I was just looking how to make some attenuators and I run into you again! You are just a wealth of great information and practical implementation ! I want to be sitting right next to you and try to catch some of your breadcrumbs! Thank you again! Stay safe and have a great weekend!
Wow, I always thought that making attenuators is a complicated science, especially getting a decent flat line frequency response. But your video proved to me that it can be quite easy, thanks!
One trick I picked up for similar attenuators and terminations is to stack a number of smd resistors to get the desired value. 3 150 ohm larger ones make good 50 ohm terminations. For attenuators it is a good way to get more precise values and somewhat better frequency flatness. I am not sure why it works. Probably stacking helps reduce stray inductance. Made a small cal set with a power splitter like this for my VNA that turned out almost as god as the original N sized ones for that application. For some reason I had good luck with the 1218 size resistors.
I'm going to put a notebook together (with graph paper of course) and follow along with these videos and even build some of the stuff you've got here. Thanks for taking the time to make these!
Wow! Another great lesson. I keep coming back to your channel to learn and refresh my memory. I remember my college professor always telling me when dealing with RF circuits "Keep 'em short, neat and tidy." Great to see such accuracy by keeping these same principles in the 20dB attenuator you built. Now, if I can only find a good used RF spectrum analyzer to do the same thing....
I used my Nano VNA to achieve what W2AEW showed, use log mag on Ch 1. Now where near as accurate as his spectrum analyzer, but never the less, you get a lot of "bang for buck" with the less than $100 nano VNA.
It's still a very good video though. I did build the version with the series 120R resistors and the response was 20dB +/- 0.1dB to 3GHz on a VNA. The return loss at LF was 'only' about 40dB (because 120R isn't the ideal value of course) and it degraded to about 28dB by 3GHz. However, the resistors fractured when removing it so I would also advise soldering some additional bracing around the structure to protect against PCB flexing when twisting it during fitting/removal from a test setup :)
I recently decided to add an attenuator to my 1.5Kw 50ohm dummy load. I found an online PI attenuator calculator that allowed a different input and output impedance. I thought if I used 3 stages and lowered the impedance in each stage I could start with an input of 100K as to not effect 50 ohms of the dummy load(and also not have any real power attenuation in the first stage). I copied the formulas for the page into a spreadsheet so I could quickly change the input impedance/attenuation/output impedance and have the spreadsheet calculate the total impedance and the power dissipated in each resistor. Luckily I simulated the circuit and it was off by a factor of 100. Being it was only a large resistor network, I then calculated it with a DC input by hand with EET101 skills, this matched the simulator. After asking numerous times on online forums, I only got responses that it's the wrong way to do it. I finally built an current sense RF sampler with an PI attenuator(50 ohms in and out) on the output, I again missed and got 68db instead of 60, but the thing works and I'm tired of fighting it. But I've never been able to comprehend why the previous method didn't work(at least mathematically), my spreadsheet outputs the same values as the online calculators. Someone suggested that the online calculators are incorrect when the input and output impedance are not the same(even though they specify that is what they do) I finally left it as the calculators are wrong so I could sleep at night. chemandy.com/calculators/matching-pi-attenuator-calculator.htm care to take a shot at this? It's purely academic at this point, but somewhat important to my sanity. KO4SWI QRZ
Thanks for the heads up once again Alan - almost purchased an attenuator on eBay but now I'll make my own. Thanks for explaining and your valuable time. 73 de GI8WFA.
Hello Alan, thanks for the clear video! I have a 20dB attenuator and had it connected on a function generator with 50 Ohm output impedance and the other side on my scope. But... no 20 dB attenuation! A bit of a surprise, but of course you have to load it with 50 ohms. Not unimportant to think about!😉
Alan, thanks for your generous contributions to the electronics and ham communities! I recently scored a Spinner 745359 100 W 40 dB attenuator. It measures an accurate 50 ohms at DC and NanoVNA sweeps show 50 ohms and excellent SWR in both directions. Does anyone know if this attenuator can accept high power input to either port? The datasheet I found is almost illegible and the sheet for the modern equivalent doesn't mention in/output, so I'm guessing they're symmetrical with regard to power handling but thought I's ask before using.
@@w2aew Spinner replied with datasheet and confirmation that the attenuator is bidirectional wrt power. Thanks for the nudge to contact them. After 30 years in science I don't even bother asking questions of manufacturers since they rarely reply unless I have a $ervice contract on the equipment. Good to see that some of the folks on the engineering side are friendlier.
That's a good design, and a great construction technique! Thanks for sharing! I would not have thought to suspend surface mount components in the air like that, like old style leaded resistors, but it clearly works well.
Very nice. Wonderful. Sometimes the simple things are so rewarding .... ! Make several videos this way for us, the guys like me, that sometimes need go back to the basics ....
