal Design? Salut, Mr Peterson ❤
Hi Zach. Because of recent projects I've been working on, I've found that most class D amplifier boards have DC input resistances of about 9K. This is, of course, a couple orders of magnitude lower than what you'd find in the typical op amp buffer or valve amplifier. I'm curious as to why they've standardized to 9K and not 10K, 100K, or 1M (the latter is common for op amp buffers and valve gain stages.) Obviously not high speed design, but I know you have an analog background, so you might have some inside baseball info. Also, I know that most CMoy headphone amplifiers have a series resistor on the output. Presumably this is to match the output of the (typically Burr-Brown) op amp to the impedance of the headphones? EDIT: Dave Jones has said that the typical "basic bitch" CRO probe is essentially a distributed LC transmission line. Concur? Dispute? Discuss.
No. The output resistor is current limiting. The input to any amplifier is used in conjunction with a capacitor to set a pole frequency to reduce sensitivity to RF signals that could cause nonlinearities. 9k dc resistance isn’t characteristic impedance. Back to school.
You mean at the receiver end? Usually no unless you are both suppressing reflections and dividing down from a high voltage signal to a lower voltage signal. It basically does both.
@@Zachariah-Peterson Yes I was talking about the receiver side. But in this video, you not only use a series termination resistor, but also add a 50Ohm parallel termination resistor at the receiver side (without wanting to divide the voltage)
@@AlbertRei3424 Sure I understand why there might be confusion. Whether you apply both depends on a few factors. You usually choose one of these for the following reasons. You can use both if the line is long and you are okay with dividing down at the input and output as an R/2R type of divider. Sometimes when you use series and apply that R/2R type of division, it was technically being divided down by a non-zero source impedance anyways so it may not appreciably affect the signal level that is injected into the line (this is sometimes the case for SPI), it just depends on the source impedance. If the line is still not within the electrically short limit, then there can also be a reflection off of the end of the line. To prevent that reflection you can then add a parallel resistor, but you also have to consider the division at the input and whether it will toggle the receiver. If the division brings the signal too low (meaning voltage sunk across the termination resistor is too low), then you will not toggle the receiver. For high-speed interfaces that have an impedance requirement, there is already that kind of parallel resistive termination anyways to prevent reflection from impedance mismatch, just look at the on-die differential termination circuit options for input buffers on an FPGA and you will see what I mean.
This is such a great video, thanks for posting this.
What is the reason, that RF ICs usually don't have output impedance of 50ohm? As you said, with lower frequencies and with frequencies more than 6ghz, impedance is matched in die, why can't manufacturers do the same with RF ICs?
I don't know the answer definitively. This is something I have seen on MCUs or ASICs with an RF interface, not always on RFICs like amplifiers, attenuators, etc... The moral of the story is to check the datasheet and make sure the appropriate termination is applied.
Did you already or you will make a video about front end input for analog signals and how we can specify the cut-off frequency of rc low pass filter and select the value of capacitor that the pin can drive ? I'm starting my career in Hardware Design as a Mixed Signal Board Designer and need knowing these topics and also if you have made a video about tips and tricks in Mixed Signal Design?
Salut, Mr Peterson ❤
Hi Zach. Because of recent projects I've been working on, I've found that most class D amplifier boards have DC input resistances of about 9K. This is, of course, a couple orders of magnitude lower than what you'd find in the typical op amp buffer or valve amplifier. I'm curious as to why they've standardized to 9K and not 10K, 100K, or 1M (the latter is common for op amp buffers and valve gain stages.) Obviously not high speed design, but I know you have an analog background, so you might have some inside baseball info.
Also, I know that most CMoy headphone amplifiers have a series resistor on the output. Presumably this is to match the output of the (typically Burr-Brown) op amp to the impedance of the headphones?
EDIT: Dave Jones has said that the typical "basic bitch" CRO probe is essentially a distributed LC transmission line. Concur? Dispute? Discuss.
No. The output resistor is current limiting. The input to any amplifier is used in conjunction with a capacitor to set a pole frequency to reduce sensitivity to RF signals that could cause nonlinearities. 9k dc resistance isn’t characteristic impedance. Back to school.
@@envisionelectronics Could have left out the last sentence there.
I thought that when a serie termination was used near an output driver, nothing needed to be added at the input driver side
You mean at the receiver end? Usually no unless you are both suppressing reflections and dividing down from a high voltage signal to a lower voltage signal. It basically does both.
@@Zachariah-Peterson Yes I was talking about the receiver side.
But in this video, you not only use a series termination resistor, but also add a 50Ohm parallel termination resistor at the receiver side (without wanting to divide the voltage)
@@AlbertRei3424 Sure I understand why there might be confusion. Whether you apply both depends on a few factors. You usually choose one of these for the following reasons. You can use both if the line is long and you are okay with dividing down at the input and output as an R/2R type of divider. Sometimes when you use series and apply that R/2R type of division, it was technically being divided down by a non-zero source impedance anyways so it may not appreciably affect the signal level that is injected into the line (this is sometimes the case for SPI), it just depends on the source impedance. If the line is still not within the electrically short limit, then there can also be a reflection off of the end of the line. To prevent that reflection you can then add a parallel resistor, but you also have to consider the division at the input and whether it will toggle the receiver. If the division brings the signal too low (meaning voltage sunk across the termination resistor is too low), then you will not toggle the receiver.
For high-speed interfaces that have an impedance requirement, there is already that kind of parallel resistive termination anyways to prevent reflection from impedance mismatch, just look at the on-die differential termination circuit options for input buffers on an FPGA and you will see what I mean.