Trying to find high value BNC attenuators.
All I can find range from -1dB to -30dB and only for 50ohms
Any suggestions on how to make a -60dB and -80dB attenuator using a couple of resistors inside an old BNC coupling?
I don't want to cut into a BNC to BNC cable
They are for audio frequencies, so 50ohms impedance to match characteristic impedance of connecting cables is unimportant.
All I can find range from -1dB to -30dB and only for 50ohms
Any suggestions on how to make a -60dB and -80dB attenuator using a couple of resistors inside an old BNC coupling?
I don't want to cut into a BNC to BNC cable
They are for audio frequencies, so 50ohms impedance to match characteristic impedance of connecting cables is unimportant.
Hi Andrew,
The impedance of RG-58 lies around 62 ohms. Of course, this is only important at high frequencies.
Your attenuator design can be at any impedance level. All you need to know is the output impedance of your generator(s) and input impedance of your load. The audio standard is 600R on a cable, Video is 75R and test & measurement is 50R as I'm sure you know.
So the most important thing to know is what the input impedance is for your test device. If it is unknown, you have one option. You need to make a buffer to preserve your voltage divider ratio. Personally, I can't see anyone figuring out the resistance ratios needed every single time you need to attenuate a signal on a different piece of equipment. That's one reason why the T&M industry has a common input impedance, along with a rough SWR rating (range or impedance error).
It looks like you're building an attenuator box. I use HP 3750A attenuators designed for a 75R RF system. If I work at audio frequencies, I have to trust the attenuators in the equipment, which being HP is a pretty good bet they are accurate. However, that is the same thing you need, but built into my generators. That also allows them to be leveled. Can you pick up a used HP 3325A,3325A or 3336B (for sine only). Those would be accurate. On the analog side, a 334A would be very accurate and probably pretty inexpensive. It is a generator that has a meter on the output and an pretty accurate attenuator. It's easy to adjust to 0.25 dB differences as the output meter is +/- 2 dB with the centre 0 being the setting on the fixed attenuators. The meter will show if the load drags the signal down, so you can correct it. It's also leveled I think.
These might be overkill for what you want to accomplish, I don't know.
-Best, Chris
The impedance of RG-58 lies around 62 ohms. Of course, this is only important at high frequencies.
Your attenuator design can be at any impedance level. All you need to know is the output impedance of your generator(s) and input impedance of your load. The audio standard is 600R on a cable, Video is 75R and test & measurement is 50R as I'm sure you know.
So the most important thing to know is what the input impedance is for your test device. If it is unknown, you have one option. You need to make a buffer to preserve your voltage divider ratio. Personally, I can't see anyone figuring out the resistance ratios needed every single time you need to attenuate a signal on a different piece of equipment. That's one reason why the T&M industry has a common input impedance, along with a rough SWR rating (range or impedance error).
It looks like you're building an attenuator box. I use HP 3750A attenuators designed for a 75R RF system. If I work at audio frequencies, I have to trust the attenuators in the equipment, which being HP is a pretty good bet they are accurate. However, that is the same thing you need, but built into my generators. That also allows them to be leveled. Can you pick up a used HP 3325A,3325A or 3336B (for sine only). Those would be accurate. On the analog side, a 334A would be very accurate and probably pretty inexpensive. It is a generator that has a meter on the output and an pretty accurate attenuator. It's easy to adjust to 0.25 dB differences as the output meter is +/- 2 dB with the centre 0 being the setting on the fixed attenuators. The meter will show if the load drags the signal down, so you can correct it. It's also leveled I think.
These might be overkill for what you want to accomplish, I don't know.
-Best, Chris
I have just assembled some RCA/Phono attenuators that need to be jury rigged into the BNC input of a Low Noise Amplifier.
I have -40dB, -60dB and -80dB RCA attenuators.
I would like to build some BNC attenuators but inside the screening of the BNC shell.
The commercial versions for 50ohms loading look like two BNC with a screwed connection between them and that the attenuator is inside that screwed connection.
I don't have facilities to convert one half of a BNC into the male screw fitting nor to the female screw fitting on the other half.
I'm looking for ideas that I can build/assemble.
But trying to avoid an attenuator screening box with BNC sockets for the IN/OUT.
The signal range at the input to the LNA is 1uVac to 350uVac so very susceptible to picking up interference. Maybe a steel/iron box is the only solution.
I have -40dB, -60dB and -80dB RCA attenuators.
I would like to build some BNC attenuators but inside the screening of the BNC shell.
The commercial versions for 50ohms loading look like two BNC with a screwed connection between them and that the attenuator is inside that screwed connection.
I don't have facilities to convert one half of a BNC into the male screw fitting nor to the female screw fitting on the other half.
I'm looking for ideas that I can build/assemble.
But trying to avoid an attenuator screening box with BNC sockets for the IN/OUT.
The signal range at the input to the LNA is 1uVac to 350uVac so very susceptible to picking up interference. Maybe a steel/iron box is the only solution.