Thanks for your input. Traditionally, I have bought mostly HP and Textronix but Advantest seems like it's a top end unit from my limited observation. I too like the ole Tek 465B scopes. I was once a FE for DEC and traveled around the nation with the VAX/VMS platforms. During that time we ALWAYS drug 465B's around. Rumor was leaving it at the office could get you dismissed, mot sure that was true. Hi I appreciate your videos, some of it proved to be a good review and I have learned as well. 73
Is there a reason to use matching parallel resistors or would it be better getting the higher accuracy of using 330 || 75 for the 61.1111Ω and 270 || 3000 for the 247.5Ω resistor? To reinforce the information presented here, I created a spreadsheet that calculates Pi and T attenuator resistor values and helps me select the best parallel or series resistors from my personal collection.
Fabulous as usual. How about this idea: use three quality (cermet) trimpots and then adjust them as precisely as indicated by the equations? It would be interesting to see the resulting bandwidth performance.
Another nice video!! I have done the same thing to make various values. I use the same type of connectors, but round. The attenuators fit perfectly inside brass tubing available at most hobby shops.
Needless to say, it is very very clear and easy to follow, from design to test. NOW MY QUESTION. Is it mandatory to use the SMD resistors if I aim to use the attenuator at 500 Mhz? Can I use regular resistors? Thanks in advance. Manfred.
So clear and simple, I think I can do it. I need to cover 0-6ghz at 3watt 10 db 50ohms. how do I alter your example? eg instead of smd resistors do I use 2w resistors (old school) and how do i cover the 6ghz extent? Note I am adding this to a 30db unit to give me 40 db. thanks Troy
If you already have a 30dB pad, you can put the additional 10dB "after" that - by doing this, the 10dB pad doesn't need to be rated for the full power. Home-building a high power 6GHz attenuator is not a trivial task.
Good video with perfect explanation. But your attenuator needs to be shielded, in order to protect it from unwanted oscillations/background RF signals. Otherwise your attenuator will not work properly. Also resistors need to be mounted correctly to make it more effective. Thanks!
Great video! If you need to calibrate this to get an exact attenuation figure, what's the best way to do that? Would little trim pots be too lossy or noisy?
Excellent explanation as always , a very useful bit of diy . Similar video on DC blockers would be great , should you be looking for suggestions , given how expensive they are to buy off the shelf.
Can we use (glass fiber boards) for rf pcb designs ??? how much is its dielectric constant I wanna design a coplanar pcb transmission line with glass fiber boards and I don't know how much is its dielectric constant....
Similar techniques can be used. Here is a basic write-up: www.electronics-notes.com/articles/radio/rf-attenuators/balanced-resistive-attenuator-pads.php
nice video. you have selected smd resistor I think 0805 package. I want to know do we have to consider wattage of resistor during selection. how it will impact the design.?
I'm very impressed. Super great work ... I'm wondering if we measure the reflection S11 or s22 ... Are they gonna be good as well?? Or should they be close to zero as possible for an ideal attenuator or not? I really don't know.
I wasn't actually thinking the values would related to *bandwidth*. I just thought trimmers would be easier and more precise than paralelling resistors. My thought was *also* that trimpots (and the even the shortest means of connecting them together) would be higher in parasitics (both inductive and capacitive) than your smt-resisitors on a thumbnail. That might limit the bandwidth some. Maybe? Personally, I leave anything above 30MHz to professionals anyway.
Hi 👋 I have an adc module with 50 ohm input impedance it says it's input voltage should be below 5v p-p if I use an amplifier with 5v supply like lna moduls before my adc should I use attenuator before my adc???
As long as the LNA output can't exceed its own supply rails, then no attenuator should be needed. This is assuming that the ADC's full input range is 0-5V.
If you want to seal it off. just build an enclosure with thin metal or find some old junk CRT monitors and strip them for parts. Or do what Raytheon does and build inside freight containers.
In general, no - because the 50 ohm impedance dominates. At very very high frequencies, there may be some compensation or special design techniques employed to maintain a flat frequency response.
I have a bird sampler/variable attenuator #4273. How do I know that fully screwed in is -40 db attenuation, How do I verify where -55db attenuation is? Is there a way to sweep these with an oscilloscope or spectrum scope? If so, a video would be very appreciated...
The short answer is - you don't know what attenuation you actually have at any setting of the thumbscrew without measuring it. The further complicating factor is that the attenuation at any given setting will vary with frequency (low frequencies are attenuated more than high frequencies), and the attenuation level will depend on the load placed on the sample port. They can be swept with a spectrum analyzer with a tracking generator, or with a VNA like the NanoVNA. You can also make measurements on a scope, although that gets tricky because the sample port attenuation is quite high, so the signal level to be measured can be quite low unless you're sending a lot of power through the main path.
@@w2aew Yeah, I'm trying to verify the accuracy of my watt meters using the formula Vpp/(2 times the square root of 2) then that number squared divided by 50 ohms for scope watts on 28.400. I just need to verify that I'm at -40 db attenuation.
