60dB attenuator prior to spectrum analyzer, how to?
I want to build a 60dB 1W 50R I/O attenuator which will be connected to the input of the spectrum analyzer.
I am confused a bit about the next things:
1. What is the difference between the two attenuation pads, T-pads and Pi-pads? Do Pi-pads shunt the signal to ground and T-pads reflect it?
2. Should I calculate the pad for 50R I/O impedance and just connect it to the input of the analyzer? I worry because for example on a Pi-pad, the output shunt resistor will be effectively connected in parallel to the 51R resistor which is inside the analyzer input, so won't this be 25R instead?
Are you really after the experience of building, or do you
just have a measurement job to do?
I have an assortment of SMA "bullet" attenuators, of course
these are lesser (3dB, 6db) but you can stack them inline.
I'd imagine they are available in higher ratios. The frequency
band shown should be non-challenging for just about anything
still sold these days.
But if you really want to build it yourself, and know you did
it right, somebody else is going to have to help you with that.
1W will be handled by any off-the-shelf coaxial SMA or BNC attenuator. Maximum available attenuation is 30 dB in most series, so you'll cascade two of it.
RF attenuators are impedance matched on both sides, matching can be theoretically achieved in pi or t configuration, there are practical reasons why pi respectively cascaded multistage pi is preferred for higher attenuation factors.
There are online calculators like https://chemandy.com/calculators/mat...calculator.htm
Regarding question 2, if you connect a 50 ohm generator to a 50 ohm load, the junction node impedance will be 25 ohms. Nothing specific about attenuators.
Building attenuator networks for the VHF or even UHF range is no rocket technology, but you should be aware of parasitic component parameters and know how to estimate it. Preferably have a network analyzer to verify the results.
The purpose of the -60 dB attenuator isn't quite clear. Most spectrum analyzers have 20 dBm (0.1 W) or even 30 dBm (1 W) input rating by internal attenuator networks. When working with 1 W level equipment, an external 20 dB attenuator may be useful as protection means, but 60 dB would be useless. It's a different matter, if you are measuring high power transmitters, but then you are looking for higher attenuator rating.
Is that normal, the node impedance (junction between the attenuator and the input of the SA) to be 25 ohms? In other words should I build it like this, or should I design the attenuator for a high output impedance, so that when connected to the 50R shunt resistor of the SA it will bring the node impedance down to 50R? Which of the two ways should I follow for correct 50R measurements?
I am only interested up to 30MHz max, no UHF here, so building something out of discrete resistors is not that difficult.
The purpose of the -60db attenuator is to increase the level range of the SA to +30dBm max. The model is the Tektronix 491, which has a linear usable range of -90dBm to -30dBm. More than that and it becomes non linear. At about +15dBm is the maximum input where you will start damaging the equipment.
So, without the attenuator you can measure linearly -90 to -30dBm and with the 60dB attenuator, you can measure -30 to +30dBm.
But the first paragraph is what I need to know to correctly design the attenuator (and every attenuator pad basically). I am always confused about this so I need to make it clear to me.
The attenuator is calculated to have 50 Ohm at input and output, if the other side is terminated with 50 Ohm.
Just use this calculator: https://www.microwaves101.com/calcul...tor-calculator
Play with values, starting from small attenuation, and you will understand how it works.
Thanks for the link, I already have seen such calculators and in fact I use these already.
However the page does not show what you mentioned, which is "The attenuator is calculated to have 50 Ohm at input and output, if the other side is terminated with 50 Ohm".
Ok, so when designing the attenuator based on these calculators, I should not care about the "node impedance" as FvM mentioned. I should only care to ensure that BOTH the output of the attenuator is designed for 50R and the input of the spectrum analyzer is terminated to 50R (shunt resistor in my case).
So for example, if I am to connect an external power meter to the attenuator, I shall ensure there is a shunt 50R resistor to ground inside the power meter (i.e terminated to 50R) and design the input attenuator based on such calculators for 50R output impedance.
Is that the correct way to do it?
And to be more specific I attach you the actual schematic including tthe attenuator in place.
The 50R output resistor of the attenuator is connected in parallel to the 50R resistor inside the analyzer.
However based on your answer this should be done that way indeed?
Yes. If you start calculating PI attentuators with low value, you will notice a small series resistance and large shunt resistors. With increasing attenuation, the series resistance increases, and the shunt resistance decreases, to keep the 50 Ohm port impedance.
Extreme cases:
0 dB PI attentuator has Rseries = 0, Rshunt = infinite
infinite dB PI attenuator has Rseries = infinite, Rshunt = 50
Your 60dB attenuator is close to the infinite dB case.
Yes, correct. Both attenuator and power meter must present 50 Ohm impedance when measuring into the device.
It's difficult to achieve a flat attenuation curve for a single stage attenuator with 25 k series resistor up to 30 MHz, the parasitic parallel capacitance must be < 0.2 pF. Better to split it into two 30 dB stages.
Ok that was really helpful!
Thanks a lot for the tip! I will do so. Two 56R shunts and one series 820R for each 30db Pi attenuator.
How about resistor wattage?
I believe only the first shunt resistor (the one in the input side of the first Pi attenuator) needs to be of high wattage to handle the +30dbm. Is that true?
Yes, only the first shunt resistor needs 1W rating, all others < 100 mW. Usual solution is multiple resistors in parallel.