Hi Andrew,
Tektronix made some cool boxes with BNC connectors already mounted. I know that someone else is still making them too. I can't remember the brand, but I have bought them before. Have a look at the big dealers in your area, like Newark or Digikey. These are available for purchase. The box is all metal for the reasons you are concerned over.
-Chris
Tektronix made some cool boxes with BNC connectors already mounted. I know that someone else is still making them too. I can't remember the brand, but I have bought them before. Have a look at the big dealers in your area, like Newark or Digikey. These are available for purchase. The box is all metal for the reasons you are concerned over.
-Chris
Those boxes are made by Pomona Electronics Pomona Electronics -. Not cheap, but usually available from Digikey.
Bill
Bill
Hi Bill,
Thanks. I remembered they were Pomona and was trying to find them on the Digikey site.
They are expensive, but worth the money when you think about it. There are few options if you don't use those.
-Chris
Thanks. I remembered they were Pomona and was trying to find them on the Digikey site.
They are expensive, but worth the money when you think about it. There are few options if you don't use those.
-Chris
I have found 50ohms BNC attenuators from -1dB to -30dB.
I cannot find any for -60dB nor -80dB.
I cannot find any for higher resistance/impedance.
10k:1r0 would give me a -80dB attenuation feeding into the 100ohm impedance of the LNA.
I am trying to drill out a couple of cable BNC connectors. The metal plating seems to be solderable, so I'm hopeful.
I cannot find any for -60dB nor -80dB.
I cannot find any for higher resistance/impedance.
10k:1r0 would give me a -80dB attenuation feeding into the 100ohm impedance of the LNA.
I am trying to drill out a couple of cable BNC connectors. The metal plating seems to be solderable, so I'm hopeful.
When making an attenuator the spacing of typ 20dB per inch. You may need to add shields inside the box or the leakage may make the BW of the circuit not work as you may need. For Audio BW the spacing can be shorter.
Duke🙂
Duke🙂
read that three times and I still don't understand.
Since I am only working in the audio band and a little beyond, then maybe I don't need 4" long attenuator for -80dB.
Hi Andrew,
No you don't. Just straight resistors, but you might need to compensate the capacitance with a few pF across the dropping resistor. You can test which value you need with a variable capacitor. At 10 ~ 20 KHz, the amount of correction might surprise you, so use the longest resistive element you have access to.
-Chris
No you don't. Just straight resistors, but you might need to compensate the capacitance with a few pF across the dropping resistor. You can test which value you need with a variable capacitor. At 10 ~ 20 KHz, the amount of correction might surprise you, so use the longest resistive element you have access to.
-Chris
Hi Andrew
We are talking about inside a box where the resistors are placed to make the attenuator not the length of connecting cable.
If the attenuator INPUT to OUTPUT path is very short and the attenuation is LARGE, you can have coupling IN to OUT. This is the stray coupling as anatech is talking about.
Duke
We are talking about inside a box where the resistors are placed to make the attenuator not the length of connecting cable.
If the attenuator INPUT to OUTPUT path is very short and the attenuation is LARGE, you can have coupling IN to OUT. This is the stray coupling as anatech is talking about.
Duke
Hello Andrew,
I'm always on the lookout for junk inline BNC attenuators. Here are examples of how they're put together.
The three holes in the threaded barrel can be configured for pi attenuation values that are reasonably flat into the uhf range.
30dB per section is possible but I would stick to 20dB sections and cascade them to get the attenuation required.
Murray
I'm always on the lookout for junk inline BNC attenuators. Here are examples of how they're put together.
The three holes in the threaded barrel can be configured for pi attenuation values that are reasonably flat into the uhf range.
30dB per section is possible but I would stick to 20dB sections and cascade them to get the attenuation required.
Murray
Those Pomona boxes are so expense, I when shopping at mouser and saw these
SMA-KIT-1.5MF Crystek Corporation | Boxes, Enclosures, Racks | DigiKey
they also have some attenuators, max 20dB, so for more you have to gang them together.
http://www.crystek.com/home/attenuator/atten.aspx
SMA-KIT-1.5MF Crystek Corporation | Boxes, Enclosures, Racks | DigiKey
they also have some attenuators, max 20dB, so for more you have to gang them together.
http://www.crystek.com/home/attenuator/atten.aspx
Last edited:
Thanks, I began to realise it was a parasitic capacitance issue after reading Anatech's post.
I think I will do the simple two resistor only version first and see if I can detect the reduction in attenuation as I approach 100kHz (the LNA's upper limit).
Last edited:
You still need the shielding, especially if you are making a 600 Ohm attenuator.
is the load low impedance or high impedance? How much attenuation do you need? How accurate do you need it?
I have built a vareity into the little pomona boxes. usually that are straight voltage dividers. figure a 600 Ohm source and then what voltage you want at the output.
If 80 dB then you are after 10,000 to 1 ratio. the first 600 Ohms is in the source. make the aritemetic simple and figure a 600 Ohm load on the source (to not exceed the load rating) Then you have 1200 Ohms total and the small resistance is 1200/10,000 or .12 Ohm. That's small enough that you can use a 600 Ohm resistor and a .12 Ohm resistor (make a series parallel combination) but the wire becomes significant at .12 Ohm. Keep it short.