@@imken2392 40dB represents a factor of 10,000:1 difference in power, which is a factor 100:1 in voltage. So, if you have a way of generating a 28.4MHz signal at something like 1Vpp, then send that thru the main path (properly terminated), then measure to see what the sample port is giving you. 10mVpp would be 40dBm down. You can see the dilemma - measuring a small level like that...
Hi, what is the wattage on those resisters? Could you suggest a good way to seal the attenuator for use in a wireless test bed? And thank you for an excellent, helpful video.
Thanks for video! What's the maximum bandwidth that normal 0603/0402 resistors can go to before the ESL is a an issue? I've been looking in Farnell at microwave resistors and they look quite expensive. Are they only required for over 4GHz for example?
Okay, I may be missing something. The O Scope attenuator should show 1 volt on the scope for every 10 in? Your paper kind of makes it look like the opposite, 1 volt in would show 10 volts on the scope. I have a couple differential probe sets. One is 20KV rated, with enormous probes. One of those deals that was just too good to pass up, and it is Tektronix, so that made it even harder to pass up. I will need attenuators. Thanks! 73 KI7AQJ -AE (passed the extra last week, & now they want me to take the VE)
Attenuator ratings are typically given in positive dB values, which is typically used for gain. Thus, you have to consciously remember to negate the attenuator value when performing the calculation.
Thanks for sharing. Seems I had not seen this one. What controls how flatness of the response of the attenuator? Is this mainly the construction characteristics? Would shielding the whole thing help even more than just the solid ground plane? Also, are SMD resistors better for this purpose vs. through hole? If so, just due to their size? Thanks so much for the great videos you produce.
Physical construction is what will determine bandwidth and flatness. SMT is generally better because leaded parts typically have higher parasitic inductance. Design of good RF attenuators is an art.
ive this sussed out thank you but there is one question still remains if i had a 27db 5 watt pi attenuator an wanted to shift it to a 10 watt 30 db pi attenuator must my R1 AND R3 risistor increase and alter not only in value but double up in wattage ??
There are a lot of pi-attenuator calculators available online to make the resistor value calculation easy: www.microwaves101.com/calculators/858-attenuator-calculator www.pasternack.com/t-calculator-pi-attn.aspx m0ukd.com/calculators/pi-attenuator-calculator/ www.qorvo.com/design-tools-pad-attenuator-calculator etc. Since the shut resistor is nearly 50 ohms, it's going to have to dissipate nearly the full power, so they'll have to be rated for the full power (10W). The series resistor will dissipate less - just a simple exercise in ohms law...
A good video but you haven't made the best use of the SMD technology if you want the 20dB attenuator to work well up to 2.5GHz. By using two 510R resistors in parallel you have doubled the capacitance across this node and this will cause the loss of the attenuator to decrease up at 2.5GHz. If you had used two 120R resistors in series here you would have had better accuracy and also a MUCH flatter attenuator response to 2.5GHz. No extra cost either and you just need one resistor value.
Would you consider to make an example? "Just to show truly green amateurs like me!! I don't even have a radio yet but I'm looking around! thanks in advance.
I built this in LTSpice with a 1V generator and 50Ohm at the input and output and I'm getting 50mV at the output. This is a 20 ratio. What am I doing wrong??
sir im in a bit of a muddle i have a rf radio output 10 watts and a unit that takes 100 mw input or less to operate if not damage will be so can you give me a pi simpe circut values please
Not a dumb question, you just use the same technique as used there with a BNC male and female soldered together on a piece of board, and a pair of 100R chip resistors like on the attenuator. Will be quite a good 200mW inline dummy load. Otherwise for less than 50MHz a T connector and a 50R terminator works well. If you need a little more power then simply solder 20 1k chip resistors in a circle a few layers deep in the terminator, that will then do 1W.
Hi, a little off topic but I have been looking for a spectrum analyzer with a tracking gen without the typical high cost. I have a great HP 8566B that covers almost DC to 20gigs but no TG. I can built an interface adapter and Phase Lock an external sig gen to the first LO but at best, I may have 80 dB of dynamic range. Are you pleased with the Advantest SA and any recommendations on a good SA that will go to 2 or 3 gigs?? Oh, I need the low range to be good at about 100 KHz 73,, Glenn WA4AOS
@@w2aew I still not get a point as well as your explanation, 1 watt equal to 30 dBm as your sample that calculate by dB, how can i know my circuit Pi of each resistor R1-R2 and R3, with my ZO 50 Ohm, how you know you 10x as a example, please let me know, i wish to implement it to my circuit, can you please explain me what is value for those R1-R2 and R3 .
+marcos At last part of the video you can see that the attenuator was pretty flat up to the maximum frequency tested which was 2.5GHz. You should be able to easily build one for 430MHz - just keep the connections as short as possible.
Absolutely brilliant Alan. Thanks for sharing. Incidentally, I'm curious; some speccies have the option of a tg. On your Advantest how easy/difficult is it to install a tg? Or does the tg have to be installed during manufacture?
@w2aew I love your videos they are very informing and explains everything in an excellent explanatory way. I have a question. How do you calculate the maximum effect these attenuators can handle? Regards Jacob, OZ2LT.
+Jacob Lindquist In general, the shunt resistors will dissipate the most power since they see the full power voltage on the input side of things. So, if you make all resistors of the same power rating, you can get a good estimate of total power rating by calculating the power in the first shunt resistor.
Hi Allen, I am going to hook up my rig 870 IF out put to a RTL dongle with a rf 20 db attenuation do you think it is enough I am not going to damage the RTL's front end??
sir i want to make a photo-diode detector circuit for my photo-diode SM05PD1A,on which a laser beam from 0.8mw HE-NE laser thorlabs is incident on it.what should be the bandwidth for the detector and how to calculate this? thank you
The bandwidth needed by the detector would depend entirely on your application. If the optical signal from your laser is continuous (i.e. not pulsed or modulated), then you don't need very much bandwidth at all. If it is pulsed and you need to measure things like rise/fall times, etc, then the bandwidth you'll need depends on how fast the optical rise and fall times are.
Alan, I can't tell from the video. Is that copper cladding on a glass substrate? I have a Radio Shack double clad board but it looks like copper over phenolic. I doubt it matters at the frequencies I'm using it at but I would like to make sure before a I commit a couple of connectors to building up a few attenuators. Rather do it right the first time. Btw my call is N1ABE.
Yes, it is an FR-4 board, which is basically fiberglass/epoxy resin. Since the PCB isn't being used to carry signals in a transmission line (it is only a ground plane in this case), I doubt that the loss of the substrate will affect operation at all.
I was just looking how to make some attenuators and I run into you again! You are just a wealth of great information and practical implementation ! I want to be sitting right next to you and try to catch some of your breadcrumbs! Thank you again! Stay safe and have a great weekend!
LOL! Same! i was just going to type this into the search bar, when this video popped on next! Serendipity!
doesnt matter if you tell few words or a lot....it is allways informative and direct to the point....teaching talent
Wow, I always thought that making attenuators is a complicated science, especially getting a decent flat line frequency response. But your video proved to me that it can be quite easy, thanks!
he's good, but now the hardest part is you can't say tee-pee lol
Fun: Yes that was ! I'm 77 years old and still enjoy tinkering around. 73 Leo
One trick I picked up for similar attenuators and terminations is to stack a number of smd resistors to get the desired value. 3 150 ohm larger ones make good 50 ohm terminations. For attenuators it is a good way to get more precise values and somewhat better frequency flatness. I am not sure why it works. Probably stacking helps reduce stray inductance. Made a small cal set with a power splitter like this for my VNA that turned out almost as god as the original N sized ones for that application. For some reason I had good luck with the 1218 size resistors.
PI & T? Yes, I'd love some.
On a more serious note, another excellent video.
I'm going to put a notebook together (with graph paper of course) and follow along with these videos and even build some of the stuff you've got here. Thanks for taking the time to make these!
WOW!!!!!!! THANKS FOR WALKING US THRU YOUR NOTES,... EXCELLENT DESCRIPTIONS!!!... GREAT DEMO!!! THANKS!!!
Wow! Another great lesson. I keep coming back to your channel to learn and refresh my memory. I remember my college professor always telling me when dealing with RF circuits "Keep 'em short, neat and tidy." Great to see such accuracy by keeping these same principles in the 20dB attenuator you built. Now, if I can only find a good used RF spectrum analyzer to do the same thing....
I used my Nano VNA to achieve what W2AEW showed, use log mag on Ch 1. Now where near as accurate as his spectrum analyzer, but never the less, you get a lot of "bang for buck" with the less than $100 nano VNA.
Excellent video! Thanks for taking the time to make them. Your explanations are very clear and concise.
It's still a very good video though. I did build the version with the series 120R resistors and the response was 20dB +/- 0.1dB to 3GHz on a VNA.
The return loss at LF was 'only' about 40dB (because 120R isn't the ideal value of course) and it degraded to about 28dB by 3GHz. However, the resistors fractured when removing it so I would also advise soldering some additional bracing around the structure to protect against PCB flexing when twisting it during fitting/removal from a test setup :)
I recently decided to add an attenuator to my 1.5Kw 50ohm dummy load. I found an online PI attenuator calculator that allowed a different input and output impedance. I thought if I used 3 stages and lowered the impedance in each stage I could start with an input of 100K as to not effect 50 ohms of the dummy load(and also not have any real power attenuation in the first stage). I copied the formulas for the page into a spreadsheet so I could quickly change the input impedance/attenuation/output impedance and have the spreadsheet calculate the total impedance and the power dissipated in each resistor. Luckily I simulated the circuit and it was off by a factor of 100. Being it was only a large resistor network, I then calculated it with a DC input by hand with EET101 skills, this matched the simulator. After asking numerous times on online forums, I only got responses that it's the wrong way to do it. I finally built an current sense RF sampler with an PI attenuator(50 ohms in and out) on the output, I again missed and got 68db instead of 60, but the thing works and I'm tired of fighting it. But I've never been able to comprehend why the previous method didn't work(at least mathematically), my spreadsheet outputs the same values as the online calculators. Someone suggested that the online calculators are incorrect when the input and output impedance are not the same(even though they specify that is what they do) I finally left it as the calculators are wrong so I could sleep at night. chemandy.com/calculators/matching-pi-attenuator-calculator.htm care to take a shot at this? It's purely academic at this point, but somewhat important to my sanity. KO4SWI QRZ
Thanks for the heads up once again Alan - almost purchased an attenuator on eBay but now I'll make my own. Thanks for explaining and your valuable time. 73 de GI8WFA.