The stray capacitance is an issue shunting the 600 + 600 Ohms and the inductance of the .12 Ohm is also an issue. These are issues at high frequencies. The good news is that your load impedance is inconsequential with a .12 Ohm source impedance at the output of the attenuator.
If you don't need absolute accuracy and can calibrate it (you will need a sensitive voltmeter and a high voltage source) then a 600 Ohm resistor and a .1 Ohm resistor may get you there.
is the load low impedance or high impedance? How much attenuation do you need? How accurate do you need it?
I have built a vareity into the little pomona boxes. usually that are straight voltage dividers. figure a 600 Ohm source and then what voltage you want at the output.
If 80 dB then you are after 10,000 to 1 ratio. the first 600 Ohms is in the source. make the aritemetic simple and figure a 600 Ohm load on the source (to not exceed the load rating) Then you have 1200 Ohms total and the small resistance is 1200/10,000 or .12 Ohm. That's small enough that you can use a 600 Ohm resistor and a .12 Ohm resistor (make a series parallel combination) but the wire becomes significant at .12 Ohm. Keep it short.
The stray capacitance is an issue shunting the 600 + 600 Ohms and the inductance of the .12 Ohm is also an issue. These are issues at high frequencies. The good news is that your load impedance is inconsequential with a .12 Ohm source impedance at the output of the attenuator.
If you don't need absolute accuracy and can calibrate it (you will need a sensitive voltmeter and a high voltage source) then a 600 Ohm resistor and a .1 Ohm resistor may get you there.
Yes I need screening. That's why I want to fit it inside the BNC connector. The body will be the screen.
I want two attenuators -60dB and -80dB.
That way when I input the -60dB into the +60dB LNA the input signal should be almost identical to the output signal and similarly when I input to the +80dB LNA from the -80dB attenuator again I will have almost identical signals at the input and output.
Neither need absolute precision. +-5% will probably do.
But I don't want capacitive coupling reducing the attenuation and the leaking HF overloading the input of either LNA and I don't want to let in avoidable interference that screws up comparison voltages.
The +60dB LNA is the Wurcer. The +80dB LNA is the Frex erms and it has a 100ohms input impedance. So I was going to load it with 10k:1r0 as I did with the jury rigged RCA attenuator. I am seeing ~40mVac 1kHz sine from the oscillator which reduces to ~4uVac after the attenuator and the output of the LNA is ~46.2mVac of 1kHz sinewave plus quite a high level of noise. But that is without any screening so I'd expect to see lots of interference.
When I increase the input signal to 3200mVac, I get 3200mVac from the output of the LNA. I have set the LNA gain to exactly match the 10k:1r0 reduction (the resistors are both +-1% but I did not check for capacitance induced error). At this signal level (just below clipping of the LNA) the noise is swamped.
I want two attenuators -60dB and -80dB.
That way when I input the -60dB into the +60dB LNA the input signal should be almost identical to the output signal and similarly when I input to the +80dB LNA from the -80dB attenuator again I will have almost identical signals at the input and output.
Neither need absolute precision. +-5% will probably do.
But I don't want capacitive coupling reducing the attenuation and the leaking HF overloading the input of either LNA and I don't want to let in avoidable interference that screws up comparison voltages.
The +60dB LNA is the Wurcer. The +80dB LNA is the Frex erms and it has a 100ohms input impedance. So I was going to load it with 10k:1r0 as I did with the jury rigged RCA attenuator. I am seeing ~40mVac 1kHz sine from the oscillator which reduces to ~4uVac after the attenuator and the output of the LNA is ~46.2mVac of 1kHz sinewave plus quite a high level of noise. But that is without any screening so I'd expect to see lots of interference.
When I increase the input signal to 3200mVac, I get 3200mVac from the output of the LNA. I have set the LNA gain to exactly match the 10k:1r0 reduction (the resistors are both +-1% but I did not check for capacitance induced error). At this signal level (just below clipping of the LNA) the noise is swamped.
Last edited:
Here are some examples of DIY construction methods (only two of them are actually attenuators, but it doesn't matter whether the inside is an attenuator, a diff-gain filter, a modulator, etc: you can always use resistors instead).
For 80dB, even at low frequencies, it is preferable to use at least two sections, preferably implemented in the "correct" manner, Pi or Tee, not simply an L.
Use an online calculator like this one for example: PI and T Pad Attenuator Calculator
To minimize the effect of the input resistor's capacitance, a good technique is to put a ground ring around the resistor's body:
For 80dB, even at low frequencies, it is preferable to use at least two sections, preferably implemented in the "correct" manner, Pi or Tee, not simply an L.
Use an online calculator like this one for example: PI and T Pad Attenuator Calculator
To minimize the effect of the input resistor's capacitance, a good technique is to put a ground ring around the resistor's body:
Attachments
- Status
- Not open for further replies.