Thank you for your time in a full explanation, your patience and knowledge is appreciated! 73 ' David K8KEM
Excellent. Amazing the bandwidth acheived!
Hello Alan, thanks for the clear video! I have a 20dB attenuator and had it connected
on a function generator with 50 Ohm output impedance and the other side on my scope.
But... no 20 dB attenuation! A bit of a surprise, but of course you have to load it with 50 ohms.
Not unimportant to think about!😉
Yes, passive attenuators rely on the fact that the impedance connected to both ends is 50 ohms (or 75 ohms for to 75 ohm variety).
Nice! I always thought that resistor has stray reactance that affect sigificantly when you vary the frequency...
Yes, all components do have some parasitic reactance - the physical construction of the device will generally dictate the magnitude of this.
Alan, thanks for your generous contributions to the electronics and ham communities! I recently scored a Spinner 745359 100 W 40 dB attenuator. It measures an accurate 50 ohms at DC and NanoVNA sweeps show 50 ohms and excellent SWR in both directions. Does anyone know if this attenuator can accept high power input to either port? The datasheet I found is almost illegible and the sheet for the modern equivalent doesn't mention in/output, so I'm guessing they're symmetrical with regard to power handling but thought I's ask before using.
I would think that if it isn't clearly marked, then it is symmetrical. Spinner is still in business, so you could email them and ask the question too.
@@w2aew Spinner replied with datasheet and confirmation that the attenuator is bidirectional wrt power. Thanks for the nudge to contact them. After 30 years in science I don't even bother asking questions of manufacturers since they rarely reply unless I have a $ervice contract on the equipment. Good to see that some of the folks on the engineering side are friendlier.
What is fun is to do the exact same circuit with Long THC leads and see your high end response skew. Another nice video!
That's a good design, and a great construction technique! Thanks for sharing! I would not have thought to suspend surface mount components in the air like that, like old style leaded resistors, but it clearly works well.
Yes, but not as robust mechanically - it's easy to delaminate a termination on a small SMT part with a little mechanical force.
just found the channel, thanks for all the education you provide
Very nice. Wonderful. Sometimes the simple things are so rewarding .... ! Make several videos this way for us, the guys like me, that sometimes need go back to the basics ....
Great Job and thank you. I give you an A++
Thanks for your input. Traditionally, I have bought mostly HP and Textronix but Advantest seems like it's a top end unit from my limited observation. I too like the ole Tek 465B scopes. I was once a FE for DEC and traveled around the nation with the VAX/VMS platforms. During that time we ALWAYS drug 465B's around. Rumor was leaving it at the office could get you dismissed, mot sure that was true. Hi
I appreciate your videos, some of it proved to be a good review and I have learned as well.
73
Love the humor! Good job!
The best channel in youtube. Keep up the good work.
Is there a reason to use matching parallel resistors or would it be better getting the higher accuracy of using 330 || 75 for the 61.1111Ω and 270 || 3000 for the 247.5Ω resistor? To reinforce the information presented here, I created a spreadsheet that calculates Pi and T attenuator resistor values and helps me select the best parallel or series resistors from my personal collection.
Thanks for the video, I really need an attenuator by tomorrow and amazon wouldn't get it to me soon enough. So this will work temporarily.
Very informative! Never thought it could be so simple!
Fabulous as usual. How about this idea: use three quality (cermet) trimpots and then adjust them as precisely as indicated by the equations? It would be interesting to see the resulting bandwidth performance.
Another nice video!! I have done the same thing to make various values. I use the same type of connectors, but round. The attenuators fit perfectly inside brass tubing available at most hobby shops.
Good example of construction technique for RF stuff!
amazing all that math looked so boring in the book..... but now it all makes sense now that you brought it to life. Thanks
Great video!! This video made me think about making a diy feedthrough 50ohm termination for my oscilloscope...🤔
Thank you for your video. Excellent! I need to make an attenuator. It so much cheaper than buying one. Thanks for the inspiration.
Que buen video, muy bien explicado. Necesito hacer un atenuador y me sirvió bastante tu enseñanza. Gracias
Great explanation!
Thanks for the video! Your soldering job was pretty interesting - surprised you managed to make that work lol.
are there any textbooks or resources that you would recommend to understand the theory behind all of this?
Needless to say, it is very very clear and easy to follow, from design to test.
NOW MY QUESTION. Is it mandatory to use the SMD resistors if I aim to use the attenuator at 500 Mhz? Can I use regular resistors?
Thanks in advance.
Manfred.
Hmmm - good question. If the leads are kept extremely short, you *might* be able to get away with leaded components. Worth a try....
What exactly is the difference between these networks and a voltage divider?
These are voltage dividers that also present the same impedance (matched to the source and load) at both ends.
İnsertion loss just.
So clear and simple, I think I can do it. I need to cover 0-6ghz at 3watt 10 db 50ohms. how do I alter your example? eg instead of smd resistors do I use 2w resistors (old school) and how do i cover the 6ghz extent? Note I am adding this to a 30db unit to give me 40 db. thanks Troy
If you already have a 30dB pad, you can put the additional 10dB "after" that - by doing this, the 10dB pad doesn't need to be rated for the full power. Home-building a high power 6GHz attenuator is not a trivial task.
Loving how you constructed the attenuator!
I have a 6GHz dummy load. How can I get a proportional sample into the TinySA from the transmission line to get an accurate power measurement?
Take a look at the RF Samplers video
Good video with perfect explanation. But your attenuator needs to be shielded, in order to protect it from unwanted oscillations/background RF signals. Otherwise your attenuator will not work properly.
Also resistors need to be mounted correctly to make it more effective. Thanks!
the RF attenuator will work well if you mounted it correctly, because stray components (inductance/capacitance) are critical
Great video! If you need to calibrate this to get an exact attenuation figure, what's the best way to do that? Would little trim pots be too lossy or noisy?
I used a commercial 10db attenuator on my Nano VNA to check calibration.
Excellent explanation as always , a very useful bit of diy . Similar video on DC blockers would be great , should you be looking for suggestions , given how expensive they are to buy off the shelf.
thank you very much Sir for this video!
kindly explain how to measure self resonating frequency of capacitor
Can we use (glass fiber boards) for rf pcb designs ??? how much is its dielectric constant I wanna design a coplanar pcb transmission line with glass fiber boards and I don't know how much is its dielectric constant....
Can u pleas make the 2400Mhz band pass filter
can we use this with a balanced audio signal to attenuate it before the adc?
Similar techniques can be used. Here is a basic write-up: www.electronics-notes.com/articles/radio/rf-attenuators/balanced-resistive-attenuator-pads.php
@@w2aew thanks
Keep up the good work!
nice video. you have selected smd resistor I think 0805 package. I want to know do we have to consider wattage of resistor during selection. how it will impact the design.?
Yes, you certainly have to take into account the power rating of the resistors if you need an attenuator that works are high power levels.
I'm very impressed. Super great work ... I'm wondering if we measure the reflection S11 or s22 ... Are they gonna be good as well?? Or should they be close to zero as possible for an ideal attenuator or not? I really don't know.
Yes, the S11 and S22 should be a very low reflection coefficient also.
I wasn't actually thinking the values would related to *bandwidth*. I just thought trimmers would be easier and more precise than paralelling resistors. My thought was *also* that trimpots (and the even the shortest means of connecting them together) would be higher in parasitics (both inductive and capacitive) than your smt-resisitors on a thumbnail. That might limit the bandwidth some. Maybe? Personally, I leave anything above 30MHz to professionals anyway.
Hi 👋 I have an adc module with 50 ohm input impedance it says it's input voltage should be below 5v p-p if I use an amplifier with 5v supply like lna moduls before my adc should I use attenuator before my adc???
As long as the LNA output can't exceed its own supply rails, then no attenuator should be needed. This is assuming that the ADC's full input range is 0-5V.
@@w2aewthanks a lot.
If you want to seal it off. just build an enclosure with thin metal or find some old junk CRT monitors and strip them for parts. Or do what Raytheon does and build inside freight containers.
I must agree your videos are excellent I always look forward to the next one. Thanx Gary
Very interesting video, wonder how u soldered that !
I thought attenuation db rating is based on power, ie 20db (n=100), those n-formulas are for voltage db only?
You use 20*LOG() when calculating using voltages, and 10*LOG() when calculating using power.
@@w2aew and because power is prportional to voltage squared and the log takes the square outside the ten turns into twenty.....
Do these not require capacitive compensation as is used in scope probes?
In general, no - because the 50 ohm impedance dominates. At very very high frequencies, there may be some compensation or special design techniques employed to maintain a flat frequency response.
How about the noise contribution. My guess is for RF it may be contributing too much noise
I have a bird sampler/variable attenuator #4273. How do I know that fully screwed in is -40 db attenuation, How do I verify where -55db attenuation is? Is there a way to sweep these with an oscilloscope or spectrum scope? If so, a video would be very appreciated...
The short answer is - you don't know what attenuation you actually have at any setting of the thumbscrew without measuring it. The further complicating factor is that the attenuation at any given setting will vary with frequency (low frequencies are attenuated more than high frequencies), and the attenuation level will depend on the load placed on the sample port. They can be swept with a spectrum analyzer with a tracking generator, or with a VNA like the NanoVNA. You can also make measurements on a scope, although that gets tricky because the sample port attenuation is quite high, so the signal level to be measured can be quite low unless you're sending a lot of power through the main path.
@@w2aew Yeah, I'm trying to verify the accuracy of my watt meters using the formula Vpp/(2 times the square root of 2) then that number squared divided by 50 ohms for scope watts on 28.400. I just need to verify that I'm at -40 db attenuation.
@@imken2392 40dB represents a factor of 10,000:1 difference in power, which is a factor 100:1 in voltage. So, if you have a way of generating a 28.4MHz signal at something like 1Vpp, then send that thru the main path (properly terminated), then measure to see what the sample port is giving you. 10mVpp would be 40dBm down. You can see the dilemma - measuring a small level like that...
@@w2aew Fortunately I have an HP 8920A communications test set....I'll get it figured out.
@@imken2392 OK - if that's got a tracking generator on it, and can cover the frequency range you need, then you're all set.
Hi, what is the wattage on those resisters? Could you suggest a good way to seal the attenuator for use in a wireless test bed? And thank you for an excellent, helpful video.
Thanks for video! What's the maximum bandwidth that normal 0603/0402 resistors can go to before the ESL is a an issue? I've been looking in Farnell at microwave resistors and they look quite expensive. Are they only required for over 4GHz for example?
It will depend on your layout, but I have seen them used up to several GHz.
Okay, I may be missing something. The O Scope attenuator should show 1 volt on the scope for every 10 in? Your paper kind of makes it look like the opposite, 1 volt in would show 10 volts on the scope. I have a couple differential probe sets. One is 20KV rated, with enormous probes. One of those deals that was just too good to pass up, and it is Tektronix, so that made it even harder to pass up. I will need attenuators. Thanks! 73 KI7AQJ -AE (passed the extra last week, & now they want me to take the VE)
Attenuator ratings are typically given in positive dB values, which is typically used for gain. Thus, you have to consciously remember to negate the attenuator value when performing the calculation.
Thanks. When you speak of "going to ground", is that soldering to the chassis?
Thanks for sharing. Seems I had not seen this one.
What controls how flatness of the response of the attenuator? Is this mainly the construction characteristics? Would shielding the whole thing help even more than just the solid ground plane? Also, are SMD resistors better for this purpose vs. through hole? If so, just due to their size?
Thanks so much for the great videos you produce.
Physical construction is what will determine bandwidth and flatness. SMT is generally better because leaded parts typically have higher parasitic inductance. Design of good RF attenuators is an art.
Thank you. I just wondered about the derivation of the equations used; could anybody point me to where I might find the derivations?
@elias deverent Yes, cou;d you post a link please?
@elias deverent Thats so kind. khwp.01@googlemail.com. Thanks you!
Very nice..next topic on circulators pls
ive this sussed out thank you but there is one question still remains if i had a 27db 5 watt pi attenuator an wanted to shift it to a 10 watt 30 db pi attenuator must my R1 AND R3 risistor increase and alter not only in value but double up in wattage ??
All three resistor values will change when you change the attenuation value. You'll have to calculate new power ratings as well.
thats was my point sir how will i calculate 5 w to 10 in put ?
There are a lot of pi-attenuator calculators available online to make the resistor value calculation easy:
www.microwaves101.com/calculators/858-attenuator-calculator
www.pasternack.com/t-calculator-pi-attn.aspx
m0ukd.com/calculators/pi-attenuator-calculator/
www.qorvo.com/design-tools-pad-attenuator-calculator
etc.
Since the shut resistor is nearly 50 ohms, it's going to have to dissipate nearly the full power, so they'll have to be rated for the full power (10W). The series resistor will dissipate less - just a simple exercise in ohms law...
A good video but you haven't made the best use of the SMD technology if you want the 20dB attenuator to work well up to 2.5GHz. By using two 510R resistors in parallel you have doubled the capacitance across this node and this will cause the loss of the attenuator to decrease up at 2.5GHz. If you had used two 120R resistors in series here you would have had better accuracy and also a MUCH flatter attenuator response to 2.5GHz. No extra cost either and you just need one resistor value.
Would you consider to make an example? "Just to show truly green amateurs like me!! I don't even have a radio yet but I'm looking around! thanks in advance.
This is awesome! Thanks so much for explaining this.
I built this in LTSpice with a 1V generator and 50Ohm at the input and output and I'm getting 50mV at the output. This is a 20 ratio. What am I doing wrong??
What happens at lower frequencies? Is it linear all the way to DC?
Hello W2AEW, can you pl explain how to design a DC block.
Great video ! Thanks so much
sir im in a bit of a muddle i have a rf radio output 10 watts and a unit that takes 100 mw input or less to operate if not damage will be so can you give me a pi simpe circut values please
Nice video, thanks for sharing it :)
Because for SMD, capacitive effects are more important than inductance?
Not a dumb question, you just use the same technique as used there with a BNC male and female soldered together on a piece of board, and a pair of 100R chip resistors like on the attenuator. Will be quite a good 200mW inline dummy load. Otherwise for less than 50MHz a T connector and a 50R terminator works well. If you need a little more power then simply solder 20 1k chip resistors in a circle a few layers deep in the terminator, that will then do 1W.
Can you check a attenuator using a signal generator and a oscilloscope
Hi, a little off topic but I have been looking for a spectrum analyzer with a tracking gen without the typical high cost. I have a great HP 8566B that covers almost DC to 20gigs but no TG. I can built an interface adapter and Phase Lock an external sig gen to the first LO but at best, I may have 80 dB of dynamic range.
Are you pleased with the Advantest SA and any recommendations on a good SA that will go to 2 or 3 gigs?? Oh, I need the low range to be good at about 100 KHz
73,,
Glenn WA4AOS
If my input is 1000 mw or 1 watt for 50 Ohm, does 1 watt is equal to 30db, what are the value for R1-R2 and R3, awaiting your answer.
1watt = 30dBm (not 30dB). Resistor values don't change based on power, only the power rating of the resistors change.
@@w2aew I still not get a point as well as your explanation, 1 watt equal to 30 dBm as your sample that calculate by dB, how can i know my circuit Pi of each resistor R1-R2 and R3, with my ZO 50 Ohm, how you know you 10x as a example, please let me know, i wish to implement it to my circuit, can you please explain me what is value for those R1-R2 and R3 .
Sooo very true .. I strive to make one as good some day .. :) Practice makes perfect lol
can i use this attenuator in a 430 MHz transmitter? can i use normal resistors? Thank for this video
+marcos At last part of the video you can see that the attenuator was pretty flat up to the maximum frequency tested which was 2.5GHz. You should be able to easily build one for 430MHz - just keep the connections as short as possible.
u re right, thanks
Great video, thank you for the info and saving me a couple of bucks :)
Absolutely brilliant Alan. Thanks for sharing. Incidentally, I'm curious; some speccies have the option of a tg. On your Advantest how easy/difficult is it to install a tg? Or does the tg have to be installed during manufacture?
I think it depends on the particular brand/model.
@@w2aew Specifically on the Advantest model you have
@@Testequip I don't know, I purchased this one used and it already had the TG installed.
@w2aew I love your videos they are very informing and explains everything in an excellent explanatory way. I have a question. How do you calculate the maximum effect these attenuators can handle? Regards Jacob, OZ2LT.
+Jacob Lindquist In general, the shunt resistors will dissipate the most power since they see the full power voltage on the input side of things. So, if you make all resistors of the same power rating, you can get a good estimate of total power rating by calculating the power in the first shunt resistor.
That is excellent and helps me alot thankyou!
Hi Allen, I am going to hook up my rig 870 IF out put to a RTL dongle with a rf 20 db attenuation do you think it is enough I am not going to damage the RTL's front end??
Based on the signal levels observed here, 20dB should be sufficient.
th-cam.com/video/y1d06pXGdMc/w-d-xo.html
sir
i want to make a photo-diode detector circuit for my photo-diode SM05PD1A,on which a laser beam from 0.8mw HE-NE laser thorlabs is incident on it.what should be the bandwidth for the detector and how to calculate this?
thank you
The bandwidth needed by the detector would depend entirely on your application. If the optical signal from your laser is continuous (i.e. not pulsed or modulated), then you don't need very much bandwidth at all. If it is pulsed and you need to measure things like rise/fall times, etc, then the bandwidth you'll need depends on how fast the optical rise and fall times are.
@@w2aew thanks a lot sir...
How much power would an attenuator like this be good for on the input?
It depends entirely on the power rating of the resistors used. In this case, probably only 100mW or less.
Alan, I can't tell from the video. Is that copper cladding on a glass substrate? I have a Radio Shack double clad board but it looks like copper over phenolic. I doubt it matters at the frequencies I'm using it at but I would like to make sure before a I commit a couple of connectors to building up a few attenuators. Rather do it right the first time. Btw my call is N1ABE.
Yes, it is an FR-4 board, which is basically fiberglass/epoxy resin. Since the PCB isn't being used to carry signals in a transmission line (it is only a ground plane in this case), I doubt that the loss of the substrate will affect operation at all.
Thanks for the reply.
Does it work at frequencies greater than 2.5GHz?
Many thanks
perfect! please do more vids on RF stuff
Hi, thank for your works.
Can i mesure an attenuator without spectrum analyser ??
Thanks 73's
Sure, as long as you have a signal generator and a means of measuring the signal amplitude with the proper termination (scope, etc.).
lol thank you. Colorimetry and details are tipical from Canon DSLR.
nice video quality and nice stuff. You have a nice workshop...